{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 2706, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0007393715341959334, "grad_norm": 28.20969009399414, "learning_rate": 0.0, "loss": 0.3435014486312866, "step": 1 }, { "epoch": 0.0014787430683918669, "grad_norm": 30.056684494018555, "learning_rate": 4.098360655737705e-08, "loss": 0.34631726145744324, "step": 2 }, { "epoch": 0.0022181146025878, "grad_norm": 32.46834182739258, "learning_rate": 8.19672131147541e-08, "loss": 0.3467066287994385, "step": 3 }, { "epoch": 0.0029574861367837337, "grad_norm": 30.077939987182617, "learning_rate": 1.2295081967213116e-07, "loss": 0.3564053773880005, "step": 4 }, { "epoch": 0.0036968576709796672, "grad_norm": 42.33125686645508, "learning_rate": 1.639344262295082e-07, "loss": 0.4083036184310913, "step": 5 }, { "epoch": 0.0044362292051756, "grad_norm": 28.65184211730957, "learning_rate": 2.0491803278688524e-07, "loss": 0.35457539558410645, "step": 6 }, { "epoch": 0.005175600739371534, "grad_norm": 22.890384674072266, "learning_rate": 2.459016393442623e-07, "loss": 0.27444490790367126, "step": 7 }, { "epoch": 0.005914972273567467, "grad_norm": 27.329212188720703, "learning_rate": 2.8688524590163937e-07, "loss": 0.31163328886032104, "step": 8 }, { "epoch": 0.006654343807763401, "grad_norm": 21.27843475341797, "learning_rate": 3.278688524590164e-07, "loss": 0.32700788974761963, "step": 9 }, { "epoch": 0.0073937153419593345, "grad_norm": 20.725561141967773, "learning_rate": 3.6885245901639347e-07, "loss": 0.361122727394104, "step": 10 }, { "epoch": 0.008133086876155268, "grad_norm": 24.032840728759766, "learning_rate": 4.0983606557377047e-07, "loss": 0.32949545979499817, "step": 11 }, { "epoch": 0.0088724584103512, "grad_norm": 24.132570266723633, "learning_rate": 4.508196721311476e-07, "loss": 0.3082168996334076, "step": 12 }, { "epoch": 0.009611829944547136, "grad_norm": 28.13865852355957, "learning_rate": 4.918032786885246e-07, "loss": 0.29127371311187744, "step": 13 }, { "epoch": 0.010351201478743069, "grad_norm": 20.27391815185547, "learning_rate": 5.327868852459017e-07, "loss": 0.27674365043640137, "step": 14 }, { "epoch": 0.011090573012939002, "grad_norm": 21.951541900634766, "learning_rate": 5.737704918032787e-07, "loss": 0.2990384101867676, "step": 15 }, { "epoch": 0.011829944547134935, "grad_norm": 14.360541343688965, "learning_rate": 6.147540983606558e-07, "loss": 0.2588816285133362, "step": 16 }, { "epoch": 0.012569316081330868, "grad_norm": 10.020234107971191, "learning_rate": 6.557377049180328e-07, "loss": 0.2672607898712158, "step": 17 }, { "epoch": 0.013308687615526803, "grad_norm": 12.917086601257324, "learning_rate": 6.967213114754098e-07, "loss": 0.27728068828582764, "step": 18 }, { "epoch": 0.014048059149722736, "grad_norm": 7.0010504722595215, "learning_rate": 7.377049180327869e-07, "loss": 0.2868509292602539, "step": 19 }, { "epoch": 0.014787430683918669, "grad_norm": 11.55745792388916, "learning_rate": 7.78688524590164e-07, "loss": 0.2841397523880005, "step": 20 }, { "epoch": 0.015526802218114602, "grad_norm": 10.659294128417969, "learning_rate": 8.196721311475409e-07, "loss": 0.2802894711494446, "step": 21 }, { "epoch": 0.016266173752310535, "grad_norm": 12.874957084655762, "learning_rate": 8.606557377049181e-07, "loss": 0.31349998712539673, "step": 22 }, { "epoch": 0.017005545286506468, "grad_norm": 13.544585227966309, "learning_rate": 9.016393442622952e-07, "loss": 0.2503204345703125, "step": 23 }, { "epoch": 0.0177449168207024, "grad_norm": 15.765387535095215, "learning_rate": 9.426229508196721e-07, "loss": 0.2727370262145996, "step": 24 }, { "epoch": 0.018484288354898338, "grad_norm": 21.7735538482666, "learning_rate": 9.836065573770493e-07, "loss": 0.2920353412628174, "step": 25 }, { "epoch": 0.01922365988909427, "grad_norm": 15.257044792175293, "learning_rate": 1.0245901639344263e-06, "loss": 0.26537156105041504, "step": 26 }, { "epoch": 0.019963031423290204, "grad_norm": 6.232316493988037, "learning_rate": 1.0655737704918034e-06, "loss": 0.2417980432510376, "step": 27 }, { "epoch": 0.020702402957486137, "grad_norm": 13.658885955810547, "learning_rate": 1.1065573770491804e-06, "loss": 0.2511620819568634, "step": 28 }, { "epoch": 0.02144177449168207, "grad_norm": 11.676366806030273, "learning_rate": 1.1475409836065575e-06, "loss": 0.2729436755180359, "step": 29 }, { "epoch": 0.022181146025878003, "grad_norm": 19.675003051757812, "learning_rate": 1.1885245901639345e-06, "loss": 0.29085803031921387, "step": 30 }, { "epoch": 0.022920517560073937, "grad_norm": 17.207317352294922, "learning_rate": 1.2295081967213116e-06, "loss": 0.2936568856239319, "step": 31 }, { "epoch": 0.02365988909426987, "grad_norm": 10.088744163513184, "learning_rate": 1.2704918032786886e-06, "loss": 0.2574964165687561, "step": 32 }, { "epoch": 0.024399260628465803, "grad_norm": 5.626157760620117, "learning_rate": 1.3114754098360657e-06, "loss": 0.2631929814815521, "step": 33 }, { "epoch": 0.025138632162661736, "grad_norm": 8.492873191833496, "learning_rate": 1.352459016393443e-06, "loss": 0.24888934195041656, "step": 34 }, { "epoch": 0.025878003696857672, "grad_norm": 9.208476066589355, "learning_rate": 1.3934426229508196e-06, "loss": 0.24017128348350525, "step": 35 }, { "epoch": 0.026617375231053605, "grad_norm": 12.99432373046875, "learning_rate": 1.4344262295081968e-06, "loss": 0.2400883287191391, "step": 36 }, { "epoch": 0.02735674676524954, "grad_norm": 12.167765617370605, "learning_rate": 1.4754098360655739e-06, "loss": 0.25405463576316833, "step": 37 }, { "epoch": 0.02809611829944547, "grad_norm": 13.971226692199707, "learning_rate": 1.516393442622951e-06, "loss": 0.26724952459335327, "step": 38 }, { "epoch": 0.028835489833641405, "grad_norm": 5.68267822265625, "learning_rate": 1.557377049180328e-06, "loss": 0.25977668166160583, "step": 39 }, { "epoch": 0.029574861367837338, "grad_norm": 9.206181526184082, "learning_rate": 1.5983606557377053e-06, "loss": 0.24900484085083008, "step": 40 }, { "epoch": 0.03031423290203327, "grad_norm": 11.085528373718262, "learning_rate": 1.6393442622950819e-06, "loss": 0.2276933789253235, "step": 41 }, { "epoch": 0.031053604436229204, "grad_norm": 15.284082412719727, "learning_rate": 1.6803278688524592e-06, "loss": 0.25084948539733887, "step": 42 }, { "epoch": 0.03179297597042514, "grad_norm": 5.6036272048950195, "learning_rate": 1.7213114754098362e-06, "loss": 0.2582319974899292, "step": 43 }, { "epoch": 0.03253234750462107, "grad_norm": 3.8713016510009766, "learning_rate": 1.7622950819672133e-06, "loss": 0.23309245705604553, "step": 44 }, { "epoch": 0.033271719038817, "grad_norm": 4.740746974945068, "learning_rate": 1.8032786885245903e-06, "loss": 0.23871830105781555, "step": 45 }, { "epoch": 0.034011090573012936, "grad_norm": 6.922915458679199, "learning_rate": 1.8442622950819674e-06, "loss": 0.2267535924911499, "step": 46 }, { "epoch": 0.03475046210720887, "grad_norm": 10.260639190673828, "learning_rate": 1.8852459016393442e-06, "loss": 0.21310696005821228, "step": 47 }, { "epoch": 0.0354898336414048, "grad_norm": 4.459721565246582, "learning_rate": 1.9262295081967215e-06, "loss": 0.21494004130363464, "step": 48 }, { "epoch": 0.03622920517560074, "grad_norm": 8.453474044799805, "learning_rate": 1.9672131147540985e-06, "loss": 0.2108994573354721, "step": 49 }, { "epoch": 0.036968576709796676, "grad_norm": 7.217679500579834, "learning_rate": 2.0081967213114756e-06, "loss": 0.2561081051826477, "step": 50 }, { "epoch": 0.03770794824399261, "grad_norm": 6.249298095703125, "learning_rate": 2.0491803278688526e-06, "loss": 0.24657168984413147, "step": 51 }, { "epoch": 0.03844731977818854, "grad_norm": 6.808646202087402, "learning_rate": 2.0901639344262297e-06, "loss": 0.26588645577430725, "step": 52 }, { "epoch": 0.039186691312384475, "grad_norm": 10.094840049743652, "learning_rate": 2.1311475409836067e-06, "loss": 0.20153748989105225, "step": 53 }, { "epoch": 0.03992606284658041, "grad_norm": 3.169302463531494, "learning_rate": 2.1721311475409838e-06, "loss": 0.23092766106128693, "step": 54 }, { "epoch": 0.04066543438077634, "grad_norm": 3.960937738418579, "learning_rate": 2.213114754098361e-06, "loss": 0.215471088886261, "step": 55 }, { "epoch": 0.041404805914972274, "grad_norm": 4.973490238189697, "learning_rate": 2.254098360655738e-06, "loss": 0.23059183359146118, "step": 56 }, { "epoch": 0.04214417744916821, "grad_norm": 2.1337392330169678, "learning_rate": 2.295081967213115e-06, "loss": 0.21520796418190002, "step": 57 }, { "epoch": 0.04288354898336414, "grad_norm": 3.863631010055542, "learning_rate": 2.336065573770492e-06, "loss": 0.2167186141014099, "step": 58 }, { "epoch": 0.043622920517560074, "grad_norm": 2.5238311290740967, "learning_rate": 2.377049180327869e-06, "loss": 0.211829274892807, "step": 59 }, { "epoch": 0.04436229205175601, "grad_norm": 4.17237663269043, "learning_rate": 2.418032786885246e-06, "loss": 0.24345429241657257, "step": 60 }, { "epoch": 0.04510166358595194, "grad_norm": 2.299755096435547, "learning_rate": 2.459016393442623e-06, "loss": 0.2165399193763733, "step": 61 }, { "epoch": 0.04584103512014787, "grad_norm": 2.8945083618164062, "learning_rate": 2.5e-06, "loss": 0.23959916830062866, "step": 62 }, { "epoch": 0.046580406654343806, "grad_norm": 6.814939975738525, "learning_rate": 2.5409836065573773e-06, "loss": 0.231363907456398, "step": 63 }, { "epoch": 0.04731977818853974, "grad_norm": 4.236604690551758, "learning_rate": 2.5819672131147543e-06, "loss": 0.21419793367385864, "step": 64 }, { "epoch": 0.04805914972273567, "grad_norm": 5.936826229095459, "learning_rate": 2.6229508196721314e-06, "loss": 0.24576932191848755, "step": 65 }, { "epoch": 0.048798521256931605, "grad_norm": 3.419132709503174, "learning_rate": 2.6639344262295084e-06, "loss": 0.21549925208091736, "step": 66 }, { "epoch": 0.04953789279112754, "grad_norm": 3.9782159328460693, "learning_rate": 2.704918032786886e-06, "loss": 0.22133967280387878, "step": 67 }, { "epoch": 0.05027726432532347, "grad_norm": 3.587904691696167, "learning_rate": 2.745901639344263e-06, "loss": 0.24038229882717133, "step": 68 }, { "epoch": 0.05101663585951941, "grad_norm": 4.746975421905518, "learning_rate": 2.786885245901639e-06, "loss": 0.2418273389339447, "step": 69 }, { "epoch": 0.051756007393715345, "grad_norm": 5.241803169250488, "learning_rate": 2.8278688524590166e-06, "loss": 0.2080921232700348, "step": 70 }, { "epoch": 0.05249537892791128, "grad_norm": 2.7427096366882324, "learning_rate": 2.8688524590163937e-06, "loss": 0.22268177568912506, "step": 71 }, { "epoch": 0.05323475046210721, "grad_norm": 6.013676166534424, "learning_rate": 2.9098360655737707e-06, "loss": 0.24857483804225922, "step": 72 }, { "epoch": 0.053974121996303144, "grad_norm": 7.389932632446289, "learning_rate": 2.9508196721311478e-06, "loss": 0.248174250125885, "step": 73 }, { "epoch": 0.05471349353049908, "grad_norm": 9.882954597473145, "learning_rate": 2.991803278688525e-06, "loss": 0.217016339302063, "step": 74 }, { "epoch": 0.05545286506469501, "grad_norm": 8.985603332519531, "learning_rate": 3.032786885245902e-06, "loss": 0.2003176212310791, "step": 75 }, { "epoch": 0.05619223659889094, "grad_norm": 11.459195137023926, "learning_rate": 3.073770491803279e-06, "loss": 0.2398378998041153, "step": 76 }, { "epoch": 0.056931608133086876, "grad_norm": 8.19696044921875, "learning_rate": 3.114754098360656e-06, "loss": 0.2080891728401184, "step": 77 }, { "epoch": 0.05767097966728281, "grad_norm": 5.222296237945557, "learning_rate": 3.155737704918033e-06, "loss": 0.2421354353427887, "step": 78 }, { "epoch": 0.05841035120147874, "grad_norm": 5.266373157501221, "learning_rate": 3.1967213114754105e-06, "loss": 0.19913814961910248, "step": 79 }, { "epoch": 0.059149722735674676, "grad_norm": 5.215465545654297, "learning_rate": 3.2377049180327876e-06, "loss": 0.23360499739646912, "step": 80 }, { "epoch": 0.05988909426987061, "grad_norm": 3.0602304935455322, "learning_rate": 3.2786885245901638e-06, "loss": 0.2106657475233078, "step": 81 }, { "epoch": 0.06062846580406654, "grad_norm": 10.758472442626953, "learning_rate": 3.3196721311475413e-06, "loss": 0.224166139960289, "step": 82 }, { "epoch": 0.061367837338262475, "grad_norm": 8.338835716247559, "learning_rate": 3.3606557377049183e-06, "loss": 0.22351481020450592, "step": 83 }, { "epoch": 0.06210720887245841, "grad_norm": 6.37253999710083, "learning_rate": 3.4016393442622954e-06, "loss": 0.2120596468448639, "step": 84 }, { "epoch": 0.06284658040665435, "grad_norm": 4.617480754852295, "learning_rate": 3.4426229508196724e-06, "loss": 0.22747653722763062, "step": 85 }, { "epoch": 0.06358595194085027, "grad_norm": 7.08601188659668, "learning_rate": 3.4836065573770495e-06, "loss": 0.2010892629623413, "step": 86 }, { "epoch": 0.06432532347504621, "grad_norm": 4.797047138214111, "learning_rate": 3.5245901639344265e-06, "loss": 0.19760149717330933, "step": 87 }, { "epoch": 0.06506469500924214, "grad_norm": 3.1087045669555664, "learning_rate": 3.5655737704918036e-06, "loss": 0.199735626578331, "step": 88 }, { "epoch": 0.06580406654343808, "grad_norm": 6.299823760986328, "learning_rate": 3.6065573770491806e-06, "loss": 0.19973841309547424, "step": 89 }, { "epoch": 0.066543438077634, "grad_norm": 3.580165147781372, "learning_rate": 3.6475409836065577e-06, "loss": 0.21495819091796875, "step": 90 }, { "epoch": 0.06728280961182995, "grad_norm": 6.880489826202393, "learning_rate": 3.6885245901639347e-06, "loss": 0.2116369903087616, "step": 91 }, { "epoch": 0.06802218114602587, "grad_norm": 6.350500106811523, "learning_rate": 3.729508196721312e-06, "loss": 0.2031579464673996, "step": 92 }, { "epoch": 0.06876155268022181, "grad_norm": 5.7083940505981445, "learning_rate": 3.7704918032786884e-06, "loss": 0.2042863368988037, "step": 93 }, { "epoch": 0.06950092421441774, "grad_norm": 3.3069581985473633, "learning_rate": 3.811475409836066e-06, "loss": 0.20683512091636658, "step": 94 }, { "epoch": 0.07024029574861368, "grad_norm": 4.842657089233398, "learning_rate": 3.852459016393443e-06, "loss": 0.23477721214294434, "step": 95 }, { "epoch": 0.0709796672828096, "grad_norm": 6.163395404815674, "learning_rate": 3.8934426229508196e-06, "loss": 0.2075931280851364, "step": 96 }, { "epoch": 0.07171903881700555, "grad_norm": 4.934092998504639, "learning_rate": 3.934426229508197e-06, "loss": 0.23072142899036407, "step": 97 }, { "epoch": 0.07245841035120149, "grad_norm": 3.756474018096924, "learning_rate": 3.975409836065574e-06, "loss": 0.21308976411819458, "step": 98 }, { "epoch": 0.07319778188539741, "grad_norm": 3.4766643047332764, "learning_rate": 4.016393442622951e-06, "loss": 0.19323647022247314, "step": 99 }, { "epoch": 0.07393715341959335, "grad_norm": 3.348733425140381, "learning_rate": 4.057377049180329e-06, "loss": 0.2084563821554184, "step": 100 }, { "epoch": 0.07467652495378928, "grad_norm": 3.4332685470581055, "learning_rate": 4.098360655737705e-06, "loss": 0.2137387990951538, "step": 101 }, { "epoch": 0.07541589648798522, "grad_norm": 2.661440134048462, "learning_rate": 4.139344262295083e-06, "loss": 0.19090895354747772, "step": 102 }, { "epoch": 0.07615526802218114, "grad_norm": 8.872175216674805, "learning_rate": 4.180327868852459e-06, "loss": 0.24043507874011993, "step": 103 }, { "epoch": 0.07689463955637708, "grad_norm": 9.602208137512207, "learning_rate": 4.221311475409837e-06, "loss": 0.22823621332645416, "step": 104 }, { "epoch": 0.07763401109057301, "grad_norm": 3.878533363342285, "learning_rate": 4.2622950819672135e-06, "loss": 0.22249910235404968, "step": 105 }, { "epoch": 0.07837338262476895, "grad_norm": 10.086756706237793, "learning_rate": 4.30327868852459e-06, "loss": 0.2282578945159912, "step": 106 }, { "epoch": 0.07911275415896488, "grad_norm": 3.955077648162842, "learning_rate": 4.3442622950819676e-06, "loss": 0.2100994884967804, "step": 107 }, { "epoch": 0.07985212569316082, "grad_norm": 3.912059783935547, "learning_rate": 4.385245901639344e-06, "loss": 0.20815467834472656, "step": 108 }, { "epoch": 0.08059149722735674, "grad_norm": 12.097794532775879, "learning_rate": 4.426229508196722e-06, "loss": 0.20556432008743286, "step": 109 }, { "epoch": 0.08133086876155268, "grad_norm": 4.8796610832214355, "learning_rate": 4.467213114754098e-06, "loss": 0.23375600576400757, "step": 110 }, { "epoch": 0.08207024029574861, "grad_norm": 8.581815719604492, "learning_rate": 4.508196721311476e-06, "loss": 0.2341940701007843, "step": 111 }, { "epoch": 0.08280961182994455, "grad_norm": 3.9816882610321045, "learning_rate": 4.549180327868853e-06, "loss": 0.17916610836982727, "step": 112 }, { "epoch": 0.08354898336414047, "grad_norm": 5.734149932861328, "learning_rate": 4.59016393442623e-06, "loss": 0.22267544269561768, "step": 113 }, { "epoch": 0.08428835489833642, "grad_norm": 5.327974796295166, "learning_rate": 4.631147540983607e-06, "loss": 0.2290513515472412, "step": 114 }, { "epoch": 0.08502772643253234, "grad_norm": 4.79965877532959, "learning_rate": 4.672131147540984e-06, "loss": 0.20292600989341736, "step": 115 }, { "epoch": 0.08576709796672828, "grad_norm": 3.0385940074920654, "learning_rate": 4.7131147540983615e-06, "loss": 0.22854262590408325, "step": 116 }, { "epoch": 0.08650646950092422, "grad_norm": 5.024645805358887, "learning_rate": 4.754098360655738e-06, "loss": 0.21665805578231812, "step": 117 }, { "epoch": 0.08724584103512015, "grad_norm": 5.881789684295654, "learning_rate": 4.795081967213115e-06, "loss": 0.19748708605766296, "step": 118 }, { "epoch": 0.08798521256931609, "grad_norm": 4.2605085372924805, "learning_rate": 4.836065573770492e-06, "loss": 0.2076157182455063, "step": 119 }, { "epoch": 0.08872458410351201, "grad_norm": 10.676498413085938, "learning_rate": 4.877049180327869e-06, "loss": 0.22034040093421936, "step": 120 }, { "epoch": 0.08946395563770795, "grad_norm": 9.042420387268066, "learning_rate": 4.918032786885246e-06, "loss": 0.2072756290435791, "step": 121 }, { "epoch": 0.09020332717190388, "grad_norm": 7.659818172454834, "learning_rate": 4.959016393442623e-06, "loss": 0.22422261536121368, "step": 122 }, { "epoch": 0.09094269870609982, "grad_norm": 7.716509819030762, "learning_rate": 5e-06, "loss": 0.20881998538970947, "step": 123 }, { "epoch": 0.09168207024029575, "grad_norm": 7.7445173263549805, "learning_rate": 4.999999283656351e-06, "loss": 0.24182188510894775, "step": 124 }, { "epoch": 0.09242144177449169, "grad_norm": 8.004592895507812, "learning_rate": 4.999997134625859e-06, "loss": 0.20841506123542786, "step": 125 }, { "epoch": 0.09316081330868761, "grad_norm": 6.48236608505249, "learning_rate": 4.999993552909891e-06, "loss": 0.20456591248512268, "step": 126 }, { "epoch": 0.09390018484288355, "grad_norm": 6.797680377960205, "learning_rate": 4.99998853851073e-06, "loss": 0.20418843626976013, "step": 127 }, { "epoch": 0.09463955637707948, "grad_norm": 5.090514183044434, "learning_rate": 4.9999820914315674e-06, "loss": 0.20935019850730896, "step": 128 }, { "epoch": 0.09537892791127542, "grad_norm": 2.1391026973724365, "learning_rate": 4.999974211676508e-06, "loss": 0.2211488038301468, "step": 129 }, { "epoch": 0.09611829944547134, "grad_norm": 6.025007724761963, "learning_rate": 4.999964899250572e-06, "loss": 0.22729283571243286, "step": 130 }, { "epoch": 0.09685767097966728, "grad_norm": 7.152823448181152, "learning_rate": 4.999954154159686e-06, "loss": 0.19743826985359192, "step": 131 }, { "epoch": 0.09759704251386321, "grad_norm": 4.205631256103516, "learning_rate": 4.999941976410694e-06, "loss": 0.23450744152069092, "step": 132 }, { "epoch": 0.09833641404805915, "grad_norm": 8.204564094543457, "learning_rate": 4.9999283660113485e-06, "loss": 0.2338467687368393, "step": 133 }, { "epoch": 0.09907578558225508, "grad_norm": 3.3347606658935547, "learning_rate": 4.999913322970318e-06, "loss": 0.23568445444107056, "step": 134 }, { "epoch": 0.09981515711645102, "grad_norm": 3.5187604427337646, "learning_rate": 4.999896847297178e-06, "loss": 0.20000651478767395, "step": 135 }, { "epoch": 0.10055452865064694, "grad_norm": 15.043497085571289, "learning_rate": 4.999878939002423e-06, "loss": 0.22715622186660767, "step": 136 }, { "epoch": 0.10129390018484288, "grad_norm": 1.9297523498535156, "learning_rate": 4.999859598097453e-06, "loss": 0.21605083346366882, "step": 137 }, { "epoch": 0.10203327171903882, "grad_norm": 6.430673122406006, "learning_rate": 4.999838824594586e-06, "loss": 0.2158619612455368, "step": 138 }, { "epoch": 0.10277264325323475, "grad_norm": 10.023754119873047, "learning_rate": 4.999816618507047e-06, "loss": 0.22568903863430023, "step": 139 }, { "epoch": 0.10351201478743069, "grad_norm": 3.9803414344787598, "learning_rate": 4.999792979848979e-06, "loss": 0.17261339724063873, "step": 140 }, { "epoch": 0.10425138632162662, "grad_norm": 9.742910385131836, "learning_rate": 4.99976790863543e-06, "loss": 0.2254084050655365, "step": 141 }, { "epoch": 0.10499075785582256, "grad_norm": 7.166801452636719, "learning_rate": 4.9997414048823665e-06, "loss": 0.2113988995552063, "step": 142 }, { "epoch": 0.10573012939001848, "grad_norm": 2.6640477180480957, "learning_rate": 4.999713468606664e-06, "loss": 0.1783626824617386, "step": 143 }, { "epoch": 0.10646950092421442, "grad_norm": 10.168929100036621, "learning_rate": 4.999684099826112e-06, "loss": 0.22603170573711395, "step": 144 }, { "epoch": 0.10720887245841035, "grad_norm": 6.008780002593994, "learning_rate": 4.999653298559409e-06, "loss": 0.20697957277297974, "step": 145 }, { "epoch": 0.10794824399260629, "grad_norm": 2.8306233882904053, "learning_rate": 4.9996210648261685e-06, "loss": 0.21823719143867493, "step": 146 }, { "epoch": 0.10868761552680221, "grad_norm": 6.534702301025391, "learning_rate": 4.999587398646918e-06, "loss": 0.22830304503440857, "step": 147 }, { "epoch": 0.10942698706099815, "grad_norm": 3.725764751434326, "learning_rate": 4.999552300043091e-06, "loss": 0.2207079827785492, "step": 148 }, { "epoch": 0.11016635859519408, "grad_norm": 3.8723645210266113, "learning_rate": 4.9995157690370374e-06, "loss": 0.2065512239933014, "step": 149 }, { "epoch": 0.11090573012939002, "grad_norm": 3.52687406539917, "learning_rate": 4.999477805652019e-06, "loss": 0.18676112592220306, "step": 150 }, { "epoch": 0.11164510166358595, "grad_norm": 7.469851493835449, "learning_rate": 4.9994384099122086e-06, "loss": 0.21065548062324524, "step": 151 }, { "epoch": 0.11238447319778189, "grad_norm": 3.6667792797088623, "learning_rate": 4.999397581842691e-06, "loss": 0.2181682139635086, "step": 152 }, { "epoch": 0.11312384473197781, "grad_norm": 4.797725677490234, "learning_rate": 4.999355321469464e-06, "loss": 0.1876065731048584, "step": 153 }, { "epoch": 0.11386321626617375, "grad_norm": 4.861723899841309, "learning_rate": 4.9993116288194374e-06, "loss": 0.19971875846385956, "step": 154 }, { "epoch": 0.11460258780036968, "grad_norm": 1.7892388105392456, "learning_rate": 4.999266503920432e-06, "loss": 0.21427346765995026, "step": 155 }, { "epoch": 0.11534195933456562, "grad_norm": 3.6066734790802, "learning_rate": 4.99921994680118e-06, "loss": 0.20555222034454346, "step": 156 }, { "epoch": 0.11608133086876156, "grad_norm": 2.5167219638824463, "learning_rate": 4.9991719574913276e-06, "loss": 0.19464248418807983, "step": 157 }, { "epoch": 0.11682070240295749, "grad_norm": 3.415402889251709, "learning_rate": 4.999122536021434e-06, "loss": 0.2183443009853363, "step": 158 }, { "epoch": 0.11756007393715343, "grad_norm": 4.245021820068359, "learning_rate": 4.999071682422964e-06, "loss": 0.1938258409500122, "step": 159 }, { "epoch": 0.11829944547134935, "grad_norm": 2.013408899307251, "learning_rate": 4.999019396728302e-06, "loss": 0.2131974697113037, "step": 160 }, { "epoch": 0.11903881700554529, "grad_norm": 6.069725036621094, "learning_rate": 4.99896567897074e-06, "loss": 0.20909833908081055, "step": 161 }, { "epoch": 0.11977818853974122, "grad_norm": 9.454623222351074, "learning_rate": 4.998910529184482e-06, "loss": 0.200724259018898, "step": 162 }, { "epoch": 0.12051756007393716, "grad_norm": 5.46451473236084, "learning_rate": 4.998853947404646e-06, "loss": 0.20431353151798248, "step": 163 }, { "epoch": 0.12125693160813308, "grad_norm": 11.093324661254883, "learning_rate": 4.998795933667259e-06, "loss": 0.23856434226036072, "step": 164 }, { "epoch": 0.12199630314232902, "grad_norm": 2.7444450855255127, "learning_rate": 4.998736488009262e-06, "loss": 0.20140336453914642, "step": 165 }, { "epoch": 0.12273567467652495, "grad_norm": 4.153026580810547, "learning_rate": 4.998675610468508e-06, "loss": 0.2260104864835739, "step": 166 }, { "epoch": 0.12347504621072089, "grad_norm": 11.857220649719238, "learning_rate": 4.998613301083759e-06, "loss": 0.21431300044059753, "step": 167 }, { "epoch": 0.12421441774491682, "grad_norm": 1.8211326599121094, "learning_rate": 4.998549559894691e-06, "loss": 0.1953921914100647, "step": 168 }, { "epoch": 0.12495378927911276, "grad_norm": 8.21317195892334, "learning_rate": 4.998484386941892e-06, "loss": 0.20154309272766113, "step": 169 }, { "epoch": 0.1256931608133087, "grad_norm": 4.826503753662109, "learning_rate": 4.998417782266859e-06, "loss": 0.17201445996761322, "step": 170 }, { "epoch": 0.12643253234750462, "grad_norm": 3.370664119720459, "learning_rate": 4.998349745912004e-06, "loss": 0.2008911669254303, "step": 171 }, { "epoch": 0.12717190388170055, "grad_norm": 3.427187442779541, "learning_rate": 4.99828027792065e-06, "loss": 0.20942333340644836, "step": 172 }, { "epoch": 0.12791127541589647, "grad_norm": 11.606083869934082, "learning_rate": 4.998209378337029e-06, "loss": 0.22189611196517944, "step": 173 }, { "epoch": 0.12865064695009243, "grad_norm": 5.491366863250732, "learning_rate": 4.998137047206288e-06, "loss": 0.19459503889083862, "step": 174 }, { "epoch": 0.12939001848428835, "grad_norm": 2.2893073558807373, "learning_rate": 4.998063284574482e-06, "loss": 0.2056599259376526, "step": 175 }, { "epoch": 0.13012939001848428, "grad_norm": 7.509938716888428, "learning_rate": 4.9979880904885805e-06, "loss": 0.1777127981185913, "step": 176 }, { "epoch": 0.13086876155268024, "grad_norm": 8.536020278930664, "learning_rate": 4.997911464996462e-06, "loss": 0.20062540471553802, "step": 177 }, { "epoch": 0.13160813308687616, "grad_norm": 5.545294284820557, "learning_rate": 4.99783340814692e-06, "loss": 0.21459735929965973, "step": 178 }, { "epoch": 0.1323475046210721, "grad_norm": 4.776959419250488, "learning_rate": 4.9977539199896556e-06, "loss": 0.18448832631111145, "step": 179 }, { "epoch": 0.133086876155268, "grad_norm": 9.121623039245605, "learning_rate": 4.9976730005752835e-06, "loss": 0.22773566842079163, "step": 180 }, { "epoch": 0.13382624768946397, "grad_norm": 4.554732799530029, "learning_rate": 4.997590649955328e-06, "loss": 0.19444584846496582, "step": 181 }, { "epoch": 0.1345656192236599, "grad_norm": 5.279799461364746, "learning_rate": 4.997506868182228e-06, "loss": 0.21681542694568634, "step": 182 }, { "epoch": 0.13530499075785582, "grad_norm": 3.889777898788452, "learning_rate": 4.997421655309331e-06, "loss": 0.2028026133775711, "step": 183 }, { "epoch": 0.13604436229205175, "grad_norm": 3.160100221633911, "learning_rate": 4.9973350113908935e-06, "loss": 0.19253882765769958, "step": 184 }, { "epoch": 0.1367837338262477, "grad_norm": 2.183661699295044, "learning_rate": 4.997246936482088e-06, "loss": 0.18883726000785828, "step": 185 }, { "epoch": 0.13752310536044363, "grad_norm": 8.732416152954102, "learning_rate": 4.9971574306389976e-06, "loss": 0.21693305671215057, "step": 186 }, { "epoch": 0.13826247689463955, "grad_norm": 4.159419059753418, "learning_rate": 4.9970664939186135e-06, "loss": 0.24000497162342072, "step": 187 }, { "epoch": 0.13900184842883548, "grad_norm": 3.013119697570801, "learning_rate": 4.996974126378839e-06, "loss": 0.209400475025177, "step": 188 }, { "epoch": 0.13974121996303143, "grad_norm": 6.901010990142822, "learning_rate": 4.996880328078491e-06, "loss": 0.2435654103755951, "step": 189 }, { "epoch": 0.14048059149722736, "grad_norm": 4.0845513343811035, "learning_rate": 4.996785099077295e-06, "loss": 0.19311289489269257, "step": 190 }, { "epoch": 0.14121996303142328, "grad_norm": 5.842734336853027, "learning_rate": 4.996688439435887e-06, "loss": 0.2088090479373932, "step": 191 }, { "epoch": 0.1419593345656192, "grad_norm": 4.06777286529541, "learning_rate": 4.996590349215815e-06, "loss": 0.19047898054122925, "step": 192 }, { "epoch": 0.14269870609981516, "grad_norm": 4.351992130279541, "learning_rate": 4.99649082847954e-06, "loss": 0.23591892421245575, "step": 193 }, { "epoch": 0.1434380776340111, "grad_norm": 6.226611137390137, "learning_rate": 4.9963898772904295e-06, "loss": 0.20714104175567627, "step": 194 }, { "epoch": 0.14417744916820702, "grad_norm": 3.3315727710723877, "learning_rate": 4.9962874957127655e-06, "loss": 0.2250680923461914, "step": 195 }, { "epoch": 0.14491682070240297, "grad_norm": 3.878664255142212, "learning_rate": 4.996183683811739e-06, "loss": 0.2087586224079132, "step": 196 }, { "epoch": 0.1456561922365989, "grad_norm": 6.630451679229736, "learning_rate": 4.996078441653453e-06, "loss": 0.2055913656949997, "step": 197 }, { "epoch": 0.14639556377079482, "grad_norm": 3.6534416675567627, "learning_rate": 4.995971769304919e-06, "loss": 0.17155134677886963, "step": 198 }, { "epoch": 0.14713493530499075, "grad_norm": 5.370640277862549, "learning_rate": 4.9958636668340615e-06, "loss": 0.20528919994831085, "step": 199 }, { "epoch": 0.1478743068391867, "grad_norm": 5.47586727142334, "learning_rate": 4.995754134309715e-06, "loss": 0.19093063473701477, "step": 200 }, { "epoch": 0.14861367837338263, "grad_norm": 3.69317364692688, "learning_rate": 4.995643171801623e-06, "loss": 0.1792043149471283, "step": 201 }, { "epoch": 0.14935304990757856, "grad_norm": 1.8511996269226074, "learning_rate": 4.995530779380442e-06, "loss": 0.2025895118713379, "step": 202 }, { "epoch": 0.15009242144177448, "grad_norm": 8.300798416137695, "learning_rate": 4.995416957117738e-06, "loss": 0.21189391613006592, "step": 203 }, { "epoch": 0.15083179297597044, "grad_norm": 5.054575443267822, "learning_rate": 4.995301705085987e-06, "loss": 0.19224077463150024, "step": 204 }, { "epoch": 0.15157116451016636, "grad_norm": 3.6741697788238525, "learning_rate": 4.995185023358576e-06, "loss": 0.2107357382774353, "step": 205 }, { "epoch": 0.1523105360443623, "grad_norm": 7.290289878845215, "learning_rate": 4.995066912009801e-06, "loss": 0.22415247559547424, "step": 206 }, { "epoch": 0.15304990757855821, "grad_norm": 4.428752899169922, "learning_rate": 4.994947371114871e-06, "loss": 0.21634747087955475, "step": 207 }, { "epoch": 0.15378927911275417, "grad_norm": 3.1207001209259033, "learning_rate": 4.994826400749901e-06, "loss": 0.20806077122688293, "step": 208 }, { "epoch": 0.1545286506469501, "grad_norm": 8.726855278015137, "learning_rate": 4.994704000991922e-06, "loss": 0.2170030027627945, "step": 209 }, { "epoch": 0.15526802218114602, "grad_norm": 8.585246086120605, "learning_rate": 4.994580171918871e-06, "loss": 0.17840728163719177, "step": 210 }, { "epoch": 0.15600739371534195, "grad_norm": 4.869527816772461, "learning_rate": 4.994454913609596e-06, "loss": 0.1750405728816986, "step": 211 }, { "epoch": 0.1567467652495379, "grad_norm": 10.767374992370605, "learning_rate": 4.9943282261438544e-06, "loss": 0.2183046191930771, "step": 212 }, { "epoch": 0.15748613678373383, "grad_norm": 6.926781177520752, "learning_rate": 4.994200109602316e-06, "loss": 0.19309481978416443, "step": 213 }, { "epoch": 0.15822550831792975, "grad_norm": 2.3617966175079346, "learning_rate": 4.994070564066557e-06, "loss": 0.19992075860500336, "step": 214 }, { "epoch": 0.1589648798521257, "grad_norm": 6.474941253662109, "learning_rate": 4.993939589619067e-06, "loss": 0.19902080297470093, "step": 215 }, { "epoch": 0.15970425138632163, "grad_norm": 7.6652631759643555, "learning_rate": 4.993807186343243e-06, "loss": 0.21584287285804749, "step": 216 }, { "epoch": 0.16044362292051756, "grad_norm": 2.87681245803833, "learning_rate": 4.9936733543233945e-06, "loss": 0.20390164852142334, "step": 217 }, { "epoch": 0.16118299445471349, "grad_norm": 3.3492889404296875, "learning_rate": 4.9935380936447375e-06, "loss": 0.19411179423332214, "step": 218 }, { "epoch": 0.16192236598890944, "grad_norm": 8.996016502380371, "learning_rate": 4.993401404393399e-06, "loss": 0.2624683380126953, "step": 219 }, { "epoch": 0.16266173752310537, "grad_norm": 3.854630470275879, "learning_rate": 4.993263286656418e-06, "loss": 0.20680753886699677, "step": 220 }, { "epoch": 0.1634011090573013, "grad_norm": 2.1955208778381348, "learning_rate": 4.993123740521737e-06, "loss": 0.20369084179401398, "step": 221 }, { "epoch": 0.16414048059149722, "grad_norm": 6.356298446655273, "learning_rate": 4.992982766078216e-06, "loss": 0.21157515048980713, "step": 222 }, { "epoch": 0.16487985212569317, "grad_norm": 3.5238497257232666, "learning_rate": 4.9928403634156194e-06, "loss": 0.20293402671813965, "step": 223 }, { "epoch": 0.1656192236598891, "grad_norm": 2.00789475440979, "learning_rate": 4.99269653262462e-06, "loss": 0.2008260190486908, "step": 224 }, { "epoch": 0.16635859519408502, "grad_norm": 1.447245478630066, "learning_rate": 4.992551273796805e-06, "loss": 0.20095610618591309, "step": 225 }, { "epoch": 0.16709796672828095, "grad_norm": 2.8336122035980225, "learning_rate": 4.992404587024666e-06, "loss": 0.21512047946453094, "step": 226 }, { "epoch": 0.1678373382624769, "grad_norm": 5.293940544128418, "learning_rate": 4.992256472401606e-06, "loss": 0.20414477586746216, "step": 227 }, { "epoch": 0.16857670979667283, "grad_norm": 2.7756428718566895, "learning_rate": 4.992106930021938e-06, "loss": 0.20077075064182281, "step": 228 }, { "epoch": 0.16931608133086876, "grad_norm": 3.7689459323883057, "learning_rate": 4.991955959980882e-06, "loss": 0.19798599183559418, "step": 229 }, { "epoch": 0.17005545286506468, "grad_norm": 6.501115798950195, "learning_rate": 4.991803562374568e-06, "loss": 0.20230010151863098, "step": 230 }, { "epoch": 0.17079482439926064, "grad_norm": 4.261605262756348, "learning_rate": 4.991649737300037e-06, "loss": 0.22297659516334534, "step": 231 }, { "epoch": 0.17153419593345656, "grad_norm": 1.7686529159545898, "learning_rate": 4.991494484855235e-06, "loss": 0.19962067902088165, "step": 232 }, { "epoch": 0.1722735674676525, "grad_norm": 6.591203212738037, "learning_rate": 4.99133780513902e-06, "loss": 0.22126823663711548, "step": 233 }, { "epoch": 0.17301293900184844, "grad_norm": 1.7832783460617065, "learning_rate": 4.9911796982511576e-06, "loss": 0.20793044567108154, "step": 234 }, { "epoch": 0.17375231053604437, "grad_norm": 4.711670398712158, "learning_rate": 4.991020164292322e-06, "loss": 0.2003735601902008, "step": 235 }, { "epoch": 0.1744916820702403, "grad_norm": 3.7317111492156982, "learning_rate": 4.990859203364098e-06, "loss": 0.20356100797653198, "step": 236 }, { "epoch": 0.17523105360443622, "grad_norm": 2.455528974533081, "learning_rate": 4.990696815568975e-06, "loss": 0.1997373104095459, "step": 237 }, { "epoch": 0.17597042513863217, "grad_norm": 1.4064719676971436, "learning_rate": 4.9905330010103555e-06, "loss": 0.17668284475803375, "step": 238 }, { "epoch": 0.1767097966728281, "grad_norm": 4.904377460479736, "learning_rate": 4.990367759792547e-06, "loss": 0.21554915606975555, "step": 239 }, { "epoch": 0.17744916820702403, "grad_norm": 2.447503089904785, "learning_rate": 4.990201092020768e-06, "loss": 0.1815405786037445, "step": 240 }, { "epoch": 0.17818853974121995, "grad_norm": 3.287691831588745, "learning_rate": 4.990032997801143e-06, "loss": 0.22017735242843628, "step": 241 }, { "epoch": 0.1789279112754159, "grad_norm": 4.097412109375, "learning_rate": 4.989863477240707e-06, "loss": 0.1979101598262787, "step": 242 }, { "epoch": 0.17966728280961183, "grad_norm": 1.2077791690826416, "learning_rate": 4.989692530447402e-06, "loss": 0.18745893239974976, "step": 243 }, { "epoch": 0.18040665434380776, "grad_norm": 2.1230628490448, "learning_rate": 4.989520157530077e-06, "loss": 0.19807478785514832, "step": 244 }, { "epoch": 0.18114602587800369, "grad_norm": 6.088411808013916, "learning_rate": 4.989346358598493e-06, "loss": 0.1851760447025299, "step": 245 }, { "epoch": 0.18188539741219964, "grad_norm": 3.149026393890381, "learning_rate": 4.989171133763314e-06, "loss": 0.1751485913991928, "step": 246 }, { "epoch": 0.18262476894639557, "grad_norm": 6.697902679443359, "learning_rate": 4.988994483136115e-06, "loss": 0.20846855640411377, "step": 247 }, { "epoch": 0.1833641404805915, "grad_norm": 2.725067138671875, "learning_rate": 4.988816406829379e-06, "loss": 0.20738132297992706, "step": 248 }, { "epoch": 0.18410351201478742, "grad_norm": 1.9915666580200195, "learning_rate": 4.988636904956497e-06, "loss": 0.17252105474472046, "step": 249 }, { "epoch": 0.18484288354898337, "grad_norm": 5.8566718101501465, "learning_rate": 4.988455977631764e-06, "loss": 0.21974192559719086, "step": 250 }, { "epoch": 0.1855822550831793, "grad_norm": 3.0231456756591797, "learning_rate": 4.988273624970388e-06, "loss": 0.1627470701932907, "step": 251 }, { "epoch": 0.18632162661737522, "grad_norm": 2.773123264312744, "learning_rate": 4.988089847088481e-06, "loss": 0.17384780943393707, "step": 252 }, { "epoch": 0.18706099815157118, "grad_norm": 3.836524724960327, "learning_rate": 4.987904644103063e-06, "loss": 0.21001049876213074, "step": 253 }, { "epoch": 0.1878003696857671, "grad_norm": 3.7989351749420166, "learning_rate": 4.987718016132063e-06, "loss": 0.18146644532680511, "step": 254 }, { "epoch": 0.18853974121996303, "grad_norm": 2.7386744022369385, "learning_rate": 4.9875299632943165e-06, "loss": 0.20317289233207703, "step": 255 }, { "epoch": 0.18927911275415896, "grad_norm": 3.3862056732177734, "learning_rate": 4.9873404857095645e-06, "loss": 0.1905624270439148, "step": 256 }, { "epoch": 0.1900184842883549, "grad_norm": 8.02692985534668, "learning_rate": 4.987149583498459e-06, "loss": 0.23594623804092407, "step": 257 }, { "epoch": 0.19075785582255084, "grad_norm": 4.152957439422607, "learning_rate": 4.986957256782555e-06, "loss": 0.1737653762102127, "step": 258 }, { "epoch": 0.19149722735674676, "grad_norm": 6.937053680419922, "learning_rate": 4.986763505684317e-06, "loss": 0.21992862224578857, "step": 259 }, { "epoch": 0.1922365988909427, "grad_norm": 5.229133605957031, "learning_rate": 4.986568330327118e-06, "loss": 0.19341173768043518, "step": 260 }, { "epoch": 0.19297597042513864, "grad_norm": 4.96560525894165, "learning_rate": 4.986371730835233e-06, "loss": 0.20328712463378906, "step": 261 }, { "epoch": 0.19371534195933457, "grad_norm": 9.03918170928955, "learning_rate": 4.986173707333848e-06, "loss": 0.19401070475578308, "step": 262 }, { "epoch": 0.1944547134935305, "grad_norm": 3.2892279624938965, "learning_rate": 4.985974259949056e-06, "loss": 0.20594069361686707, "step": 263 }, { "epoch": 0.19519408502772642, "grad_norm": 6.467484474182129, "learning_rate": 4.985773388807852e-06, "loss": 0.20840777456760406, "step": 264 }, { "epoch": 0.19593345656192238, "grad_norm": 9.916982650756836, "learning_rate": 4.985571094038143e-06, "loss": 0.21257367730140686, "step": 265 }, { "epoch": 0.1966728280961183, "grad_norm": 2.8090975284576416, "learning_rate": 4.985367375768738e-06, "loss": 0.17601069808006287, "step": 266 }, { "epoch": 0.19741219963031423, "grad_norm": 3.6637356281280518, "learning_rate": 4.985162234129357e-06, "loss": 0.18365761637687683, "step": 267 }, { "epoch": 0.19815157116451015, "grad_norm": 6.63193416595459, "learning_rate": 4.984955669250623e-06, "loss": 0.19654522836208344, "step": 268 }, { "epoch": 0.1988909426987061, "grad_norm": 5.122948169708252, "learning_rate": 4.984747681264066e-06, "loss": 0.20467603206634521, "step": 269 }, { "epoch": 0.19963031423290203, "grad_norm": 2.052618980407715, "learning_rate": 4.984538270302123e-06, "loss": 0.2092115879058838, "step": 270 }, { "epoch": 0.20036968576709796, "grad_norm": 4.634403228759766, "learning_rate": 4.984327436498135e-06, "loss": 0.2088620513677597, "step": 271 }, { "epoch": 0.2011090573012939, "grad_norm": 6.077898979187012, "learning_rate": 4.984115179986352e-06, "loss": 0.1845199167728424, "step": 272 }, { "epoch": 0.20184842883548984, "grad_norm": 1.7117427587509155, "learning_rate": 4.9839015009019255e-06, "loss": 0.19528141617774963, "step": 273 }, { "epoch": 0.20258780036968577, "grad_norm": 3.516902446746826, "learning_rate": 4.98368639938092e-06, "loss": 0.2160193771123886, "step": 274 }, { "epoch": 0.2033271719038817, "grad_norm": 3.411320924758911, "learning_rate": 4.983469875560297e-06, "loss": 0.19210456311702728, "step": 275 }, { "epoch": 0.20406654343807765, "grad_norm": 6.692568302154541, "learning_rate": 4.983251929577932e-06, "loss": 0.21587666869163513, "step": 276 }, { "epoch": 0.20480591497227357, "grad_norm": 3.1736679077148438, "learning_rate": 4.9830325615726e-06, "loss": 0.20437613129615784, "step": 277 }, { "epoch": 0.2055452865064695, "grad_norm": 5.157447814941406, "learning_rate": 4.982811771683982e-06, "loss": 0.1943303644657135, "step": 278 }, { "epoch": 0.20628465804066543, "grad_norm": 6.810500621795654, "learning_rate": 4.982589560052668e-06, "loss": 0.18354395031929016, "step": 279 }, { "epoch": 0.20702402957486138, "grad_norm": 4.875326156616211, "learning_rate": 4.9823659268201526e-06, "loss": 0.15934759378433228, "step": 280 }, { "epoch": 0.2077634011090573, "grad_norm": 5.021531105041504, "learning_rate": 4.982140872128831e-06, "loss": 0.22825074195861816, "step": 281 }, { "epoch": 0.20850277264325323, "grad_norm": 4.189813613891602, "learning_rate": 4.981914396122009e-06, "loss": 0.19403614103794098, "step": 282 }, { "epoch": 0.20924214417744916, "grad_norm": 4.390859127044678, "learning_rate": 4.981686498943895e-06, "loss": 0.21228596568107605, "step": 283 }, { "epoch": 0.2099815157116451, "grad_norm": 1.5592443943023682, "learning_rate": 4.981457180739601e-06, "loss": 0.18647845089435577, "step": 284 }, { "epoch": 0.21072088724584104, "grad_norm": 2.7587664127349854, "learning_rate": 4.9812264416551465e-06, "loss": 0.16063226759433746, "step": 285 }, { "epoch": 0.21146025878003696, "grad_norm": 10.106371879577637, "learning_rate": 4.980994281837455e-06, "loss": 0.21723011136054993, "step": 286 }, { "epoch": 0.2121996303142329, "grad_norm": 3.6956052780151367, "learning_rate": 4.980760701434353e-06, "loss": 0.21702319383621216, "step": 287 }, { "epoch": 0.21293900184842884, "grad_norm": 6.041219711303711, "learning_rate": 4.980525700594573e-06, "loss": 0.1681990623474121, "step": 288 }, { "epoch": 0.21367837338262477, "grad_norm": 5.66995906829834, "learning_rate": 4.980289279467753e-06, "loss": 0.195545956492424, "step": 289 }, { "epoch": 0.2144177449168207, "grad_norm": 3.6900999546051025, "learning_rate": 4.980051438204433e-06, "loss": 0.19705148041248322, "step": 290 }, { "epoch": 0.21515711645101662, "grad_norm": 3.436933755874634, "learning_rate": 4.979812176956059e-06, "loss": 0.17405083775520325, "step": 291 }, { "epoch": 0.21589648798521258, "grad_norm": 3.901014566421509, "learning_rate": 4.97957149587498e-06, "loss": 0.22101682424545288, "step": 292 }, { "epoch": 0.2166358595194085, "grad_norm": 4.328439712524414, "learning_rate": 4.979329395114451e-06, "loss": 0.16530074179172516, "step": 293 }, { "epoch": 0.21737523105360443, "grad_norm": 4.81765604019165, "learning_rate": 4.979085874828628e-06, "loss": 0.18608534336090088, "step": 294 }, { "epoch": 0.21811460258780038, "grad_norm": 3.900667190551758, "learning_rate": 4.978840935172573e-06, "loss": 0.19320952892303467, "step": 295 }, { "epoch": 0.2188539741219963, "grad_norm": 7.104184627532959, "learning_rate": 4.978594576302252e-06, "loss": 0.20164072513580322, "step": 296 }, { "epoch": 0.21959334565619223, "grad_norm": 4.423066139221191, "learning_rate": 4.9783467983745325e-06, "loss": 0.2018386870622635, "step": 297 }, { "epoch": 0.22033271719038816, "grad_norm": 3.0070528984069824, "learning_rate": 4.978097601547188e-06, "loss": 0.20285794138908386, "step": 298 }, { "epoch": 0.22107208872458411, "grad_norm": 2.575495719909668, "learning_rate": 4.977846985978895e-06, "loss": 0.21943402290344238, "step": 299 }, { "epoch": 0.22181146025878004, "grad_norm": 1.4534478187561035, "learning_rate": 4.977594951829233e-06, "loss": 0.1806921362876892, "step": 300 }, { "epoch": 0.22255083179297597, "grad_norm": 4.882898807525635, "learning_rate": 4.9773414992586835e-06, "loss": 0.20885509252548218, "step": 301 }, { "epoch": 0.2232902033271719, "grad_norm": 1.4090415239334106, "learning_rate": 4.977086628428634e-06, "loss": 0.18108311295509338, "step": 302 }, { "epoch": 0.22402957486136785, "grad_norm": 2.6148059368133545, "learning_rate": 4.976830339501372e-06, "loss": 0.20696879923343658, "step": 303 }, { "epoch": 0.22476894639556377, "grad_norm": 5.0142059326171875, "learning_rate": 4.97657263264009e-06, "loss": 0.20448508858680725, "step": 304 }, { "epoch": 0.2255083179297597, "grad_norm": 3.375950813293457, "learning_rate": 4.976313508008882e-06, "loss": 0.2123827338218689, "step": 305 }, { "epoch": 0.22624768946395563, "grad_norm": 3.452205181121826, "learning_rate": 4.976052965772748e-06, "loss": 0.19331833720207214, "step": 306 }, { "epoch": 0.22698706099815158, "grad_norm": 2.753509998321533, "learning_rate": 4.975791006097586e-06, "loss": 0.18583831191062927, "step": 307 }, { "epoch": 0.2277264325323475, "grad_norm": 5.09559440612793, "learning_rate": 4.9755276291501995e-06, "loss": 0.13720080256462097, "step": 308 }, { "epoch": 0.22846580406654343, "grad_norm": 2.3837342262268066, "learning_rate": 4.975262835098294e-06, "loss": 0.20531409978866577, "step": 309 }, { "epoch": 0.22920517560073936, "grad_norm": 4.245652198791504, "learning_rate": 4.974996624110477e-06, "loss": 0.18447069823741913, "step": 310 }, { "epoch": 0.2299445471349353, "grad_norm": 3.867917060852051, "learning_rate": 4.974728996356258e-06, "loss": 0.2040885090827942, "step": 311 }, { "epoch": 0.23068391866913124, "grad_norm": 2.705718755722046, "learning_rate": 4.9744599520060495e-06, "loss": 0.22290891408920288, "step": 312 }, { "epoch": 0.23142329020332716, "grad_norm": 6.047215461730957, "learning_rate": 4.9741894912311645e-06, "loss": 0.18437939882278442, "step": 313 }, { "epoch": 0.23216266173752312, "grad_norm": 4.581978797912598, "learning_rate": 4.9739176142038195e-06, "loss": 0.21812546253204346, "step": 314 }, { "epoch": 0.23290203327171904, "grad_norm": 1.290384292602539, "learning_rate": 4.973644321097132e-06, "loss": 0.1888195425271988, "step": 315 }, { "epoch": 0.23364140480591497, "grad_norm": 4.1905717849731445, "learning_rate": 4.9733696120851226e-06, "loss": 0.1866227090358734, "step": 316 }, { "epoch": 0.2343807763401109, "grad_norm": 4.627075672149658, "learning_rate": 4.973093487342711e-06, "loss": 0.19650252163410187, "step": 317 }, { "epoch": 0.23512014787430685, "grad_norm": 3.7013800144195557, "learning_rate": 4.972815947045718e-06, "loss": 0.2041383981704712, "step": 318 }, { "epoch": 0.23585951940850278, "grad_norm": 3.662653684616089, "learning_rate": 4.97253699137087e-06, "loss": 0.17552237212657928, "step": 319 }, { "epoch": 0.2365988909426987, "grad_norm": 4.362520694732666, "learning_rate": 4.9722566204957925e-06, "loss": 0.20358365774154663, "step": 320 }, { "epoch": 0.23733826247689463, "grad_norm": 3.0910589694976807, "learning_rate": 4.971974834599009e-06, "loss": 0.22397223114967346, "step": 321 }, { "epoch": 0.23807763401109058, "grad_norm": 2.742668390274048, "learning_rate": 4.971691633859947e-06, "loss": 0.172819584608078, "step": 322 }, { "epoch": 0.2388170055452865, "grad_norm": 4.824236869812012, "learning_rate": 4.971407018458937e-06, "loss": 0.16910478472709656, "step": 323 }, { "epoch": 0.23955637707948244, "grad_norm": 2.080307960510254, "learning_rate": 4.971120988577205e-06, "loss": 0.19993707537651062, "step": 324 }, { "epoch": 0.24029574861367836, "grad_norm": 4.409267902374268, "learning_rate": 4.970833544396883e-06, "loss": 0.18212445080280304, "step": 325 }, { "epoch": 0.24103512014787432, "grad_norm": 2.5548388957977295, "learning_rate": 4.970544686100999e-06, "loss": 0.22281834483146667, "step": 326 }, { "epoch": 0.24177449168207024, "grad_norm": 5.165862560272217, "learning_rate": 4.970254413873484e-06, "loss": 0.19897609949111938, "step": 327 }, { "epoch": 0.24251386321626617, "grad_norm": 3.0673162937164307, "learning_rate": 4.96996272789917e-06, "loss": 0.19602590799331665, "step": 328 }, { "epoch": 0.2432532347504621, "grad_norm": 3.4100685119628906, "learning_rate": 4.969669628363787e-06, "loss": 0.19802257418632507, "step": 329 }, { "epoch": 0.24399260628465805, "grad_norm": 5.849836826324463, "learning_rate": 4.969375115453966e-06, "loss": 0.20035341382026672, "step": 330 }, { "epoch": 0.24473197781885397, "grad_norm": 2.2142269611358643, "learning_rate": 4.969079189357238e-06, "loss": 0.20022240281105042, "step": 331 }, { "epoch": 0.2454713493530499, "grad_norm": 7.2502760887146, "learning_rate": 4.9687818502620345e-06, "loss": 0.18783003091812134, "step": 332 }, { "epoch": 0.24621072088724585, "grad_norm": 2.206679105758667, "learning_rate": 4.968483098357686e-06, "loss": 0.20760440826416016, "step": 333 }, { "epoch": 0.24695009242144178, "grad_norm": 3.9647910594940186, "learning_rate": 4.968182933834422e-06, "loss": 0.22456102073192596, "step": 334 }, { "epoch": 0.2476894639556377, "grad_norm": 1.4649746417999268, "learning_rate": 4.967881356883374e-06, "loss": 0.1662876158952713, "step": 335 }, { "epoch": 0.24842883548983363, "grad_norm": 1.8426079750061035, "learning_rate": 4.967578367696569e-06, "loss": 0.18827533721923828, "step": 336 }, { "epoch": 0.2491682070240296, "grad_norm": 2.9524762630462646, "learning_rate": 4.967273966466937e-06, "loss": 0.18616607785224915, "step": 337 }, { "epoch": 0.2499075785582255, "grad_norm": 3.8809752464294434, "learning_rate": 4.966968153388304e-06, "loss": 0.19369903206825256, "step": 338 }, { "epoch": 0.25064695009242144, "grad_norm": 2.9631190299987793, "learning_rate": 4.966660928655398e-06, "loss": 0.17391178011894226, "step": 339 }, { "epoch": 0.2513863216266174, "grad_norm": 1.863934874534607, "learning_rate": 4.966352292463843e-06, "loss": 0.17923042178153992, "step": 340 }, { "epoch": 0.2521256931608133, "grad_norm": 2.974100112915039, "learning_rate": 4.966042245010164e-06, "loss": 0.1969846487045288, "step": 341 }, { "epoch": 0.25286506469500925, "grad_norm": 2.7544326782226562, "learning_rate": 4.9657307864917825e-06, "loss": 0.17971652746200562, "step": 342 }, { "epoch": 0.2536044362292052, "grad_norm": 4.532392501831055, "learning_rate": 4.9654179171070205e-06, "loss": 0.19897158443927765, "step": 343 }, { "epoch": 0.2543438077634011, "grad_norm": 3.203714370727539, "learning_rate": 4.965103637055098e-06, "loss": 0.2183421552181244, "step": 344 }, { "epoch": 0.25508317929759705, "grad_norm": 4.025516510009766, "learning_rate": 4.9647879465361325e-06, "loss": 0.17418156564235687, "step": 345 }, { "epoch": 0.25582255083179295, "grad_norm": 3.079972982406616, "learning_rate": 4.96447084575114e-06, "loss": 0.19540292024612427, "step": 346 }, { "epoch": 0.2565619223659889, "grad_norm": 1.3878945112228394, "learning_rate": 4.964152334902033e-06, "loss": 0.18051466345787048, "step": 347 }, { "epoch": 0.25730129390018486, "grad_norm": 1.545939564704895, "learning_rate": 4.963832414191625e-06, "loss": 0.21832066774368286, "step": 348 }, { "epoch": 0.25804066543438076, "grad_norm": 2.546102523803711, "learning_rate": 4.963511083823625e-06, "loss": 0.16010339558124542, "step": 349 }, { "epoch": 0.2587800369685767, "grad_norm": 6.880647659301758, "learning_rate": 4.963188344002639e-06, "loss": 0.2270277738571167, "step": 350 }, { "epoch": 0.25951940850277266, "grad_norm": 3.9450864791870117, "learning_rate": 4.962864194934173e-06, "loss": 0.18076881766319275, "step": 351 }, { "epoch": 0.26025878003696856, "grad_norm": 9.001601219177246, "learning_rate": 4.962538636824628e-06, "loss": 0.21617156267166138, "step": 352 }, { "epoch": 0.2609981515711645, "grad_norm": 1.5321000814437866, "learning_rate": 4.962211669881305e-06, "loss": 0.16994032263755798, "step": 353 }, { "epoch": 0.26173752310536047, "grad_norm": 5.1643476486206055, "learning_rate": 4.961883294312397e-06, "loss": 0.18858814239501953, "step": 354 }, { "epoch": 0.26247689463955637, "grad_norm": 5.514026641845703, "learning_rate": 4.961553510327e-06, "loss": 0.20326799154281616, "step": 355 }, { "epoch": 0.2632162661737523, "grad_norm": 1.4135687351226807, "learning_rate": 4.9612223181351025e-06, "loss": 0.1999034583568573, "step": 356 }, { "epoch": 0.2639556377079482, "grad_norm": 3.671128749847412, "learning_rate": 4.96088971794759e-06, "loss": 0.19361166656017303, "step": 357 }, { "epoch": 0.2646950092421442, "grad_norm": 3.1278562545776367, "learning_rate": 4.9605557099762476e-06, "loss": 0.19335943460464478, "step": 358 }, { "epoch": 0.26543438077634013, "grad_norm": 3.2491109371185303, "learning_rate": 4.960220294433754e-06, "loss": 0.20872804522514343, "step": 359 }, { "epoch": 0.266173752310536, "grad_norm": 2.4866161346435547, "learning_rate": 4.959883471533685e-06, "loss": 0.1972534954547882, "step": 360 }, { "epoch": 0.266913123844732, "grad_norm": 2.4362189769744873, "learning_rate": 4.959545241490514e-06, "loss": 0.1855696588754654, "step": 361 }, { "epoch": 0.26765249537892793, "grad_norm": 1.4062579870224, "learning_rate": 4.959205604519608e-06, "loss": 0.1712094098329544, "step": 362 }, { "epoch": 0.26839186691312383, "grad_norm": 5.277360916137695, "learning_rate": 4.958864560837229e-06, "loss": 0.13842511177062988, "step": 363 }, { "epoch": 0.2691312384473198, "grad_norm": 4.513322830200195, "learning_rate": 4.958522110660538e-06, "loss": 0.1801295131444931, "step": 364 }, { "epoch": 0.2698706099815157, "grad_norm": 3.9188873767852783, "learning_rate": 4.95817825420759e-06, "loss": 0.20560815930366516, "step": 365 }, { "epoch": 0.27060998151571164, "grad_norm": 3.109624147415161, "learning_rate": 4.957832991697337e-06, "loss": 0.2042866349220276, "step": 366 }, { "epoch": 0.2713493530499076, "grad_norm": 3.8909871578216553, "learning_rate": 4.957486323349622e-06, "loss": 0.18892249464988708, "step": 367 }, { "epoch": 0.2720887245841035, "grad_norm": 1.7890655994415283, "learning_rate": 4.95713824938519e-06, "loss": 0.2239435613155365, "step": 368 }, { "epoch": 0.27282809611829945, "grad_norm": 2.4576032161712646, "learning_rate": 4.956788770025673e-06, "loss": 0.205025315284729, "step": 369 }, { "epoch": 0.2735674676524954, "grad_norm": 1.9039106369018555, "learning_rate": 4.956437885493603e-06, "loss": 0.18697002530097961, "step": 370 }, { "epoch": 0.2743068391866913, "grad_norm": 1.8320542573928833, "learning_rate": 4.956085596012407e-06, "loss": 0.18404042720794678, "step": 371 }, { "epoch": 0.27504621072088725, "grad_norm": 3.330578565597534, "learning_rate": 4.955731901806405e-06, "loss": 0.21784764528274536, "step": 372 }, { "epoch": 0.2757855822550832, "grad_norm": 5.931859493255615, "learning_rate": 4.955376803100811e-06, "loss": 0.20153406262397766, "step": 373 }, { "epoch": 0.2765249537892791, "grad_norm": 1.6254774332046509, "learning_rate": 4.955020300121735e-06, "loss": 0.19227978587150574, "step": 374 }, { "epoch": 0.27726432532347506, "grad_norm": 3.379150867462158, "learning_rate": 4.954662393096179e-06, "loss": 0.146018385887146, "step": 375 }, { "epoch": 0.27800369685767096, "grad_norm": 2.3373329639434814, "learning_rate": 4.9543030822520415e-06, "loss": 0.15471619367599487, "step": 376 }, { "epoch": 0.2787430683918669, "grad_norm": 2.792217493057251, "learning_rate": 4.953942367818113e-06, "loss": 0.18618158996105194, "step": 377 }, { "epoch": 0.27948243992606286, "grad_norm": 3.6047966480255127, "learning_rate": 4.953580250024079e-06, "loss": 0.220870703458786, "step": 378 }, { "epoch": 0.28022181146025876, "grad_norm": 3.3589179515838623, "learning_rate": 4.953216729100516e-06, "loss": 0.1973685324192047, "step": 379 }, { "epoch": 0.2809611829944547, "grad_norm": 3.0283615589141846, "learning_rate": 4.952851805278898e-06, "loss": 0.13302838802337646, "step": 380 }, { "epoch": 0.28170055452865067, "grad_norm": 3.4742496013641357, "learning_rate": 4.952485478791591e-06, "loss": 0.17402079701423645, "step": 381 }, { "epoch": 0.28243992606284657, "grad_norm": 2.698174476623535, "learning_rate": 4.9521177498718505e-06, "loss": 0.19752776622772217, "step": 382 }, { "epoch": 0.2831792975970425, "grad_norm": 3.239231586456299, "learning_rate": 4.951748618753829e-06, "loss": 0.17603503167629242, "step": 383 }, { "epoch": 0.2839186691312384, "grad_norm": 2.2759969234466553, "learning_rate": 4.951378085672572e-06, "loss": 0.16738224029541016, "step": 384 }, { "epoch": 0.2846580406654344, "grad_norm": 3.01668381690979, "learning_rate": 4.9510061508640145e-06, "loss": 0.19361895322799683, "step": 385 }, { "epoch": 0.28539741219963033, "grad_norm": 2.795328140258789, "learning_rate": 4.950632814564988e-06, "loss": 0.18197861313819885, "step": 386 }, { "epoch": 0.2861367837338262, "grad_norm": 2.949355125427246, "learning_rate": 4.950258077013212e-06, "loss": 0.1901976466178894, "step": 387 }, { "epoch": 0.2868761552680222, "grad_norm": 3.3545072078704834, "learning_rate": 4.949881938447303e-06, "loss": 0.2126903533935547, "step": 388 }, { "epoch": 0.28761552680221814, "grad_norm": 2.9221582412719727, "learning_rate": 4.949504399106765e-06, "loss": 0.19287967681884766, "step": 389 }, { "epoch": 0.28835489833641403, "grad_norm": 2.407536268234253, "learning_rate": 4.949125459231998e-06, "loss": 0.19694918394088745, "step": 390 }, { "epoch": 0.28909426987061, "grad_norm": 1.9233505725860596, "learning_rate": 4.948745119064291e-06, "loss": 0.1657274216413498, "step": 391 }, { "epoch": 0.28983364140480594, "grad_norm": 3.315389394760132, "learning_rate": 4.948363378845825e-06, "loss": 0.19870537519454956, "step": 392 }, { "epoch": 0.29057301293900184, "grad_norm": 1.751921534538269, "learning_rate": 4.947980238819674e-06, "loss": 0.18633981049060822, "step": 393 }, { "epoch": 0.2913123844731978, "grad_norm": 2.8910748958587646, "learning_rate": 4.947595699229803e-06, "loss": 0.17477154731750488, "step": 394 }, { "epoch": 0.2920517560073937, "grad_norm": 2.2841148376464844, "learning_rate": 4.947209760321066e-06, "loss": 0.19347456097602844, "step": 395 }, { "epoch": 0.29279112754158965, "grad_norm": 1.7701276540756226, "learning_rate": 4.94682242233921e-06, "loss": 0.16910573840141296, "step": 396 }, { "epoch": 0.2935304990757856, "grad_norm": 2.517714262008667, "learning_rate": 4.946433685530873e-06, "loss": 0.1877957433462143, "step": 397 }, { "epoch": 0.2942698706099815, "grad_norm": 1.6047612428665161, "learning_rate": 4.946043550143582e-06, "loss": 0.16637498140335083, "step": 398 }, { "epoch": 0.29500924214417745, "grad_norm": 3.306480646133423, "learning_rate": 4.945652016425757e-06, "loss": 0.19013050198554993, "step": 399 }, { "epoch": 0.2957486136783734, "grad_norm": 2.7501003742218018, "learning_rate": 4.945259084626706e-06, "loss": 0.18749837577342987, "step": 400 }, { "epoch": 0.2964879852125693, "grad_norm": 3.562255620956421, "learning_rate": 4.94486475499663e-06, "loss": 0.21013903617858887, "step": 401 }, { "epoch": 0.29722735674676526, "grad_norm": 1.5612611770629883, "learning_rate": 4.944469027786616e-06, "loss": 0.1757989227771759, "step": 402 }, { "epoch": 0.29796672828096116, "grad_norm": 1.4557161331176758, "learning_rate": 4.944071903248646e-06, "loss": 0.18925026059150696, "step": 403 }, { "epoch": 0.2987060998151571, "grad_norm": 1.2884657382965088, "learning_rate": 4.943673381635586e-06, "loss": 0.2027808427810669, "step": 404 }, { "epoch": 0.29944547134935307, "grad_norm": 2.1903083324432373, "learning_rate": 4.9432734632011955e-06, "loss": 0.1938842386007309, "step": 405 }, { "epoch": 0.30018484288354896, "grad_norm": 3.596497058868408, "learning_rate": 4.942872148200124e-06, "loss": 0.17932522296905518, "step": 406 }, { "epoch": 0.3009242144177449, "grad_norm": 1.6900584697723389, "learning_rate": 4.9424694368879076e-06, "loss": 0.17114651203155518, "step": 407 }, { "epoch": 0.30166358595194087, "grad_norm": 2.3050434589385986, "learning_rate": 4.942065329520973e-06, "loss": 0.20005150139331818, "step": 408 }, { "epoch": 0.30240295748613677, "grad_norm": 1.6271713972091675, "learning_rate": 4.941659826356636e-06, "loss": 0.21347790956497192, "step": 409 }, { "epoch": 0.3031423290203327, "grad_norm": 3.326235771179199, "learning_rate": 4.941252927653099e-06, "loss": 0.1901698112487793, "step": 410 }, { "epoch": 0.3038817005545287, "grad_norm": 1.6994389295578003, "learning_rate": 4.9408446336694565e-06, "loss": 0.20140168070793152, "step": 411 }, { "epoch": 0.3046210720887246, "grad_norm": 4.4089179039001465, "learning_rate": 4.94043494466569e-06, "loss": 0.20247846841812134, "step": 412 }, { "epoch": 0.30536044362292053, "grad_norm": 2.2888059616088867, "learning_rate": 4.940023860902667e-06, "loss": 0.1706225574016571, "step": 413 }, { "epoch": 0.30609981515711643, "grad_norm": 4.313504219055176, "learning_rate": 4.939611382642145e-06, "loss": 0.17735162377357483, "step": 414 }, { "epoch": 0.3068391866913124, "grad_norm": 1.262676477432251, "learning_rate": 4.939197510146772e-06, "loss": 0.18521173298358917, "step": 415 }, { "epoch": 0.30757855822550834, "grad_norm": 3.3992981910705566, "learning_rate": 4.9387822436800795e-06, "loss": 0.17279919981956482, "step": 416 }, { "epoch": 0.30831792975970423, "grad_norm": 4.168759346008301, "learning_rate": 4.938365583506489e-06, "loss": 0.19164159893989563, "step": 417 }, { "epoch": 0.3090573012939002, "grad_norm": 3.923553943634033, "learning_rate": 4.937947529891308e-06, "loss": 0.12282617390155792, "step": 418 }, { "epoch": 0.30979667282809614, "grad_norm": 3.8974716663360596, "learning_rate": 4.937528083100733e-06, "loss": 0.19642212986946106, "step": 419 }, { "epoch": 0.31053604436229204, "grad_norm": 2.260380983352661, "learning_rate": 4.937107243401847e-06, "loss": 0.20441368222236633, "step": 420 }, { "epoch": 0.311275415896488, "grad_norm": 2.0485382080078125, "learning_rate": 4.9366850110626165e-06, "loss": 0.1878972351551056, "step": 421 }, { "epoch": 0.3120147874306839, "grad_norm": 2.2595975399017334, "learning_rate": 4.936261386351902e-06, "loss": 0.19753524661064148, "step": 422 }, { "epoch": 0.31275415896487985, "grad_norm": 3.378420829772949, "learning_rate": 4.935836369539446e-06, "loss": 0.19001618027687073, "step": 423 }, { "epoch": 0.3134935304990758, "grad_norm": 2.842827558517456, "learning_rate": 4.935409960895875e-06, "loss": 0.21350106596946716, "step": 424 }, { "epoch": 0.3142329020332717, "grad_norm": 3.6550257205963135, "learning_rate": 4.934982160692707e-06, "loss": 0.18161126971244812, "step": 425 }, { "epoch": 0.31497227356746765, "grad_norm": 1.5455166101455688, "learning_rate": 4.934552969202343e-06, "loss": 0.17453379929065704, "step": 426 }, { "epoch": 0.3157116451016636, "grad_norm": 2.4665050506591797, "learning_rate": 4.934122386698071e-06, "loss": 0.18739360570907593, "step": 427 }, { "epoch": 0.3164510166358595, "grad_norm": 3.473987340927124, "learning_rate": 4.9336904134540644e-06, "loss": 0.20507827401161194, "step": 428 }, { "epoch": 0.31719038817005546, "grad_norm": 2.872896432876587, "learning_rate": 4.933257049745382e-06, "loss": 0.1690932661294937, "step": 429 }, { "epoch": 0.3179297597042514, "grad_norm": 4.161239147186279, "learning_rate": 4.932822295847967e-06, "loss": 0.21778708696365356, "step": 430 }, { "epoch": 0.3186691312384473, "grad_norm": 3.754774808883667, "learning_rate": 4.932386152038651e-06, "loss": 0.17505313456058502, "step": 431 }, { "epoch": 0.31940850277264327, "grad_norm": 2.8470230102539062, "learning_rate": 4.931948618595147e-06, "loss": 0.1954474151134491, "step": 432 }, { "epoch": 0.32014787430683916, "grad_norm": 2.725842237472534, "learning_rate": 4.931509695796055e-06, "loss": 0.16789510846138, "step": 433 }, { "epoch": 0.3208872458410351, "grad_norm": 3.7293572425842285, "learning_rate": 4.931069383920859e-06, "loss": 0.1872147023677826, "step": 434 }, { "epoch": 0.32162661737523107, "grad_norm": 3.126220226287842, "learning_rate": 4.930627683249928e-06, "loss": 0.18535178899765015, "step": 435 }, { "epoch": 0.32236598890942697, "grad_norm": 3.3590991497039795, "learning_rate": 4.930184594064513e-06, "loss": 0.1817149519920349, "step": 436 }, { "epoch": 0.3231053604436229, "grad_norm": 6.456763744354248, "learning_rate": 4.929740116646753e-06, "loss": 0.2003297656774521, "step": 437 }, { "epoch": 0.3238447319778189, "grad_norm": 2.0418624877929688, "learning_rate": 4.929294251279668e-06, "loss": 0.19090133905410767, "step": 438 }, { "epoch": 0.3245841035120148, "grad_norm": 3.6235079765319824, "learning_rate": 4.928846998247164e-06, "loss": 0.17229098081588745, "step": 439 }, { "epoch": 0.32532347504621073, "grad_norm": 3.027684211730957, "learning_rate": 4.928398357834028e-06, "loss": 0.16972586512565613, "step": 440 }, { "epoch": 0.32606284658040663, "grad_norm": 3.51938533782959, "learning_rate": 4.927948330325931e-06, "loss": 0.19878186285495758, "step": 441 }, { "epoch": 0.3268022181146026, "grad_norm": 3.184513807296753, "learning_rate": 4.927496916009429e-06, "loss": 0.19489428400993347, "step": 442 }, { "epoch": 0.32754158964879854, "grad_norm": 2.111138105392456, "learning_rate": 4.927044115171961e-06, "loss": 0.21555860340595245, "step": 443 }, { "epoch": 0.32828096118299444, "grad_norm": 2.7417244911193848, "learning_rate": 4.926589928101847e-06, "loss": 0.2102731168270111, "step": 444 }, { "epoch": 0.3290203327171904, "grad_norm": 3.44394850730896, "learning_rate": 4.92613435508829e-06, "loss": 0.20277199149131775, "step": 445 }, { "epoch": 0.32975970425138634, "grad_norm": 3.2249069213867188, "learning_rate": 4.925677396421378e-06, "loss": 0.19095033407211304, "step": 446 }, { "epoch": 0.33049907578558224, "grad_norm": 3.2694175243377686, "learning_rate": 4.925219052392076e-06, "loss": 0.1959449052810669, "step": 447 }, { "epoch": 0.3312384473197782, "grad_norm": 2.4543354511260986, "learning_rate": 4.924759323292237e-06, "loss": 0.1881168931722641, "step": 448 }, { "epoch": 0.33197781885397415, "grad_norm": 4.519211769104004, "learning_rate": 4.924298209414594e-06, "loss": 0.20120590925216675, "step": 449 }, { "epoch": 0.33271719038817005, "grad_norm": 5.576292991638184, "learning_rate": 4.92383571105276e-06, "loss": 0.19449052214622498, "step": 450 }, { "epoch": 0.333456561922366, "grad_norm": 3.489123821258545, "learning_rate": 4.923371828501231e-06, "loss": 0.17418664693832397, "step": 451 }, { "epoch": 0.3341959334565619, "grad_norm": 4.678319454193115, "learning_rate": 4.922906562055383e-06, "loss": 0.191522479057312, "step": 452 }, { "epoch": 0.33493530499075785, "grad_norm": 4.944378852844238, "learning_rate": 4.922439912011476e-06, "loss": 0.17443710565567017, "step": 453 }, { "epoch": 0.3356746765249538, "grad_norm": 3.5776729583740234, "learning_rate": 4.92197187866665e-06, "loss": 0.20157679915428162, "step": 454 }, { "epoch": 0.3364140480591497, "grad_norm": 4.248790264129639, "learning_rate": 4.921502462318923e-06, "loss": 0.1750725358724594, "step": 455 }, { "epoch": 0.33715341959334566, "grad_norm": 3.941281318664551, "learning_rate": 4.9210316632671976e-06, "loss": 0.1962079256772995, "step": 456 }, { "epoch": 0.3378927911275416, "grad_norm": 4.940640926361084, "learning_rate": 4.920559481811253e-06, "loss": 0.21210484206676483, "step": 457 }, { "epoch": 0.3386321626617375, "grad_norm": 4.511053562164307, "learning_rate": 4.920085918251754e-06, "loss": 0.16779550909996033, "step": 458 }, { "epoch": 0.33937153419593347, "grad_norm": 5.283036231994629, "learning_rate": 4.91961097289024e-06, "loss": 0.18102189898490906, "step": 459 }, { "epoch": 0.34011090573012936, "grad_norm": 3.537224054336548, "learning_rate": 4.919134646029132e-06, "loss": 0.2005159556865692, "step": 460 }, { "epoch": 0.3408502772643253, "grad_norm": 1.9376684427261353, "learning_rate": 4.918656937971731e-06, "loss": 0.20104727149009705, "step": 461 }, { "epoch": 0.3415896487985213, "grad_norm": 4.289368152618408, "learning_rate": 4.9181778490222196e-06, "loss": 0.18661120533943176, "step": 462 }, { "epoch": 0.34232902033271717, "grad_norm": 4.873020172119141, "learning_rate": 4.917697379485655e-06, "loss": 0.20539036393165588, "step": 463 }, { "epoch": 0.3430683918669131, "grad_norm": 2.0268120765686035, "learning_rate": 4.9172155296679795e-06, "loss": 0.2115393579006195, "step": 464 }, { "epoch": 0.3438077634011091, "grad_norm": 5.200189113616943, "learning_rate": 4.916732299876007e-06, "loss": 0.18942251801490784, "step": 465 }, { "epoch": 0.344547134935305, "grad_norm": 6.176848411560059, "learning_rate": 4.916247690417437e-06, "loss": 0.1909760683774948, "step": 466 }, { "epoch": 0.34528650646950093, "grad_norm": 1.4998761415481567, "learning_rate": 4.915761701600843e-06, "loss": 0.192938432097435, "step": 467 }, { "epoch": 0.3460258780036969, "grad_norm": 3.8685660362243652, "learning_rate": 4.915274333735679e-06, "loss": 0.19205079972743988, "step": 468 }, { "epoch": 0.3467652495378928, "grad_norm": 2.626343011856079, "learning_rate": 4.9147855871322755e-06, "loss": 0.17065462470054626, "step": 469 }, { "epoch": 0.34750462107208874, "grad_norm": 5.095709323883057, "learning_rate": 4.914295462101843e-06, "loss": 0.1737920343875885, "step": 470 }, { "epoch": 0.34824399260628464, "grad_norm": 2.335280656814575, "learning_rate": 4.913803958956469e-06, "loss": 0.16478238999843597, "step": 471 }, { "epoch": 0.3489833641404806, "grad_norm": 6.868231296539307, "learning_rate": 4.913311078009116e-06, "loss": 0.1932230293750763, "step": 472 }, { "epoch": 0.34972273567467654, "grad_norm": 2.8917229175567627, "learning_rate": 4.912816819573628e-06, "loss": 0.14274708926677704, "step": 473 }, { "epoch": 0.35046210720887244, "grad_norm": 2.769686222076416, "learning_rate": 4.912321183964724e-06, "loss": 0.1786687672138214, "step": 474 }, { "epoch": 0.3512014787430684, "grad_norm": 2.4529917240142822, "learning_rate": 4.911824171497998e-06, "loss": 0.193831205368042, "step": 475 }, { "epoch": 0.35194085027726435, "grad_norm": 4.159771919250488, "learning_rate": 4.911325782489923e-06, "loss": 0.1755894273519516, "step": 476 }, { "epoch": 0.35268022181146025, "grad_norm": 2.9535415172576904, "learning_rate": 4.91082601725785e-06, "loss": 0.14007362723350525, "step": 477 }, { "epoch": 0.3534195933456562, "grad_norm": 2.038978338241577, "learning_rate": 4.910324876120004e-06, "loss": 0.19608092308044434, "step": 478 }, { "epoch": 0.3541589648798521, "grad_norm": 4.6805338859558105, "learning_rate": 4.909822359395484e-06, "loss": 0.1898474097251892, "step": 479 }, { "epoch": 0.35489833641404805, "grad_norm": 5.300136566162109, "learning_rate": 4.90931846740427e-06, "loss": 0.1947241872549057, "step": 480 }, { "epoch": 0.355637707948244, "grad_norm": 2.227041006088257, "learning_rate": 4.908813200467216e-06, "loss": 0.20272713899612427, "step": 481 }, { "epoch": 0.3563770794824399, "grad_norm": 2.729194164276123, "learning_rate": 4.908306558906048e-06, "loss": 0.18599915504455566, "step": 482 }, { "epoch": 0.35711645101663586, "grad_norm": 3.1959433555603027, "learning_rate": 4.907798543043372e-06, "loss": 0.19570805132389069, "step": 483 }, { "epoch": 0.3578558225508318, "grad_norm": 2.2035648822784424, "learning_rate": 4.907289153202665e-06, "loss": 0.18361464142799377, "step": 484 }, { "epoch": 0.3585951940850277, "grad_norm": 1.674584150314331, "learning_rate": 4.9067783897082835e-06, "loss": 0.16498896479606628, "step": 485 }, { "epoch": 0.35933456561922367, "grad_norm": 2.7351181507110596, "learning_rate": 4.906266252885454e-06, "loss": 0.17017021775245667, "step": 486 }, { "epoch": 0.3600739371534196, "grad_norm": 3.016345262527466, "learning_rate": 4.9057527430602815e-06, "loss": 0.16729816794395447, "step": 487 }, { "epoch": 0.3608133086876155, "grad_norm": 2.94889760017395, "learning_rate": 4.905237860559741e-06, "loss": 0.18914532661437988, "step": 488 }, { "epoch": 0.3615526802218115, "grad_norm": 1.787521481513977, "learning_rate": 4.9047216057116866e-06, "loss": 0.15992596745491028, "step": 489 }, { "epoch": 0.36229205175600737, "grad_norm": 3.4122729301452637, "learning_rate": 4.904203978844841e-06, "loss": 0.1670689582824707, "step": 490 }, { "epoch": 0.3630314232902033, "grad_norm": 5.430705547332764, "learning_rate": 4.903684980288805e-06, "loss": 0.20110470056533813, "step": 491 }, { "epoch": 0.3637707948243993, "grad_norm": 2.498366594314575, "learning_rate": 4.903164610374048e-06, "loss": 0.17977997660636902, "step": 492 }, { "epoch": 0.3645101663585952, "grad_norm": 2.661695957183838, "learning_rate": 4.902642869431919e-06, "loss": 0.1749214231967926, "step": 493 }, { "epoch": 0.36524953789279113, "grad_norm": 1.9155484437942505, "learning_rate": 4.902119757794635e-06, "loss": 0.15549886226654053, "step": 494 }, { "epoch": 0.3659889094269871, "grad_norm": 2.7752366065979004, "learning_rate": 4.901595275795287e-06, "loss": 0.1847965568304062, "step": 495 }, { "epoch": 0.366728280961183, "grad_norm": 3.22649884223938, "learning_rate": 4.90106942376784e-06, "loss": 0.17237883806228638, "step": 496 }, { "epoch": 0.36746765249537894, "grad_norm": 3.672759771347046, "learning_rate": 4.900542202047128e-06, "loss": 0.14139653742313385, "step": 497 }, { "epoch": 0.36820702402957484, "grad_norm": 1.7752732038497925, "learning_rate": 4.900013610968861e-06, "loss": 0.17842869460582733, "step": 498 }, { "epoch": 0.3689463955637708, "grad_norm": 3.662363290786743, "learning_rate": 4.899483650869621e-06, "loss": 0.16319534182548523, "step": 499 }, { "epoch": 0.36968576709796674, "grad_norm": 3.29624342918396, "learning_rate": 4.898952322086857e-06, "loss": 0.1359483003616333, "step": 500 }, { "epoch": 0.37042513863216264, "grad_norm": 4.195192813873291, "learning_rate": 4.898419624958893e-06, "loss": 0.18541277945041656, "step": 501 }, { "epoch": 0.3711645101663586, "grad_norm": 4.873483180999756, "learning_rate": 4.897885559824925e-06, "loss": 0.17912454903125763, "step": 502 }, { "epoch": 0.37190388170055455, "grad_norm": 2.618875026702881, "learning_rate": 4.897350127025019e-06, "loss": 0.14924466609954834, "step": 503 }, { "epoch": 0.37264325323475045, "grad_norm": 3.4020743370056152, "learning_rate": 4.89681332690011e-06, "loss": 0.15018168091773987, "step": 504 }, { "epoch": 0.3733826247689464, "grad_norm": 3.1516058444976807, "learning_rate": 4.896275159792007e-06, "loss": 0.17922906577587128, "step": 505 }, { "epoch": 0.37412199630314236, "grad_norm": 1.2430545091629028, "learning_rate": 4.895735626043389e-06, "loss": 0.18154628574848175, "step": 506 }, { "epoch": 0.37486136783733826, "grad_norm": 1.721394658088684, "learning_rate": 4.895194725997801e-06, "loss": 0.18758463859558105, "step": 507 }, { "epoch": 0.3756007393715342, "grad_norm": 1.5942250490188599, "learning_rate": 4.8946524599996636e-06, "loss": 0.16313239932060242, "step": 508 }, { "epoch": 0.3763401109057301, "grad_norm": 2.368898868560791, "learning_rate": 4.894108828394264e-06, "loss": 0.2073643058538437, "step": 509 }, { "epoch": 0.37707948243992606, "grad_norm": 3.522796154022217, "learning_rate": 4.893563831527758e-06, "loss": 0.14792346954345703, "step": 510 }, { "epoch": 0.377818853974122, "grad_norm": 1.963173270225525, "learning_rate": 4.893017469747175e-06, "loss": 0.16947436332702637, "step": 511 }, { "epoch": 0.3785582255083179, "grad_norm": 2.4875857830047607, "learning_rate": 4.892469743400409e-06, "loss": 0.19672825932502747, "step": 512 }, { "epoch": 0.37929759704251387, "grad_norm": 2.111938238143921, "learning_rate": 4.8919206528362245e-06, "loss": 0.20601886510849, "step": 513 }, { "epoch": 0.3800369685767098, "grad_norm": 3.870346784591675, "learning_rate": 4.891370198404256e-06, "loss": 0.19080619513988495, "step": 514 }, { "epoch": 0.3807763401109057, "grad_norm": 2.388782024383545, "learning_rate": 4.890818380455005e-06, "loss": 0.18508875370025635, "step": 515 }, { "epoch": 0.3815157116451017, "grad_norm": 4.513940811157227, "learning_rate": 4.890265199339841e-06, "loss": 0.19175559282302856, "step": 516 }, { "epoch": 0.3822550831792976, "grad_norm": 1.8588372468948364, "learning_rate": 4.889710655411002e-06, "loss": 0.17021694779396057, "step": 517 }, { "epoch": 0.3829944547134935, "grad_norm": 2.3641581535339355, "learning_rate": 4.889154749021593e-06, "loss": 0.1683879941701889, "step": 518 }, { "epoch": 0.3837338262476895, "grad_norm": 1.6540082693099976, "learning_rate": 4.888597480525591e-06, "loss": 0.18171462416648865, "step": 519 }, { "epoch": 0.3844731977818854, "grad_norm": 2.263054609298706, "learning_rate": 4.888038850277833e-06, "loss": 0.16475945711135864, "step": 520 }, { "epoch": 0.38521256931608133, "grad_norm": 5.135030746459961, "learning_rate": 4.887478858634027e-06, "loss": 0.2104199230670929, "step": 521 }, { "epoch": 0.3859519408502773, "grad_norm": 3.631883144378662, "learning_rate": 4.8869175059507505e-06, "loss": 0.18258056044578552, "step": 522 }, { "epoch": 0.3866913123844732, "grad_norm": 2.292091131210327, "learning_rate": 4.886354792585441e-06, "loss": 0.17052476108074188, "step": 523 }, { "epoch": 0.38743068391866914, "grad_norm": 6.142096519470215, "learning_rate": 4.8857907188964085e-06, "loss": 0.1181279867887497, "step": 524 }, { "epoch": 0.38817005545286504, "grad_norm": 3.5677828788757324, "learning_rate": 4.885225285242827e-06, "loss": 0.1823442131280899, "step": 525 }, { "epoch": 0.388909426987061, "grad_norm": 4.943482398986816, "learning_rate": 4.884658491984735e-06, "loss": 0.17542597651481628, "step": 526 }, { "epoch": 0.38964879852125694, "grad_norm": 4.233567714691162, "learning_rate": 4.884090339483039e-06, "loss": 0.1837269812822342, "step": 527 }, { "epoch": 0.39038817005545284, "grad_norm": 3.1174087524414062, "learning_rate": 4.883520828099511e-06, "loss": 0.20580413937568665, "step": 528 }, { "epoch": 0.3911275415896488, "grad_norm": 4.667779445648193, "learning_rate": 4.882949958196784e-06, "loss": 0.20776045322418213, "step": 529 }, { "epoch": 0.39186691312384475, "grad_norm": 3.6076111793518066, "learning_rate": 4.882377730138362e-06, "loss": 0.16494110226631165, "step": 530 }, { "epoch": 0.39260628465804065, "grad_norm": 2.9697000980377197, "learning_rate": 4.8818041442886114e-06, "loss": 0.18925270438194275, "step": 531 }, { "epoch": 0.3933456561922366, "grad_norm": 4.898978233337402, "learning_rate": 4.881229201012762e-06, "loss": 0.19672900438308716, "step": 532 }, { "epoch": 0.39408502772643256, "grad_norm": 5.18672513961792, "learning_rate": 4.880652900676909e-06, "loss": 0.18659251928329468, "step": 533 }, { "epoch": 0.39482439926062846, "grad_norm": 4.211703777313232, "learning_rate": 4.880075243648012e-06, "loss": 0.17279843986034393, "step": 534 }, { "epoch": 0.3955637707948244, "grad_norm": 3.9313645362854004, "learning_rate": 4.879496230293893e-06, "loss": 0.21086424589157104, "step": 535 }, { "epoch": 0.3963031423290203, "grad_norm": 3.0819432735443115, "learning_rate": 4.8789158609832404e-06, "loss": 0.1821284294128418, "step": 536 }, { "epoch": 0.39704251386321626, "grad_norm": 2.8224122524261475, "learning_rate": 4.8783341360856025e-06, "loss": 0.1821846216917038, "step": 537 }, { "epoch": 0.3977818853974122, "grad_norm": 2.411010265350342, "learning_rate": 4.877751055971393e-06, "loss": 0.17668360471725464, "step": 538 }, { "epoch": 0.3985212569316081, "grad_norm": 6.427002429962158, "learning_rate": 4.877166621011889e-06, "loss": 0.2088906317949295, "step": 539 }, { "epoch": 0.39926062846580407, "grad_norm": 4.623311519622803, "learning_rate": 4.876580831579229e-06, "loss": 0.18372967839241028, "step": 540 }, { "epoch": 0.4, "grad_norm": 3.702808141708374, "learning_rate": 4.875993688046414e-06, "loss": 0.19196008145809174, "step": 541 }, { "epoch": 0.4007393715341959, "grad_norm": 4.659071922302246, "learning_rate": 4.875405190787308e-06, "loss": 0.20254270732402802, "step": 542 }, { "epoch": 0.4014787430683919, "grad_norm": 6.3780293464660645, "learning_rate": 4.874815340176637e-06, "loss": 0.16161353886127472, "step": 543 }, { "epoch": 0.4022181146025878, "grad_norm": 2.799818992614746, "learning_rate": 4.874224136589988e-06, "loss": 0.14977577328681946, "step": 544 }, { "epoch": 0.4029574861367837, "grad_norm": 2.2714004516601562, "learning_rate": 4.8736315804038086e-06, "loss": 0.16189119219779968, "step": 545 }, { "epoch": 0.4036968576709797, "grad_norm": 3.7751646041870117, "learning_rate": 4.87303767199541e-06, "loss": 0.17450234293937683, "step": 546 }, { "epoch": 0.4044362292051756, "grad_norm": 4.89046049118042, "learning_rate": 4.872442411742964e-06, "loss": 0.2071964144706726, "step": 547 }, { "epoch": 0.40517560073937153, "grad_norm": 3.923156499862671, "learning_rate": 4.8718458000255006e-06, "loss": 0.1691572517156601, "step": 548 }, { "epoch": 0.4059149722735675, "grad_norm": 3.6698544025421143, "learning_rate": 4.871247837222915e-06, "loss": 0.1895117312669754, "step": 549 }, { "epoch": 0.4066543438077634, "grad_norm": 2.4097533226013184, "learning_rate": 4.8706485237159575e-06, "loss": 0.17249655723571777, "step": 550 }, { "epoch": 0.40739371534195934, "grad_norm": 2.937636613845825, "learning_rate": 4.870047859886241e-06, "loss": 0.17791570723056793, "step": 551 }, { "epoch": 0.4081330868761553, "grad_norm": 3.479900360107422, "learning_rate": 4.86944584611624e-06, "loss": 0.13473932445049286, "step": 552 }, { "epoch": 0.4088724584103512, "grad_norm": 2.7332003116607666, "learning_rate": 4.868842482789285e-06, "loss": 0.16670075058937073, "step": 553 }, { "epoch": 0.40961182994454715, "grad_norm": 3.6123313903808594, "learning_rate": 4.868237770289569e-06, "loss": 0.18228423595428467, "step": 554 }, { "epoch": 0.41035120147874304, "grad_norm": 4.321937561035156, "learning_rate": 4.867631709002142e-06, "loss": 0.1945151388645172, "step": 555 }, { "epoch": 0.411090573012939, "grad_norm": 4.009243011474609, "learning_rate": 4.867024299312912e-06, "loss": 0.1898512840270996, "step": 556 }, { "epoch": 0.41182994454713495, "grad_norm": 3.1275603771209717, "learning_rate": 4.866415541608649e-06, "loss": 0.17757806181907654, "step": 557 }, { "epoch": 0.41256931608133085, "grad_norm": 1.9743423461914062, "learning_rate": 4.865805436276979e-06, "loss": 0.1627628207206726, "step": 558 }, { "epoch": 0.4133086876155268, "grad_norm": 2.1861090660095215, "learning_rate": 4.865193983706386e-06, "loss": 0.1676299124956131, "step": 559 }, { "epoch": 0.41404805914972276, "grad_norm": 1.4164756536483765, "learning_rate": 4.864581184286213e-06, "loss": 0.19306384027004242, "step": 560 }, { "epoch": 0.41478743068391866, "grad_norm": 5.857456207275391, "learning_rate": 4.863967038406659e-06, "loss": 0.15408867597579956, "step": 561 }, { "epoch": 0.4155268022181146, "grad_norm": 2.7450852394104004, "learning_rate": 4.863351546458783e-06, "loss": 0.18276607990264893, "step": 562 }, { "epoch": 0.4162661737523105, "grad_norm": 2.5199315547943115, "learning_rate": 4.862734708834497e-06, "loss": 0.15018776059150696, "step": 563 }, { "epoch": 0.41700554528650646, "grad_norm": 5.711653709411621, "learning_rate": 4.8621165259265745e-06, "loss": 0.1764090657234192, "step": 564 }, { "epoch": 0.4177449168207024, "grad_norm": 5.6009063720703125, "learning_rate": 4.861496998128643e-06, "loss": 0.2030719518661499, "step": 565 }, { "epoch": 0.4184842883548983, "grad_norm": 2.271238088607788, "learning_rate": 4.8608761258351855e-06, "loss": 0.17857134342193604, "step": 566 }, { "epoch": 0.41922365988909427, "grad_norm": 5.852835655212402, "learning_rate": 4.860253909441543e-06, "loss": 0.18937352299690247, "step": 567 }, { "epoch": 0.4199630314232902, "grad_norm": 9.181206703186035, "learning_rate": 4.859630349343913e-06, "loss": 0.15087956190109253, "step": 568 }, { "epoch": 0.4207024029574861, "grad_norm": 2.3028366565704346, "learning_rate": 4.8590054459393466e-06, "loss": 0.18140892684459686, "step": 569 }, { "epoch": 0.4214417744916821, "grad_norm": 4.69753360748291, "learning_rate": 4.85837919962575e-06, "loss": 0.17112240195274353, "step": 570 }, { "epoch": 0.42218114602587803, "grad_norm": 4.830615043640137, "learning_rate": 4.857751610801887e-06, "loss": 0.17245914041996002, "step": 571 }, { "epoch": 0.4229205175600739, "grad_norm": 6.464344024658203, "learning_rate": 4.857122679867373e-06, "loss": 0.1981751024723053, "step": 572 }, { "epoch": 0.4236598890942699, "grad_norm": 1.2698463201522827, "learning_rate": 4.856492407222681e-06, "loss": 0.1615900993347168, "step": 573 }, { "epoch": 0.4243992606284658, "grad_norm": 4.632750511169434, "learning_rate": 4.855860793269136e-06, "loss": 0.189609557390213, "step": 574 }, { "epoch": 0.42513863216266173, "grad_norm": 5.69913387298584, "learning_rate": 4.8552278384089205e-06, "loss": 0.17114976048469543, "step": 575 }, { "epoch": 0.4258780036968577, "grad_norm": 2.123908519744873, "learning_rate": 4.854593543045066e-06, "loss": 0.18548008799552917, "step": 576 }, { "epoch": 0.4266173752310536, "grad_norm": 3.17171311378479, "learning_rate": 4.85395790758146e-06, "loss": 0.15605610609054565, "step": 577 }, { "epoch": 0.42735674676524954, "grad_norm": 1.3894273042678833, "learning_rate": 4.853320932422845e-06, "loss": 0.1902095526456833, "step": 578 }, { "epoch": 0.4280961182994455, "grad_norm": 1.8296966552734375, "learning_rate": 4.852682617974813e-06, "loss": 0.1898842453956604, "step": 579 }, { "epoch": 0.4288354898336414, "grad_norm": 2.6010353565216064, "learning_rate": 4.852042964643813e-06, "loss": 0.18566036224365234, "step": 580 }, { "epoch": 0.42957486136783735, "grad_norm": 2.5525448322296143, "learning_rate": 4.851401972837142e-06, "loss": 0.18922948837280273, "step": 581 }, { "epoch": 0.43031423290203324, "grad_norm": 2.1477468013763428, "learning_rate": 4.850759642962953e-06, "loss": 0.17380322515964508, "step": 582 }, { "epoch": 0.4310536044362292, "grad_norm": 3.2113258838653564, "learning_rate": 4.850115975430248e-06, "loss": 0.18408995866775513, "step": 583 }, { "epoch": 0.43179297597042515, "grad_norm": 3.3463635444641113, "learning_rate": 4.849470970648883e-06, "loss": 0.15865588188171387, "step": 584 }, { "epoch": 0.43253234750462105, "grad_norm": 2.5083394050598145, "learning_rate": 4.848824629029566e-06, "loss": 0.1569748818874359, "step": 585 }, { "epoch": 0.433271719038817, "grad_norm": 3.0956215858459473, "learning_rate": 4.848176950983853e-06, "loss": 0.1608296036720276, "step": 586 }, { "epoch": 0.43401109057301296, "grad_norm": 3.71537709236145, "learning_rate": 4.847527936924154e-06, "loss": 0.17386090755462646, "step": 587 }, { "epoch": 0.43475046210720886, "grad_norm": 1.6988158226013184, "learning_rate": 4.846877587263728e-06, "loss": 0.1663358509540558, "step": 588 }, { "epoch": 0.4354898336414048, "grad_norm": 2.1743204593658447, "learning_rate": 4.846225902416686e-06, "loss": 0.16862967610359192, "step": 589 }, { "epoch": 0.43622920517560076, "grad_norm": 1.8015776872634888, "learning_rate": 4.845572882797987e-06, "loss": 0.18719495832920074, "step": 590 }, { "epoch": 0.43696857670979666, "grad_norm": 5.341803073883057, "learning_rate": 4.8449185288234425e-06, "loss": 0.15915217995643616, "step": 591 }, { "epoch": 0.4377079482439926, "grad_norm": 2.735485792160034, "learning_rate": 4.844262840909712e-06, "loss": 0.17008912563323975, "step": 592 }, { "epoch": 0.4384473197781885, "grad_norm": 3.155380964279175, "learning_rate": 4.8436058194743034e-06, "loss": 0.1866861879825592, "step": 593 }, { "epoch": 0.43918669131238447, "grad_norm": 3.1706371307373047, "learning_rate": 4.842947464935577e-06, "loss": 0.18546411395072937, "step": 594 }, { "epoch": 0.4399260628465804, "grad_norm": 4.940435409545898, "learning_rate": 4.842287777712739e-06, "loss": 0.17565804719924927, "step": 595 }, { "epoch": 0.4406654343807763, "grad_norm": 1.4294342994689941, "learning_rate": 4.841626758225845e-06, "loss": 0.15728192031383514, "step": 596 }, { "epoch": 0.4414048059149723, "grad_norm": 2.8830206394195557, "learning_rate": 4.8409644068958e-06, "loss": 0.17907655239105225, "step": 597 }, { "epoch": 0.44214417744916823, "grad_norm": 5.274194240570068, "learning_rate": 4.8403007241443545e-06, "loss": 0.186065673828125, "step": 598 }, { "epoch": 0.44288354898336413, "grad_norm": 3.0492947101593018, "learning_rate": 4.8396357103941096e-06, "loss": 0.14108574390411377, "step": 599 }, { "epoch": 0.4436229205175601, "grad_norm": 3.5136914253234863, "learning_rate": 4.838969366068513e-06, "loss": 0.19868043065071106, "step": 600 }, { "epoch": 0.444362292051756, "grad_norm": 4.653659820556641, "learning_rate": 4.838301691591859e-06, "loss": 0.2041354477405548, "step": 601 }, { "epoch": 0.44510166358595193, "grad_norm": 5.009173393249512, "learning_rate": 4.8376326873892885e-06, "loss": 0.1561928689479828, "step": 602 }, { "epoch": 0.4458410351201479, "grad_norm": 4.728450298309326, "learning_rate": 4.836962353886792e-06, "loss": 0.17004457116127014, "step": 603 }, { "epoch": 0.4465804066543438, "grad_norm": 4.365522861480713, "learning_rate": 4.836290691511202e-06, "loss": 0.18689453601837158, "step": 604 }, { "epoch": 0.44731977818853974, "grad_norm": 2.3493268489837646, "learning_rate": 4.835617700690201e-06, "loss": 0.156974196434021, "step": 605 }, { "epoch": 0.4480591497227357, "grad_norm": 2.0357296466827393, "learning_rate": 4.834943381852315e-06, "loss": 0.14403943717479706, "step": 606 }, { "epoch": 0.4487985212569316, "grad_norm": 4.1241068840026855, "learning_rate": 4.8342677354269164e-06, "loss": 0.18279887735843658, "step": 607 }, { "epoch": 0.44953789279112755, "grad_norm": 4.2877655029296875, "learning_rate": 4.833590761844224e-06, "loss": 0.1579398214817047, "step": 608 }, { "epoch": 0.4502772643253235, "grad_norm": 5.036591053009033, "learning_rate": 4.832912461535299e-06, "loss": 0.15637795627117157, "step": 609 }, { "epoch": 0.4510166358595194, "grad_norm": 3.262542247772217, "learning_rate": 4.832232834932051e-06, "loss": 0.19110111892223358, "step": 610 }, { "epoch": 0.45175600739371535, "grad_norm": 4.060462474822998, "learning_rate": 4.83155188246723e-06, "loss": 0.1499650776386261, "step": 611 }, { "epoch": 0.45249537892791125, "grad_norm": 5.050755977630615, "learning_rate": 4.830869604574434e-06, "loss": 0.18508405983448029, "step": 612 }, { "epoch": 0.4532347504621072, "grad_norm": 3.669936180114746, "learning_rate": 4.830186001688104e-06, "loss": 0.16692817211151123, "step": 613 }, { "epoch": 0.45397412199630316, "grad_norm": 2.127078056335449, "learning_rate": 4.8295010742435215e-06, "loss": 0.15445223450660706, "step": 614 }, { "epoch": 0.45471349353049906, "grad_norm": 3.4250833988189697, "learning_rate": 4.828814822676817e-06, "loss": 0.15857501327991486, "step": 615 }, { "epoch": 0.455452865064695, "grad_norm": 4.682999134063721, "learning_rate": 4.828127247424959e-06, "loss": 0.17373526096343994, "step": 616 }, { "epoch": 0.45619223659889097, "grad_norm": 2.1424851417541504, "learning_rate": 4.827438348925762e-06, "loss": 0.15541434288024902, "step": 617 }, { "epoch": 0.45693160813308686, "grad_norm": 2.0315818786621094, "learning_rate": 4.826748127617881e-06, "loss": 0.13082262873649597, "step": 618 }, { "epoch": 0.4576709796672828, "grad_norm": 2.5078721046447754, "learning_rate": 4.826056583940816e-06, "loss": 0.17770272493362427, "step": 619 }, { "epoch": 0.4584103512014787, "grad_norm": 2.0080971717834473, "learning_rate": 4.825363718334905e-06, "loss": 0.1381104290485382, "step": 620 }, { "epoch": 0.45914972273567467, "grad_norm": 3.8514301776885986, "learning_rate": 4.824669531241333e-06, "loss": 0.1652662456035614, "step": 621 }, { "epoch": 0.4598890942698706, "grad_norm": 2.6427738666534424, "learning_rate": 4.823974023102122e-06, "loss": 0.17247578501701355, "step": 622 }, { "epoch": 0.4606284658040665, "grad_norm": 3.852588176727295, "learning_rate": 4.8232771943601365e-06, "loss": 0.18326795101165771, "step": 623 }, { "epoch": 0.4613678373382625, "grad_norm": 2.1233651638031006, "learning_rate": 4.822579045459083e-06, "loss": 0.15889766812324524, "step": 624 }, { "epoch": 0.46210720887245843, "grad_norm": 1.2998524904251099, "learning_rate": 4.821879576843507e-06, "loss": 0.1389409452676773, "step": 625 }, { "epoch": 0.46284658040665433, "grad_norm": 1.8767200708389282, "learning_rate": 4.821178788958795e-06, "loss": 0.1443537473678589, "step": 626 }, { "epoch": 0.4635859519408503, "grad_norm": 4.217147350311279, "learning_rate": 4.820476682251174e-06, "loss": 0.1464684009552002, "step": 627 }, { "epoch": 0.46432532347504624, "grad_norm": 2.6972694396972656, "learning_rate": 4.8197732571677115e-06, "loss": 0.1492946445941925, "step": 628 }, { "epoch": 0.46506469500924214, "grad_norm": 2.5312769412994385, "learning_rate": 4.819068514156311e-06, "loss": 0.14108282327651978, "step": 629 }, { "epoch": 0.4658040665434381, "grad_norm": 4.668809413909912, "learning_rate": 4.818362453665719e-06, "loss": 0.20964300632476807, "step": 630 }, { "epoch": 0.466543438077634, "grad_norm": 2.4359426498413086, "learning_rate": 4.817655076145519e-06, "loss": 0.17611706256866455, "step": 631 }, { "epoch": 0.46728280961182994, "grad_norm": 2.197355031967163, "learning_rate": 4.816946382046135e-06, "loss": 0.14925548434257507, "step": 632 }, { "epoch": 0.4680221811460259, "grad_norm": 4.263963222503662, "learning_rate": 4.816236371818825e-06, "loss": 0.18305569887161255, "step": 633 }, { "epoch": 0.4687615526802218, "grad_norm": 4.3361077308654785, "learning_rate": 4.815525045915691e-06, "loss": 0.17148344218730927, "step": 634 }, { "epoch": 0.46950092421441775, "grad_norm": 2.3554129600524902, "learning_rate": 4.814812404789667e-06, "loss": 0.1490202248096466, "step": 635 }, { "epoch": 0.4702402957486137, "grad_norm": 5.680135250091553, "learning_rate": 4.81409844889453e-06, "loss": 0.1951674520969391, "step": 636 }, { "epoch": 0.4709796672828096, "grad_norm": 4.607978343963623, "learning_rate": 4.81338317868489e-06, "loss": 0.17289328575134277, "step": 637 }, { "epoch": 0.47171903881700555, "grad_norm": 2.492432117462158, "learning_rate": 4.812666594616195e-06, "loss": 0.17770680785179138, "step": 638 }, { "epoch": 0.47245841035120145, "grad_norm": 7.372415065765381, "learning_rate": 4.811948697144729e-06, "loss": 0.171659454703331, "step": 639 }, { "epoch": 0.4731977818853974, "grad_norm": 4.640161514282227, "learning_rate": 4.811229486727614e-06, "loss": 0.18462872505187988, "step": 640 }, { "epoch": 0.47393715341959336, "grad_norm": 2.0780885219573975, "learning_rate": 4.810508963822808e-06, "loss": 0.15288013219833374, "step": 641 }, { "epoch": 0.47467652495378926, "grad_norm": 3.73551082611084, "learning_rate": 4.809787128889103e-06, "loss": 0.16165347397327423, "step": 642 }, { "epoch": 0.4754158964879852, "grad_norm": 6.247860908508301, "learning_rate": 4.809063982386127e-06, "loss": 0.2100682258605957, "step": 643 }, { "epoch": 0.47615526802218117, "grad_norm": 3.9838383197784424, "learning_rate": 4.808339524774345e-06, "loss": 0.16113927960395813, "step": 644 }, { "epoch": 0.47689463955637706, "grad_norm": 1.9657275676727295, "learning_rate": 4.8076137565150535e-06, "loss": 0.17443963885307312, "step": 645 }, { "epoch": 0.477634011090573, "grad_norm": 4.524256229400635, "learning_rate": 4.8068866780703874e-06, "loss": 0.19584642350673676, "step": 646 }, { "epoch": 0.478373382624769, "grad_norm": 4.598405361175537, "learning_rate": 4.806158289903311e-06, "loss": 0.14032676815986633, "step": 647 }, { "epoch": 0.47911275415896487, "grad_norm": 3.7696003913879395, "learning_rate": 4.805428592477628e-06, "loss": 0.17653219401836395, "step": 648 }, { "epoch": 0.4798521256931608, "grad_norm": 2.1558001041412354, "learning_rate": 4.804697586257972e-06, "loss": 0.12954163551330566, "step": 649 }, { "epoch": 0.4805914972273567, "grad_norm": 3.112332582473755, "learning_rate": 4.803965271709811e-06, "loss": 0.13160648941993713, "step": 650 }, { "epoch": 0.4813308687615527, "grad_norm": 7.2848429679870605, "learning_rate": 4.803231649299447e-06, "loss": 0.1866053342819214, "step": 651 }, { "epoch": 0.48207024029574863, "grad_norm": 4.359170913696289, "learning_rate": 4.802496719494014e-06, "loss": 0.16787825524806976, "step": 652 }, { "epoch": 0.48280961182994453, "grad_norm": 2.489879608154297, "learning_rate": 4.801760482761477e-06, "loss": 0.16540026664733887, "step": 653 }, { "epoch": 0.4835489833641405, "grad_norm": 4.734084129333496, "learning_rate": 4.801022939570636e-06, "loss": 0.153651162981987, "step": 654 }, { "epoch": 0.48428835489833644, "grad_norm": 4.840785980224609, "learning_rate": 4.800284090391122e-06, "loss": 0.17076314985752106, "step": 655 }, { "epoch": 0.48502772643253234, "grad_norm": 2.382577419281006, "learning_rate": 4.799543935693396e-06, "loss": 0.14944642782211304, "step": 656 }, { "epoch": 0.4857670979667283, "grad_norm": 2.5587995052337646, "learning_rate": 4.798802475948751e-06, "loss": 0.19000697135925293, "step": 657 }, { "epoch": 0.4865064695009242, "grad_norm": 4.527267932891846, "learning_rate": 4.798059711629313e-06, "loss": 0.15456947684288025, "step": 658 }, { "epoch": 0.48724584103512014, "grad_norm": 3.2326266765594482, "learning_rate": 4.797315643208036e-06, "loss": 0.17799514532089233, "step": 659 }, { "epoch": 0.4879852125693161, "grad_norm": 1.8261770009994507, "learning_rate": 4.796570271158707e-06, "loss": 0.1838194876909256, "step": 660 }, { "epoch": 0.488724584103512, "grad_norm": 2.139300584793091, "learning_rate": 4.795823595955939e-06, "loss": 0.14556211233139038, "step": 661 }, { "epoch": 0.48946395563770795, "grad_norm": 5.549762725830078, "learning_rate": 4.7950756180751804e-06, "loss": 0.16418114304542542, "step": 662 }, { "epoch": 0.4902033271719039, "grad_norm": 3.079294204711914, "learning_rate": 4.794326337992703e-06, "loss": 0.16837550699710846, "step": 663 }, { "epoch": 0.4909426987060998, "grad_norm": 1.3725556135177612, "learning_rate": 4.793575756185614e-06, "loss": 0.15919502079486847, "step": 664 }, { "epoch": 0.49168207024029575, "grad_norm": 4.0683417320251465, "learning_rate": 4.792823873131844e-06, "loss": 0.16488151252269745, "step": 665 }, { "epoch": 0.4924214417744917, "grad_norm": 3.8263964653015137, "learning_rate": 4.792070689310154e-06, "loss": 0.18885064125061035, "step": 666 }, { "epoch": 0.4931608133086876, "grad_norm": 2.560678482055664, "learning_rate": 4.791316205200136e-06, "loss": 0.14624285697937012, "step": 667 }, { "epoch": 0.49390018484288356, "grad_norm": 1.6900960206985474, "learning_rate": 4.790560421282207e-06, "loss": 0.19251729547977448, "step": 668 }, { "epoch": 0.49463955637707946, "grad_norm": 1.9133824110031128, "learning_rate": 4.789803338037611e-06, "loss": 0.1783493459224701, "step": 669 }, { "epoch": 0.4953789279112754, "grad_norm": 4.269384860992432, "learning_rate": 4.789044955948421e-06, "loss": 0.16687452793121338, "step": 670 }, { "epoch": 0.49611829944547137, "grad_norm": 2.582716226577759, "learning_rate": 4.788285275497538e-06, "loss": 0.17614001035690308, "step": 671 }, { "epoch": 0.49685767097966727, "grad_norm": 1.8547046184539795, "learning_rate": 4.787524297168688e-06, "loss": 0.15722380578517914, "step": 672 }, { "epoch": 0.4975970425138632, "grad_norm": 4.041539192199707, "learning_rate": 4.7867620214464235e-06, "loss": 0.1447826623916626, "step": 673 }, { "epoch": 0.4983364140480592, "grad_norm": 3.2038493156433105, "learning_rate": 4.785998448816123e-06, "loss": 0.1750241369009018, "step": 674 }, { "epoch": 0.49907578558225507, "grad_norm": 1.9863206148147583, "learning_rate": 4.785233579763993e-06, "loss": 0.1638374626636505, "step": 675 }, { "epoch": 0.499815157116451, "grad_norm": 3.0003461837768555, "learning_rate": 4.7844674147770615e-06, "loss": 0.16986197233200073, "step": 676 }, { "epoch": 0.500554528650647, "grad_norm": 1.837052583694458, "learning_rate": 4.7836999543431865e-06, "loss": 0.16603896021842957, "step": 677 }, { "epoch": 0.5012939001848429, "grad_norm": 4.0716233253479, "learning_rate": 4.782931198951046e-06, "loss": 0.14635348320007324, "step": 678 }, { "epoch": 0.5020332717190388, "grad_norm": 2.682547092437744, "learning_rate": 4.7821611490901465e-06, "loss": 0.15590140223503113, "step": 679 }, { "epoch": 0.5027726432532348, "grad_norm": 3.531830310821533, "learning_rate": 4.7813898052508165e-06, "loss": 0.163869708776474, "step": 680 }, { "epoch": 0.5035120147874307, "grad_norm": 2.8275809288024902, "learning_rate": 4.780617167924209e-06, "loss": 0.1575053632259369, "step": 681 }, { "epoch": 0.5042513863216266, "grad_norm": 3.349299907684326, "learning_rate": 4.779843237602303e-06, "loss": 0.18957555294036865, "step": 682 }, { "epoch": 0.5049907578558226, "grad_norm": 5.3836669921875, "learning_rate": 4.779068014777894e-06, "loss": 0.15763241052627563, "step": 683 }, { "epoch": 0.5057301293900185, "grad_norm": 2.8582265377044678, "learning_rate": 4.778291499944609e-06, "loss": 0.14066645503044128, "step": 684 }, { "epoch": 0.5064695009242144, "grad_norm": 1.9966505765914917, "learning_rate": 4.777513693596892e-06, "loss": 0.15318961441516876, "step": 685 }, { "epoch": 0.5072088724584104, "grad_norm": 3.8183205127716064, "learning_rate": 4.776734596230012e-06, "loss": 0.16952334344387054, "step": 686 }, { "epoch": 0.5079482439926063, "grad_norm": 2.752988338470459, "learning_rate": 4.775954208340059e-06, "loss": 0.1533605307340622, "step": 687 }, { "epoch": 0.5086876155268022, "grad_norm": 1.6982992887496948, "learning_rate": 4.775172530423944e-06, "loss": 0.14522695541381836, "step": 688 }, { "epoch": 0.5094269870609982, "grad_norm": 1.2992998361587524, "learning_rate": 4.774389562979401e-06, "loss": 0.15216994285583496, "step": 689 }, { "epoch": 0.5101663585951941, "grad_norm": 2.475571870803833, "learning_rate": 4.773605306504984e-06, "loss": 0.14560532569885254, "step": 690 }, { "epoch": 0.51090573012939, "grad_norm": 4.23562479019165, "learning_rate": 4.772819761500069e-06, "loss": 0.17575529217720032, "step": 691 }, { "epoch": 0.5116451016635859, "grad_norm": 2.1457931995391846, "learning_rate": 4.772032928464853e-06, "loss": 0.13310635089874268, "step": 692 }, { "epoch": 0.5123844731977819, "grad_norm": 2.746450185775757, "learning_rate": 4.771244807900348e-06, "loss": 0.1369253545999527, "step": 693 }, { "epoch": 0.5131238447319778, "grad_norm": 6.4124274253845215, "learning_rate": 4.770455400308393e-06, "loss": 0.18119290471076965, "step": 694 }, { "epoch": 0.5138632162661737, "grad_norm": 3.598480701446533, "learning_rate": 4.769664706191642e-06, "loss": 0.16992652416229248, "step": 695 }, { "epoch": 0.5146025878003697, "grad_norm": 2.6366045475006104, "learning_rate": 4.76887272605357e-06, "loss": 0.20640172064304352, "step": 696 }, { "epoch": 0.5153419593345656, "grad_norm": 3.3892807960510254, "learning_rate": 4.7680794603984695e-06, "loss": 0.147743821144104, "step": 697 }, { "epoch": 0.5160813308687615, "grad_norm": 6.361764907836914, "learning_rate": 4.767284909731453e-06, "loss": 0.1751834750175476, "step": 698 }, { "epoch": 0.5168207024029575, "grad_norm": 2.021069049835205, "learning_rate": 4.7664890745584505e-06, "loss": 0.1462891399860382, "step": 699 }, { "epoch": 0.5175600739371534, "grad_norm": 3.420747995376587, "learning_rate": 4.765691955386208e-06, "loss": 0.1864093691110611, "step": 700 }, { "epoch": 0.5182994454713493, "grad_norm": 4.140267372131348, "learning_rate": 4.764893552722295e-06, "loss": 0.16056779026985168, "step": 701 }, { "epoch": 0.5190388170055453, "grad_norm": 2.561947822570801, "learning_rate": 4.764093867075091e-06, "loss": 0.16064240038394928, "step": 702 }, { "epoch": 0.5197781885397412, "grad_norm": 2.457017183303833, "learning_rate": 4.763292898953797e-06, "loss": 0.16763383150100708, "step": 703 }, { "epoch": 0.5205175600739371, "grad_norm": 2.8869576454162598, "learning_rate": 4.7624906488684295e-06, "loss": 0.09965041279792786, "step": 704 }, { "epoch": 0.5212569316081331, "grad_norm": 2.9010937213897705, "learning_rate": 4.7616871173298205e-06, "loss": 0.1461082398891449, "step": 705 }, { "epoch": 0.521996303142329, "grad_norm": 2.274061679840088, "learning_rate": 4.76088230484962e-06, "loss": 0.19074402749538422, "step": 706 }, { "epoch": 0.5227356746765249, "grad_norm": 3.9505417346954346, "learning_rate": 4.76007621194029e-06, "loss": 0.165589839220047, "step": 707 }, { "epoch": 0.5234750462107209, "grad_norm": 2.6640408039093018, "learning_rate": 4.759268839115112e-06, "loss": 0.18663503229618073, "step": 708 }, { "epoch": 0.5242144177449168, "grad_norm": 2.0465333461761475, "learning_rate": 4.758460186888181e-06, "loss": 0.16357293725013733, "step": 709 }, { "epoch": 0.5249537892791127, "grad_norm": 3.9147684574127197, "learning_rate": 4.757650255774404e-06, "loss": 0.14295531809329987, "step": 710 }, { "epoch": 0.5256931608133086, "grad_norm": 2.0661113262176514, "learning_rate": 4.756839046289506e-06, "loss": 0.11934880912303925, "step": 711 }, { "epoch": 0.5264325323475046, "grad_norm": 2.221226215362549, "learning_rate": 4.756026558950025e-06, "loss": 0.15016141533851624, "step": 712 }, { "epoch": 0.5271719038817005, "grad_norm": 2.8608486652374268, "learning_rate": 4.7552127942733115e-06, "loss": 0.15610262751579285, "step": 713 }, { "epoch": 0.5279112754158964, "grad_norm": 2.8770039081573486, "learning_rate": 4.75439775277753e-06, "loss": 0.14757013320922852, "step": 714 }, { "epoch": 0.5286506469500925, "grad_norm": 2.376265287399292, "learning_rate": 4.7535814349816586e-06, "loss": 0.19148889183998108, "step": 715 }, { "epoch": 0.5293900184842883, "grad_norm": 2.970752477645874, "learning_rate": 4.7527638414054874e-06, "loss": 0.16932906210422516, "step": 716 }, { "epoch": 0.5301293900184842, "grad_norm": 5.098373889923096, "learning_rate": 4.751944972569618e-06, "loss": 0.19912059605121613, "step": 717 }, { "epoch": 0.5308687615526803, "grad_norm": 3.4155285358428955, "learning_rate": 4.751124828995466e-06, "loss": 0.1735413819551468, "step": 718 }, { "epoch": 0.5316081330868762, "grad_norm": 2.703457832336426, "learning_rate": 4.750303411205259e-06, "loss": 0.17557120323181152, "step": 719 }, { "epoch": 0.532347504621072, "grad_norm": 3.501183032989502, "learning_rate": 4.749480719722033e-06, "loss": 0.1648527979850769, "step": 720 }, { "epoch": 0.5330868761552681, "grad_norm": 5.627288818359375, "learning_rate": 4.748656755069637e-06, "loss": 0.18210983276367188, "step": 721 }, { "epoch": 0.533826247689464, "grad_norm": 3.086840867996216, "learning_rate": 4.74783151777273e-06, "loss": 0.16540925204753876, "step": 722 }, { "epoch": 0.5345656192236599, "grad_norm": 3.589855909347534, "learning_rate": 4.747005008356783e-06, "loss": 0.14442303776741028, "step": 723 }, { "epoch": 0.5353049907578559, "grad_norm": 3.9827327728271484, "learning_rate": 4.746177227348075e-06, "loss": 0.19366826117038727, "step": 724 }, { "epoch": 0.5360443622920518, "grad_norm": 4.948014259338379, "learning_rate": 4.745348175273696e-06, "loss": 0.1639503538608551, "step": 725 }, { "epoch": 0.5367837338262477, "grad_norm": 2.058845281600952, "learning_rate": 4.744517852661544e-06, "loss": 0.1403340995311737, "step": 726 }, { "epoch": 0.5375231053604437, "grad_norm": 3.582648992538452, "learning_rate": 4.743686260040327e-06, "loss": 0.1661899983882904, "step": 727 }, { "epoch": 0.5382624768946396, "grad_norm": 4.425167560577393, "learning_rate": 4.742853397939562e-06, "loss": 0.17407336831092834, "step": 728 }, { "epoch": 0.5390018484288355, "grad_norm": 2.927203416824341, "learning_rate": 4.742019266889575e-06, "loss": 0.1354059875011444, "step": 729 }, { "epoch": 0.5397412199630314, "grad_norm": 3.626664638519287, "learning_rate": 4.741183867421497e-06, "loss": 0.16258402168750763, "step": 730 }, { "epoch": 0.5404805914972274, "grad_norm": 3.6221466064453125, "learning_rate": 4.740347200067268e-06, "loss": 0.14312878251075745, "step": 731 }, { "epoch": 0.5412199630314233, "grad_norm": 2.8593246936798096, "learning_rate": 4.7395092653596385e-06, "loss": 0.1622759997844696, "step": 732 }, { "epoch": 0.5419593345656192, "grad_norm": 2.6975250244140625, "learning_rate": 4.73867006383216e-06, "loss": 0.16512002050876617, "step": 733 }, { "epoch": 0.5426987060998152, "grad_norm": 3.7378289699554443, "learning_rate": 4.737829596019198e-06, "loss": 0.11877238750457764, "step": 734 }, { "epoch": 0.5434380776340111, "grad_norm": 3.975489854812622, "learning_rate": 4.736987862455916e-06, "loss": 0.20991036295890808, "step": 735 }, { "epoch": 0.544177449168207, "grad_norm": 1.9186402559280396, "learning_rate": 4.736144863678291e-06, "loss": 0.16109146177768707, "step": 736 }, { "epoch": 0.544916820702403, "grad_norm": 2.3307924270629883, "learning_rate": 4.735300600223102e-06, "loss": 0.11596597731113434, "step": 737 }, { "epoch": 0.5456561922365989, "grad_norm": 3.470093011856079, "learning_rate": 4.7344550726279324e-06, "loss": 0.14143389463424683, "step": 738 }, { "epoch": 0.5463955637707948, "grad_norm": 3.068608283996582, "learning_rate": 4.733608281431172e-06, "loss": 0.17365147173404694, "step": 739 }, { "epoch": 0.5471349353049908, "grad_norm": 2.0672249794006348, "learning_rate": 4.732760227172016e-06, "loss": 0.1508428454399109, "step": 740 }, { "epoch": 0.5478743068391867, "grad_norm": 3.5433614253997803, "learning_rate": 4.731910910390461e-06, "loss": 0.17056798934936523, "step": 741 }, { "epoch": 0.5486136783733826, "grad_norm": 2.6028504371643066, "learning_rate": 4.731060331627313e-06, "loss": 0.13013440370559692, "step": 742 }, { "epoch": 0.5493530499075786, "grad_norm": 2.489981174468994, "learning_rate": 4.730208491424174e-06, "loss": 0.1499713808298111, "step": 743 }, { "epoch": 0.5500924214417745, "grad_norm": 2.8025598526000977, "learning_rate": 4.729355390323456e-06, "loss": 0.17507296800613403, "step": 744 }, { "epoch": 0.5508317929759704, "grad_norm": 3.620969533920288, "learning_rate": 4.7285010288683685e-06, "loss": 0.1440659761428833, "step": 745 }, { "epoch": 0.5515711645101664, "grad_norm": 2.036953926086426, "learning_rate": 4.7276454076029286e-06, "loss": 0.16161805391311646, "step": 746 }, { "epoch": 0.5523105360443623, "grad_norm": 2.4746549129486084, "learning_rate": 4.72678852707195e-06, "loss": 0.14864866435527802, "step": 747 }, { "epoch": 0.5530499075785582, "grad_norm": 4.962723255157471, "learning_rate": 4.7259303878210554e-06, "loss": 0.16390742361545563, "step": 748 }, { "epoch": 0.5537892791127541, "grad_norm": 4.4101057052612305, "learning_rate": 4.725070990396662e-06, "loss": 0.16961067914962769, "step": 749 }, { "epoch": 0.5545286506469501, "grad_norm": 3.579427480697632, "learning_rate": 4.724210335345992e-06, "loss": 0.15413346886634827, "step": 750 }, { "epoch": 0.555268022181146, "grad_norm": 3.647165060043335, "learning_rate": 4.723348423217066e-06, "loss": 0.14261889457702637, "step": 751 }, { "epoch": 0.5560073937153419, "grad_norm": 3.7783424854278564, "learning_rate": 4.722485254558709e-06, "loss": 0.18903255462646484, "step": 752 }, { "epoch": 0.5567467652495379, "grad_norm": 2.9593396186828613, "learning_rate": 4.721620829920541e-06, "loss": 0.12314149737358093, "step": 753 }, { "epoch": 0.5574861367837338, "grad_norm": 4.466017723083496, "learning_rate": 4.720755149852986e-06, "loss": 0.16140368580818176, "step": 754 }, { "epoch": 0.5582255083179297, "grad_norm": 2.1652283668518066, "learning_rate": 4.7198882149072664e-06, "loss": 0.14149397611618042, "step": 755 }, { "epoch": 0.5589648798521257, "grad_norm": 4.591597557067871, "learning_rate": 4.7190200256354e-06, "loss": 0.1508958339691162, "step": 756 }, { "epoch": 0.5597042513863216, "grad_norm": 5.1699628829956055, "learning_rate": 4.71815058259021e-06, "loss": 0.18371352553367615, "step": 757 }, { "epoch": 0.5604436229205175, "grad_norm": 3.0858964920043945, "learning_rate": 4.71727988632531e-06, "loss": 0.13068190217018127, "step": 758 }, { "epoch": 0.5611829944547135, "grad_norm": 3.0122194290161133, "learning_rate": 4.7164079373951194e-06, "loss": 0.20142771303653717, "step": 759 }, { "epoch": 0.5619223659889094, "grad_norm": 3.839064598083496, "learning_rate": 4.71553473635485e-06, "loss": 0.1812341809272766, "step": 760 }, { "epoch": 0.5626617375231053, "grad_norm": 3.6308064460754395, "learning_rate": 4.714660283760512e-06, "loss": 0.16326898336410522, "step": 761 }, { "epoch": 0.5634011090573013, "grad_norm": 2.588911533355713, "learning_rate": 4.7137845801689145e-06, "loss": 0.14729513227939606, "step": 762 }, { "epoch": 0.5641404805914972, "grad_norm": 2.833604097366333, "learning_rate": 4.7129076261376595e-06, "loss": 0.17238138616085052, "step": 763 }, { "epoch": 0.5648798521256931, "grad_norm": 3.550633668899536, "learning_rate": 4.712029422225149e-06, "loss": 0.19326528906822205, "step": 764 }, { "epoch": 0.5656192236598891, "grad_norm": 5.893483638763428, "learning_rate": 4.71114996899058e-06, "loss": 0.16156089305877686, "step": 765 }, { "epoch": 0.566358595194085, "grad_norm": 3.8947107791900635, "learning_rate": 4.710269266993943e-06, "loss": 0.1531086266040802, "step": 766 }, { "epoch": 0.5670979667282809, "grad_norm": 3.553295135498047, "learning_rate": 4.709387316796025e-06, "loss": 0.15872901678085327, "step": 767 }, { "epoch": 0.5678373382624768, "grad_norm": 4.978425979614258, "learning_rate": 4.708504118958408e-06, "loss": 0.13273049890995026, "step": 768 }, { "epoch": 0.5685767097966729, "grad_norm": 4.3065338134765625, "learning_rate": 4.7076196740434685e-06, "loss": 0.17127755284309387, "step": 769 }, { "epoch": 0.5693160813308688, "grad_norm": 4.653750419616699, "learning_rate": 4.706733982614375e-06, "loss": 0.15254057943820953, "step": 770 }, { "epoch": 0.5700554528650646, "grad_norm": 3.244551181793213, "learning_rate": 4.705847045235094e-06, "loss": 0.14277026057243347, "step": 771 }, { "epoch": 0.5707948243992607, "grad_norm": 6.004458904266357, "learning_rate": 4.704958862470381e-06, "loss": 0.16850589215755463, "step": 772 }, { "epoch": 0.5715341959334566, "grad_norm": 4.126717567443848, "learning_rate": 4.704069434885787e-06, "loss": 0.19929200410842896, "step": 773 }, { "epoch": 0.5722735674676525, "grad_norm": 2.7319836616516113, "learning_rate": 4.703178763047654e-06, "loss": 0.16919048130512238, "step": 774 }, { "epoch": 0.5730129390018485, "grad_norm": 2.0547428131103516, "learning_rate": 4.702286847523118e-06, "loss": 0.13484343886375427, "step": 775 }, { "epoch": 0.5737523105360444, "grad_norm": 4.243725299835205, "learning_rate": 4.7013936888801074e-06, "loss": 0.1611710786819458, "step": 776 }, { "epoch": 0.5744916820702403, "grad_norm": 2.091276168823242, "learning_rate": 4.700499287687339e-06, "loss": 0.14918309450149536, "step": 777 }, { "epoch": 0.5752310536044363, "grad_norm": 1.865685224533081, "learning_rate": 4.699603644514323e-06, "loss": 0.1252010017633438, "step": 778 }, { "epoch": 0.5759704251386322, "grad_norm": 2.927765369415283, "learning_rate": 4.69870675993136e-06, "loss": 0.10881543159484863, "step": 779 }, { "epoch": 0.5767097966728281, "grad_norm": 2.24281644821167, "learning_rate": 4.697808634509541e-06, "loss": 0.15992532670497894, "step": 780 }, { "epoch": 0.5774491682070241, "grad_norm": 2.353679895401001, "learning_rate": 4.6969092688207495e-06, "loss": 0.17222705483436584, "step": 781 }, { "epoch": 0.57818853974122, "grad_norm": 2.7806553840637207, "learning_rate": 4.696008663437654e-06, "loss": 0.14250144362449646, "step": 782 }, { "epoch": 0.5789279112754159, "grad_norm": 2.598917245864868, "learning_rate": 4.695106818933716e-06, "loss": 0.12392956018447876, "step": 783 }, { "epoch": 0.5796672828096119, "grad_norm": 2.391997814178467, "learning_rate": 4.694203735883185e-06, "loss": 0.16084200143814087, "step": 784 }, { "epoch": 0.5804066543438078, "grad_norm": 3.03591251373291, "learning_rate": 4.693299414861097e-06, "loss": 0.16314697265625, "step": 785 }, { "epoch": 0.5811460258780037, "grad_norm": 2.498542547225952, "learning_rate": 4.692393856443281e-06, "loss": 0.13383112847805023, "step": 786 }, { "epoch": 0.5818853974121996, "grad_norm": 2.824936628341675, "learning_rate": 4.691487061206351e-06, "loss": 0.12116478383541107, "step": 787 }, { "epoch": 0.5826247689463956, "grad_norm": 3.075801134109497, "learning_rate": 4.690579029727707e-06, "loss": 0.12529975175857544, "step": 788 }, { "epoch": 0.5833641404805915, "grad_norm": 2.224318742752075, "learning_rate": 4.68966976258554e-06, "loss": 0.13426576554775238, "step": 789 }, { "epoch": 0.5841035120147874, "grad_norm": 3.8500564098358154, "learning_rate": 4.688759260358824e-06, "loss": 0.20451095700263977, "step": 790 }, { "epoch": 0.5848428835489834, "grad_norm": 3.3103363513946533, "learning_rate": 4.687847523627323e-06, "loss": 0.15620622038841248, "step": 791 }, { "epoch": 0.5855822550831793, "grad_norm": 3.9935896396636963, "learning_rate": 4.686934552971582e-06, "loss": 0.09807637333869934, "step": 792 }, { "epoch": 0.5863216266173752, "grad_norm": 4.404787063598633, "learning_rate": 4.6860203489729395e-06, "loss": 0.15312685072422028, "step": 793 }, { "epoch": 0.5870609981515712, "grad_norm": 2.5484912395477295, "learning_rate": 4.685104912213512e-06, "loss": 0.14565953612327576, "step": 794 }, { "epoch": 0.5878003696857671, "grad_norm": 6.020355701446533, "learning_rate": 4.6841882432762045e-06, "loss": 0.17845411598682404, "step": 795 }, { "epoch": 0.588539741219963, "grad_norm": 2.270612955093384, "learning_rate": 4.683270342744705e-06, "loss": 0.15276987850666046, "step": 796 }, { "epoch": 0.589279112754159, "grad_norm": 2.5641887187957764, "learning_rate": 4.682351211203488e-06, "loss": 0.15517903864383698, "step": 797 }, { "epoch": 0.5900184842883549, "grad_norm": 2.432572603225708, "learning_rate": 4.68143084923781e-06, "loss": 0.15443310141563416, "step": 798 }, { "epoch": 0.5907578558225508, "grad_norm": 2.874272108078003, "learning_rate": 4.68050925743371e-06, "loss": 0.13221290707588196, "step": 799 }, { "epoch": 0.5914972273567468, "grad_norm": 1.8766229152679443, "learning_rate": 4.679586436378013e-06, "loss": 0.16370995342731476, "step": 800 }, { "epoch": 0.5922365988909427, "grad_norm": 3.3797097206115723, "learning_rate": 4.678662386658325e-06, "loss": 0.17400145530700684, "step": 801 }, { "epoch": 0.5929759704251386, "grad_norm": 2.9587244987487793, "learning_rate": 4.677737108863035e-06, "loss": 0.13921616971492767, "step": 802 }, { "epoch": 0.5937153419593346, "grad_norm": 3.381242513656616, "learning_rate": 4.676810603581312e-06, "loss": 0.13810670375823975, "step": 803 }, { "epoch": 0.5944547134935305, "grad_norm": 4.834743499755859, "learning_rate": 4.67588287140311e-06, "loss": 0.17131520807743073, "step": 804 }, { "epoch": 0.5951940850277264, "grad_norm": 2.3767356872558594, "learning_rate": 4.674953912919161e-06, "loss": 0.17324703931808472, "step": 805 }, { "epoch": 0.5959334565619223, "grad_norm": 3.6876204013824463, "learning_rate": 4.67402372872098e-06, "loss": 0.1311509609222412, "step": 806 }, { "epoch": 0.5966728280961183, "grad_norm": 3.735725164413452, "learning_rate": 4.673092319400862e-06, "loss": 0.10795502364635468, "step": 807 }, { "epoch": 0.5974121996303142, "grad_norm": 5.9450459480285645, "learning_rate": 4.67215968555188e-06, "loss": 0.18074136972427368, "step": 808 }, { "epoch": 0.5981515711645101, "grad_norm": 5.384657382965088, "learning_rate": 4.671225827767891e-06, "loss": 0.12841591238975525, "step": 809 }, { "epoch": 0.5988909426987061, "grad_norm": 3.5342044830322266, "learning_rate": 4.670290746643527e-06, "loss": 0.12454596161842346, "step": 810 }, { "epoch": 0.599630314232902, "grad_norm": 4.005172252655029, "learning_rate": 4.669354442774201e-06, "loss": 0.16052979230880737, "step": 811 }, { "epoch": 0.6003696857670979, "grad_norm": 5.312797546386719, "learning_rate": 4.668416916756106e-06, "loss": 0.11460230499505997, "step": 812 }, { "epoch": 0.6011090573012939, "grad_norm": 2.7572813034057617, "learning_rate": 4.66747816918621e-06, "loss": 0.11400257050991058, "step": 813 }, { "epoch": 0.6018484288354898, "grad_norm": 3.9107675552368164, "learning_rate": 4.66653820066226e-06, "loss": 0.14539238810539246, "step": 814 }, { "epoch": 0.6025878003696857, "grad_norm": 2.630927085876465, "learning_rate": 4.665597011782782e-06, "loss": 0.18871790170669556, "step": 815 }, { "epoch": 0.6033271719038817, "grad_norm": 3.9844391345977783, "learning_rate": 4.664654603147078e-06, "loss": 0.187518909573555, "step": 816 }, { "epoch": 0.6040665434380776, "grad_norm": 2.7100141048431396, "learning_rate": 4.6637109753552274e-06, "loss": 0.16740864515304565, "step": 817 }, { "epoch": 0.6048059149722735, "grad_norm": 3.70308256149292, "learning_rate": 4.662766129008083e-06, "loss": 0.16847898066043854, "step": 818 }, { "epoch": 0.6055452865064695, "grad_norm": 4.185943603515625, "learning_rate": 4.661820064707277e-06, "loss": 0.1302773654460907, "step": 819 }, { "epoch": 0.6062846580406654, "grad_norm": 4.646094799041748, "learning_rate": 4.660872783055215e-06, "loss": 0.15050533413887024, "step": 820 }, { "epoch": 0.6070240295748613, "grad_norm": 3.269235849380493, "learning_rate": 4.65992428465508e-06, "loss": 0.116980642080307, "step": 821 }, { "epoch": 0.6077634011090574, "grad_norm": 1.8627281188964844, "learning_rate": 4.658974570110826e-06, "loss": 0.16046127676963806, "step": 822 }, { "epoch": 0.6085027726432533, "grad_norm": 3.825092077255249, "learning_rate": 4.658023640027186e-06, "loss": 0.1651129275560379, "step": 823 }, { "epoch": 0.6092421441774492, "grad_norm": 3.8717854022979736, "learning_rate": 4.6570714950096616e-06, "loss": 0.14412778615951538, "step": 824 }, { "epoch": 0.609981515711645, "grad_norm": 2.5130889415740967, "learning_rate": 4.656118135664534e-06, "loss": 0.14149588346481323, "step": 825 }, { "epoch": 0.6107208872458411, "grad_norm": 1.8306896686553955, "learning_rate": 4.6551635625988545e-06, "loss": 0.14308232069015503, "step": 826 }, { "epoch": 0.611460258780037, "grad_norm": 3.007230043411255, "learning_rate": 4.654207776420446e-06, "loss": 0.13278841972351074, "step": 827 }, { "epoch": 0.6121996303142329, "grad_norm": 6.800658226013184, "learning_rate": 4.653250777737905e-06, "loss": 0.16868703067302704, "step": 828 }, { "epoch": 0.6129390018484289, "grad_norm": 3.59250807762146, "learning_rate": 4.652292567160603e-06, "loss": 0.1285383701324463, "step": 829 }, { "epoch": 0.6136783733826248, "grad_norm": 2.744007110595703, "learning_rate": 4.6513331452986775e-06, "loss": 0.14459793269634247, "step": 830 }, { "epoch": 0.6144177449168207, "grad_norm": 5.441387176513672, "learning_rate": 4.650372512763041e-06, "loss": 0.16824068129062653, "step": 831 }, { "epoch": 0.6151571164510167, "grad_norm": 5.671798229217529, "learning_rate": 4.649410670165377e-06, "loss": 0.18649187684059143, "step": 832 }, { "epoch": 0.6158964879852126, "grad_norm": 4.311167240142822, "learning_rate": 4.648447618118137e-06, "loss": 0.13841019570827484, "step": 833 }, { "epoch": 0.6166358595194085, "grad_norm": 3.111867666244507, "learning_rate": 4.647483357234547e-06, "loss": 0.13049234449863434, "step": 834 }, { "epoch": 0.6173752310536045, "grad_norm": 2.286797285079956, "learning_rate": 4.646517888128599e-06, "loss": 0.1257648468017578, "step": 835 }, { "epoch": 0.6181146025878004, "grad_norm": 2.611382246017456, "learning_rate": 4.645551211415055e-06, "loss": 0.1424291729927063, "step": 836 }, { "epoch": 0.6188539741219963, "grad_norm": 4.665614604949951, "learning_rate": 4.644583327709446e-06, "loss": 0.14314186573028564, "step": 837 }, { "epoch": 0.6195933456561923, "grad_norm": 3.91795015335083, "learning_rate": 4.643614237628071e-06, "loss": 0.13295093178749084, "step": 838 }, { "epoch": 0.6203327171903882, "grad_norm": 2.7669074535369873, "learning_rate": 4.642643941788e-06, "loss": 0.17946338653564453, "step": 839 }, { "epoch": 0.6210720887245841, "grad_norm": 4.917020797729492, "learning_rate": 4.641672440807068e-06, "loss": 0.15382787585258484, "step": 840 }, { "epoch": 0.6218114602587801, "grad_norm": 4.084968566894531, "learning_rate": 4.640699735303879e-06, "loss": 0.10625254362821579, "step": 841 }, { "epoch": 0.622550831792976, "grad_norm": 3.862807035446167, "learning_rate": 4.639725825897801e-06, "loss": 0.14555761218070984, "step": 842 }, { "epoch": 0.6232902033271719, "grad_norm": 3.296131134033203, "learning_rate": 4.638750713208973e-06, "loss": 0.16281816363334656, "step": 843 }, { "epoch": 0.6240295748613678, "grad_norm": 7.079187870025635, "learning_rate": 4.637774397858297e-06, "loss": 0.11450199037790298, "step": 844 }, { "epoch": 0.6247689463955638, "grad_norm": 4.877682685852051, "learning_rate": 4.636796880467441e-06, "loss": 0.15248212218284607, "step": 845 }, { "epoch": 0.6255083179297597, "grad_norm": 2.8867483139038086, "learning_rate": 4.63581816165884e-06, "loss": 0.14363549649715424, "step": 846 }, { "epoch": 0.6262476894639556, "grad_norm": 2.6870007514953613, "learning_rate": 4.634838242055693e-06, "loss": 0.15628795325756073, "step": 847 }, { "epoch": 0.6269870609981516, "grad_norm": 5.705078601837158, "learning_rate": 4.633857122281963e-06, "loss": 0.15535646677017212, "step": 848 }, { "epoch": 0.6277264325323475, "grad_norm": 4.876985549926758, "learning_rate": 4.632874802962379e-06, "loss": 0.1665460467338562, "step": 849 }, { "epoch": 0.6284658040665434, "grad_norm": 4.290618896484375, "learning_rate": 4.6318912847224325e-06, "loss": 0.11197323352098465, "step": 850 }, { "epoch": 0.6292051756007394, "grad_norm": 3.367198944091797, "learning_rate": 4.630906568188378e-06, "loss": 0.14486274123191833, "step": 851 }, { "epoch": 0.6299445471349353, "grad_norm": 4.793536186218262, "learning_rate": 4.629920653987234e-06, "loss": 0.10955233126878738, "step": 852 }, { "epoch": 0.6306839186691312, "grad_norm": 6.078773498535156, "learning_rate": 4.628933542746781e-06, "loss": 0.15265001356601715, "step": 853 }, { "epoch": 0.6314232902033272, "grad_norm": 5.251503944396973, "learning_rate": 4.627945235095563e-06, "loss": 0.17625001072883606, "step": 854 }, { "epoch": 0.6321626617375231, "grad_norm": 2.8956634998321533, "learning_rate": 4.626955731662884e-06, "loss": 0.12384217977523804, "step": 855 }, { "epoch": 0.632902033271719, "grad_norm": 4.361636638641357, "learning_rate": 4.6259650330788104e-06, "loss": 0.12767034769058228, "step": 856 }, { "epoch": 0.633641404805915, "grad_norm": 4.696187496185303, "learning_rate": 4.62497313997417e-06, "loss": 0.1590440273284912, "step": 857 }, { "epoch": 0.6343807763401109, "grad_norm": 4.904086112976074, "learning_rate": 4.623980052980549e-06, "loss": 0.14212676882743835, "step": 858 }, { "epoch": 0.6351201478743068, "grad_norm": 4.309323787689209, "learning_rate": 4.6229857727302975e-06, "loss": 0.1344265341758728, "step": 859 }, { "epoch": 0.6358595194085028, "grad_norm": 3.4507253170013428, "learning_rate": 4.621990299856521e-06, "loss": 0.14179135859012604, "step": 860 }, { "epoch": 0.6365988909426987, "grad_norm": 3.324286937713623, "learning_rate": 4.62099363499309e-06, "loss": 0.12610816955566406, "step": 861 }, { "epoch": 0.6373382624768946, "grad_norm": 4.735670566558838, "learning_rate": 4.619995778774627e-06, "loss": 0.15670350193977356, "step": 862 }, { "epoch": 0.6380776340110905, "grad_norm": 4.5814409255981445, "learning_rate": 4.618996731836521e-06, "loss": 0.18869367241859436, "step": 863 }, { "epoch": 0.6388170055452865, "grad_norm": 2.6500649452209473, "learning_rate": 4.617996494814911e-06, "loss": 0.12681332230567932, "step": 864 }, { "epoch": 0.6395563770794824, "grad_norm": 3.2556118965148926, "learning_rate": 4.6169950683467e-06, "loss": 0.09333169460296631, "step": 865 }, { "epoch": 0.6402957486136783, "grad_norm": 5.0627593994140625, "learning_rate": 4.615992453069546e-06, "loss": 0.16567069292068481, "step": 866 }, { "epoch": 0.6410351201478743, "grad_norm": 3.248217821121216, "learning_rate": 4.614988649621864e-06, "loss": 0.13380670547485352, "step": 867 }, { "epoch": 0.6417744916820702, "grad_norm": 3.892122745513916, "learning_rate": 4.613983658642826e-06, "loss": 0.10441601276397705, "step": 868 }, { "epoch": 0.6425138632162661, "grad_norm": 4.899961948394775, "learning_rate": 4.6129774807723596e-06, "loss": 0.1817372739315033, "step": 869 }, { "epoch": 0.6432532347504621, "grad_norm": 2.8200302124023438, "learning_rate": 4.611970116651149e-06, "loss": 0.14664994180202484, "step": 870 }, { "epoch": 0.643992606284658, "grad_norm": 4.0175604820251465, "learning_rate": 4.610961566920632e-06, "loss": 0.16752851009368896, "step": 871 }, { "epoch": 0.6447319778188539, "grad_norm": 5.294934272766113, "learning_rate": 4.609951832223004e-06, "loss": 0.17403966188430786, "step": 872 }, { "epoch": 0.64547134935305, "grad_norm": 2.433837652206421, "learning_rate": 4.608940913201212e-06, "loss": 0.16837018728256226, "step": 873 }, { "epoch": 0.6462107208872458, "grad_norm": 2.7489590644836426, "learning_rate": 4.60792881049896e-06, "loss": 0.10922932624816895, "step": 874 }, { "epoch": 0.6469500924214417, "grad_norm": 3.4473342895507812, "learning_rate": 4.606915524760704e-06, "loss": 0.14771488308906555, "step": 875 }, { "epoch": 0.6476894639556378, "grad_norm": 3.2592828273773193, "learning_rate": 4.605901056631653e-06, "loss": 0.14853042364120483, "step": 876 }, { "epoch": 0.6484288354898337, "grad_norm": 2.981815814971924, "learning_rate": 4.60488540675777e-06, "loss": 0.10845884680747986, "step": 877 }, { "epoch": 0.6491682070240296, "grad_norm": 3.191302537918091, "learning_rate": 4.60386857578577e-06, "loss": 0.13122737407684326, "step": 878 }, { "epoch": 0.6499075785582256, "grad_norm": 3.1937096118927, "learning_rate": 4.602850564363119e-06, "loss": 0.13606278598308563, "step": 879 }, { "epoch": 0.6506469500924215, "grad_norm": 3.4065868854522705, "learning_rate": 4.6018313731380375e-06, "loss": 0.13637031614780426, "step": 880 }, { "epoch": 0.6513863216266174, "grad_norm": 2.6559529304504395, "learning_rate": 4.600811002759494e-06, "loss": 0.1326475739479065, "step": 881 }, { "epoch": 0.6521256931608133, "grad_norm": 3.824077606201172, "learning_rate": 4.5997894538772095e-06, "loss": 0.13607358932495117, "step": 882 }, { "epoch": 0.6528650646950093, "grad_norm": 4.274715423583984, "learning_rate": 4.5987667271416544e-06, "loss": 0.12092474848031998, "step": 883 }, { "epoch": 0.6536044362292052, "grad_norm": 4.3679633140563965, "learning_rate": 4.597742823204053e-06, "loss": 0.12280683219432831, "step": 884 }, { "epoch": 0.6543438077634011, "grad_norm": 3.8415067195892334, "learning_rate": 4.596717742716372e-06, "loss": 0.17076697945594788, "step": 885 }, { "epoch": 0.6550831792975971, "grad_norm": 4.529657363891602, "learning_rate": 4.595691486331334e-06, "loss": 0.15193352103233337, "step": 886 }, { "epoch": 0.655822550831793, "grad_norm": 4.221188068389893, "learning_rate": 4.594664054702407e-06, "loss": 0.12977463006973267, "step": 887 }, { "epoch": 0.6565619223659889, "grad_norm": 4.04974889755249, "learning_rate": 4.593635448483807e-06, "loss": 0.11564896255731583, "step": 888 }, { "epoch": 0.6573012939001849, "grad_norm": 3.5890543460845947, "learning_rate": 4.5926056683305e-06, "loss": 0.13755643367767334, "step": 889 }, { "epoch": 0.6580406654343808, "grad_norm": 4.615071773529053, "learning_rate": 4.591574714898199e-06, "loss": 0.11539887636899948, "step": 890 }, { "epoch": 0.6587800369685767, "grad_norm": 3.399475574493408, "learning_rate": 4.590542588843361e-06, "loss": 0.1781056523323059, "step": 891 }, { "epoch": 0.6595194085027727, "grad_norm": 2.362799644470215, "learning_rate": 4.589509290823195e-06, "loss": 0.13189618289470673, "step": 892 }, { "epoch": 0.6602587800369686, "grad_norm": 3.3037302494049072, "learning_rate": 4.588474821495652e-06, "loss": 0.10927771031856537, "step": 893 }, { "epoch": 0.6609981515711645, "grad_norm": 3.9652740955352783, "learning_rate": 4.587439181519431e-06, "loss": 0.07969710230827332, "step": 894 }, { "epoch": 0.6617375231053605, "grad_norm": 3.7434260845184326, "learning_rate": 4.586402371553974e-06, "loss": 0.1448361575603485, "step": 895 }, { "epoch": 0.6624768946395564, "grad_norm": 3.7640929222106934, "learning_rate": 4.585364392259472e-06, "loss": 0.15153451263904572, "step": 896 }, { "epoch": 0.6632162661737523, "grad_norm": 3.790480136871338, "learning_rate": 4.584325244296857e-06, "loss": 0.14495888352394104, "step": 897 }, { "epoch": 0.6639556377079483, "grad_norm": 3.1762747764587402, "learning_rate": 4.583284928327805e-06, "loss": 0.1590115875005722, "step": 898 }, { "epoch": 0.6646950092421442, "grad_norm": 3.470430612564087, "learning_rate": 4.58224344501474e-06, "loss": 0.15011745691299438, "step": 899 }, { "epoch": 0.6654343807763401, "grad_norm": 3.131564140319824, "learning_rate": 4.581200795020824e-06, "loss": 0.1339944452047348, "step": 900 }, { "epoch": 0.666173752310536, "grad_norm": 3.4234955310821533, "learning_rate": 4.580156979009965e-06, "loss": 0.13204757869243622, "step": 901 }, { "epoch": 0.666913123844732, "grad_norm": 3.6548311710357666, "learning_rate": 4.579111997646814e-06, "loss": 0.1402699202299118, "step": 902 }, { "epoch": 0.6676524953789279, "grad_norm": 3.8409335613250732, "learning_rate": 4.578065851596761e-06, "loss": 0.13123875856399536, "step": 903 }, { "epoch": 0.6683918669131238, "grad_norm": 3.441267728805542, "learning_rate": 4.57701854152594e-06, "loss": 0.15350517630577087, "step": 904 }, { "epoch": 0.6691312384473198, "grad_norm": 2.040598154067993, "learning_rate": 4.575970068101227e-06, "loss": 0.16457492113113403, "step": 905 }, { "epoch": 0.6698706099815157, "grad_norm": 2.7849159240722656, "learning_rate": 4.574920431990235e-06, "loss": 0.12542587518692017, "step": 906 }, { "epoch": 0.6706099815157116, "grad_norm": 2.167228937149048, "learning_rate": 4.5738696338613205e-06, "loss": 0.1633736491203308, "step": 907 }, { "epoch": 0.6713493530499076, "grad_norm": 3.334031820297241, "learning_rate": 4.5728176743835814e-06, "loss": 0.11475145816802979, "step": 908 }, { "epoch": 0.6720887245841035, "grad_norm": 2.8139688968658447, "learning_rate": 4.571764554226848e-06, "loss": 0.07561016827821732, "step": 909 }, { "epoch": 0.6728280961182994, "grad_norm": 3.0561816692352295, "learning_rate": 4.5707102740616995e-06, "loss": 0.15085852146148682, "step": 910 }, { "epoch": 0.6735674676524954, "grad_norm": 2.5357742309570312, "learning_rate": 4.569654834559446e-06, "loss": 0.17744961380958557, "step": 911 }, { "epoch": 0.6743068391866913, "grad_norm": 2.3239829540252686, "learning_rate": 4.568598236392139e-06, "loss": 0.14179649949073792, "step": 912 }, { "epoch": 0.6750462107208872, "grad_norm": 2.6410136222839355, "learning_rate": 4.567540480232567e-06, "loss": 0.12341522425413132, "step": 913 }, { "epoch": 0.6757855822550832, "grad_norm": 3.4236292839050293, "learning_rate": 4.566481566754255e-06, "loss": 0.11541327834129333, "step": 914 }, { "epoch": 0.6765249537892791, "grad_norm": 3.4030954837799072, "learning_rate": 4.565421496631469e-06, "loss": 0.10436497628688812, "step": 915 }, { "epoch": 0.677264325323475, "grad_norm": 4.889449119567871, "learning_rate": 4.564360270539207e-06, "loss": 0.11563746631145477, "step": 916 }, { "epoch": 0.678003696857671, "grad_norm": 3.1410131454467773, "learning_rate": 4.563297889153205e-06, "loss": 0.16103772819042206, "step": 917 }, { "epoch": 0.6787430683918669, "grad_norm": 3.3436050415039062, "learning_rate": 4.562234353149934e-06, "loss": 0.0962660163640976, "step": 918 }, { "epoch": 0.6794824399260628, "grad_norm": 2.823788642883301, "learning_rate": 4.561169663206599e-06, "loss": 0.14485594630241394, "step": 919 }, { "epoch": 0.6802218114602587, "grad_norm": 3.6501622200012207, "learning_rate": 4.560103820001144e-06, "loss": 0.1366327702999115, "step": 920 }, { "epoch": 0.6809611829944547, "grad_norm": 4.576330661773682, "learning_rate": 4.559036824212242e-06, "loss": 0.13583794236183167, "step": 921 }, { "epoch": 0.6817005545286506, "grad_norm": 3.1599533557891846, "learning_rate": 4.5579686765193046e-06, "loss": 0.1210615411400795, "step": 922 }, { "epoch": 0.6824399260628465, "grad_norm": 3.2068991661071777, "learning_rate": 4.556899377602474e-06, "loss": 0.14454463124275208, "step": 923 }, { "epoch": 0.6831792975970425, "grad_norm": 3.5472817420959473, "learning_rate": 4.5558289281426255e-06, "loss": 0.13894306123256683, "step": 924 }, { "epoch": 0.6839186691312384, "grad_norm": 4.4314656257629395, "learning_rate": 4.554757328821368e-06, "loss": 0.05132925882935524, "step": 925 }, { "epoch": 0.6846580406654343, "grad_norm": 3.6451871395111084, "learning_rate": 4.553684580321042e-06, "loss": 0.14502431452274323, "step": 926 }, { "epoch": 0.6853974121996304, "grad_norm": 4.387892723083496, "learning_rate": 4.552610683324721e-06, "loss": 0.16045036911964417, "step": 927 }, { "epoch": 0.6861367837338262, "grad_norm": 3.359744071960449, "learning_rate": 4.551535638516208e-06, "loss": 0.12066791951656342, "step": 928 }, { "epoch": 0.6868761552680221, "grad_norm": 3.3421378135681152, "learning_rate": 4.550459446580039e-06, "loss": 0.1668878197669983, "step": 929 }, { "epoch": 0.6876155268022182, "grad_norm": 5.443203449249268, "learning_rate": 4.549382108201477e-06, "loss": 0.14286752045154572, "step": 930 }, { "epoch": 0.6883548983364141, "grad_norm": 4.112909317016602, "learning_rate": 4.548303624066518e-06, "loss": 0.10513077676296234, "step": 931 }, { "epoch": 0.68909426987061, "grad_norm": 3.0984623432159424, "learning_rate": 4.547223994861888e-06, "loss": 0.12037807703018188, "step": 932 }, { "epoch": 0.689833641404806, "grad_norm": 4.596905708312988, "learning_rate": 4.54614322127504e-06, "loss": 0.14312095940113068, "step": 933 }, { "epoch": 0.6905730129390019, "grad_norm": 3.8356096744537354, "learning_rate": 4.545061303994155e-06, "loss": 0.14708495140075684, "step": 934 }, { "epoch": 0.6913123844731978, "grad_norm": 2.4087865352630615, "learning_rate": 4.543978243708145e-06, "loss": 0.17291653156280518, "step": 935 }, { "epoch": 0.6920517560073938, "grad_norm": 3.6104929447174072, "learning_rate": 4.542894041106648e-06, "loss": 0.13233229517936707, "step": 936 }, { "epoch": 0.6927911275415897, "grad_norm": 4.06725549697876, "learning_rate": 4.541808696880031e-06, "loss": 0.12034182250499725, "step": 937 }, { "epoch": 0.6935304990757856, "grad_norm": 2.201307773590088, "learning_rate": 4.540722211719386e-06, "loss": 0.16851699352264404, "step": 938 }, { "epoch": 0.6942698706099815, "grad_norm": 2.990849733352661, "learning_rate": 4.539634586316532e-06, "loss": 0.10556219518184662, "step": 939 }, { "epoch": 0.6950092421441775, "grad_norm": 2.8756422996520996, "learning_rate": 4.538545821364016e-06, "loss": 0.12532076239585876, "step": 940 }, { "epoch": 0.6957486136783734, "grad_norm": 3.3327078819274902, "learning_rate": 4.5374559175551075e-06, "loss": 0.1259128898382187, "step": 941 }, { "epoch": 0.6964879852125693, "grad_norm": 3.710946559906006, "learning_rate": 4.536364875583802e-06, "loss": 0.07917308062314987, "step": 942 }, { "epoch": 0.6972273567467653, "grad_norm": 3.0749738216400146, "learning_rate": 4.535272696144823e-06, "loss": 0.08811580389738083, "step": 943 }, { "epoch": 0.6979667282809612, "grad_norm": 2.6856727600097656, "learning_rate": 4.534179379933614e-06, "loss": 0.13855117559432983, "step": 944 }, { "epoch": 0.6987060998151571, "grad_norm": 4.939313888549805, "learning_rate": 4.533084927646343e-06, "loss": 0.11320143938064575, "step": 945 }, { "epoch": 0.6994454713493531, "grad_norm": 3.231778383255005, "learning_rate": 4.531989339979904e-06, "loss": 0.11996599286794662, "step": 946 }, { "epoch": 0.700184842883549, "grad_norm": 4.916162490844727, "learning_rate": 4.530892617631911e-06, "loss": 0.08725371211767197, "step": 947 }, { "epoch": 0.7009242144177449, "grad_norm": 4.063236236572266, "learning_rate": 4.529794761300704e-06, "loss": 0.11653236299753189, "step": 948 }, { "epoch": 0.7016635859519409, "grad_norm": 3.992332696914673, "learning_rate": 4.528695771685341e-06, "loss": 0.1441875398159027, "step": 949 }, { "epoch": 0.7024029574861368, "grad_norm": 4.271751403808594, "learning_rate": 4.527595649485603e-06, "loss": 0.11827435344457626, "step": 950 }, { "epoch": 0.7031423290203327, "grad_norm": 7.479649543762207, "learning_rate": 4.526494395401996e-06, "loss": 0.15513616800308228, "step": 951 }, { "epoch": 0.7038817005545287, "grad_norm": 2.8579702377319336, "learning_rate": 4.52539201013574e-06, "loss": 0.12465555965900421, "step": 952 }, { "epoch": 0.7046210720887246, "grad_norm": 6.537776947021484, "learning_rate": 4.52428849438878e-06, "loss": 0.20274174213409424, "step": 953 }, { "epoch": 0.7053604436229205, "grad_norm": 4.624680995941162, "learning_rate": 4.52318384886378e-06, "loss": 0.11686702817678452, "step": 954 }, { "epoch": 0.7060998151571165, "grad_norm": 6.075769901275635, "learning_rate": 4.522078074264122e-06, "loss": 0.15062734484672546, "step": 955 }, { "epoch": 0.7068391866913124, "grad_norm": 3.9062342643737793, "learning_rate": 4.520971171293909e-06, "loss": 0.1135726198554039, "step": 956 }, { "epoch": 0.7075785582255083, "grad_norm": 3.917595863342285, "learning_rate": 4.51986314065796e-06, "loss": 0.14332154393196106, "step": 957 }, { "epoch": 0.7083179297597042, "grad_norm": 3.163081407546997, "learning_rate": 4.518753983061814e-06, "loss": 0.13744205236434937, "step": 958 }, { "epoch": 0.7090573012939002, "grad_norm": 5.609212398529053, "learning_rate": 4.517643699211728e-06, "loss": 0.12954027950763702, "step": 959 }, { "epoch": 0.7097966728280961, "grad_norm": 4.273238658905029, "learning_rate": 4.516532289814674e-06, "loss": 0.13329319655895233, "step": 960 }, { "epoch": 0.710536044362292, "grad_norm": 2.7175307273864746, "learning_rate": 4.515419755578343e-06, "loss": 0.1089073121547699, "step": 961 }, { "epoch": 0.711275415896488, "grad_norm": 2.768371105194092, "learning_rate": 4.514306097211138e-06, "loss": 0.13509899377822876, "step": 962 }, { "epoch": 0.7120147874306839, "grad_norm": 4.836801528930664, "learning_rate": 4.513191315422184e-06, "loss": 0.1299794316291809, "step": 963 }, { "epoch": 0.7127541589648798, "grad_norm": 3.889954090118408, "learning_rate": 4.5120754109213185e-06, "loss": 0.14094999432563782, "step": 964 }, { "epoch": 0.7134935304990758, "grad_norm": 2.7400331497192383, "learning_rate": 4.510958384419091e-06, "loss": 0.08131387829780579, "step": 965 }, { "epoch": 0.7142329020332717, "grad_norm": 2.727675437927246, "learning_rate": 4.50984023662677e-06, "loss": 0.10789898782968521, "step": 966 }, { "epoch": 0.7149722735674676, "grad_norm": 3.480168342590332, "learning_rate": 4.508720968256335e-06, "loss": 0.15534761548042297, "step": 967 }, { "epoch": 0.7157116451016636, "grad_norm": 5.621164321899414, "learning_rate": 4.507600580020481e-06, "loss": 0.10739950835704803, "step": 968 }, { "epoch": 0.7164510166358595, "grad_norm": 4.695189952850342, "learning_rate": 4.506479072632614e-06, "loss": 0.12257075309753418, "step": 969 }, { "epoch": 0.7171903881700554, "grad_norm": 2.397080659866333, "learning_rate": 4.505356446806855e-06, "loss": 0.1081814393401146, "step": 970 }, { "epoch": 0.7179297597042514, "grad_norm": 3.0632669925689697, "learning_rate": 4.504232703258034e-06, "loss": 0.15462034940719604, "step": 971 }, { "epoch": 0.7186691312384473, "grad_norm": 2.5709829330444336, "learning_rate": 4.503107842701696e-06, "loss": 0.1404099464416504, "step": 972 }, { "epoch": 0.7194085027726432, "grad_norm": 3.408996105194092, "learning_rate": 4.501981865854096e-06, "loss": 0.10105183720588684, "step": 973 }, { "epoch": 0.7201478743068392, "grad_norm": 3.7999813556671143, "learning_rate": 4.500854773432199e-06, "loss": 0.09952839463949203, "step": 974 }, { "epoch": 0.7208872458410351, "grad_norm": 3.5327067375183105, "learning_rate": 4.499726566153682e-06, "loss": 0.1118134930729866, "step": 975 }, { "epoch": 0.721626617375231, "grad_norm": 3.47570538520813, "learning_rate": 4.49859724473693e-06, "loss": 0.09614289551973343, "step": 976 }, { "epoch": 0.7223659889094269, "grad_norm": 3.012096405029297, "learning_rate": 4.497466809901038e-06, "loss": 0.13388845324516296, "step": 977 }, { "epoch": 0.723105360443623, "grad_norm": 3.3246681690216064, "learning_rate": 4.496335262365811e-06, "loss": 0.15537630021572113, "step": 978 }, { "epoch": 0.7238447319778188, "grad_norm": 4.180583477020264, "learning_rate": 4.495202602851762e-06, "loss": 0.12088309973478317, "step": 979 }, { "epoch": 0.7245841035120147, "grad_norm": 4.137762546539307, "learning_rate": 4.494068832080111e-06, "loss": 0.14920644462108612, "step": 980 }, { "epoch": 0.7253234750462108, "grad_norm": 3.8928377628326416, "learning_rate": 4.492933950772786e-06, "loss": 0.1349623203277588, "step": 981 }, { "epoch": 0.7260628465804067, "grad_norm": 3.332488775253296, "learning_rate": 4.491797959652424e-06, "loss": 0.15278375148773193, "step": 982 }, { "epoch": 0.7268022181146025, "grad_norm": 4.639747619628906, "learning_rate": 4.490660859442368e-06, "loss": 0.13657760620117188, "step": 983 }, { "epoch": 0.7275415896487986, "grad_norm": 3.9793009757995605, "learning_rate": 4.489522650866663e-06, "loss": 0.11405781656503677, "step": 984 }, { "epoch": 0.7282809611829945, "grad_norm": 4.0632100105285645, "learning_rate": 4.4883833346500664e-06, "loss": 0.14820992946624756, "step": 985 }, { "epoch": 0.7290203327171904, "grad_norm": 2.9828133583068848, "learning_rate": 4.487242911518036e-06, "loss": 0.1161801666021347, "step": 986 }, { "epoch": 0.7297597042513864, "grad_norm": 3.2644383907318115, "learning_rate": 4.486101382196737e-06, "loss": 0.0412287563085556, "step": 987 }, { "epoch": 0.7304990757855823, "grad_norm": 3.360593318939209, "learning_rate": 4.484958747413038e-06, "loss": 0.10909684747457504, "step": 988 }, { "epoch": 0.7312384473197782, "grad_norm": 4.385203838348389, "learning_rate": 4.483815007894511e-06, "loss": 0.14807087182998657, "step": 989 }, { "epoch": 0.7319778188539742, "grad_norm": 3.8708910942077637, "learning_rate": 4.4826701643694335e-06, "loss": 0.13878567516803741, "step": 990 }, { "epoch": 0.7327171903881701, "grad_norm": 2.9203641414642334, "learning_rate": 4.481524217566783e-06, "loss": 0.14720484614372253, "step": 991 }, { "epoch": 0.733456561922366, "grad_norm": 4.6046905517578125, "learning_rate": 4.480377168216241e-06, "loss": 0.1157422885298729, "step": 992 }, { "epoch": 0.734195933456562, "grad_norm": 3.610866069793701, "learning_rate": 4.479229017048192e-06, "loss": 0.11628878861665726, "step": 993 }, { "epoch": 0.7349353049907579, "grad_norm": 3.1202094554901123, "learning_rate": 4.47807976479372e-06, "loss": 0.11745578050613403, "step": 994 }, { "epoch": 0.7356746765249538, "grad_norm": 2.968994617462158, "learning_rate": 4.476929412184613e-06, "loss": 0.1345207542181015, "step": 995 }, { "epoch": 0.7364140480591497, "grad_norm": 3.070164203643799, "learning_rate": 4.4757779599533565e-06, "loss": 0.16769666969776154, "step": 996 }, { "epoch": 0.7371534195933457, "grad_norm": 3.8851921558380127, "learning_rate": 4.474625408833138e-06, "loss": 0.10556286573410034, "step": 997 }, { "epoch": 0.7378927911275416, "grad_norm": 4.144081115722656, "learning_rate": 4.473471759557845e-06, "loss": 0.09202352166175842, "step": 998 }, { "epoch": 0.7386321626617375, "grad_norm": 4.622758865356445, "learning_rate": 4.472317012862062e-06, "loss": 0.15082745254039764, "step": 999 }, { "epoch": 0.7393715341959335, "grad_norm": 4.131690502166748, "learning_rate": 4.4711611694810755e-06, "loss": 0.0928441733121872, "step": 1000 }, { "epoch": 0.7401109057301294, "grad_norm": 4.318033695220947, "learning_rate": 4.4700042301508674e-06, "loss": 0.11782021820545197, "step": 1001 }, { "epoch": 0.7408502772643253, "grad_norm": 3.8475229740142822, "learning_rate": 4.468846195608119e-06, "loss": 0.1355893462896347, "step": 1002 }, { "epoch": 0.7415896487985213, "grad_norm": 3.302737236022949, "learning_rate": 4.467687066590209e-06, "loss": 0.09406667947769165, "step": 1003 }, { "epoch": 0.7423290203327172, "grad_norm": 3.169285535812378, "learning_rate": 4.466526843835214e-06, "loss": 0.1248103678226471, "step": 1004 }, { "epoch": 0.7430683918669131, "grad_norm": 3.8648622035980225, "learning_rate": 4.465365528081903e-06, "loss": 0.10561999678611755, "step": 1005 }, { "epoch": 0.7438077634011091, "grad_norm": 3.4125852584838867, "learning_rate": 4.464203120069746e-06, "loss": 0.08859232068061829, "step": 1006 }, { "epoch": 0.744547134935305, "grad_norm": 2.778193712234497, "learning_rate": 4.463039620538904e-06, "loss": 0.107968270778656, "step": 1007 }, { "epoch": 0.7452865064695009, "grad_norm": 4.290040969848633, "learning_rate": 4.461875030230238e-06, "loss": 0.07091116160154343, "step": 1008 }, { "epoch": 0.7460258780036969, "grad_norm": 5.645488262176514, "learning_rate": 4.460709349885298e-06, "loss": 0.11916011571884155, "step": 1009 }, { "epoch": 0.7467652495378928, "grad_norm": 2.6184372901916504, "learning_rate": 4.4595425802463326e-06, "loss": 0.15330058336257935, "step": 1010 }, { "epoch": 0.7475046210720887, "grad_norm": 3.3194804191589355, "learning_rate": 4.458374722056282e-06, "loss": 0.1511726677417755, "step": 1011 }, { "epoch": 0.7482439926062847, "grad_norm": 4.385641098022461, "learning_rate": 4.45720577605878e-06, "loss": 0.13421355187892914, "step": 1012 }, { "epoch": 0.7489833641404806, "grad_norm": 3.936274290084839, "learning_rate": 4.456035742998152e-06, "loss": 0.158488929271698, "step": 1013 }, { "epoch": 0.7497227356746765, "grad_norm": 3.2508444786071777, "learning_rate": 4.454864623619417e-06, "loss": 0.11031194031238556, "step": 1014 }, { "epoch": 0.7504621072088724, "grad_norm": 5.590580463409424, "learning_rate": 4.453692418668285e-06, "loss": 0.1707010567188263, "step": 1015 }, { "epoch": 0.7512014787430684, "grad_norm": 4.513060569763184, "learning_rate": 4.4525191288911575e-06, "loss": 0.15660960972309113, "step": 1016 }, { "epoch": 0.7519408502772643, "grad_norm": 3.635286331176758, "learning_rate": 4.451344755035126e-06, "loss": 0.10081051290035248, "step": 1017 }, { "epoch": 0.7526802218114602, "grad_norm": 4.105156421661377, "learning_rate": 4.450169297847975e-06, "loss": 0.14389489591121674, "step": 1018 }, { "epoch": 0.7534195933456562, "grad_norm": 3.5091850757598877, "learning_rate": 4.448992758078173e-06, "loss": 0.09724768996238708, "step": 1019 }, { "epoch": 0.7541589648798521, "grad_norm": 4.017787456512451, "learning_rate": 4.447815136474886e-06, "loss": 0.10924763977527618, "step": 1020 }, { "epoch": 0.754898336414048, "grad_norm": 2.4758358001708984, "learning_rate": 4.446636433787963e-06, "loss": 0.14961928129196167, "step": 1021 }, { "epoch": 0.755637707948244, "grad_norm": 4.248450756072998, "learning_rate": 4.44545665076794e-06, "loss": 0.07977153360843658, "step": 1022 }, { "epoch": 0.7563770794824399, "grad_norm": 2.389699935913086, "learning_rate": 4.444275788166047e-06, "loss": 0.10636080056428909, "step": 1023 }, { "epoch": 0.7571164510166358, "grad_norm": 2.88844633102417, "learning_rate": 4.443093846734197e-06, "loss": 0.12921306490898132, "step": 1024 }, { "epoch": 0.7578558225508318, "grad_norm": 2.9868359565734863, "learning_rate": 4.441910827224992e-06, "loss": 0.11540296673774719, "step": 1025 }, { "epoch": 0.7585951940850277, "grad_norm": 4.222764015197754, "learning_rate": 4.440726730391717e-06, "loss": 0.1593322455883026, "step": 1026 }, { "epoch": 0.7593345656192236, "grad_norm": 4.21931266784668, "learning_rate": 4.4395415569883485e-06, "loss": 0.13664610683918, "step": 1027 }, { "epoch": 0.7600739371534196, "grad_norm": 4.310160160064697, "learning_rate": 4.438355307769545e-06, "loss": 0.09575330466032028, "step": 1028 }, { "epoch": 0.7608133086876155, "grad_norm": 3.6469929218292236, "learning_rate": 4.437167983490649e-06, "loss": 0.13698357343673706, "step": 1029 }, { "epoch": 0.7615526802218114, "grad_norm": 3.6871531009674072, "learning_rate": 4.43597958490769e-06, "loss": 0.10674195736646652, "step": 1030 }, { "epoch": 0.7622920517560073, "grad_norm": 3.0672287940979004, "learning_rate": 4.4347901127773814e-06, "loss": 0.13451142609119415, "step": 1031 }, { "epoch": 0.7630314232902033, "grad_norm": 6.35833215713501, "learning_rate": 4.433599567857118e-06, "loss": 0.1453242003917694, "step": 1032 }, { "epoch": 0.7637707948243992, "grad_norm": 3.715650796890259, "learning_rate": 4.4324079509049805e-06, "loss": 0.12706822156906128, "step": 1033 }, { "epoch": 0.7645101663585951, "grad_norm": 3.641895055770874, "learning_rate": 4.431215262679731e-06, "loss": 0.13312478363513947, "step": 1034 }, { "epoch": 0.7652495378927912, "grad_norm": 3.34460186958313, "learning_rate": 4.430021503940812e-06, "loss": 0.098685622215271, "step": 1035 }, { "epoch": 0.765988909426987, "grad_norm": 4.137783527374268, "learning_rate": 4.428826675448351e-06, "loss": 0.09973786771297455, "step": 1036 }, { "epoch": 0.766728280961183, "grad_norm": 6.355085372924805, "learning_rate": 4.4276307779631535e-06, "loss": 0.14205795526504517, "step": 1037 }, { "epoch": 0.767467652495379, "grad_norm": 4.551486015319824, "learning_rate": 4.426433812246709e-06, "loss": 0.13335782289505005, "step": 1038 }, { "epoch": 0.7682070240295749, "grad_norm": 4.7518463134765625, "learning_rate": 4.425235779061184e-06, "loss": 0.08735991269350052, "step": 1039 }, { "epoch": 0.7689463955637708, "grad_norm": 4.884884357452393, "learning_rate": 4.424036679169425e-06, "loss": 0.1528855860233307, "step": 1040 }, { "epoch": 0.7696857670979668, "grad_norm": 3.0694806575775146, "learning_rate": 4.42283651333496e-06, "loss": 0.14952296018600464, "step": 1041 }, { "epoch": 0.7704251386321627, "grad_norm": 5.099454879760742, "learning_rate": 4.421635282321996e-06, "loss": 0.13428297638893127, "step": 1042 }, { "epoch": 0.7711645101663586, "grad_norm": 3.2187917232513428, "learning_rate": 4.420432986895415e-06, "loss": 0.11045783758163452, "step": 1043 }, { "epoch": 0.7719038817005546, "grad_norm": 4.289616584777832, "learning_rate": 4.419229627820777e-06, "loss": 0.10439257323741913, "step": 1044 }, { "epoch": 0.7726432532347505, "grad_norm": 4.351758003234863, "learning_rate": 4.418025205864324e-06, "loss": 0.09375796467065811, "step": 1045 }, { "epoch": 0.7733826247689464, "grad_norm": 3.8626911640167236, "learning_rate": 4.416819721792969e-06, "loss": 0.14191950857639313, "step": 1046 }, { "epoch": 0.7741219963031424, "grad_norm": 2.9727189540863037, "learning_rate": 4.415613176374306e-06, "loss": 0.13614343106746674, "step": 1047 }, { "epoch": 0.7748613678373383, "grad_norm": 4.687735557556152, "learning_rate": 4.414405570376601e-06, "loss": 0.11992155015468597, "step": 1048 }, { "epoch": 0.7756007393715342, "grad_norm": 3.6259965896606445, "learning_rate": 4.413196904568798e-06, "loss": 0.11179658770561218, "step": 1049 }, { "epoch": 0.7763401109057301, "grad_norm": 3.255819082260132, "learning_rate": 4.411987179720513e-06, "loss": 0.14662602543830872, "step": 1050 }, { "epoch": 0.7770794824399261, "grad_norm": 3.291825532913208, "learning_rate": 4.410776396602041e-06, "loss": 0.1281317174434662, "step": 1051 }, { "epoch": 0.777818853974122, "grad_norm": 4.145272254943848, "learning_rate": 4.409564555984347e-06, "loss": 0.11451210826635361, "step": 1052 }, { "epoch": 0.7785582255083179, "grad_norm": 3.7757816314697266, "learning_rate": 4.408351658639069e-06, "loss": 0.16336993873119354, "step": 1053 }, { "epoch": 0.7792975970425139, "grad_norm": 2.9482007026672363, "learning_rate": 4.407137705338521e-06, "loss": 0.11121069639921188, "step": 1054 }, { "epoch": 0.7800369685767098, "grad_norm": 2.5280957221984863, "learning_rate": 4.405922696855688e-06, "loss": 0.09960369765758514, "step": 1055 }, { "epoch": 0.7807763401109057, "grad_norm": 3.138637065887451, "learning_rate": 4.404706633964223e-06, "loss": 0.06405548751354218, "step": 1056 }, { "epoch": 0.7815157116451017, "grad_norm": 3.522660970687866, "learning_rate": 4.4034895174384575e-06, "loss": 0.09651178121566772, "step": 1057 }, { "epoch": 0.7822550831792976, "grad_norm": 3.7279775142669678, "learning_rate": 4.40227134805339e-06, "loss": 0.14216922223567963, "step": 1058 }, { "epoch": 0.7829944547134935, "grad_norm": 3.3054893016815186, "learning_rate": 4.401052126584689e-06, "loss": 0.11242197453975677, "step": 1059 }, { "epoch": 0.7837338262476895, "grad_norm": 3.506132125854492, "learning_rate": 4.399831853808694e-06, "loss": 0.15510648488998413, "step": 1060 }, { "epoch": 0.7844731977818854, "grad_norm": 4.938169479370117, "learning_rate": 4.398610530502414e-06, "loss": 0.0803990513086319, "step": 1061 }, { "epoch": 0.7852125693160813, "grad_norm": 4.779224395751953, "learning_rate": 4.397388157443524e-06, "loss": 0.09809288382530212, "step": 1062 }, { "epoch": 0.7859519408502773, "grad_norm": 4.612921714782715, "learning_rate": 4.396164735410372e-06, "loss": 0.1603783518075943, "step": 1063 }, { "epoch": 0.7866913123844732, "grad_norm": 4.075824737548828, "learning_rate": 4.394940265181973e-06, "loss": 0.10593071579933167, "step": 1064 }, { "epoch": 0.7874306839186691, "grad_norm": 3.6433308124542236, "learning_rate": 4.393714747538006e-06, "loss": 0.07010030746459961, "step": 1065 }, { "epoch": 0.7881700554528651, "grad_norm": 4.160884857177734, "learning_rate": 4.392488183258821e-06, "loss": 0.1257454752922058, "step": 1066 }, { "epoch": 0.788909426987061, "grad_norm": 4.0547637939453125, "learning_rate": 4.391260573125431e-06, "loss": 0.1045047864317894, "step": 1067 }, { "epoch": 0.7896487985212569, "grad_norm": 4.485244274139404, "learning_rate": 4.390031917919517e-06, "loss": 0.0846809670329094, "step": 1068 }, { "epoch": 0.7903881700554528, "grad_norm": 4.450351238250732, "learning_rate": 4.388802218423427e-06, "loss": 0.07518786191940308, "step": 1069 }, { "epoch": 0.7911275415896488, "grad_norm": 4.760242462158203, "learning_rate": 4.38757147542017e-06, "loss": 0.08815909922122955, "step": 1070 }, { "epoch": 0.7918669131238447, "grad_norm": 4.190087795257568, "learning_rate": 4.386339689693423e-06, "loss": 0.10284540057182312, "step": 1071 }, { "epoch": 0.7926062846580406, "grad_norm": 5.740088939666748, "learning_rate": 4.385106862027523e-06, "loss": 0.08730830997228622, "step": 1072 }, { "epoch": 0.7933456561922366, "grad_norm": 4.3095479011535645, "learning_rate": 4.383872993207476e-06, "loss": 0.07502824068069458, "step": 1073 }, { "epoch": 0.7940850277264325, "grad_norm": 5.062305450439453, "learning_rate": 4.382638084018946e-06, "loss": 0.06505836546421051, "step": 1074 }, { "epoch": 0.7948243992606284, "grad_norm": 5.639937877655029, "learning_rate": 4.381402135248262e-06, "loss": 0.07292391359806061, "step": 1075 }, { "epoch": 0.7955637707948244, "grad_norm": 5.596986770629883, "learning_rate": 4.380165147682414e-06, "loss": 0.16185149550437927, "step": 1076 }, { "epoch": 0.7963031423290203, "grad_norm": 4.430655002593994, "learning_rate": 4.3789271221090526e-06, "loss": 0.09099143743515015, "step": 1077 }, { "epoch": 0.7970425138632162, "grad_norm": 5.833934307098389, "learning_rate": 4.377688059316493e-06, "loss": 0.09594988822937012, "step": 1078 }, { "epoch": 0.7977818853974122, "grad_norm": 4.660792350769043, "learning_rate": 4.376447960093705e-06, "loss": 0.1363900750875473, "step": 1079 }, { "epoch": 0.7985212569316081, "grad_norm": 4.1517014503479, "learning_rate": 4.375206825230326e-06, "loss": 0.11537504941225052, "step": 1080 }, { "epoch": 0.799260628465804, "grad_norm": 3.193995237350464, "learning_rate": 4.373964655516644e-06, "loss": 0.13342374563217163, "step": 1081 }, { "epoch": 0.8, "grad_norm": 4.769655227661133, "learning_rate": 4.372721451743614e-06, "loss": 0.1343742460012436, "step": 1082 }, { "epoch": 0.8007393715341959, "grad_norm": 4.519454479217529, "learning_rate": 4.3714772147028435e-06, "loss": 0.12650378048419952, "step": 1083 }, { "epoch": 0.8014787430683918, "grad_norm": 3.2109344005584717, "learning_rate": 4.370231945186601e-06, "loss": 0.14660406112670898, "step": 1084 }, { "epoch": 0.8022181146025879, "grad_norm": 3.860455274581909, "learning_rate": 4.368985643987813e-06, "loss": 0.09913087636232376, "step": 1085 }, { "epoch": 0.8029574861367837, "grad_norm": 3.4974863529205322, "learning_rate": 4.36773831190006e-06, "loss": 0.07878021150827408, "step": 1086 }, { "epoch": 0.8036968576709796, "grad_norm": 4.235291957855225, "learning_rate": 4.366489949717582e-06, "loss": 0.09740541130304337, "step": 1087 }, { "epoch": 0.8044362292051755, "grad_norm": 5.459899425506592, "learning_rate": 4.365240558235273e-06, "loss": 0.1231757253408432, "step": 1088 }, { "epoch": 0.8051756007393716, "grad_norm": 3.76308012008667, "learning_rate": 4.363990138248682e-06, "loss": 0.12674230337142944, "step": 1089 }, { "epoch": 0.8059149722735675, "grad_norm": 5.744166851043701, "learning_rate": 4.362738690554014e-06, "loss": 0.14928017556667328, "step": 1090 }, { "epoch": 0.8066543438077634, "grad_norm": 4.7592644691467285, "learning_rate": 4.361486215948129e-06, "loss": 0.17292439937591553, "step": 1091 }, { "epoch": 0.8073937153419594, "grad_norm": 3.8683133125305176, "learning_rate": 4.36023271522854e-06, "loss": 0.11135274916887283, "step": 1092 }, { "epoch": 0.8081330868761553, "grad_norm": 3.470438003540039, "learning_rate": 4.358978189193414e-06, "loss": 0.09727613627910614, "step": 1093 }, { "epoch": 0.8088724584103512, "grad_norm": 4.6206254959106445, "learning_rate": 4.357722638641568e-06, "loss": 0.15000253915786743, "step": 1094 }, { "epoch": 0.8096118299445472, "grad_norm": 5.094419002532959, "learning_rate": 4.356466064372477e-06, "loss": 0.08137444406747818, "step": 1095 }, { "epoch": 0.8103512014787431, "grad_norm": 4.375420570373535, "learning_rate": 4.355208467186262e-06, "loss": 0.10380066186189651, "step": 1096 }, { "epoch": 0.811090573012939, "grad_norm": 4.179043292999268, "learning_rate": 4.353949847883699e-06, "loss": 0.11566860228776932, "step": 1097 }, { "epoch": 0.811829944547135, "grad_norm": 4.208347320556641, "learning_rate": 4.352690207266214e-06, "loss": 0.09243634343147278, "step": 1098 }, { "epoch": 0.8125693160813309, "grad_norm": 3.22833514213562, "learning_rate": 4.351429546135882e-06, "loss": 0.11217519640922546, "step": 1099 }, { "epoch": 0.8133086876155268, "grad_norm": 4.24362850189209, "learning_rate": 4.35016786529543e-06, "loss": 0.14460645616054535, "step": 1100 }, { "epoch": 0.8140480591497228, "grad_norm": 5.144036769866943, "learning_rate": 4.348905165548232e-06, "loss": 0.09959906339645386, "step": 1101 }, { "epoch": 0.8147874306839187, "grad_norm": 4.998846530914307, "learning_rate": 4.347641447698312e-06, "loss": 0.10640853643417358, "step": 1102 }, { "epoch": 0.8155268022181146, "grad_norm": 4.6822919845581055, "learning_rate": 4.346376712550343e-06, "loss": 0.08978576213121414, "step": 1103 }, { "epoch": 0.8162661737523106, "grad_norm": 3.7299253940582275, "learning_rate": 4.345110960909646e-06, "loss": 0.11477670818567276, "step": 1104 }, { "epoch": 0.8170055452865065, "grad_norm": 4.123977184295654, "learning_rate": 4.343844193582186e-06, "loss": 0.13858568668365479, "step": 1105 }, { "epoch": 0.8177449168207024, "grad_norm": 3.7561981678009033, "learning_rate": 4.342576411374578e-06, "loss": 0.12420995533466339, "step": 1106 }, { "epoch": 0.8184842883548983, "grad_norm": 5.773029327392578, "learning_rate": 4.341307615094083e-06, "loss": 0.13800518214702606, "step": 1107 }, { "epoch": 0.8192236598890943, "grad_norm": 4.099851131439209, "learning_rate": 4.340037805548604e-06, "loss": 0.1121729239821434, "step": 1108 }, { "epoch": 0.8199630314232902, "grad_norm": 5.013585090637207, "learning_rate": 4.338766983546696e-06, "loss": 0.05948600172996521, "step": 1109 }, { "epoch": 0.8207024029574861, "grad_norm": 4.083653450012207, "learning_rate": 4.337495149897554e-06, "loss": 0.12183589488267899, "step": 1110 }, { "epoch": 0.8214417744916821, "grad_norm": 4.592380046844482, "learning_rate": 4.336222305411015e-06, "loss": 0.1189546212553978, "step": 1111 }, { "epoch": 0.822181146025878, "grad_norm": 4.099369525909424, "learning_rate": 4.334948450897566e-06, "loss": 0.12251915037631989, "step": 1112 }, { "epoch": 0.8229205175600739, "grad_norm": 5.412926197052002, "learning_rate": 4.333673587168331e-06, "loss": 0.08992078900337219, "step": 1113 }, { "epoch": 0.8236598890942699, "grad_norm": 3.350243330001831, "learning_rate": 4.332397715035082e-06, "loss": 0.13348829746246338, "step": 1114 }, { "epoch": 0.8243992606284658, "grad_norm": 4.499497890472412, "learning_rate": 4.331120835310228e-06, "loss": 0.10105817019939423, "step": 1115 }, { "epoch": 0.8251386321626617, "grad_norm": 4.149452209472656, "learning_rate": 4.329842948806824e-06, "loss": 0.10513957589864731, "step": 1116 }, { "epoch": 0.8258780036968577, "grad_norm": 4.505613327026367, "learning_rate": 4.328564056338564e-06, "loss": 0.09821823239326477, "step": 1117 }, { "epoch": 0.8266173752310536, "grad_norm": 4.835047721862793, "learning_rate": 4.32728415871978e-06, "loss": 0.10941477119922638, "step": 1118 }, { "epoch": 0.8273567467652495, "grad_norm": 5.696407318115234, "learning_rate": 4.326003256765451e-06, "loss": 0.1524343490600586, "step": 1119 }, { "epoch": 0.8280961182994455, "grad_norm": 4.641087532043457, "learning_rate": 4.3247213512911865e-06, "loss": 0.09340010583400726, "step": 1120 }, { "epoch": 0.8288354898336414, "grad_norm": 4.749405384063721, "learning_rate": 4.323438443113244e-06, "loss": 0.06019987538456917, "step": 1121 }, { "epoch": 0.8295748613678373, "grad_norm": 4.346889972686768, "learning_rate": 4.322154533048513e-06, "loss": 0.10649069398641586, "step": 1122 }, { "epoch": 0.8303142329020333, "grad_norm": 3.9889936447143555, "learning_rate": 4.320869621914523e-06, "loss": 0.11654748022556305, "step": 1123 }, { "epoch": 0.8310536044362292, "grad_norm": 3.4693710803985596, "learning_rate": 4.319583710529441e-06, "loss": 0.0825081616640091, "step": 1124 }, { "epoch": 0.8317929759704251, "grad_norm": 3.307853937149048, "learning_rate": 4.318296799712072e-06, "loss": 0.10760924220085144, "step": 1125 }, { "epoch": 0.832532347504621, "grad_norm": 3.3743722438812256, "learning_rate": 4.3170088902818555e-06, "loss": 0.11097480356693268, "step": 1126 }, { "epoch": 0.833271719038817, "grad_norm": 3.204616069793701, "learning_rate": 4.315719983058869e-06, "loss": 0.1082085445523262, "step": 1127 }, { "epoch": 0.8340110905730129, "grad_norm": 3.353914260864258, "learning_rate": 4.314430078863822e-06, "loss": 0.1267772912979126, "step": 1128 }, { "epoch": 0.8347504621072088, "grad_norm": 4.226475715637207, "learning_rate": 4.313139178518061e-06, "loss": 0.059496477246284485, "step": 1129 }, { "epoch": 0.8354898336414048, "grad_norm": 3.5461621284484863, "learning_rate": 4.311847282843569e-06, "loss": 0.07821174710988998, "step": 1130 }, { "epoch": 0.8362292051756007, "grad_norm": 4.146816253662109, "learning_rate": 4.310554392662959e-06, "loss": 0.09618723392486572, "step": 1131 }, { "epoch": 0.8369685767097966, "grad_norm": 4.0911641120910645, "learning_rate": 4.309260508799478e-06, "loss": 0.04347590357065201, "step": 1132 }, { "epoch": 0.8377079482439926, "grad_norm": 4.65231466293335, "learning_rate": 4.307965632077008e-06, "loss": 0.10232586413621902, "step": 1133 }, { "epoch": 0.8384473197781885, "grad_norm": 5.217278003692627, "learning_rate": 4.30666976332006e-06, "loss": 0.08495929837226868, "step": 1134 }, { "epoch": 0.8391866913123844, "grad_norm": 4.0834174156188965, "learning_rate": 4.30537290335378e-06, "loss": 0.07926441729068756, "step": 1135 }, { "epoch": 0.8399260628465804, "grad_norm": 5.392715930938721, "learning_rate": 4.304075053003941e-06, "loss": 0.14881646633148193, "step": 1136 }, { "epoch": 0.8406654343807763, "grad_norm": 4.900696277618408, "learning_rate": 4.302776213096953e-06, "loss": 0.1625932902097702, "step": 1137 }, { "epoch": 0.8414048059149722, "grad_norm": 5.491786956787109, "learning_rate": 4.301476384459849e-06, "loss": 0.1250731647014618, "step": 1138 }, { "epoch": 0.8421441774491683, "grad_norm": 4.5717291831970215, "learning_rate": 4.300175567920295e-06, "loss": 0.08523121476173401, "step": 1139 }, { "epoch": 0.8428835489833642, "grad_norm": 2.773526430130005, "learning_rate": 4.2988737643065875e-06, "loss": 0.14581814408302307, "step": 1140 }, { "epoch": 0.84362292051756, "grad_norm": 3.085679769515991, "learning_rate": 4.297570974447648e-06, "loss": 0.0958675891160965, "step": 1141 }, { "epoch": 0.8443622920517561, "grad_norm": 2.998152494430542, "learning_rate": 4.2962671991730285e-06, "loss": 0.13574102520942688, "step": 1142 }, { "epoch": 0.845101663585952, "grad_norm": 4.334967136383057, "learning_rate": 4.2949624393129086e-06, "loss": 0.09472116082906723, "step": 1143 }, { "epoch": 0.8458410351201479, "grad_norm": 3.11582612991333, "learning_rate": 4.293656695698093e-06, "loss": 0.11980532854795456, "step": 1144 }, { "epoch": 0.8465804066543438, "grad_norm": 4.6548590660095215, "learning_rate": 4.292349969160014e-06, "loss": 0.08757999539375305, "step": 1145 }, { "epoch": 0.8473197781885398, "grad_norm": 4.42149543762207, "learning_rate": 4.291042260530731e-06, "loss": 0.11318723857402802, "step": 1146 }, { "epoch": 0.8480591497227357, "grad_norm": 4.860963821411133, "learning_rate": 4.289733570642925e-06, "loss": 0.0971972644329071, "step": 1147 }, { "epoch": 0.8487985212569316, "grad_norm": 3.4016273021698, "learning_rate": 4.288423900329905e-06, "loss": 0.11181987822055817, "step": 1148 }, { "epoch": 0.8495378927911276, "grad_norm": 4.837772846221924, "learning_rate": 4.287113250425604e-06, "loss": 0.15177515149116516, "step": 1149 }, { "epoch": 0.8502772643253235, "grad_norm": 5.395931243896484, "learning_rate": 4.285801621764578e-06, "loss": 0.11296069622039795, "step": 1150 }, { "epoch": 0.8510166358595194, "grad_norm": 3.791879177093506, "learning_rate": 4.284489015182007e-06, "loss": 0.08023515343666077, "step": 1151 }, { "epoch": 0.8517560073937154, "grad_norm": 4.529984474182129, "learning_rate": 4.283175431513693e-06, "loss": 0.1374194175004959, "step": 1152 }, { "epoch": 0.8524953789279113, "grad_norm": 4.568609237670898, "learning_rate": 4.2818608715960584e-06, "loss": 0.11850523948669434, "step": 1153 }, { "epoch": 0.8532347504621072, "grad_norm": 3.4900598526000977, "learning_rate": 4.280545336266151e-06, "loss": 0.13443690538406372, "step": 1154 }, { "epoch": 0.8539741219963032, "grad_norm": 3.5371007919311523, "learning_rate": 4.279228826361638e-06, "loss": 0.12925179302692413, "step": 1155 }, { "epoch": 0.8547134935304991, "grad_norm": 4.309397220611572, "learning_rate": 4.2779113427208056e-06, "loss": 0.12049844861030579, "step": 1156 }, { "epoch": 0.855452865064695, "grad_norm": 3.443669319152832, "learning_rate": 4.276592886182562e-06, "loss": 0.12183625251054764, "step": 1157 }, { "epoch": 0.856192236598891, "grad_norm": 3.2949330806732178, "learning_rate": 4.275273457586435e-06, "loss": 0.08956754207611084, "step": 1158 }, { "epoch": 0.8569316081330869, "grad_norm": 5.292552471160889, "learning_rate": 4.273953057772569e-06, "loss": 0.06444615125656128, "step": 1159 }, { "epoch": 0.8576709796672828, "grad_norm": 3.629683494567871, "learning_rate": 4.272631687581731e-06, "loss": 0.1334766298532486, "step": 1160 }, { "epoch": 0.8584103512014788, "grad_norm": 3.397022008895874, "learning_rate": 4.2713093478553e-06, "loss": 0.09166721254587173, "step": 1161 }, { "epoch": 0.8591497227356747, "grad_norm": 4.713347911834717, "learning_rate": 4.269986039435279e-06, "loss": 0.10261302441358566, "step": 1162 }, { "epoch": 0.8598890942698706, "grad_norm": 4.277237415313721, "learning_rate": 4.268661763164283e-06, "loss": 0.0771171897649765, "step": 1163 }, { "epoch": 0.8606284658040665, "grad_norm": 4.085428714752197, "learning_rate": 4.267336519885544e-06, "loss": 0.06878191232681274, "step": 1164 }, { "epoch": 0.8613678373382625, "grad_norm": 4.007308483123779, "learning_rate": 4.266010310442912e-06, "loss": 0.07790796458721161, "step": 1165 }, { "epoch": 0.8621072088724584, "grad_norm": 5.290504455566406, "learning_rate": 4.26468313568085e-06, "loss": 0.12028934806585312, "step": 1166 }, { "epoch": 0.8628465804066543, "grad_norm": 4.15550422668457, "learning_rate": 4.263354996444436e-06, "loss": 0.09509509801864624, "step": 1167 }, { "epoch": 0.8635859519408503, "grad_norm": 4.0749053955078125, "learning_rate": 4.262025893579364e-06, "loss": 0.09429234266281128, "step": 1168 }, { "epoch": 0.8643253234750462, "grad_norm": 4.43678092956543, "learning_rate": 4.260695827931939e-06, "loss": 0.08479442447423935, "step": 1169 }, { "epoch": 0.8650646950092421, "grad_norm": 4.59469747543335, "learning_rate": 4.25936480034908e-06, "loss": 0.07726822793483734, "step": 1170 }, { "epoch": 0.8658040665434381, "grad_norm": 5.544498443603516, "learning_rate": 4.258032811678318e-06, "loss": 0.13192571699619293, "step": 1171 }, { "epoch": 0.866543438077634, "grad_norm": 4.242419242858887, "learning_rate": 4.256699862767799e-06, "loss": 0.1212301105260849, "step": 1172 }, { "epoch": 0.8672828096118299, "grad_norm": 3.2229247093200684, "learning_rate": 4.255365954466277e-06, "loss": 0.14079955220222473, "step": 1173 }, { "epoch": 0.8680221811460259, "grad_norm": 3.5402562618255615, "learning_rate": 4.254031087623118e-06, "loss": 0.14209944009780884, "step": 1174 }, { "epoch": 0.8687615526802218, "grad_norm": 4.389798641204834, "learning_rate": 4.2526952630882975e-06, "loss": 0.11701329797506332, "step": 1175 }, { "epoch": 0.8695009242144177, "grad_norm": 3.4088661670684814, "learning_rate": 4.251358481712403e-06, "loss": 0.13094139099121094, "step": 1176 }, { "epoch": 0.8702402957486137, "grad_norm": 3.6841607093811035, "learning_rate": 4.2500207443466294e-06, "loss": 0.12017621099948883, "step": 1177 }, { "epoch": 0.8709796672828096, "grad_norm": 3.584477424621582, "learning_rate": 4.24868205184278e-06, "loss": 0.14217346906661987, "step": 1178 }, { "epoch": 0.8717190388170055, "grad_norm": 4.014596939086914, "learning_rate": 4.247342405053268e-06, "loss": 0.1099638119339943, "step": 1179 }, { "epoch": 0.8724584103512015, "grad_norm": 3.8312196731567383, "learning_rate": 4.246001804831114e-06, "loss": 0.08588013052940369, "step": 1180 }, { "epoch": 0.8731977818853974, "grad_norm": 2.865461826324463, "learning_rate": 4.244660252029943e-06, "loss": 0.12784135341644287, "step": 1181 }, { "epoch": 0.8739371534195933, "grad_norm": 4.5479888916015625, "learning_rate": 4.24331774750399e-06, "loss": 0.11991754174232483, "step": 1182 }, { "epoch": 0.8746765249537892, "grad_norm": 4.578449249267578, "learning_rate": 4.2419742921080955e-06, "loss": 0.09516175091266632, "step": 1183 }, { "epoch": 0.8754158964879852, "grad_norm": 4.088892459869385, "learning_rate": 4.240629886697703e-06, "loss": 0.13725392520427704, "step": 1184 }, { "epoch": 0.8761552680221811, "grad_norm": 3.672952175140381, "learning_rate": 4.239284532128864e-06, "loss": 0.11205307394266129, "step": 1185 }, { "epoch": 0.876894639556377, "grad_norm": 4.5370192527771, "learning_rate": 4.237938229258231e-06, "loss": 0.06802000105381012, "step": 1186 }, { "epoch": 0.877634011090573, "grad_norm": 3.404069423675537, "learning_rate": 4.236590978943064e-06, "loss": 0.09558530151844025, "step": 1187 }, { "epoch": 0.8783733826247689, "grad_norm": 5.77680778503418, "learning_rate": 4.2352427820412235e-06, "loss": 0.12876060605049133, "step": 1188 }, { "epoch": 0.8791127541589648, "grad_norm": 3.4506001472473145, "learning_rate": 4.233893639411174e-06, "loss": 0.06644639372825623, "step": 1189 }, { "epoch": 0.8798521256931608, "grad_norm": 4.216806888580322, "learning_rate": 4.232543551911983e-06, "loss": 0.09841654449701309, "step": 1190 }, { "epoch": 0.8805914972273567, "grad_norm": 3.168968439102173, "learning_rate": 4.231192520403317e-06, "loss": 0.1014893501996994, "step": 1191 }, { "epoch": 0.8813308687615526, "grad_norm": 3.7611169815063477, "learning_rate": 4.2298405457454455e-06, "loss": 0.12180839478969574, "step": 1192 }, { "epoch": 0.8820702402957487, "grad_norm": 3.637988567352295, "learning_rate": 4.228487628799239e-06, "loss": 0.08095245808362961, "step": 1193 }, { "epoch": 0.8828096118299446, "grad_norm": 3.6542165279388428, "learning_rate": 4.227133770426166e-06, "loss": 0.0929374247789383, "step": 1194 }, { "epoch": 0.8835489833641405, "grad_norm": 4.065385818481445, "learning_rate": 4.225778971488297e-06, "loss": 0.09481123089790344, "step": 1195 }, { "epoch": 0.8842883548983365, "grad_norm": 3.878265857696533, "learning_rate": 4.2244232328483e-06, "loss": 0.08546587079763412, "step": 1196 }, { "epoch": 0.8850277264325324, "grad_norm": 3.068464994430542, "learning_rate": 4.223066555369442e-06, "loss": 0.08463460952043533, "step": 1197 }, { "epoch": 0.8857670979667283, "grad_norm": 3.874769449234009, "learning_rate": 4.221708939915585e-06, "loss": 0.10389642417430878, "step": 1198 }, { "epoch": 0.8865064695009243, "grad_norm": 6.3651275634765625, "learning_rate": 4.220350387351192e-06, "loss": 0.10650073736906052, "step": 1199 }, { "epoch": 0.8872458410351202, "grad_norm": 3.959449052810669, "learning_rate": 4.2189908985413205e-06, "loss": 0.13688693940639496, "step": 1200 }, { "epoch": 0.8879852125693161, "grad_norm": 4.233499050140381, "learning_rate": 4.217630474351626e-06, "loss": 0.10445420444011688, "step": 1201 }, { "epoch": 0.888724584103512, "grad_norm": 4.601794719696045, "learning_rate": 4.2162691156483585e-06, "loss": 0.1194075345993042, "step": 1202 }, { "epoch": 0.889463955637708, "grad_norm": 4.140490531921387, "learning_rate": 4.214906823298362e-06, "loss": 0.06718821823596954, "step": 1203 }, { "epoch": 0.8902033271719039, "grad_norm": 4.659821033477783, "learning_rate": 4.213543598169077e-06, "loss": 0.11478680372238159, "step": 1204 }, { "epoch": 0.8909426987060998, "grad_norm": 5.876899719238281, "learning_rate": 4.212179441128536e-06, "loss": 0.11243575066328049, "step": 1205 }, { "epoch": 0.8916820702402958, "grad_norm": 4.685629367828369, "learning_rate": 4.210814353045366e-06, "loss": 0.10399587452411652, "step": 1206 }, { "epoch": 0.8924214417744917, "grad_norm": 4.908453464508057, "learning_rate": 4.209448334788787e-06, "loss": 0.08657746016979218, "step": 1207 }, { "epoch": 0.8931608133086876, "grad_norm": 4.820060729980469, "learning_rate": 4.2080813872286116e-06, "loss": 0.10951615869998932, "step": 1208 }, { "epoch": 0.8939001848428836, "grad_norm": 5.461979866027832, "learning_rate": 4.206713511235243e-06, "loss": 0.08679677546024323, "step": 1209 }, { "epoch": 0.8946395563770795, "grad_norm": 4.82766056060791, "learning_rate": 4.205344707679676e-06, "loss": 0.11240117251873016, "step": 1210 }, { "epoch": 0.8953789279112754, "grad_norm": 6.631044864654541, "learning_rate": 4.203974977433497e-06, "loss": 0.12286077439785004, "step": 1211 }, { "epoch": 0.8961182994454714, "grad_norm": 4.5834832191467285, "learning_rate": 4.2026043213688796e-06, "loss": 0.1155431866645813, "step": 1212 }, { "epoch": 0.8968576709796673, "grad_norm": 3.32370662689209, "learning_rate": 4.201232740358591e-06, "loss": 0.09225167334079742, "step": 1213 }, { "epoch": 0.8975970425138632, "grad_norm": 4.695135116577148, "learning_rate": 4.199860235275985e-06, "loss": 0.09303955733776093, "step": 1214 }, { "epoch": 0.8983364140480592, "grad_norm": 3.3509750366210938, "learning_rate": 4.198486806995003e-06, "loss": 0.12233513593673706, "step": 1215 }, { "epoch": 0.8990757855822551, "grad_norm": 4.982052803039551, "learning_rate": 4.197112456390176e-06, "loss": 0.10937395691871643, "step": 1216 }, { "epoch": 0.899815157116451, "grad_norm": 3.9571213722229004, "learning_rate": 4.195737184336621e-06, "loss": 0.09689034521579742, "step": 1217 }, { "epoch": 0.900554528650647, "grad_norm": 5.1216816902160645, "learning_rate": 4.194360991710043e-06, "loss": 0.12414836883544922, "step": 1218 }, { "epoch": 0.9012939001848429, "grad_norm": 4.9327287673950195, "learning_rate": 4.192983879386733e-06, "loss": 0.09931027889251709, "step": 1219 }, { "epoch": 0.9020332717190388, "grad_norm": 4.988483905792236, "learning_rate": 4.191605848243567e-06, "loss": 0.10050538182258606, "step": 1220 }, { "epoch": 0.9027726432532347, "grad_norm": 3.5163230895996094, "learning_rate": 4.190226899158005e-06, "loss": 0.09773328900337219, "step": 1221 }, { "epoch": 0.9035120147874307, "grad_norm": 4.393080234527588, "learning_rate": 4.188847033008093e-06, "loss": 0.10089230537414551, "step": 1222 }, { "epoch": 0.9042513863216266, "grad_norm": 4.85678243637085, "learning_rate": 4.187466250672462e-06, "loss": 0.08683080971240997, "step": 1223 }, { "epoch": 0.9049907578558225, "grad_norm": 3.0256028175354004, "learning_rate": 4.186084553030324e-06, "loss": 0.09773094952106476, "step": 1224 }, { "epoch": 0.9057301293900185, "grad_norm": 4.333364009857178, "learning_rate": 4.1847019409614745e-06, "loss": 0.10258297622203827, "step": 1225 }, { "epoch": 0.9064695009242144, "grad_norm": 3.728494167327881, "learning_rate": 4.183318415346293e-06, "loss": 0.077602319419384, "step": 1226 }, { "epoch": 0.9072088724584103, "grad_norm": 4.309238910675049, "learning_rate": 4.181933977065739e-06, "loss": 0.1436232477426529, "step": 1227 }, { "epoch": 0.9079482439926063, "grad_norm": 4.306010723114014, "learning_rate": 4.1805486270013505e-06, "loss": 0.10629457235336304, "step": 1228 }, { "epoch": 0.9086876155268022, "grad_norm": 3.8146021366119385, "learning_rate": 4.179162366035253e-06, "loss": 0.13241106271743774, "step": 1229 }, { "epoch": 0.9094269870609981, "grad_norm": 3.382673501968384, "learning_rate": 4.177775195050147e-06, "loss": 0.08619637042284012, "step": 1230 }, { "epoch": 0.9101663585951941, "grad_norm": 3.5418856143951416, "learning_rate": 4.176387114929312e-06, "loss": 0.11826753616333008, "step": 1231 }, { "epoch": 0.91090573012939, "grad_norm": 3.7916886806488037, "learning_rate": 4.17499812655661e-06, "loss": 0.06805799156427383, "step": 1232 }, { "epoch": 0.9116451016635859, "grad_norm": 4.111300945281982, "learning_rate": 4.173608230816477e-06, "loss": 0.10466843843460083, "step": 1233 }, { "epoch": 0.9123844731977819, "grad_norm": 3.663238048553467, "learning_rate": 4.172217428593931e-06, "loss": 0.09523216634988785, "step": 1234 }, { "epoch": 0.9131238447319778, "grad_norm": 3.9000613689422607, "learning_rate": 4.1708257207745626e-06, "loss": 0.06594498455524445, "step": 1235 }, { "epoch": 0.9138632162661737, "grad_norm": 4.055272579193115, "learning_rate": 4.169433108244544e-06, "loss": 0.08028724044561386, "step": 1236 }, { "epoch": 0.9146025878003697, "grad_norm": 3.842741012573242, "learning_rate": 4.168039591890621e-06, "loss": 0.09036500751972198, "step": 1237 }, { "epoch": 0.9153419593345656, "grad_norm": 3.715682029724121, "learning_rate": 4.166645172600114e-06, "loss": 0.08146093785762787, "step": 1238 }, { "epoch": 0.9160813308687615, "grad_norm": 3.7646493911743164, "learning_rate": 4.165249851260921e-06, "loss": 0.09951554983854294, "step": 1239 }, { "epoch": 0.9168207024029574, "grad_norm": 3.493722438812256, "learning_rate": 4.1638536287615106e-06, "loss": 0.11436507105827332, "step": 1240 }, { "epoch": 0.9175600739371534, "grad_norm": 5.539567470550537, "learning_rate": 4.162456505990927e-06, "loss": 0.0811745822429657, "step": 1241 }, { "epoch": 0.9182994454713493, "grad_norm": 4.578609943389893, "learning_rate": 4.1610584838387914e-06, "loss": 0.0626993328332901, "step": 1242 }, { "epoch": 0.9190388170055452, "grad_norm": 5.424917697906494, "learning_rate": 4.159659563195291e-06, "loss": 0.09291280806064606, "step": 1243 }, { "epoch": 0.9197781885397412, "grad_norm": 5.393265724182129, "learning_rate": 4.15825974495119e-06, "loss": 0.1047685369849205, "step": 1244 }, { "epoch": 0.9205175600739371, "grad_norm": 6.572983741760254, "learning_rate": 4.156859029997822e-06, "loss": 0.08550649881362915, "step": 1245 }, { "epoch": 0.921256931608133, "grad_norm": 4.584841251373291, "learning_rate": 4.155457419227092e-06, "loss": 0.0753939226269722, "step": 1246 }, { "epoch": 0.921996303142329, "grad_norm": 3.9199461936950684, "learning_rate": 4.154054913531477e-06, "loss": 0.10660315304994583, "step": 1247 }, { "epoch": 0.922735674676525, "grad_norm": 4.957601547241211, "learning_rate": 4.152651513804019e-06, "loss": 0.08097516000270844, "step": 1248 }, { "epoch": 0.9234750462107209, "grad_norm": 4.350642204284668, "learning_rate": 4.151247220938336e-06, "loss": 0.087107352912426, "step": 1249 }, { "epoch": 0.9242144177449169, "grad_norm": 4.438808917999268, "learning_rate": 4.14984203582861e-06, "loss": 0.12166370451450348, "step": 1250 }, { "epoch": 0.9249537892791128, "grad_norm": 4.26732873916626, "learning_rate": 4.1484359593695915e-06, "loss": 0.07875093817710876, "step": 1251 }, { "epoch": 0.9256931608133087, "grad_norm": 3.9393012523651123, "learning_rate": 4.147028992456602e-06, "loss": 0.08533179759979248, "step": 1252 }, { "epoch": 0.9264325323475047, "grad_norm": 4.251672744750977, "learning_rate": 4.145621135985526e-06, "loss": 0.09056583046913147, "step": 1253 }, { "epoch": 0.9271719038817006, "grad_norm": 6.198204040527344, "learning_rate": 4.144212390852816e-06, "loss": 0.08777090162038803, "step": 1254 }, { "epoch": 0.9279112754158965, "grad_norm": 5.609645366668701, "learning_rate": 4.142802757955491e-06, "loss": 0.11775058507919312, "step": 1255 }, { "epoch": 0.9286506469500925, "grad_norm": 5.441091537475586, "learning_rate": 4.141392238191134e-06, "loss": 0.10015446692705154, "step": 1256 }, { "epoch": 0.9293900184842884, "grad_norm": 4.611072063446045, "learning_rate": 4.139980832457893e-06, "loss": 0.08257386088371277, "step": 1257 }, { "epoch": 0.9301293900184843, "grad_norm": 3.8205137252807617, "learning_rate": 4.138568541654481e-06, "loss": 0.09945394098758698, "step": 1258 }, { "epoch": 0.9308687615526802, "grad_norm": 4.229599475860596, "learning_rate": 4.137155366680176e-06, "loss": 0.10981465876102448, "step": 1259 }, { "epoch": 0.9316081330868762, "grad_norm": 4.974471092224121, "learning_rate": 4.135741308434812e-06, "loss": 0.061925265938043594, "step": 1260 }, { "epoch": 0.9323475046210721, "grad_norm": 4.433976173400879, "learning_rate": 4.134326367818795e-06, "loss": 0.10514142364263535, "step": 1261 }, { "epoch": 0.933086876155268, "grad_norm": 3.386148691177368, "learning_rate": 4.132910545733087e-06, "loss": 0.06907913088798523, "step": 1262 }, { "epoch": 0.933826247689464, "grad_norm": 3.217339515686035, "learning_rate": 4.131493843079213e-06, "loss": 0.1272631287574768, "step": 1263 }, { "epoch": 0.9345656192236599, "grad_norm": 5.5113205909729, "learning_rate": 4.130076260759256e-06, "loss": 0.09658055007457733, "step": 1264 }, { "epoch": 0.9353049907578558, "grad_norm": 3.9208343029022217, "learning_rate": 4.128657799675864e-06, "loss": 0.0945916548371315, "step": 1265 }, { "epoch": 0.9360443622920518, "grad_norm": 4.007211208343506, "learning_rate": 4.127238460732242e-06, "loss": 0.07331337034702301, "step": 1266 }, { "epoch": 0.9367837338262477, "grad_norm": 4.3939433097839355, "learning_rate": 4.1258182448321515e-06, "loss": 0.09612911939620972, "step": 1267 }, { "epoch": 0.9375231053604436, "grad_norm": 5.030735969543457, "learning_rate": 4.124397152879918e-06, "loss": 0.09950044751167297, "step": 1268 }, { "epoch": 0.9382624768946396, "grad_norm": 3.539287805557251, "learning_rate": 4.122975185780419e-06, "loss": 0.09777024388313293, "step": 1269 }, { "epoch": 0.9390018484288355, "grad_norm": 4.895534515380859, "learning_rate": 4.121552344439093e-06, "loss": 0.03619559854269028, "step": 1270 }, { "epoch": 0.9397412199630314, "grad_norm": 4.7412590980529785, "learning_rate": 4.120128629761935e-06, "loss": 0.05294866859912872, "step": 1271 }, { "epoch": 0.9404805914972274, "grad_norm": 4.485791206359863, "learning_rate": 4.118704042655493e-06, "loss": 0.12009914964437485, "step": 1272 }, { "epoch": 0.9412199630314233, "grad_norm": 5.124727249145508, "learning_rate": 4.1172785840268744e-06, "loss": 0.04898155853152275, "step": 1273 }, { "epoch": 0.9419593345656192, "grad_norm": 5.7515869140625, "learning_rate": 4.115852254783741e-06, "loss": 0.07676545530557632, "step": 1274 }, { "epoch": 0.9426987060998152, "grad_norm": 5.201990604400635, "learning_rate": 4.114425055834304e-06, "loss": 0.06933489441871643, "step": 1275 }, { "epoch": 0.9434380776340111, "grad_norm": 6.6892547607421875, "learning_rate": 4.112996988087336e-06, "loss": 0.13698887825012207, "step": 1276 }, { "epoch": 0.944177449168207, "grad_norm": 5.98723840713501, "learning_rate": 4.111568052452156e-06, "loss": 0.07249816507101059, "step": 1277 }, { "epoch": 0.9449168207024029, "grad_norm": 3.9719531536102295, "learning_rate": 4.1101382498386405e-06, "loss": 0.09347586333751678, "step": 1278 }, { "epoch": 0.9456561922365989, "grad_norm": 4.638999938964844, "learning_rate": 4.1087075811572165e-06, "loss": 0.09085911512374878, "step": 1279 }, { "epoch": 0.9463955637707948, "grad_norm": 4.368131637573242, "learning_rate": 4.107276047318861e-06, "loss": 0.10805981606245041, "step": 1280 }, { "epoch": 0.9471349353049907, "grad_norm": 4.054360389709473, "learning_rate": 4.105843649235104e-06, "loss": 0.07774773240089417, "step": 1281 }, { "epoch": 0.9478743068391867, "grad_norm": 4.047018051147461, "learning_rate": 4.104410387818024e-06, "loss": 0.0813901424407959, "step": 1282 }, { "epoch": 0.9486136783733826, "grad_norm": 4.8320770263671875, "learning_rate": 4.10297626398025e-06, "loss": 0.08886587619781494, "step": 1283 }, { "epoch": 0.9493530499075785, "grad_norm": 3.640439510345459, "learning_rate": 4.101541278634962e-06, "loss": 0.10193506628274918, "step": 1284 }, { "epoch": 0.9500924214417745, "grad_norm": 3.9127964973449707, "learning_rate": 4.100105432695884e-06, "loss": 0.12419667840003967, "step": 1285 }, { "epoch": 0.9508317929759704, "grad_norm": 4.430679798126221, "learning_rate": 4.0986687270772934e-06, "loss": 0.09100896865129471, "step": 1286 }, { "epoch": 0.9515711645101663, "grad_norm": 3.773420572280884, "learning_rate": 4.09723116269401e-06, "loss": 0.06787562370300293, "step": 1287 }, { "epoch": 0.9523105360443623, "grad_norm": 3.0453920364379883, "learning_rate": 4.095792740461405e-06, "loss": 0.07871933281421661, "step": 1288 }, { "epoch": 0.9530499075785582, "grad_norm": 3.77889084815979, "learning_rate": 4.094353461295393e-06, "loss": 0.043538376688957214, "step": 1289 }, { "epoch": 0.9537892791127541, "grad_norm": 5.164119720458984, "learning_rate": 4.092913326112434e-06, "loss": 0.08838309347629547, "step": 1290 }, { "epoch": 0.9545286506469501, "grad_norm": 5.940572261810303, "learning_rate": 4.091472335829534e-06, "loss": 0.08913789689540863, "step": 1291 }, { "epoch": 0.955268022181146, "grad_norm": 4.726081848144531, "learning_rate": 4.090030491364244e-06, "loss": 0.07146337628364563, "step": 1292 }, { "epoch": 0.9560073937153419, "grad_norm": 7.037106990814209, "learning_rate": 4.088587793634658e-06, "loss": 0.09538397192955017, "step": 1293 }, { "epoch": 0.956746765249538, "grad_norm": 5.5326690673828125, "learning_rate": 4.087144243559414e-06, "loss": 0.07475347816944122, "step": 1294 }, { "epoch": 0.9574861367837338, "grad_norm": 4.927600860595703, "learning_rate": 4.085699842057693e-06, "loss": 0.1126512736082077, "step": 1295 }, { "epoch": 0.9582255083179297, "grad_norm": 3.9789552688598633, "learning_rate": 4.084254590049216e-06, "loss": 0.055631399154663086, "step": 1296 }, { "epoch": 0.9589648798521256, "grad_norm": 4.878199577331543, "learning_rate": 4.082808488454248e-06, "loss": 0.08683672547340393, "step": 1297 }, { "epoch": 0.9597042513863216, "grad_norm": 4.572041034698486, "learning_rate": 4.081361538193595e-06, "loss": 0.13889378309249878, "step": 1298 }, { "epoch": 0.9604436229205175, "grad_norm": 3.6683974266052246, "learning_rate": 4.0799137401885995e-06, "loss": 0.06948162615299225, "step": 1299 }, { "epoch": 0.9611829944547134, "grad_norm": 3.6569597721099854, "learning_rate": 4.07846509536115e-06, "loss": 0.06934677064418793, "step": 1300 }, { "epoch": 0.9619223659889095, "grad_norm": 3.4164679050445557, "learning_rate": 4.077015604633669e-06, "loss": 0.03840772062540054, "step": 1301 }, { "epoch": 0.9626617375231054, "grad_norm": 4.617073059082031, "learning_rate": 4.07556526892912e-06, "loss": 0.09075620770454407, "step": 1302 }, { "epoch": 0.9634011090573013, "grad_norm": 3.8203790187835693, "learning_rate": 4.074114089171006e-06, "loss": 0.11782803386449814, "step": 1303 }, { "epoch": 0.9641404805914973, "grad_norm": 3.702298402786255, "learning_rate": 4.0726620662833616e-06, "loss": 0.1098349392414093, "step": 1304 }, { "epoch": 0.9648798521256932, "grad_norm": 3.672513008117676, "learning_rate": 4.071209201190766e-06, "loss": 0.10353456437587738, "step": 1305 }, { "epoch": 0.9656192236598891, "grad_norm": 5.232205390930176, "learning_rate": 4.069755494818329e-06, "loss": 0.11133196204900742, "step": 1306 }, { "epoch": 0.9663585951940851, "grad_norm": 3.9812822341918945, "learning_rate": 4.068300948091698e-06, "loss": 0.12189260125160217, "step": 1307 }, { "epoch": 0.967097966728281, "grad_norm": 4.096288204193115, "learning_rate": 4.066845561937057e-06, "loss": 0.04673321172595024, "step": 1308 }, { "epoch": 0.9678373382624769, "grad_norm": 3.7782952785491943, "learning_rate": 4.065389337281121e-06, "loss": 0.12715455889701843, "step": 1309 }, { "epoch": 0.9685767097966729, "grad_norm": 4.584633827209473, "learning_rate": 4.063932275051143e-06, "loss": 0.07155602425336838, "step": 1310 }, { "epoch": 0.9693160813308688, "grad_norm": 3.613093376159668, "learning_rate": 4.0624743761749045e-06, "loss": 0.07733263075351715, "step": 1311 }, { "epoch": 0.9700554528650647, "grad_norm": 3.766472578048706, "learning_rate": 4.061015641580726e-06, "loss": 0.10558874905109406, "step": 1312 }, { "epoch": 0.9707948243992607, "grad_norm": 4.334668159484863, "learning_rate": 4.059556072197453e-06, "loss": 0.09266974031925201, "step": 1313 }, { "epoch": 0.9715341959334566, "grad_norm": 4.107821464538574, "learning_rate": 4.058095668954468e-06, "loss": 0.14506253600120544, "step": 1314 }, { "epoch": 0.9722735674676525, "grad_norm": 4.466830730438232, "learning_rate": 4.056634432781683e-06, "loss": 0.06958738714456558, "step": 1315 }, { "epoch": 0.9730129390018484, "grad_norm": 4.183511734008789, "learning_rate": 4.05517236460954e-06, "loss": 0.11150211095809937, "step": 1316 }, { "epoch": 0.9737523105360444, "grad_norm": 4.001382827758789, "learning_rate": 4.0537094653690105e-06, "loss": 0.1088789775967598, "step": 1317 }, { "epoch": 0.9744916820702403, "grad_norm": 4.70311164855957, "learning_rate": 4.052245735991595e-06, "loss": 0.08169393986463547, "step": 1318 }, { "epoch": 0.9752310536044362, "grad_norm": 4.318610668182373, "learning_rate": 4.050781177409323e-06, "loss": 0.0718044638633728, "step": 1319 }, { "epoch": 0.9759704251386322, "grad_norm": 3.7538394927978516, "learning_rate": 4.049315790554752e-06, "loss": 0.08584800362586975, "step": 1320 }, { "epoch": 0.9767097966728281, "grad_norm": 3.212029457092285, "learning_rate": 4.047849576360968e-06, "loss": 0.07662405073642731, "step": 1321 }, { "epoch": 0.977449168207024, "grad_norm": 3.960224151611328, "learning_rate": 4.0463825357615815e-06, "loss": 0.09749943017959595, "step": 1322 }, { "epoch": 0.97818853974122, "grad_norm": 5.274507522583008, "learning_rate": 4.044914669690732e-06, "loss": 0.052690379321575165, "step": 1323 }, { "epoch": 0.9789279112754159, "grad_norm": 4.026221752166748, "learning_rate": 4.043445979083081e-06, "loss": 0.08483590185642242, "step": 1324 }, { "epoch": 0.9796672828096118, "grad_norm": 4.826830863952637, "learning_rate": 4.041976464873819e-06, "loss": 0.06050415709614754, "step": 1325 }, { "epoch": 0.9804066543438078, "grad_norm": 5.789690017700195, "learning_rate": 4.040506127998658e-06, "loss": 0.07601231336593628, "step": 1326 }, { "epoch": 0.9811460258780037, "grad_norm": 3.9995474815368652, "learning_rate": 4.039034969393836e-06, "loss": 0.07332482933998108, "step": 1327 }, { "epoch": 0.9818853974121996, "grad_norm": 6.921424388885498, "learning_rate": 4.037562989996112e-06, "loss": 0.1290704905986786, "step": 1328 }, { "epoch": 0.9826247689463956, "grad_norm": 4.82431697845459, "learning_rate": 4.036090190742768e-06, "loss": 0.10626563429832458, "step": 1329 }, { "epoch": 0.9833641404805915, "grad_norm": 4.520009994506836, "learning_rate": 4.034616572571611e-06, "loss": 0.05628187954425812, "step": 1330 }, { "epoch": 0.9841035120147874, "grad_norm": 5.5009684562683105, "learning_rate": 4.033142136420965e-06, "loss": 0.0951467752456665, "step": 1331 }, { "epoch": 0.9848428835489834, "grad_norm": 4.682956218719482, "learning_rate": 4.031666883229678e-06, "loss": 0.06576283276081085, "step": 1332 }, { "epoch": 0.9855822550831793, "grad_norm": 5.203146457672119, "learning_rate": 4.0301908139371196e-06, "loss": 0.0924161970615387, "step": 1333 }, { "epoch": 0.9863216266173752, "grad_norm": 3.3276848793029785, "learning_rate": 4.028713929483173e-06, "loss": 0.0758313313126564, "step": 1334 }, { "epoch": 0.9870609981515711, "grad_norm": 4.787983417510986, "learning_rate": 4.027236230808246e-06, "loss": 0.06754524260759354, "step": 1335 }, { "epoch": 0.9878003696857671, "grad_norm": 4.544626712799072, "learning_rate": 4.0257577188532635e-06, "loss": 0.0633925050497055, "step": 1336 }, { "epoch": 0.988539741219963, "grad_norm": 3.8388311862945557, "learning_rate": 4.024278394559667e-06, "loss": 0.08474404364824295, "step": 1337 }, { "epoch": 0.9892791127541589, "grad_norm": 4.243175029754639, "learning_rate": 4.022798258869417e-06, "loss": 0.09176161140203476, "step": 1338 }, { "epoch": 0.9900184842883549, "grad_norm": 5.340390205383301, "learning_rate": 4.021317312724989e-06, "loss": 0.08944704383611679, "step": 1339 }, { "epoch": 0.9907578558225508, "grad_norm": 4.487743854522705, "learning_rate": 4.019835557069375e-06, "loss": 0.03800221532583237, "step": 1340 }, { "epoch": 0.9914972273567467, "grad_norm": 5.833356857299805, "learning_rate": 4.018352992846084e-06, "loss": 0.09027811884880066, "step": 1341 }, { "epoch": 0.9922365988909427, "grad_norm": 5.087343215942383, "learning_rate": 4.016869620999138e-06, "loss": 0.11056815832853317, "step": 1342 }, { "epoch": 0.9929759704251386, "grad_norm": 5.863541603088379, "learning_rate": 4.015385442473073e-06, "loss": 0.10686534643173218, "step": 1343 }, { "epoch": 0.9937153419593345, "grad_norm": 4.763558387756348, "learning_rate": 4.013900458212941e-06, "loss": 0.10080383718013763, "step": 1344 }, { "epoch": 0.9944547134935305, "grad_norm": 4.233135223388672, "learning_rate": 4.012414669164305e-06, "loss": 0.08028380572795868, "step": 1345 }, { "epoch": 0.9951940850277264, "grad_norm": 3.851355791091919, "learning_rate": 4.0109280762732424e-06, "loss": 0.06436631828546524, "step": 1346 }, { "epoch": 0.9959334565619223, "grad_norm": 3.734239339828491, "learning_rate": 4.009440680486339e-06, "loss": 0.11942875385284424, "step": 1347 }, { "epoch": 0.9966728280961183, "grad_norm": 3.8997111320495605, "learning_rate": 4.007952482750693e-06, "loss": 0.0909205749630928, "step": 1348 }, { "epoch": 0.9974121996303142, "grad_norm": 3.961653470993042, "learning_rate": 4.006463484013918e-06, "loss": 0.06718957424163818, "step": 1349 }, { "epoch": 0.9981515711645101, "grad_norm": 4.4562530517578125, "learning_rate": 4.004973685224131e-06, "loss": 0.07018178701400757, "step": 1350 }, { "epoch": 0.9988909426987062, "grad_norm": 4.014961242675781, "learning_rate": 4.003483087329962e-06, "loss": 0.04467064142227173, "step": 1351 }, { "epoch": 0.999630314232902, "grad_norm": 4.203616142272949, "learning_rate": 4.001991691280549e-06, "loss": 0.07319538295269012, "step": 1352 }, { "epoch": 1.0, "grad_norm": 5.036942005157471, "learning_rate": 4.000499498025538e-06, "loss": 0.10681704431772232, "step": 1353 }, { "epoch": 1.000739371534196, "grad_norm": 4.741019248962402, "learning_rate": 3.999006508515083e-06, "loss": 0.06588105857372284, "step": 1354 }, { "epoch": 1.0014787430683918, "grad_norm": 3.6566123962402344, "learning_rate": 3.9975127236998465e-06, "loss": 0.0428168922662735, "step": 1355 }, { "epoch": 1.0022181146025877, "grad_norm": 3.1419551372528076, "learning_rate": 3.996018144530993e-06, "loss": 0.03882722556591034, "step": 1356 }, { "epoch": 1.0029574861367838, "grad_norm": 4.294229030609131, "learning_rate": 3.994522771960198e-06, "loss": 0.03644448518753052, "step": 1357 }, { "epoch": 1.0036968576709797, "grad_norm": 4.529779434204102, "learning_rate": 3.993026606939639e-06, "loss": 0.0829087346792221, "step": 1358 }, { "epoch": 1.0044362292051756, "grad_norm": 4.230042934417725, "learning_rate": 3.991529650421998e-06, "loss": 0.030598118901252747, "step": 1359 }, { "epoch": 1.0051756007393715, "grad_norm": 3.701796054840088, "learning_rate": 3.990031903360464e-06, "loss": 0.028279315680265427, "step": 1360 }, { "epoch": 1.0059149722735674, "grad_norm": 5.365969181060791, "learning_rate": 3.988533366708727e-06, "loss": 0.06232389062643051, "step": 1361 }, { "epoch": 1.0066543438077633, "grad_norm": 3.3692679405212402, "learning_rate": 3.987034041420978e-06, "loss": 0.049815207719802856, "step": 1362 }, { "epoch": 1.0073937153419594, "grad_norm": 4.879206657409668, "learning_rate": 3.985533928451914e-06, "loss": 0.06753820180892944, "step": 1363 }, { "epoch": 1.0081330868761553, "grad_norm": 3.670508861541748, "learning_rate": 3.984033028756731e-06, "loss": 0.02023305930197239, "step": 1364 }, { "epoch": 1.0088724584103512, "grad_norm": 4.177745342254639, "learning_rate": 3.982531343291126e-06, "loss": 0.07621532678604126, "step": 1365 }, { "epoch": 1.0096118299445471, "grad_norm": 3.727121591567993, "learning_rate": 3.981028873011299e-06, "loss": 0.04562678188085556, "step": 1366 }, { "epoch": 1.010351201478743, "grad_norm": 4.435227394104004, "learning_rate": 3.979525618873948e-06, "loss": 0.03926529362797737, "step": 1367 }, { "epoch": 1.011090573012939, "grad_norm": 4.8861846923828125, "learning_rate": 3.978021581836269e-06, "loss": 0.06006963551044464, "step": 1368 }, { "epoch": 1.011829944547135, "grad_norm": 4.537956237792969, "learning_rate": 3.976516762855956e-06, "loss": 0.05064978823065758, "step": 1369 }, { "epoch": 1.012569316081331, "grad_norm": 4.677010536193848, "learning_rate": 3.9750111628912045e-06, "loss": 0.053411804139614105, "step": 1370 }, { "epoch": 1.0133086876155268, "grad_norm": 4.249866485595703, "learning_rate": 3.973504782900704e-06, "loss": 0.03713822364807129, "step": 1371 }, { "epoch": 1.0140480591497227, "grad_norm": 4.085451602935791, "learning_rate": 3.971997623843642e-06, "loss": 0.04377533495426178, "step": 1372 }, { "epoch": 1.0147874306839186, "grad_norm": 4.086529731750488, "learning_rate": 3.970489686679703e-06, "loss": 0.06076490879058838, "step": 1373 }, { "epoch": 1.0155268022181145, "grad_norm": 5.284555912017822, "learning_rate": 3.968980972369064e-06, "loss": 0.08044473081827164, "step": 1374 }, { "epoch": 1.0162661737523104, "grad_norm": 4.394164085388184, "learning_rate": 3.967471481872401e-06, "loss": 0.05967220664024353, "step": 1375 }, { "epoch": 1.0170055452865066, "grad_norm": 3.9827587604522705, "learning_rate": 3.965961216150878e-06, "loss": 0.07967160642147064, "step": 1376 }, { "epoch": 1.0177449168207024, "grad_norm": 3.9739670753479004, "learning_rate": 3.964450176166161e-06, "loss": 0.017542574554681778, "step": 1377 }, { "epoch": 1.0184842883548983, "grad_norm": 5.479536056518555, "learning_rate": 3.962938362880403e-06, "loss": 0.07677479088306427, "step": 1378 }, { "epoch": 1.0192236598890942, "grad_norm": 4.370586395263672, "learning_rate": 3.96142577725625e-06, "loss": 0.041966523975133896, "step": 1379 }, { "epoch": 1.0199630314232901, "grad_norm": 4.2133941650390625, "learning_rate": 3.959912420256841e-06, "loss": 0.07239414751529694, "step": 1380 }, { "epoch": 1.020702402957486, "grad_norm": 5.301894187927246, "learning_rate": 3.9583982928458075e-06, "loss": 0.06467868387699127, "step": 1381 }, { "epoch": 1.0214417744916822, "grad_norm": 3.613269329071045, "learning_rate": 3.956883395987269e-06, "loss": 0.04065053164958954, "step": 1382 }, { "epoch": 1.022181146025878, "grad_norm": 4.049715995788574, "learning_rate": 3.9553677306458346e-06, "loss": 0.05089162290096283, "step": 1383 }, { "epoch": 1.022920517560074, "grad_norm": 3.478083610534668, "learning_rate": 3.953851297786606e-06, "loss": 0.06887378543615341, "step": 1384 }, { "epoch": 1.0236598890942699, "grad_norm": 4.052163600921631, "learning_rate": 3.9523340983751704e-06, "loss": 0.04072084277868271, "step": 1385 }, { "epoch": 1.0243992606284658, "grad_norm": 3.9868500232696533, "learning_rate": 3.950816133377606e-06, "loss": 0.04679039865732193, "step": 1386 }, { "epoch": 1.0251386321626617, "grad_norm": 4.609038352966309, "learning_rate": 3.949297403760475e-06, "loss": 0.04944848269224167, "step": 1387 }, { "epoch": 1.0258780036968578, "grad_norm": 4.457963943481445, "learning_rate": 3.94777791049083e-06, "loss": 0.035124409943819046, "step": 1388 }, { "epoch": 1.0266173752310537, "grad_norm": 3.649895668029785, "learning_rate": 3.946257654536206e-06, "loss": 0.11919085681438446, "step": 1389 }, { "epoch": 1.0273567467652496, "grad_norm": 3.285341501235962, "learning_rate": 3.9447366368646275e-06, "loss": 0.07299504429101944, "step": 1390 }, { "epoch": 1.0280961182994455, "grad_norm": 4.55879020690918, "learning_rate": 3.9432148584446035e-06, "loss": 0.06183500960469246, "step": 1391 }, { "epoch": 1.0288354898336414, "grad_norm": 2.846226453781128, "learning_rate": 3.941692320245124e-06, "loss": 0.022933870553970337, "step": 1392 }, { "epoch": 1.0295748613678373, "grad_norm": 3.4374260902404785, "learning_rate": 3.940169023235666e-06, "loss": 0.06488388031721115, "step": 1393 }, { "epoch": 1.0303142329020332, "grad_norm": 3.8827688694000244, "learning_rate": 3.938644968386188e-06, "loss": 0.044609010219573975, "step": 1394 }, { "epoch": 1.0310536044362293, "grad_norm": 4.96807336807251, "learning_rate": 3.937120156667133e-06, "loss": 0.041307590901851654, "step": 1395 }, { "epoch": 1.0317929759704252, "grad_norm": 3.2196218967437744, "learning_rate": 3.935594589049425e-06, "loss": 0.019824277609586716, "step": 1396 }, { "epoch": 1.032532347504621, "grad_norm": 4.314876556396484, "learning_rate": 3.9340682665044674e-06, "loss": 0.025385022163391113, "step": 1397 }, { "epoch": 1.033271719038817, "grad_norm": 4.840250492095947, "learning_rate": 3.932541190004147e-06, "loss": 0.04788488149642944, "step": 1398 }, { "epoch": 1.0340110905730129, "grad_norm": 3.9281272888183594, "learning_rate": 3.931013360520829e-06, "loss": 0.06176590919494629, "step": 1399 }, { "epoch": 1.0347504621072088, "grad_norm": 4.057123184204102, "learning_rate": 3.92948477902736e-06, "loss": 0.07197315990924835, "step": 1400 }, { "epoch": 1.035489833641405, "grad_norm": 4.793193340301514, "learning_rate": 3.9279554464970625e-06, "loss": 0.055562298744916916, "step": 1401 }, { "epoch": 1.0362292051756008, "grad_norm": 4.456971645355225, "learning_rate": 3.92642536390374e-06, "loss": 0.04613329470157623, "step": 1402 }, { "epoch": 1.0369685767097967, "grad_norm": 5.329600811004639, "learning_rate": 3.9248945322216715e-06, "loss": 0.04888426512479782, "step": 1403 }, { "epoch": 1.0377079482439926, "grad_norm": 4.043542385101318, "learning_rate": 3.923362952425615e-06, "loss": 0.05653994157910347, "step": 1404 }, { "epoch": 1.0384473197781885, "grad_norm": 5.43079137802124, "learning_rate": 3.921830625490803e-06, "loss": 0.05034356191754341, "step": 1405 }, { "epoch": 1.0391866913123844, "grad_norm": 4.561283588409424, "learning_rate": 3.920297552392945e-06, "loss": 0.06475773453712463, "step": 1406 }, { "epoch": 1.0399260628465805, "grad_norm": 4.936164379119873, "learning_rate": 3.918763734108225e-06, "loss": 0.06211283802986145, "step": 1407 }, { "epoch": 1.0406654343807764, "grad_norm": 4.655771255493164, "learning_rate": 3.9172291716133006e-06, "loss": 0.06735041737556458, "step": 1408 }, { "epoch": 1.0414048059149723, "grad_norm": 3.664992570877075, "learning_rate": 3.915693865885306e-06, "loss": 0.06556246429681778, "step": 1409 }, { "epoch": 1.0421441774491682, "grad_norm": 3.8677175045013428, "learning_rate": 3.9141578179018465e-06, "loss": 0.07061460614204407, "step": 1410 }, { "epoch": 1.042883548983364, "grad_norm": 4.16327428817749, "learning_rate": 3.912621028641001e-06, "loss": 0.08909595757722855, "step": 1411 }, { "epoch": 1.04362292051756, "grad_norm": 3.742260217666626, "learning_rate": 3.911083499081319e-06, "loss": 0.028425315394997597, "step": 1412 }, { "epoch": 1.044362292051756, "grad_norm": 3.4379234313964844, "learning_rate": 3.909545230201821e-06, "loss": 0.042131051421165466, "step": 1413 }, { "epoch": 1.045101663585952, "grad_norm": 3.346024990081787, "learning_rate": 3.908006222982003e-06, "loss": 0.049899570643901825, "step": 1414 }, { "epoch": 1.045841035120148, "grad_norm": 4.092625141143799, "learning_rate": 3.906466478401825e-06, "loss": 0.05895501375198364, "step": 1415 }, { "epoch": 1.0465804066543438, "grad_norm": 3.868272066116333, "learning_rate": 3.904925997441721e-06, "loss": 0.0228949673473835, "step": 1416 }, { "epoch": 1.0473197781885397, "grad_norm": 5.014120101928711, "learning_rate": 3.9033847810825895e-06, "loss": 0.06177529692649841, "step": 1417 }, { "epoch": 1.0480591497227356, "grad_norm": 3.3737688064575195, "learning_rate": 3.901842830305803e-06, "loss": 0.02094052918255329, "step": 1418 }, { "epoch": 1.0487985212569315, "grad_norm": 3.5036473274230957, "learning_rate": 3.9003001460931966e-06, "loss": 0.04771972447633743, "step": 1419 }, { "epoch": 1.0495378927911276, "grad_norm": 6.338980674743652, "learning_rate": 3.898756729427075e-06, "loss": 0.030775506049394608, "step": 1420 }, { "epoch": 1.0502772643253235, "grad_norm": 3.5010781288146973, "learning_rate": 3.897212581290208e-06, "loss": 0.052935890853405, "step": 1421 }, { "epoch": 1.0510166358595194, "grad_norm": 4.218605041503906, "learning_rate": 3.895667702665833e-06, "loss": 0.06493952870368958, "step": 1422 }, { "epoch": 1.0517560073937153, "grad_norm": 4.930954456329346, "learning_rate": 3.89412209453765e-06, "loss": 0.06427063047885895, "step": 1423 }, { "epoch": 1.0524953789279112, "grad_norm": 5.38845157623291, "learning_rate": 3.892575757889827e-06, "loss": 0.042093805968761444, "step": 1424 }, { "epoch": 1.0532347504621071, "grad_norm": 5.220769882202148, "learning_rate": 3.89102869370699e-06, "loss": 0.039854712784290314, "step": 1425 }, { "epoch": 1.0539741219963032, "grad_norm": 5.319488048553467, "learning_rate": 3.889480902974235e-06, "loss": 0.04376843944191933, "step": 1426 }, { "epoch": 1.0547134935304991, "grad_norm": 4.615324020385742, "learning_rate": 3.8879323866771165e-06, "loss": 0.04933340474963188, "step": 1427 }, { "epoch": 1.055452865064695, "grad_norm": 4.863185405731201, "learning_rate": 3.886383145801652e-06, "loss": 0.06715421378612518, "step": 1428 }, { "epoch": 1.056192236598891, "grad_norm": 3.7147064208984375, "learning_rate": 3.884833181334322e-06, "loss": 0.037244174629449844, "step": 1429 }, { "epoch": 1.0569316081330868, "grad_norm": 3.4412238597869873, "learning_rate": 3.883282494262064e-06, "loss": 0.0640559196472168, "step": 1430 }, { "epoch": 1.0576709796672827, "grad_norm": 3.620983839035034, "learning_rate": 3.88173108557228e-06, "loss": 0.054013293236494064, "step": 1431 }, { "epoch": 1.0584103512014786, "grad_norm": 3.5243310928344727, "learning_rate": 3.880178956252827e-06, "loss": 0.05671863257884979, "step": 1432 }, { "epoch": 1.0591497227356748, "grad_norm": 3.7861313819885254, "learning_rate": 3.878626107292025e-06, "loss": 0.04580123722553253, "step": 1433 }, { "epoch": 1.0598890942698707, "grad_norm": 3.9307303428649902, "learning_rate": 3.8770725396786506e-06, "loss": 0.0476737916469574, "step": 1434 }, { "epoch": 1.0606284658040666, "grad_norm": 3.5850024223327637, "learning_rate": 3.875518254401936e-06, "loss": 0.03999211639165878, "step": 1435 }, { "epoch": 1.0613678373382625, "grad_norm": 5.670407772064209, "learning_rate": 3.873963252451573e-06, "loss": 0.06335477530956268, "step": 1436 }, { "epoch": 1.0621072088724584, "grad_norm": 3.830083131790161, "learning_rate": 3.87240753481771e-06, "loss": 0.08204620331525803, "step": 1437 }, { "epoch": 1.0628465804066543, "grad_norm": 4.005023002624512, "learning_rate": 3.870851102490951e-06, "loss": 0.03009466454386711, "step": 1438 }, { "epoch": 1.0635859519408504, "grad_norm": 4.269959926605225, "learning_rate": 3.8692939564623504e-06, "loss": 0.03756941109895706, "step": 1439 }, { "epoch": 1.0643253234750463, "grad_norm": 4.429284572601318, "learning_rate": 3.867736097723424e-06, "loss": 0.08064084500074387, "step": 1440 }, { "epoch": 1.0650646950092422, "grad_norm": 2.9563121795654297, "learning_rate": 3.866177527266136e-06, "loss": 0.01585797220468521, "step": 1441 }, { "epoch": 1.065804066543438, "grad_norm": 4.001089096069336, "learning_rate": 3.864618246082908e-06, "loss": 0.05304029583930969, "step": 1442 }, { "epoch": 1.066543438077634, "grad_norm": 4.314488410949707, "learning_rate": 3.863058255166612e-06, "loss": 0.0339159220457077, "step": 1443 }, { "epoch": 1.0672828096118299, "grad_norm": 4.839524745941162, "learning_rate": 3.86149755551057e-06, "loss": 0.06313038617372513, "step": 1444 }, { "epoch": 1.0680221811460258, "grad_norm": 3.937715530395508, "learning_rate": 3.8599361481085595e-06, "loss": 0.043298352509737015, "step": 1445 }, { "epoch": 1.0687615526802219, "grad_norm": 4.339387893676758, "learning_rate": 3.8583740339548055e-06, "loss": 0.042849086225032806, "step": 1446 }, { "epoch": 1.0695009242144178, "grad_norm": 3.5561468601226807, "learning_rate": 3.856811214043984e-06, "loss": 0.02313273772597313, "step": 1447 }, { "epoch": 1.0702402957486137, "grad_norm": 4.300020217895508, "learning_rate": 3.855247689371221e-06, "loss": 0.077162966132164, "step": 1448 }, { "epoch": 1.0709796672828096, "grad_norm": 3.8480985164642334, "learning_rate": 3.853683460932091e-06, "loss": 0.05855659395456314, "step": 1449 }, { "epoch": 1.0717190388170055, "grad_norm": 4.279290199279785, "learning_rate": 3.8521185297226145e-06, "loss": 0.04996457323431969, "step": 1450 }, { "epoch": 1.0724584103512016, "grad_norm": 4.394758224487305, "learning_rate": 3.8505528967392635e-06, "loss": 0.04224211722612381, "step": 1451 }, { "epoch": 1.0731977818853975, "grad_norm": 4.699183940887451, "learning_rate": 3.848986562978954e-06, "loss": 0.09879889339208603, "step": 1452 }, { "epoch": 1.0739371534195934, "grad_norm": 3.1117947101593018, "learning_rate": 3.847419529439048e-06, "loss": 0.04117349907755852, "step": 1453 }, { "epoch": 1.0746765249537893, "grad_norm": 4.282766819000244, "learning_rate": 3.845851797117355e-06, "loss": 0.04817522317171097, "step": 1454 }, { "epoch": 1.0754158964879852, "grad_norm": 4.233505725860596, "learning_rate": 3.844283367012128e-06, "loss": 0.048714231699705124, "step": 1455 }, { "epoch": 1.076155268022181, "grad_norm": 5.243323802947998, "learning_rate": 3.842714240122064e-06, "loss": 0.0421614944934845, "step": 1456 }, { "epoch": 1.076894639556377, "grad_norm": 3.4909918308258057, "learning_rate": 3.841144417446305e-06, "loss": 0.061538536101579666, "step": 1457 }, { "epoch": 1.077634011090573, "grad_norm": 4.194746971130371, "learning_rate": 3.839573899984436e-06, "loss": 0.07142636179924011, "step": 1458 }, { "epoch": 1.078373382624769, "grad_norm": 4.430039405822754, "learning_rate": 3.838002688736484e-06, "loss": 0.06479091942310333, "step": 1459 }, { "epoch": 1.079112754158965, "grad_norm": 4.348980903625488, "learning_rate": 3.836430784702916e-06, "loss": 0.05091116204857826, "step": 1460 }, { "epoch": 1.0798521256931608, "grad_norm": 4.095023155212402, "learning_rate": 3.834858188884644e-06, "loss": 0.03941578418016434, "step": 1461 }, { "epoch": 1.0805914972273567, "grad_norm": 3.998353958129883, "learning_rate": 3.8332849022830164e-06, "loss": 0.04813533276319504, "step": 1462 }, { "epoch": 1.0813308687615526, "grad_norm": 6.3184003829956055, "learning_rate": 3.831710925899824e-06, "loss": 0.06277060508728027, "step": 1463 }, { "epoch": 1.0820702402957485, "grad_norm": 3.7152817249298096, "learning_rate": 3.830136260737296e-06, "loss": 0.030380651354789734, "step": 1464 }, { "epoch": 1.0828096118299446, "grad_norm": 2.4195539951324463, "learning_rate": 3.8285609077981005e-06, "loss": 0.006544262170791626, "step": 1465 }, { "epoch": 1.0835489833641405, "grad_norm": 5.408345699310303, "learning_rate": 3.8269848680853425e-06, "loss": 0.05076972022652626, "step": 1466 }, { "epoch": 1.0842883548983364, "grad_norm": 5.480380535125732, "learning_rate": 3.825408142602567e-06, "loss": 0.046970948576927185, "step": 1467 }, { "epoch": 1.0850277264325323, "grad_norm": 5.887692928314209, "learning_rate": 3.823830732353753e-06, "loss": 0.0266402717679739, "step": 1468 }, { "epoch": 1.0857670979667282, "grad_norm": 5.6646409034729, "learning_rate": 3.8222526383433145e-06, "loss": 0.0667831152677536, "step": 1469 }, { "epoch": 1.0865064695009243, "grad_norm": 5.054966449737549, "learning_rate": 3.820673861576105e-06, "loss": 0.049983665347099304, "step": 1470 }, { "epoch": 1.0872458410351202, "grad_norm": 4.131236553192139, "learning_rate": 3.81909440305741e-06, "loss": 0.04726223275065422, "step": 1471 }, { "epoch": 1.0879852125693161, "grad_norm": 4.2182536125183105, "learning_rate": 3.817514263792948e-06, "loss": 0.054164204746484756, "step": 1472 }, { "epoch": 1.088724584103512, "grad_norm": 5.131840705871582, "learning_rate": 3.815933444788876e-06, "loss": 0.08190158754587173, "step": 1473 }, { "epoch": 1.089463955637708, "grad_norm": 4.322277545928955, "learning_rate": 3.8143519470517763e-06, "loss": 0.04772038757801056, "step": 1474 }, { "epoch": 1.0902033271719038, "grad_norm": 3.899325132369995, "learning_rate": 3.8127697715886696e-06, "loss": 0.047164954245090485, "step": 1475 }, { "epoch": 1.0909426987060997, "grad_norm": 4.350290775299072, "learning_rate": 3.8111869194070062e-06, "loss": 0.025600440800189972, "step": 1476 }, { "epoch": 1.0916820702402958, "grad_norm": 4.65999174118042, "learning_rate": 3.809603391514667e-06, "loss": 0.0524335578083992, "step": 1477 }, { "epoch": 1.0924214417744917, "grad_norm": 4.007490158081055, "learning_rate": 3.808019188919962e-06, "loss": 0.04302043467760086, "step": 1478 }, { "epoch": 1.0931608133086876, "grad_norm": 4.942358493804932, "learning_rate": 3.806434312631633e-06, "loss": 0.029673490673303604, "step": 1479 }, { "epoch": 1.0939001848428835, "grad_norm": 4.08126974105835, "learning_rate": 3.8048487636588493e-06, "loss": 0.06169648468494415, "step": 1480 }, { "epoch": 1.0946395563770794, "grad_norm": 3.6289124488830566, "learning_rate": 3.8032625430112107e-06, "loss": 0.03594616800546646, "step": 1481 }, { "epoch": 1.0953789279112753, "grad_norm": 3.291499137878418, "learning_rate": 3.8016756516987406e-06, "loss": 0.07245972752571106, "step": 1482 }, { "epoch": 1.0961182994454712, "grad_norm": 4.070366382598877, "learning_rate": 3.8000880907318945e-06, "loss": 0.07448311150074005, "step": 1483 }, { "epoch": 1.0968576709796674, "grad_norm": 3.3727457523345947, "learning_rate": 3.798499861121549e-06, "loss": 0.07423198223114014, "step": 1484 }, { "epoch": 1.0975970425138633, "grad_norm": 4.6825480461120605, "learning_rate": 3.7969109638790106e-06, "loss": 0.07969753444194794, "step": 1485 }, { "epoch": 1.0983364140480592, "grad_norm": 3.4554505348205566, "learning_rate": 3.79532140001601e-06, "loss": 0.0352659597992897, "step": 1486 }, { "epoch": 1.099075785582255, "grad_norm": 4.660616397857666, "learning_rate": 3.7937311705447015e-06, "loss": 0.03255036473274231, "step": 1487 }, { "epoch": 1.099815157116451, "grad_norm": 3.697787284851074, "learning_rate": 3.7921402764776627e-06, "loss": 0.02977178804576397, "step": 1488 }, { "epoch": 1.1005545286506468, "grad_norm": 3.54510760307312, "learning_rate": 3.7905487188278962e-06, "loss": 0.025200925767421722, "step": 1489 }, { "epoch": 1.101293900184843, "grad_norm": 5.845457553863525, "learning_rate": 3.7889564986088253e-06, "loss": 0.04675303399562836, "step": 1490 }, { "epoch": 1.1020332717190389, "grad_norm": 2.7989249229431152, "learning_rate": 3.787363616834297e-06, "loss": 0.021130550652742386, "step": 1491 }, { "epoch": 1.1027726432532348, "grad_norm": 3.908860921859741, "learning_rate": 3.785770074518579e-06, "loss": 0.030432656407356262, "step": 1492 }, { "epoch": 1.1035120147874307, "grad_norm": 5.146832466125488, "learning_rate": 3.7841758726763573e-06, "loss": 0.09010156989097595, "step": 1493 }, { "epoch": 1.1042513863216266, "grad_norm": 5.582509994506836, "learning_rate": 3.7825810123227413e-06, "loss": 0.061894889920949936, "step": 1494 }, { "epoch": 1.1049907578558225, "grad_norm": 4.027811050415039, "learning_rate": 3.780985494473258e-06, "loss": 0.03920402377843857, "step": 1495 }, { "epoch": 1.1057301293900186, "grad_norm": 5.095418930053711, "learning_rate": 3.779389320143853e-06, "loss": 0.0563197135925293, "step": 1496 }, { "epoch": 1.1064695009242145, "grad_norm": 4.209932804107666, "learning_rate": 3.7777924903508907e-06, "loss": 0.06595513224601746, "step": 1497 }, { "epoch": 1.1072088724584104, "grad_norm": 3.9431228637695312, "learning_rate": 3.7761950061111513e-06, "loss": 0.037098344415426254, "step": 1498 }, { "epoch": 1.1079482439926063, "grad_norm": 4.145330429077148, "learning_rate": 3.774596868441834e-06, "loss": 0.05148395523428917, "step": 1499 }, { "epoch": 1.1086876155268022, "grad_norm": 2.7628045082092285, "learning_rate": 3.772998078360552e-06, "loss": 0.02619437873363495, "step": 1500 }, { "epoch": 1.109426987060998, "grad_norm": 3.7640159130096436, "learning_rate": 3.771398636885335e-06, "loss": 0.04779599979519844, "step": 1501 }, { "epoch": 1.110166358595194, "grad_norm": 4.045032978057861, "learning_rate": 3.7697985450346263e-06, "loss": 0.04915345460176468, "step": 1502 }, { "epoch": 1.11090573012939, "grad_norm": 4.301629543304443, "learning_rate": 3.7681978038272856e-06, "loss": 0.012543149292469025, "step": 1503 }, { "epoch": 1.111645101663586, "grad_norm": 3.9443776607513428, "learning_rate": 3.7665964142825837e-06, "loss": 0.0286052618175745, "step": 1504 }, { "epoch": 1.1123844731977819, "grad_norm": 4.943824768066406, "learning_rate": 3.7649943774202065e-06, "loss": 0.04453812167048454, "step": 1505 }, { "epoch": 1.1131238447319778, "grad_norm": 5.41103458404541, "learning_rate": 3.76339169426025e-06, "loss": 0.05596032366156578, "step": 1506 }, { "epoch": 1.1138632162661737, "grad_norm": 4.593433380126953, "learning_rate": 3.761788365823222e-06, "loss": 0.07106387615203857, "step": 1507 }, { "epoch": 1.1146025878003696, "grad_norm": 5.436102390289307, "learning_rate": 3.760184393130043e-06, "loss": 0.06030649691820145, "step": 1508 }, { "epoch": 1.1153419593345657, "grad_norm": 4.45260763168335, "learning_rate": 3.7585797772020416e-06, "loss": 0.048198603093624115, "step": 1509 }, { "epoch": 1.1160813308687616, "grad_norm": 4.111651420593262, "learning_rate": 3.756974519060957e-06, "loss": 0.06183779239654541, "step": 1510 }, { "epoch": 1.1168207024029575, "grad_norm": 4.714849472045898, "learning_rate": 3.7553686197289373e-06, "loss": 0.04190994054079056, "step": 1511 }, { "epoch": 1.1175600739371534, "grad_norm": 4.431219577789307, "learning_rate": 3.753762080228539e-06, "loss": 0.043491557240486145, "step": 1512 }, { "epoch": 1.1182994454713493, "grad_norm": 4.414183139801025, "learning_rate": 3.7521549015827257e-06, "loss": 0.07439300417900085, "step": 1513 }, { "epoch": 1.1190388170055452, "grad_norm": 4.55821418762207, "learning_rate": 3.7505470848148673e-06, "loss": 0.04337716102600098, "step": 1514 }, { "epoch": 1.1197781885397413, "grad_norm": 5.152626037597656, "learning_rate": 3.748938630948743e-06, "loss": 0.058400511741638184, "step": 1515 }, { "epoch": 1.1205175600739372, "grad_norm": 4.438140869140625, "learning_rate": 3.7473295410085343e-06, "loss": 0.06300479173660278, "step": 1516 }, { "epoch": 1.1212569316081331, "grad_norm": 3.669670343399048, "learning_rate": 3.7457198160188294e-06, "loss": 0.07930319756269455, "step": 1517 }, { "epoch": 1.121996303142329, "grad_norm": 4.098944187164307, "learning_rate": 3.7441094570046203e-06, "loss": 0.046091385185718536, "step": 1518 }, { "epoch": 1.122735674676525, "grad_norm": 3.9620044231414795, "learning_rate": 3.742498464991303e-06, "loss": 0.036158572882413864, "step": 1519 }, { "epoch": 1.1234750462107208, "grad_norm": 3.567861795425415, "learning_rate": 3.740886841004678e-06, "loss": 0.049262885004282, "step": 1520 }, { "epoch": 1.1242144177449167, "grad_norm": 3.4628243446350098, "learning_rate": 3.739274586070944e-06, "loss": 0.05596566200256348, "step": 1521 }, { "epoch": 1.1249537892791128, "grad_norm": 4.1450958251953125, "learning_rate": 3.7376617012167067e-06, "loss": 0.05639415234327316, "step": 1522 }, { "epoch": 1.1256931608133087, "grad_norm": 4.4603190422058105, "learning_rate": 3.736048187468968e-06, "loss": 0.06877661496400833, "step": 1523 }, { "epoch": 1.1264325323475046, "grad_norm": 5.338214874267578, "learning_rate": 3.7344340458551347e-06, "loss": 0.06740066409111023, "step": 1524 }, { "epoch": 1.1271719038817005, "grad_norm": 4.124356269836426, "learning_rate": 3.7328192774030102e-06, "loss": 0.03988737612962723, "step": 1525 }, { "epoch": 1.1279112754158964, "grad_norm": 4.348417282104492, "learning_rate": 3.7312038831407986e-06, "loss": 0.04678662121295929, "step": 1526 }, { "epoch": 1.1286506469500925, "grad_norm": 4.780818462371826, "learning_rate": 3.729587864097103e-06, "loss": 0.026174910366535187, "step": 1527 }, { "epoch": 1.1293900184842884, "grad_norm": 4.039291858673096, "learning_rate": 3.727971221300921e-06, "loss": 0.06461787223815918, "step": 1528 }, { "epoch": 1.1301293900184843, "grad_norm": 4.628822326660156, "learning_rate": 3.726353955781652e-06, "loss": 0.024545755237340927, "step": 1529 }, { "epoch": 1.1308687615526802, "grad_norm": 4.005336284637451, "learning_rate": 3.724736068569088e-06, "loss": 0.06931277364492416, "step": 1530 }, { "epoch": 1.1316081330868761, "grad_norm": 3.603341579437256, "learning_rate": 3.72311756069342e-06, "loss": 0.05838761478662491, "step": 1531 }, { "epoch": 1.132347504621072, "grad_norm": 5.5958051681518555, "learning_rate": 3.721498433185232e-06, "loss": 0.07873627543449402, "step": 1532 }, { "epoch": 1.133086876155268, "grad_norm": 3.954246997833252, "learning_rate": 3.7198786870755033e-06, "loss": 0.051265180110931396, "step": 1533 }, { "epoch": 1.133826247689464, "grad_norm": 3.7719409465789795, "learning_rate": 3.7182583233956065e-06, "loss": 0.051491670310497284, "step": 1534 }, { "epoch": 1.13456561922366, "grad_norm": 3.6958329677581787, "learning_rate": 3.71663734317731e-06, "loss": 0.07144208252429962, "step": 1535 }, { "epoch": 1.1353049907578558, "grad_norm": 3.620664358139038, "learning_rate": 3.715015747452771e-06, "loss": 0.061878785490989685, "step": 1536 }, { "epoch": 1.1360443622920517, "grad_norm": 2.8694684505462646, "learning_rate": 3.7133935372545402e-06, "loss": 0.04246073216199875, "step": 1537 }, { "epoch": 1.1367837338262476, "grad_norm": 3.0797221660614014, "learning_rate": 3.7117707136155606e-06, "loss": 0.023655645549297333, "step": 1538 }, { "epoch": 1.1375231053604435, "grad_norm": 2.9915781021118164, "learning_rate": 3.7101472775691656e-06, "loss": 0.032836705446243286, "step": 1539 }, { "epoch": 1.1382624768946394, "grad_norm": 1.431882619857788, "learning_rate": 3.7085232301490777e-06, "loss": 0.013051377609372139, "step": 1540 }, { "epoch": 1.1390018484288356, "grad_norm": 2.8968708515167236, "learning_rate": 3.706898572389408e-06, "loss": 0.028337422758340836, "step": 1541 }, { "epoch": 1.1397412199630315, "grad_norm": 2.729809522628784, "learning_rate": 3.7052733053246575e-06, "loss": 0.03947356343269348, "step": 1542 }, { "epoch": 1.1404805914972274, "grad_norm": 3.235309600830078, "learning_rate": 3.7036474299897152e-06, "loss": 0.02713494934141636, "step": 1543 }, { "epoch": 1.1412199630314233, "grad_norm": 3.896871566772461, "learning_rate": 3.702020947419857e-06, "loss": 0.04431937634944916, "step": 1544 }, { "epoch": 1.1419593345656192, "grad_norm": 3.395770311355591, "learning_rate": 3.700393858650746e-06, "loss": 0.06061399355530739, "step": 1545 }, { "epoch": 1.1426987060998153, "grad_norm": 2.5584824085235596, "learning_rate": 3.6987661647184298e-06, "loss": 0.014262707903981209, "step": 1546 }, { "epoch": 1.1434380776340112, "grad_norm": 4.159789085388184, "learning_rate": 3.6971378666593432e-06, "loss": 0.04988367110490799, "step": 1547 }, { "epoch": 1.144177449168207, "grad_norm": 3.944742202758789, "learning_rate": 3.6955089655103037e-06, "loss": 0.038242075592279434, "step": 1548 }, { "epoch": 1.144916820702403, "grad_norm": 4.041978359222412, "learning_rate": 3.6938794623085154e-06, "loss": 0.041930295526981354, "step": 1549 }, { "epoch": 1.1456561922365989, "grad_norm": 5.784710884094238, "learning_rate": 3.6922493580915635e-06, "loss": 0.051013290882110596, "step": 1550 }, { "epoch": 1.1463955637707948, "grad_norm": 4.224785327911377, "learning_rate": 3.690618653897418e-06, "loss": 0.07451000809669495, "step": 1551 }, { "epoch": 1.1471349353049907, "grad_norm": 5.2269415855407715, "learning_rate": 3.6889873507644276e-06, "loss": 0.05007501319050789, "step": 1552 }, { "epoch": 1.1478743068391868, "grad_norm": 4.179022312164307, "learning_rate": 3.6873554497313254e-06, "loss": 0.02951493114233017, "step": 1553 }, { "epoch": 1.1486136783733827, "grad_norm": 3.5029048919677734, "learning_rate": 3.6857229518372244e-06, "loss": 0.05164355784654617, "step": 1554 }, { "epoch": 1.1493530499075786, "grad_norm": 4.244418144226074, "learning_rate": 3.6840898581216173e-06, "loss": 0.07505753636360168, "step": 1555 }, { "epoch": 1.1500924214417745, "grad_norm": 3.8284716606140137, "learning_rate": 3.682456169624376e-06, "loss": 0.04691307619214058, "step": 1556 }, { "epoch": 1.1508317929759704, "grad_norm": 3.557596206665039, "learning_rate": 3.6808218873857526e-06, "loss": 0.01931578665971756, "step": 1557 }, { "epoch": 1.1515711645101663, "grad_norm": 3.8000147342681885, "learning_rate": 3.679187012446375e-06, "loss": 0.052891235798597336, "step": 1558 }, { "epoch": 1.1523105360443622, "grad_norm": 3.692012310028076, "learning_rate": 3.677551545847251e-06, "loss": 0.06712351739406586, "step": 1559 }, { "epoch": 1.1530499075785583, "grad_norm": 4.198168754577637, "learning_rate": 3.6759154886297627e-06, "loss": 0.04641050100326538, "step": 1560 }, { "epoch": 1.1537892791127542, "grad_norm": 4.69840145111084, "learning_rate": 3.6742788418356682e-06, "loss": 0.09333176910877228, "step": 1561 }, { "epoch": 1.15452865064695, "grad_norm": 5.4046759605407715, "learning_rate": 3.672641606507105e-06, "loss": 0.04709932953119278, "step": 1562 }, { "epoch": 1.155268022181146, "grad_norm": 3.0951247215270996, "learning_rate": 3.6710037836865816e-06, "loss": 0.031689636409282684, "step": 1563 }, { "epoch": 1.156007393715342, "grad_norm": 3.875154733657837, "learning_rate": 3.6693653744169803e-06, "loss": 0.02963271550834179, "step": 1564 }, { "epoch": 1.156746765249538, "grad_norm": 4.350226402282715, "learning_rate": 3.6677263797415584e-06, "loss": 0.06012594699859619, "step": 1565 }, { "epoch": 1.157486136783734, "grad_norm": 6.06222677230835, "learning_rate": 3.6660868007039474e-06, "loss": 0.06347065418958664, "step": 1566 }, { "epoch": 1.1582255083179298, "grad_norm": 3.8948123455047607, "learning_rate": 3.664446638348147e-06, "loss": 0.07032172381877899, "step": 1567 }, { "epoch": 1.1589648798521257, "grad_norm": 4.957052230834961, "learning_rate": 3.6628058937185313e-06, "loss": 0.03219733387231827, "step": 1568 }, { "epoch": 1.1597042513863216, "grad_norm": 3.1492080688476562, "learning_rate": 3.6611645678598444e-06, "loss": 0.01786056160926819, "step": 1569 }, { "epoch": 1.1604436229205175, "grad_norm": 4.222098350524902, "learning_rate": 3.6595226618172002e-06, "loss": 0.044508352875709534, "step": 1570 }, { "epoch": 1.1611829944547134, "grad_norm": 4.087658405303955, "learning_rate": 3.6578801766360826e-06, "loss": 0.019816629588603973, "step": 1571 }, { "epoch": 1.1619223659889095, "grad_norm": 4.754374980926514, "learning_rate": 3.656237113362343e-06, "loss": 0.07251189649105072, "step": 1572 }, { "epoch": 1.1626617375231054, "grad_norm": 5.316276550292969, "learning_rate": 3.654593473042202e-06, "loss": 0.06063852459192276, "step": 1573 }, { "epoch": 1.1634011090573013, "grad_norm": 4.3338212966918945, "learning_rate": 3.652949256722249e-06, "loss": 0.10667117685079575, "step": 1574 }, { "epoch": 1.1641404805914972, "grad_norm": 3.9593513011932373, "learning_rate": 3.6513044654494366e-06, "loss": 0.05591852217912674, "step": 1575 }, { "epoch": 1.1648798521256931, "grad_norm": 4.64186429977417, "learning_rate": 3.6496591002710864e-06, "loss": 0.04264893755316734, "step": 1576 }, { "epoch": 1.165619223659889, "grad_norm": 4.046307563781738, "learning_rate": 3.6480131622348854e-06, "loss": 0.07777218520641327, "step": 1577 }, { "epoch": 1.166358595194085, "grad_norm": 4.9835028648376465, "learning_rate": 3.6463666523888814e-06, "loss": 0.061341360211372375, "step": 1578 }, { "epoch": 1.167097966728281, "grad_norm": 3.5038509368896484, "learning_rate": 3.6447195717814925e-06, "loss": 0.02547246590256691, "step": 1579 }, { "epoch": 1.167837338262477, "grad_norm": 3.9165847301483154, "learning_rate": 3.6430719214614966e-06, "loss": 0.0495951771736145, "step": 1580 }, { "epoch": 1.1685767097966728, "grad_norm": 3.4221436977386475, "learning_rate": 3.6414237024780343e-06, "loss": 0.04253728687763214, "step": 1581 }, { "epoch": 1.1693160813308687, "grad_norm": 5.97763204574585, "learning_rate": 3.6397749158806085e-06, "loss": 0.042353831231594086, "step": 1582 }, { "epoch": 1.1700554528650646, "grad_norm": 3.622999906539917, "learning_rate": 3.638125562719085e-06, "loss": 0.039114974439144135, "step": 1583 }, { "epoch": 1.1707948243992607, "grad_norm": 4.133333683013916, "learning_rate": 3.636475644043688e-06, "loss": 0.08208160102367401, "step": 1584 }, { "epoch": 1.1715341959334566, "grad_norm": 5.648869514465332, "learning_rate": 3.634825160905004e-06, "loss": 0.050117265433073044, "step": 1585 }, { "epoch": 1.1722735674676525, "grad_norm": 3.7436439990997314, "learning_rate": 3.6331741143539777e-06, "loss": 0.022507671266794205, "step": 1586 }, { "epoch": 1.1730129390018484, "grad_norm": 5.343332290649414, "learning_rate": 3.6315225054419133e-06, "loss": 0.09238451719284058, "step": 1587 }, { "epoch": 1.1737523105360443, "grad_norm": 3.102790594100952, "learning_rate": 3.6298703352204723e-06, "loss": 0.028020355850458145, "step": 1588 }, { "epoch": 1.1744916820702402, "grad_norm": 4.545044422149658, "learning_rate": 3.6282176047416732e-06, "loss": 0.021875519305467606, "step": 1589 }, { "epoch": 1.1752310536044361, "grad_norm": 4.641297817230225, "learning_rate": 3.6265643150578926e-06, "loss": 0.05604306235909462, "step": 1590 }, { "epoch": 1.1759704251386323, "grad_norm": 3.284341335296631, "learning_rate": 3.624910467221864e-06, "loss": 0.04456428810954094, "step": 1591 }, { "epoch": 1.1767097966728282, "grad_norm": 4.767324447631836, "learning_rate": 3.6232560622866725e-06, "loss": 0.022826502099633217, "step": 1592 }, { "epoch": 1.177449168207024, "grad_norm": 4.743930816650391, "learning_rate": 3.621601101305762e-06, "loss": 0.04117700457572937, "step": 1593 }, { "epoch": 1.17818853974122, "grad_norm": 3.423701524734497, "learning_rate": 3.6199455853329262e-06, "loss": 0.026070907711982727, "step": 1594 }, { "epoch": 1.1789279112754159, "grad_norm": 2.8132238388061523, "learning_rate": 3.618289515422318e-06, "loss": 0.021664179861545563, "step": 1595 }, { "epoch": 1.1796672828096118, "grad_norm": 3.5409963130950928, "learning_rate": 3.6166328926284384e-06, "loss": 0.06648053973913193, "step": 1596 }, { "epoch": 1.1804066543438076, "grad_norm": 3.421830892562866, "learning_rate": 3.6149757180061413e-06, "loss": 0.021374188363552094, "step": 1597 }, { "epoch": 1.1811460258780038, "grad_norm": 2.5615041255950928, "learning_rate": 3.613317992610633e-06, "loss": 0.013270225375890732, "step": 1598 }, { "epoch": 1.1818853974121997, "grad_norm": 4.3620429039001465, "learning_rate": 3.6116597174974705e-06, "loss": 0.05150116980075836, "step": 1599 }, { "epoch": 1.1826247689463956, "grad_norm": 4.804753303527832, "learning_rate": 3.6100008937225597e-06, "loss": 0.09965303540229797, "step": 1600 }, { "epoch": 1.1833641404805915, "grad_norm": 2.813279390335083, "learning_rate": 3.608341522342156e-06, "loss": 0.030204463750123978, "step": 1601 }, { "epoch": 1.1841035120147874, "grad_norm": 3.248621940612793, "learning_rate": 3.6066816044128645e-06, "loss": 0.056374918669462204, "step": 1602 }, { "epoch": 1.1848428835489835, "grad_norm": 4.718692779541016, "learning_rate": 3.6050211409916387e-06, "loss": 0.03185797482728958, "step": 1603 }, { "epoch": 1.1855822550831794, "grad_norm": 4.294806480407715, "learning_rate": 3.6033601331357766e-06, "loss": 0.041007332503795624, "step": 1604 }, { "epoch": 1.1863216266173753, "grad_norm": 4.885090351104736, "learning_rate": 3.6016985819029265e-06, "loss": 0.07812973856925964, "step": 1605 }, { "epoch": 1.1870609981515712, "grad_norm": 4.463866233825684, "learning_rate": 3.6000364883510804e-06, "loss": 0.051631614565849304, "step": 1606 }, { "epoch": 1.187800369685767, "grad_norm": 3.4745380878448486, "learning_rate": 3.5983738535385766e-06, "loss": 0.08781878650188446, "step": 1607 }, { "epoch": 1.188539741219963, "grad_norm": 4.146370887756348, "learning_rate": 3.5967106785240963e-06, "loss": 0.05277165398001671, "step": 1608 }, { "epoch": 1.1892791127541589, "grad_norm": 4.177722454071045, "learning_rate": 3.595046964366667e-06, "loss": 0.05835288017988205, "step": 1609 }, { "epoch": 1.190018484288355, "grad_norm": 3.4609451293945312, "learning_rate": 3.593382712125659e-06, "loss": 0.027741847559809685, "step": 1610 }, { "epoch": 1.190757855822551, "grad_norm": 4.47485876083374, "learning_rate": 3.5917179228607852e-06, "loss": 0.019743382930755615, "step": 1611 }, { "epoch": 1.1914972273567468, "grad_norm": 5.819014072418213, "learning_rate": 3.5900525976320984e-06, "loss": 0.0355045460164547, "step": 1612 }, { "epoch": 1.1922365988909427, "grad_norm": 4.220027446746826, "learning_rate": 3.588386737499995e-06, "loss": 0.052725017070770264, "step": 1613 }, { "epoch": 1.1929759704251386, "grad_norm": 3.984128713607788, "learning_rate": 3.5867203435252105e-06, "loss": 0.04878327623009682, "step": 1614 }, { "epoch": 1.1937153419593345, "grad_norm": 5.145803928375244, "learning_rate": 3.585053416768823e-06, "loss": 0.07606357336044312, "step": 1615 }, { "epoch": 1.1944547134935304, "grad_norm": 4.73368501663208, "learning_rate": 3.583385958292247e-06, "loss": 0.05140469968318939, "step": 1616 }, { "epoch": 1.1951940850277265, "grad_norm": 4.06217622756958, "learning_rate": 3.5817179691572362e-06, "loss": 0.0682159811258316, "step": 1617 }, { "epoch": 1.1959334565619224, "grad_norm": 3.781529664993286, "learning_rate": 3.580049450425882e-06, "loss": 0.05463150516152382, "step": 1618 }, { "epoch": 1.1966728280961183, "grad_norm": 4.047537326812744, "learning_rate": 3.5783804031606157e-06, "loss": 0.05228657275438309, "step": 1619 }, { "epoch": 1.1974121996303142, "grad_norm": 4.011489391326904, "learning_rate": 3.5767108284242003e-06, "loss": 0.050535015761852264, "step": 1620 }, { "epoch": 1.19815157116451, "grad_norm": 3.6123383045196533, "learning_rate": 3.5750407272797395e-06, "loss": 0.07487041503190994, "step": 1621 }, { "epoch": 1.1988909426987062, "grad_norm": 4.078595161437988, "learning_rate": 3.5733701007906696e-06, "loss": 0.0672132670879364, "step": 1622 }, { "epoch": 1.1996303142329021, "grad_norm": 5.218087673187256, "learning_rate": 3.5716989500207598e-06, "loss": 0.04338410124182701, "step": 1623 }, { "epoch": 1.200369685767098, "grad_norm": 3.45326566696167, "learning_rate": 3.570027276034119e-06, "loss": 0.04988257586956024, "step": 1624 }, { "epoch": 1.201109057301294, "grad_norm": 5.764541149139404, "learning_rate": 3.5683550798951823e-06, "loss": 0.03313326835632324, "step": 1625 }, { "epoch": 1.2018484288354898, "grad_norm": 4.506944179534912, "learning_rate": 3.5666823626687213e-06, "loss": 0.028018824756145477, "step": 1626 }, { "epoch": 1.2025878003696857, "grad_norm": 3.7844510078430176, "learning_rate": 3.5650091254198392e-06, "loss": 0.032436564564704895, "step": 1627 }, { "epoch": 1.2033271719038816, "grad_norm": 4.345643043518066, "learning_rate": 3.5633353692139687e-06, "loss": 0.050059739500284195, "step": 1628 }, { "epoch": 1.2040665434380777, "grad_norm": 4.959719181060791, "learning_rate": 3.561661095116875e-06, "loss": 0.05397345870733261, "step": 1629 }, { "epoch": 1.2048059149722736, "grad_norm": 4.199836730957031, "learning_rate": 3.5599863041946512e-06, "loss": 0.08072807639837265, "step": 1630 }, { "epoch": 1.2055452865064695, "grad_norm": 4.733626365661621, "learning_rate": 3.5583109975137215e-06, "loss": 0.04808652773499489, "step": 1631 }, { "epoch": 1.2062846580406654, "grad_norm": 4.205710411071777, "learning_rate": 3.5566351761408346e-06, "loss": 0.067138671875, "step": 1632 }, { "epoch": 1.2070240295748613, "grad_norm": 5.35701322555542, "learning_rate": 3.554958841143072e-06, "loss": 0.05827876925468445, "step": 1633 }, { "epoch": 1.2077634011090572, "grad_norm": 6.332939624786377, "learning_rate": 3.5532819935878393e-06, "loss": 0.043229736387729645, "step": 1634 }, { "epoch": 1.2085027726432531, "grad_norm": 4.482143402099609, "learning_rate": 3.551604634542869e-06, "loss": 0.06546908617019653, "step": 1635 }, { "epoch": 1.2092421441774492, "grad_norm": 3.707951545715332, "learning_rate": 3.549926765076219e-06, "loss": 0.027232065796852112, "step": 1636 }, { "epoch": 1.2099815157116451, "grad_norm": 4.799881935119629, "learning_rate": 3.548248386256272e-06, "loss": 0.05478432774543762, "step": 1637 }, { "epoch": 1.210720887245841, "grad_norm": 4.637964248657227, "learning_rate": 3.5465694991517367e-06, "loss": 0.045040763914585114, "step": 1638 }, { "epoch": 1.211460258780037, "grad_norm": 5.152524471282959, "learning_rate": 3.5448901048316447e-06, "loss": 0.06047637760639191, "step": 1639 }, { "epoch": 1.2121996303142328, "grad_norm": 3.4953043460845947, "learning_rate": 3.543210204365349e-06, "loss": 0.03467964380979538, "step": 1640 }, { "epoch": 1.212939001848429, "grad_norm": 4.618036270141602, "learning_rate": 3.5415297988225268e-06, "loss": 0.05030439794063568, "step": 1641 }, { "epoch": 1.2136783733826249, "grad_norm": 4.4850358963012695, "learning_rate": 3.5398488892731747e-06, "loss": 0.04376080259680748, "step": 1642 }, { "epoch": 1.2144177449168208, "grad_norm": 5.210366249084473, "learning_rate": 3.5381674767876136e-06, "loss": 0.0422753281891346, "step": 1643 }, { "epoch": 1.2151571164510167, "grad_norm": 4.28802490234375, "learning_rate": 3.5364855624364823e-06, "loss": 0.038576193153858185, "step": 1644 }, { "epoch": 1.2158964879852125, "grad_norm": 4.583132743835449, "learning_rate": 3.534803147290739e-06, "loss": 0.038191452622413635, "step": 1645 }, { "epoch": 1.2166358595194084, "grad_norm": 4.912837982177734, "learning_rate": 3.5331202324216617e-06, "loss": 0.03159748762845993, "step": 1646 }, { "epoch": 1.2173752310536043, "grad_norm": 5.249161243438721, "learning_rate": 3.531436818900846e-06, "loss": 0.03312121331691742, "step": 1647 }, { "epoch": 1.2181146025878005, "grad_norm": 4.988842964172363, "learning_rate": 3.5297529078002056e-06, "loss": 0.014639316126704216, "step": 1648 }, { "epoch": 1.2188539741219964, "grad_norm": 5.244650840759277, "learning_rate": 3.5280685001919705e-06, "loss": 0.047078102827072144, "step": 1649 }, { "epoch": 1.2195933456561923, "grad_norm": 6.9011101722717285, "learning_rate": 3.5263835971486885e-06, "loss": 0.07607591152191162, "step": 1650 }, { "epoch": 1.2203327171903882, "grad_norm": 4.432397842407227, "learning_rate": 3.52469819974322e-06, "loss": 0.05443239212036133, "step": 1651 }, { "epoch": 1.221072088724584, "grad_norm": 4.4911580085754395, "learning_rate": 3.5230123090487413e-06, "loss": 0.03275042772293091, "step": 1652 }, { "epoch": 1.22181146025878, "grad_norm": 5.737323760986328, "learning_rate": 3.521325926138744e-06, "loss": 0.04259651526808739, "step": 1653 }, { "epoch": 1.2225508317929759, "grad_norm": 5.020395278930664, "learning_rate": 3.5196390520870326e-06, "loss": 0.06604944169521332, "step": 1654 }, { "epoch": 1.223290203327172, "grad_norm": 3.8429818153381348, "learning_rate": 3.5179516879677227e-06, "loss": 0.033076442778110504, "step": 1655 }, { "epoch": 1.2240295748613679, "grad_norm": 4.406617641448975, "learning_rate": 3.5162638348552445e-06, "loss": 0.044135525822639465, "step": 1656 }, { "epoch": 1.2247689463955638, "grad_norm": 5.199479103088379, "learning_rate": 3.5145754938243377e-06, "loss": 0.06421215832233429, "step": 1657 }, { "epoch": 1.2255083179297597, "grad_norm": 5.079687595367432, "learning_rate": 3.5128866659500534e-06, "loss": 0.04744883254170418, "step": 1658 }, { "epoch": 1.2262476894639556, "grad_norm": 4.131003379821777, "learning_rate": 3.511197352307753e-06, "loss": 0.07647845149040222, "step": 1659 }, { "epoch": 1.2269870609981517, "grad_norm": 4.536802768707275, "learning_rate": 3.509507553973107e-06, "loss": 0.037001509219408035, "step": 1660 }, { "epoch": 1.2277264325323476, "grad_norm": 4.474244117736816, "learning_rate": 3.507817272022092e-06, "loss": 0.03175854682922363, "step": 1661 }, { "epoch": 1.2284658040665435, "grad_norm": 5.868588924407959, "learning_rate": 3.5061265075309976e-06, "loss": 0.044160861521959305, "step": 1662 }, { "epoch": 1.2292051756007394, "grad_norm": 3.320564031600952, "learning_rate": 3.504435261576417e-06, "loss": 0.03750721365213394, "step": 1663 }, { "epoch": 1.2299445471349353, "grad_norm": 3.7511751651763916, "learning_rate": 3.502743535235251e-06, "loss": 0.039905428886413574, "step": 1664 }, { "epoch": 1.2306839186691312, "grad_norm": 4.706625938415527, "learning_rate": 3.501051329584707e-06, "loss": 0.03475469350814819, "step": 1665 }, { "epoch": 1.231423290203327, "grad_norm": 6.361069202423096, "learning_rate": 3.499358645702294e-06, "loss": 0.07778972387313843, "step": 1666 }, { "epoch": 1.2321626617375232, "grad_norm": 3.8132588863372803, "learning_rate": 3.4976654846658316e-06, "loss": 0.06861699372529984, "step": 1667 }, { "epoch": 1.232902033271719, "grad_norm": 3.391822099685669, "learning_rate": 3.4959718475534375e-06, "loss": 0.010053792968392372, "step": 1668 }, { "epoch": 1.233641404805915, "grad_norm": 4.740298271179199, "learning_rate": 3.4942777354435378e-06, "loss": 0.0381120927631855, "step": 1669 }, { "epoch": 1.234380776340111, "grad_norm": 3.0367586612701416, "learning_rate": 3.4925831494148556e-06, "loss": 0.02145843580365181, "step": 1670 }, { "epoch": 1.2351201478743068, "grad_norm": 3.4320263862609863, "learning_rate": 3.490888090546419e-06, "loss": 0.026334280148148537, "step": 1671 }, { "epoch": 1.2358595194085027, "grad_norm": 4.539562702178955, "learning_rate": 3.4891925599175567e-06, "loss": 0.0454818457365036, "step": 1672 }, { "epoch": 1.2365988909426986, "grad_norm": 5.070994853973389, "learning_rate": 3.487496558607898e-06, "loss": 0.04190422222018242, "step": 1673 }, { "epoch": 1.2373382624768947, "grad_norm": 4.751770496368408, "learning_rate": 3.4858000876973717e-06, "loss": 0.03426140174269676, "step": 1674 }, { "epoch": 1.2380776340110906, "grad_norm": 5.203316688537598, "learning_rate": 3.4841031482662053e-06, "loss": 0.05876597762107849, "step": 1675 }, { "epoch": 1.2388170055452865, "grad_norm": 4.415191650390625, "learning_rate": 3.4824057413949243e-06, "loss": 0.034823667258024216, "step": 1676 }, { "epoch": 1.2395563770794824, "grad_norm": 3.7515628337860107, "learning_rate": 3.480707868164353e-06, "loss": 0.053972434252500534, "step": 1677 }, { "epoch": 1.2402957486136783, "grad_norm": 3.188796043395996, "learning_rate": 3.4790095296556126e-06, "loss": 0.024889256805181503, "step": 1678 }, { "epoch": 1.2410351201478744, "grad_norm": 3.4816176891326904, "learning_rate": 3.4773107269501176e-06, "loss": 0.049791838973760605, "step": 1679 }, { "epoch": 1.2417744916820703, "grad_norm": 3.5281319618225098, "learning_rate": 3.475611461129584e-06, "loss": 0.05456764996051788, "step": 1680 }, { "epoch": 1.2425138632162662, "grad_norm": 4.483880043029785, "learning_rate": 3.473911733276016e-06, "loss": 0.0736626386642456, "step": 1681 }, { "epoch": 1.2432532347504621, "grad_norm": 4.584135055541992, "learning_rate": 3.472211544471717e-06, "loss": 0.032800376415252686, "step": 1682 }, { "epoch": 1.243992606284658, "grad_norm": 3.83801007270813, "learning_rate": 3.4705108957992818e-06, "loss": 0.036840833723545074, "step": 1683 }, { "epoch": 1.244731977818854, "grad_norm": 4.903773784637451, "learning_rate": 3.468809788341597e-06, "loss": 0.073693186044693, "step": 1684 }, { "epoch": 1.2454713493530498, "grad_norm": 5.35080623626709, "learning_rate": 3.467108223181844e-06, "loss": 0.05486017465591431, "step": 1685 }, { "epoch": 1.246210720887246, "grad_norm": 4.156801223754883, "learning_rate": 3.4654062014034946e-06, "loss": 0.0514569990336895, "step": 1686 }, { "epoch": 1.2469500924214418, "grad_norm": 4.248398780822754, "learning_rate": 3.4637037240903097e-06, "loss": 0.03426332771778107, "step": 1687 }, { "epoch": 1.2476894639556377, "grad_norm": 4.059643745422363, "learning_rate": 3.4620007923263427e-06, "loss": 0.044772014021873474, "step": 1688 }, { "epoch": 1.2484288354898336, "grad_norm": 4.4526872634887695, "learning_rate": 3.4602974071959345e-06, "loss": 0.05782069265842438, "step": 1689 }, { "epoch": 1.2491682070240295, "grad_norm": 3.7669477462768555, "learning_rate": 3.4585935697837148e-06, "loss": 0.01950206607580185, "step": 1690 }, { "epoch": 1.2499075785582254, "grad_norm": 3.671632766723633, "learning_rate": 3.456889281174603e-06, "loss": 0.021865859627723694, "step": 1691 }, { "epoch": 1.2506469500924213, "grad_norm": 3.1558001041412354, "learning_rate": 3.4551845424538055e-06, "loss": 0.015344541519880295, "step": 1692 }, { "epoch": 1.2513863216266174, "grad_norm": 4.5838823318481445, "learning_rate": 3.453479354706813e-06, "loss": 0.052923597395420074, "step": 1693 }, { "epoch": 1.2521256931608133, "grad_norm": 3.995950698852539, "learning_rate": 3.451773719019405e-06, "loss": 0.05472055822610855, "step": 1694 }, { "epoch": 1.2528650646950092, "grad_norm": 5.126954555511475, "learning_rate": 3.4500676364776424e-06, "loss": 0.09449449181556702, "step": 1695 }, { "epoch": 1.2536044362292051, "grad_norm": 4.6259942054748535, "learning_rate": 3.448361108167876e-06, "loss": 0.03852273151278496, "step": 1696 }, { "epoch": 1.254343807763401, "grad_norm": 3.654628276824951, "learning_rate": 3.4466541351767353e-06, "loss": 0.020222794264554977, "step": 1697 }, { "epoch": 1.2550831792975972, "grad_norm": 6.997132301330566, "learning_rate": 3.444946718591137e-06, "loss": 0.07769189029932022, "step": 1698 }, { "epoch": 1.2558225508317928, "grad_norm": 6.565527439117432, "learning_rate": 3.443238859498278e-06, "loss": 0.06530594825744629, "step": 1699 }, { "epoch": 1.256561922365989, "grad_norm": 4.412631034851074, "learning_rate": 3.4415305589856364e-06, "loss": 0.034206561744213104, "step": 1700 }, { "epoch": 1.2573012939001849, "grad_norm": 4.59832239151001, "learning_rate": 3.4398218181409728e-06, "loss": 0.02699708193540573, "step": 1701 }, { "epoch": 1.2580406654343808, "grad_norm": 3.8585071563720703, "learning_rate": 3.4381126380523285e-06, "loss": 0.049002841114997864, "step": 1702 }, { "epoch": 1.2587800369685767, "grad_norm": 4.627016544342041, "learning_rate": 3.4364030198080236e-06, "loss": 0.02653808519244194, "step": 1703 }, { "epoch": 1.2595194085027726, "grad_norm": 4.211055278778076, "learning_rate": 3.4346929644966563e-06, "loss": 0.05341026186943054, "step": 1704 }, { "epoch": 1.2602587800369687, "grad_norm": 3.267509698867798, "learning_rate": 3.432982473207106e-06, "loss": 0.03004801645874977, "step": 1705 }, { "epoch": 1.2609981515711646, "grad_norm": 3.987762451171875, "learning_rate": 3.4312715470285263e-06, "loss": 0.014309546910226345, "step": 1706 }, { "epoch": 1.2617375231053605, "grad_norm": 5.205948352813721, "learning_rate": 3.4295601870503496e-06, "loss": 0.09598828852176666, "step": 1707 }, { "epoch": 1.2624768946395564, "grad_norm": 3.9246180057525635, "learning_rate": 3.427848394362285e-06, "loss": 0.05853836238384247, "step": 1708 }, { "epoch": 1.2632162661737523, "grad_norm": 3.380101442337036, "learning_rate": 3.4261361700543154e-06, "loss": 0.03801044821739197, "step": 1709 }, { "epoch": 1.2639556377079482, "grad_norm": 4.1062092781066895, "learning_rate": 3.4244235152167e-06, "loss": 0.07209949940443039, "step": 1710 }, { "epoch": 1.264695009242144, "grad_norm": 4.295192718505859, "learning_rate": 3.4227104309399717e-06, "loss": 0.048080358654260635, "step": 1711 }, { "epoch": 1.2654343807763402, "grad_norm": 3.5375428199768066, "learning_rate": 3.420996918314936e-06, "loss": 0.06518437713384628, "step": 1712 }, { "epoch": 1.266173752310536, "grad_norm": 4.166370391845703, "learning_rate": 3.4192829784326732e-06, "loss": 0.04291682690382004, "step": 1713 }, { "epoch": 1.266913123844732, "grad_norm": 3.665632724761963, "learning_rate": 3.4175686123845324e-06, "loss": 0.030556289479136467, "step": 1714 }, { "epoch": 1.2676524953789279, "grad_norm": 3.9172723293304443, "learning_rate": 3.415853821262138e-06, "loss": 0.05603163689374924, "step": 1715 }, { "epoch": 1.2683918669131238, "grad_norm": 4.972112655639648, "learning_rate": 3.4141386061573822e-06, "loss": 0.06377255916595459, "step": 1716 }, { "epoch": 1.26913123844732, "grad_norm": 3.496056318283081, "learning_rate": 3.4124229681624288e-06, "loss": 0.06310956925153732, "step": 1717 }, { "epoch": 1.2698706099815156, "grad_norm": 3.5217227935791016, "learning_rate": 3.410706908369709e-06, "loss": 0.03947349637746811, "step": 1718 }, { "epoch": 1.2706099815157117, "grad_norm": 3.6419568061828613, "learning_rate": 3.4089904278719243e-06, "loss": 0.054214440286159515, "step": 1719 }, { "epoch": 1.2713493530499076, "grad_norm": 3.3619019985198975, "learning_rate": 3.4072735277620435e-06, "loss": 0.05851227045059204, "step": 1720 }, { "epoch": 1.2720887245841035, "grad_norm": 4.920952796936035, "learning_rate": 3.405556209133302e-06, "loss": 0.04873725026845932, "step": 1721 }, { "epoch": 1.2728280961182994, "grad_norm": 4.2338433265686035, "learning_rate": 3.403838473079204e-06, "loss": 0.043531984090805054, "step": 1722 }, { "epoch": 1.2735674676524953, "grad_norm": 4.196835517883301, "learning_rate": 3.402120320693516e-06, "loss": 0.03516124188899994, "step": 1723 }, { "epoch": 1.2743068391866914, "grad_norm": 4.126051425933838, "learning_rate": 3.400401753070272e-06, "loss": 0.027111945673823357, "step": 1724 }, { "epoch": 1.2750462107208873, "grad_norm": 4.380075931549072, "learning_rate": 3.3986827713037698e-06, "loss": 0.03489740937948227, "step": 1725 }, { "epoch": 1.2757855822550832, "grad_norm": 4.61604642868042, "learning_rate": 3.3969633764885707e-06, "loss": 0.024856679141521454, "step": 1726 }, { "epoch": 1.276524953789279, "grad_norm": 4.290892601013184, "learning_rate": 3.3952435697194995e-06, "loss": 0.04364258423447609, "step": 1727 }, { "epoch": 1.277264325323475, "grad_norm": 4.126423358917236, "learning_rate": 3.3935233520916432e-06, "loss": 0.03021683357656002, "step": 1728 }, { "epoch": 1.278003696857671, "grad_norm": 3.850780487060547, "learning_rate": 3.3918027247003493e-06, "loss": 0.028641577810049057, "step": 1729 }, { "epoch": 1.2787430683918668, "grad_norm": 3.756622076034546, "learning_rate": 3.390081688641228e-06, "loss": 0.033767782151699066, "step": 1730 }, { "epoch": 1.279482439926063, "grad_norm": 3.2558655738830566, "learning_rate": 3.388360245010149e-06, "loss": 0.01572289690375328, "step": 1731 }, { "epoch": 1.2802218114602588, "grad_norm": 3.882073163986206, "learning_rate": 3.3866383949032404e-06, "loss": 0.04392088204622269, "step": 1732 }, { "epoch": 1.2809611829944547, "grad_norm": 5.556154727935791, "learning_rate": 3.3849161394168907e-06, "loss": 0.049150288105010986, "step": 1733 }, { "epoch": 1.2817005545286506, "grad_norm": 4.529847145080566, "learning_rate": 3.3831934796477454e-06, "loss": 0.0345187708735466, "step": 1734 }, { "epoch": 1.2824399260628465, "grad_norm": 4.262209892272949, "learning_rate": 3.38147041669271e-06, "loss": 0.04159713536500931, "step": 1735 }, { "epoch": 1.2831792975970426, "grad_norm": 4.320438385009766, "learning_rate": 3.379746951648942e-06, "loss": 0.042020805180072784, "step": 1736 }, { "epoch": 1.2839186691312383, "grad_norm": 4.225905418395996, "learning_rate": 3.378023085613859e-06, "loss": 0.051368117332458496, "step": 1737 }, { "epoch": 1.2846580406654344, "grad_norm": 4.808729648590088, "learning_rate": 3.3762988196851313e-06, "loss": 0.04594305902719498, "step": 1738 }, { "epoch": 1.2853974121996303, "grad_norm": 5.0824761390686035, "learning_rate": 3.3745741549606877e-06, "loss": 0.05065502971410751, "step": 1739 }, { "epoch": 1.2861367837338262, "grad_norm": 5.466503143310547, "learning_rate": 3.3728490925387057e-06, "loss": 0.05452965945005417, "step": 1740 }, { "epoch": 1.2868761552680221, "grad_norm": 3.7270150184631348, "learning_rate": 3.3711236335176204e-06, "loss": 0.05116987228393555, "step": 1741 }, { "epoch": 1.287615526802218, "grad_norm": 4.10521936416626, "learning_rate": 3.369397778996116e-06, "loss": 0.06185043603181839, "step": 1742 }, { "epoch": 1.2883548983364141, "grad_norm": 3.2292590141296387, "learning_rate": 3.367671530073132e-06, "loss": 0.04634600132703781, "step": 1743 }, { "epoch": 1.28909426987061, "grad_norm": 2.914355754852295, "learning_rate": 3.365944887847855e-06, "loss": 0.03180009126663208, "step": 1744 }, { "epoch": 1.289833641404806, "grad_norm": 3.1051673889160156, "learning_rate": 3.3642178534197266e-06, "loss": 0.03441058099269867, "step": 1745 }, { "epoch": 1.2905730129390018, "grad_norm": 2.823819398880005, "learning_rate": 3.362490427888435e-06, "loss": 0.02606029622256756, "step": 1746 }, { "epoch": 1.2913123844731977, "grad_norm": 4.084022521972656, "learning_rate": 3.360762612353918e-06, "loss": 0.04997067153453827, "step": 1747 }, { "epoch": 1.2920517560073936, "grad_norm": 3.7839322090148926, "learning_rate": 3.3590344079163617e-06, "loss": 0.029296647757291794, "step": 1748 }, { "epoch": 1.2927911275415895, "grad_norm": 3.569840431213379, "learning_rate": 3.3573058156762013e-06, "loss": 0.00987438764423132, "step": 1749 }, { "epoch": 1.2935304990757857, "grad_norm": 5.210137367248535, "learning_rate": 3.3555768367341158e-06, "loss": 0.058630213141441345, "step": 1750 }, { "epoch": 1.2942698706099816, "grad_norm": 4.781435012817383, "learning_rate": 3.353847472191034e-06, "loss": 0.06960426270961761, "step": 1751 }, { "epoch": 1.2950092421441775, "grad_norm": 5.696279525756836, "learning_rate": 3.3521177231481284e-06, "loss": 0.03422160819172859, "step": 1752 }, { "epoch": 1.2957486136783734, "grad_norm": 4.133578777313232, "learning_rate": 3.3503875907068166e-06, "loss": 0.05358908697962761, "step": 1753 }, { "epoch": 1.2964879852125692, "grad_norm": 4.150849342346191, "learning_rate": 3.3486570759687597e-06, "loss": 0.06500653177499771, "step": 1754 }, { "epoch": 1.2972273567467654, "grad_norm": 3.313936710357666, "learning_rate": 3.346926180035863e-06, "loss": 0.031044665724039078, "step": 1755 }, { "epoch": 1.297966728280961, "grad_norm": 3.5787172317504883, "learning_rate": 3.3451949040102744e-06, "loss": 0.013787534087896347, "step": 1756 }, { "epoch": 1.2987060998151572, "grad_norm": 4.388860702514648, "learning_rate": 3.3434632489943836e-06, "loss": 0.04443540424108505, "step": 1757 }, { "epoch": 1.299445471349353, "grad_norm": 3.4837167263031006, "learning_rate": 3.341731216090822e-06, "loss": 0.04388009384274483, "step": 1758 }, { "epoch": 1.300184842883549, "grad_norm": 4.4583563804626465, "learning_rate": 3.3399988064024625e-06, "loss": 0.047968070954084396, "step": 1759 }, { "epoch": 1.3009242144177449, "grad_norm": 5.493947982788086, "learning_rate": 3.338266021032415e-06, "loss": 0.05226663500070572, "step": 1760 }, { "epoch": 1.3016635859519408, "grad_norm": 4.212222576141357, "learning_rate": 3.33653286108403e-06, "loss": 0.04505445808172226, "step": 1761 }, { "epoch": 1.3024029574861369, "grad_norm": 4.990040302276611, "learning_rate": 3.334799327660899e-06, "loss": 0.030473804101347923, "step": 1762 }, { "epoch": 1.3031423290203328, "grad_norm": 4.693982124328613, "learning_rate": 3.3330654218668486e-06, "loss": 0.07144170999526978, "step": 1763 }, { "epoch": 1.3038817005545287, "grad_norm": 3.511174440383911, "learning_rate": 3.3313311448059426e-06, "loss": 0.0560251846909523, "step": 1764 }, { "epoch": 1.3046210720887246, "grad_norm": 3.161210298538208, "learning_rate": 3.3295964975824823e-06, "loss": 0.023144172504544258, "step": 1765 }, { "epoch": 1.3053604436229205, "grad_norm": 5.274788856506348, "learning_rate": 3.3278614813010036e-06, "loss": 0.03979238122701645, "step": 1766 }, { "epoch": 1.3060998151571164, "grad_norm": 3.5781753063201904, "learning_rate": 3.3261260970662773e-06, "loss": 0.0458076149225235, "step": 1767 }, { "epoch": 1.3068391866913123, "grad_norm": 4.534839153289795, "learning_rate": 3.324390345983311e-06, "loss": 0.03560962900519371, "step": 1768 }, { "epoch": 1.3075785582255084, "grad_norm": 4.5514607429504395, "learning_rate": 3.3226542291573443e-06, "loss": 0.05645601451396942, "step": 1769 }, { "epoch": 1.3083179297597043, "grad_norm": 5.135239601135254, "learning_rate": 3.320917747693847e-06, "loss": 0.07008571922779083, "step": 1770 }, { "epoch": 1.3090573012939002, "grad_norm": 3.786862850189209, "learning_rate": 3.3191809026985242e-06, "loss": 0.04564914107322693, "step": 1771 }, { "epoch": 1.309796672828096, "grad_norm": 3.8029234409332275, "learning_rate": 3.3174436952773135e-06, "loss": 0.017453622072935104, "step": 1772 }, { "epoch": 1.310536044362292, "grad_norm": 5.286562442779541, "learning_rate": 3.3157061265363805e-06, "loss": 0.0631764680147171, "step": 1773 }, { "epoch": 1.311275415896488, "grad_norm": 4.614762306213379, "learning_rate": 3.3139681975821214e-06, "loss": 0.0241777915507555, "step": 1774 }, { "epoch": 1.3120147874306838, "grad_norm": 4.5521674156188965, "learning_rate": 3.312229909521163e-06, "loss": 0.044854938983917236, "step": 1775 }, { "epoch": 1.31275415896488, "grad_norm": 4.58071231842041, "learning_rate": 3.31049126346036e-06, "loss": 0.03635796159505844, "step": 1776 }, { "epoch": 1.3134935304990758, "grad_norm": 3.384355306625366, "learning_rate": 3.308752260506795e-06, "loss": 0.04358760267496109, "step": 1777 }, { "epoch": 1.3142329020332717, "grad_norm": 4.781922340393066, "learning_rate": 3.307012901767778e-06, "loss": 0.06818266212940216, "step": 1778 }, { "epoch": 1.3149722735674676, "grad_norm": 3.635669231414795, "learning_rate": 3.3052731883508454e-06, "loss": 0.043879590928554535, "step": 1779 }, { "epoch": 1.3157116451016635, "grad_norm": 3.2005958557128906, "learning_rate": 3.30353312136376e-06, "loss": 0.05378726124763489, "step": 1780 }, { "epoch": 1.3164510166358596, "grad_norm": 3.661566734313965, "learning_rate": 3.301792701914508e-06, "loss": 0.030277729034423828, "step": 1781 }, { "epoch": 1.3171903881700555, "grad_norm": 4.118501663208008, "learning_rate": 3.3000519311113033e-06, "loss": 0.056799858808517456, "step": 1782 }, { "epoch": 1.3179297597042514, "grad_norm": 3.978651523590088, "learning_rate": 3.2983108100625804e-06, "loss": 0.028682535514235497, "step": 1783 }, { "epoch": 1.3186691312384473, "grad_norm": 4.309295654296875, "learning_rate": 3.296569339876998e-06, "loss": 0.040621235966682434, "step": 1784 }, { "epoch": 1.3194085027726432, "grad_norm": 3.7804534435272217, "learning_rate": 3.294827521663437e-06, "loss": 0.03452149033546448, "step": 1785 }, { "epoch": 1.320147874306839, "grad_norm": 3.9435226917266846, "learning_rate": 3.293085356531e-06, "loss": 0.04753955081105232, "step": 1786 }, { "epoch": 1.320887245841035, "grad_norm": 3.538787603378296, "learning_rate": 3.291342845589011e-06, "loss": 0.02370123192667961, "step": 1787 }, { "epoch": 1.3216266173752311, "grad_norm": 4.119046688079834, "learning_rate": 3.2895999899470143e-06, "loss": 0.06080850213766098, "step": 1788 }, { "epoch": 1.322365988909427, "grad_norm": 3.75650954246521, "learning_rate": 3.2878567907147714e-06, "loss": 0.03982203081250191, "step": 1789 }, { "epoch": 1.323105360443623, "grad_norm": 3.866591691970825, "learning_rate": 3.2861132490022644e-06, "loss": 0.03350983187556267, "step": 1790 }, { "epoch": 1.3238447319778188, "grad_norm": 3.512105703353882, "learning_rate": 3.2843693659196944e-06, "loss": 0.03704433515667915, "step": 1791 }, { "epoch": 1.3245841035120147, "grad_norm": 5.023584365844727, "learning_rate": 3.2826251425774788e-06, "loss": 0.03492205962538719, "step": 1792 }, { "epoch": 1.3253234750462108, "grad_norm": 3.1005797386169434, "learning_rate": 3.280880580086251e-06, "loss": 0.01204571221023798, "step": 1793 }, { "epoch": 1.3260628465804065, "grad_norm": 5.75824499130249, "learning_rate": 3.2791356795568615e-06, "loss": 0.09049466252326965, "step": 1794 }, { "epoch": 1.3268022181146026, "grad_norm": 4.310823440551758, "learning_rate": 3.277390442100374e-06, "loss": 0.09420497715473175, "step": 1795 }, { "epoch": 1.3275415896487985, "grad_norm": 3.1543054580688477, "learning_rate": 3.2756448688280706e-06, "loss": 0.030765991657972336, "step": 1796 }, { "epoch": 1.3282809611829944, "grad_norm": 4.70048713684082, "learning_rate": 3.273898960851443e-06, "loss": 0.021918419748544693, "step": 1797 }, { "epoch": 1.3290203327171903, "grad_norm": 4.074340343475342, "learning_rate": 3.2721527192822e-06, "loss": 0.029995206743478775, "step": 1798 }, { "epoch": 1.3297597042513862, "grad_norm": 4.472945213317871, "learning_rate": 3.270406145232259e-06, "loss": 0.020319774746894836, "step": 1799 }, { "epoch": 1.3304990757855824, "grad_norm": 3.542631149291992, "learning_rate": 3.268659239813752e-06, "loss": 0.03948182612657547, "step": 1800 }, { "epoch": 1.3312384473197783, "grad_norm": 3.329470157623291, "learning_rate": 3.2669120041390196e-06, "loss": 0.02917860448360443, "step": 1801 }, { "epoch": 1.3319778188539741, "grad_norm": 3.4645564556121826, "learning_rate": 3.2651644393206155e-06, "loss": 0.04709656909108162, "step": 1802 }, { "epoch": 1.33271719038817, "grad_norm": 3.613194227218628, "learning_rate": 3.2634165464712998e-06, "loss": 0.03617327660322189, "step": 1803 }, { "epoch": 1.333456561922366, "grad_norm": 3.821059465408325, "learning_rate": 3.2616683267040446e-06, "loss": 0.042008932679891586, "step": 1804 }, { "epoch": 1.3341959334565618, "grad_norm": 3.449082851409912, "learning_rate": 3.259919781132027e-06, "loss": 0.032718174159526825, "step": 1805 }, { "epoch": 1.3349353049907577, "grad_norm": 3.817267894744873, "learning_rate": 3.2581709108686354e-06, "loss": 0.04439918324351311, "step": 1806 }, { "epoch": 1.3356746765249539, "grad_norm": 2.9120864868164062, "learning_rate": 3.2564217170274614e-06, "loss": 0.011206596158444881, "step": 1807 }, { "epoch": 1.3364140480591498, "grad_norm": 4.678460121154785, "learning_rate": 3.2546722007223047e-06, "loss": 0.053602516651153564, "step": 1808 }, { "epoch": 1.3371534195933457, "grad_norm": 4.894521713256836, "learning_rate": 3.252922363067168e-06, "loss": 0.051415637135505676, "step": 1809 }, { "epoch": 1.3378927911275416, "grad_norm": 4.4334259033203125, "learning_rate": 3.251172205176262e-06, "loss": 0.06196487694978714, "step": 1810 }, { "epoch": 1.3386321626617375, "grad_norm": 3.4015538692474365, "learning_rate": 3.2494217281639994e-06, "loss": 0.04067976027727127, "step": 1811 }, { "epoch": 1.3393715341959336, "grad_norm": 4.524235725402832, "learning_rate": 3.2476709331449957e-06, "loss": 0.02215833216905594, "step": 1812 }, { "epoch": 1.3401109057301293, "grad_norm": 3.951145648956299, "learning_rate": 3.2459198212340707e-06, "loss": 0.02938673458993435, "step": 1813 }, { "epoch": 1.3408502772643254, "grad_norm": 5.134332656860352, "learning_rate": 3.244168393546242e-06, "loss": 0.033017341047525406, "step": 1814 }, { "epoch": 1.3415896487985213, "grad_norm": 3.4407999515533447, "learning_rate": 3.242416651196734e-06, "loss": 0.042460668832063675, "step": 1815 }, { "epoch": 1.3423290203327172, "grad_norm": 3.742039918899536, "learning_rate": 3.240664595300967e-06, "loss": 0.030449826270341873, "step": 1816 }, { "epoch": 1.343068391866913, "grad_norm": 4.423660755157471, "learning_rate": 3.238912226974563e-06, "loss": 0.05324477702379227, "step": 1817 }, { "epoch": 1.343807763401109, "grad_norm": 3.979867935180664, "learning_rate": 3.237159547333342e-06, "loss": 0.023329421877861023, "step": 1818 }, { "epoch": 1.344547134935305, "grad_norm": 3.356274366378784, "learning_rate": 3.2354065574933225e-06, "loss": 0.04046935960650444, "step": 1819 }, { "epoch": 1.345286506469501, "grad_norm": 4.129702568054199, "learning_rate": 3.2336532585707203e-06, "loss": 0.03641985356807709, "step": 1820 }, { "epoch": 1.3460258780036969, "grad_norm": 5.111711502075195, "learning_rate": 3.2318996516819505e-06, "loss": 0.045784808695316315, "step": 1821 }, { "epoch": 1.3467652495378928, "grad_norm": 3.5794153213500977, "learning_rate": 3.2301457379436197e-06, "loss": 0.0257391557097435, "step": 1822 }, { "epoch": 1.3475046210720887, "grad_norm": 4.222311019897461, "learning_rate": 3.228391518472533e-06, "loss": 0.059843845665454865, "step": 1823 }, { "epoch": 1.3482439926062846, "grad_norm": 4.048084259033203, "learning_rate": 3.2266369943856896e-06, "loss": 0.04686416685581207, "step": 1824 }, { "epoch": 1.3489833641404805, "grad_norm": 3.213160514831543, "learning_rate": 3.2248821668002824e-06, "loss": 0.02497437782585621, "step": 1825 }, { "epoch": 1.3497227356746766, "grad_norm": 2.5641860961914062, "learning_rate": 3.2231270368336976e-06, "loss": 0.01966736651957035, "step": 1826 }, { "epoch": 1.3504621072088725, "grad_norm": 4.300053596496582, "learning_rate": 3.2213716056035148e-06, "loss": 0.024927327409386635, "step": 1827 }, { "epoch": 1.3512014787430684, "grad_norm": 4.060112953186035, "learning_rate": 3.219615874227505e-06, "loss": 0.03845725953578949, "step": 1828 }, { "epoch": 1.3519408502772643, "grad_norm": 5.539316654205322, "learning_rate": 3.2178598438236276e-06, "loss": 0.05044252797961235, "step": 1829 }, { "epoch": 1.3526802218114602, "grad_norm": 3.796105146408081, "learning_rate": 3.2161035155100372e-06, "loss": 0.036312587559223175, "step": 1830 }, { "epoch": 1.3534195933456563, "grad_norm": 4.496875762939453, "learning_rate": 3.2143468904050752e-06, "loss": 0.02534867450594902, "step": 1831 }, { "epoch": 1.354158964879852, "grad_norm": 4.877654075622559, "learning_rate": 3.212589969627271e-06, "loss": 0.024930361658334732, "step": 1832 }, { "epoch": 1.354898336414048, "grad_norm": 5.153099536895752, "learning_rate": 3.2108327542953467e-06, "loss": 0.054404277354478836, "step": 1833 }, { "epoch": 1.355637707948244, "grad_norm": 4.414557456970215, "learning_rate": 3.2090752455282053e-06, "loss": 0.02811371721327305, "step": 1834 }, { "epoch": 1.35637707948244, "grad_norm": 1.184424638748169, "learning_rate": 3.2073174444449434e-06, "loss": 0.008057434111833572, "step": 1835 }, { "epoch": 1.3571164510166358, "grad_norm": 4.189840316772461, "learning_rate": 3.20555935216484e-06, "loss": 0.0287184938788414, "step": 1836 }, { "epoch": 1.3578558225508317, "grad_norm": 3.6070220470428467, "learning_rate": 3.203800969807359e-06, "loss": 0.009721175767481327, "step": 1837 }, { "epoch": 1.3585951940850278, "grad_norm": 4.487643241882324, "learning_rate": 3.202042298492151e-06, "loss": 0.031913287937641144, "step": 1838 }, { "epoch": 1.3593345656192237, "grad_norm": 4.381163597106934, "learning_rate": 3.20028333933905e-06, "loss": 0.02599390596151352, "step": 1839 }, { "epoch": 1.3600739371534196, "grad_norm": 4.036530017852783, "learning_rate": 3.198524093468072e-06, "loss": 0.03891151025891304, "step": 1840 }, { "epoch": 1.3608133086876155, "grad_norm": 3.666299343109131, "learning_rate": 3.1967645619994194e-06, "loss": 0.045582108199596405, "step": 1841 }, { "epoch": 1.3615526802218114, "grad_norm": 3.8531851768493652, "learning_rate": 3.1950047460534706e-06, "loss": 0.04099529981613159, "step": 1842 }, { "epoch": 1.3622920517560073, "grad_norm": 4.038908004760742, "learning_rate": 3.1932446467507894e-06, "loss": 0.04748481512069702, "step": 1843 }, { "epoch": 1.3630314232902032, "grad_norm": 3.0914454460144043, "learning_rate": 3.1914842652121185e-06, "loss": 0.04399260878562927, "step": 1844 }, { "epoch": 1.3637707948243993, "grad_norm": 4.055509567260742, "learning_rate": 3.1897236025583823e-06, "loss": 0.034545592963695526, "step": 1845 }, { "epoch": 1.3645101663585952, "grad_norm": 4.436142921447754, "learning_rate": 3.187962659910681e-06, "loss": 0.04009518027305603, "step": 1846 }, { "epoch": 1.3652495378927911, "grad_norm": 2.791486978530884, "learning_rate": 3.186201438390295e-06, "loss": 0.0381295382976532, "step": 1847 }, { "epoch": 1.365988909426987, "grad_norm": 5.266452312469482, "learning_rate": 3.184439939118682e-06, "loss": 0.044539064168930054, "step": 1848 }, { "epoch": 1.366728280961183, "grad_norm": 3.9474387168884277, "learning_rate": 3.182678163217477e-06, "loss": 0.04774560034275055, "step": 1849 }, { "epoch": 1.367467652495379, "grad_norm": 3.4582157135009766, "learning_rate": 3.180916111808489e-06, "loss": 0.03164920210838318, "step": 1850 }, { "epoch": 1.3682070240295747, "grad_norm": 5.083932399749756, "learning_rate": 3.1791537860137065e-06, "loss": 0.037624672055244446, "step": 1851 }, { "epoch": 1.3689463955637708, "grad_norm": 4.276607036590576, "learning_rate": 3.1773911869552887e-06, "loss": 0.054113369435071945, "step": 1852 }, { "epoch": 1.3696857670979667, "grad_norm": 3.380979299545288, "learning_rate": 3.175628315755571e-06, "loss": 0.02465420588850975, "step": 1853 }, { "epoch": 1.3704251386321626, "grad_norm": 3.678612470626831, "learning_rate": 3.17386517353706e-06, "loss": 0.02901320904493332, "step": 1854 }, { "epoch": 1.3711645101663585, "grad_norm": 5.496438503265381, "learning_rate": 3.1721017614224386e-06, "loss": 0.05046052485704422, "step": 1855 }, { "epoch": 1.3719038817005544, "grad_norm": 6.942925930023193, "learning_rate": 3.170338080534558e-06, "loss": 0.04968785494565964, "step": 1856 }, { "epoch": 1.3726432532347506, "grad_norm": 4.484908103942871, "learning_rate": 3.1685741319964427e-06, "loss": 0.07595298439264297, "step": 1857 }, { "epoch": 1.3733826247689465, "grad_norm": 3.653970718383789, "learning_rate": 3.166809916931285e-06, "loss": 0.016920406371355057, "step": 1858 }, { "epoch": 1.3741219963031424, "grad_norm": 4.178115367889404, "learning_rate": 3.165045436462451e-06, "loss": 0.06136554107069969, "step": 1859 }, { "epoch": 1.3748613678373383, "grad_norm": 4.527619361877441, "learning_rate": 3.163280691713472e-06, "loss": 0.03319022059440613, "step": 1860 }, { "epoch": 1.3756007393715342, "grad_norm": 5.058053016662598, "learning_rate": 3.1615156838080503e-06, "loss": 0.06136186420917511, "step": 1861 }, { "epoch": 1.37634011090573, "grad_norm": 3.602701425552368, "learning_rate": 3.1597504138700517e-06, "loss": 0.012986863031983376, "step": 1862 }, { "epoch": 1.377079482439926, "grad_norm": 3.5133230686187744, "learning_rate": 3.157984883023515e-06, "loss": 0.030167371034622192, "step": 1863 }, { "epoch": 1.377818853974122, "grad_norm": 3.844728469848633, "learning_rate": 3.1562190923926406e-06, "loss": 0.03528578579425812, "step": 1864 }, { "epoch": 1.378558225508318, "grad_norm": 2.7810544967651367, "learning_rate": 3.154453043101795e-06, "loss": 0.019485440105199814, "step": 1865 }, { "epoch": 1.3792975970425139, "grad_norm": 3.7134196758270264, "learning_rate": 3.1526867362755102e-06, "loss": 0.029044032096862793, "step": 1866 }, { "epoch": 1.3800369685767098, "grad_norm": 4.405758857727051, "learning_rate": 3.150920173038482e-06, "loss": 0.04203013330698013, "step": 1867 }, { "epoch": 1.3807763401109057, "grad_norm": 2.6754424571990967, "learning_rate": 3.149153354515569e-06, "loss": 0.012488674372434616, "step": 1868 }, { "epoch": 1.3815157116451018, "grad_norm": 4.1771559715271, "learning_rate": 3.1473862818317942e-06, "loss": 0.045646607875823975, "step": 1869 }, { "epoch": 1.3822550831792975, "grad_norm": 4.4224653244018555, "learning_rate": 3.145618956112339e-06, "loss": 0.03740588575601578, "step": 1870 }, { "epoch": 1.3829944547134936, "grad_norm": 4.44866418838501, "learning_rate": 3.1438513784825497e-06, "loss": 0.023214900866150856, "step": 1871 }, { "epoch": 1.3837338262476895, "grad_norm": 3.8322079181671143, "learning_rate": 3.1420835500679297e-06, "loss": 0.014995075762271881, "step": 1872 }, { "epoch": 1.3844731977818854, "grad_norm": 4.018552780151367, "learning_rate": 3.140315471994144e-06, "loss": 0.04759146273136139, "step": 1873 }, { "epoch": 1.3852125693160813, "grad_norm": 4.341092109680176, "learning_rate": 3.138547145387017e-06, "loss": 0.03347864747047424, "step": 1874 }, { "epoch": 1.3859519408502772, "grad_norm": 6.252964973449707, "learning_rate": 3.136778571372529e-06, "loss": 0.038853202015161514, "step": 1875 }, { "epoch": 1.3866913123844733, "grad_norm": 5.277734756469727, "learning_rate": 3.1350097510768208e-06, "loss": 0.04883398860692978, "step": 1876 }, { "epoch": 1.3874306839186692, "grad_norm": 3.5425944328308105, "learning_rate": 3.1332406856261878e-06, "loss": 0.01788489893078804, "step": 1877 }, { "epoch": 1.388170055452865, "grad_norm": 3.9657859802246094, "learning_rate": 3.1314713761470825e-06, "loss": 0.029259758070111275, "step": 1878 }, { "epoch": 1.388909426987061, "grad_norm": 4.594902992248535, "learning_rate": 3.129701823766113e-06, "loss": 0.04151662811636925, "step": 1879 }, { "epoch": 1.389648798521257, "grad_norm": 4.671213150024414, "learning_rate": 3.1279320296100397e-06, "loss": 0.0368942990899086, "step": 1880 }, { "epoch": 1.3903881700554528, "grad_norm": 4.249755859375, "learning_rate": 3.1261619948057814e-06, "loss": 0.029932955279946327, "step": 1881 }, { "epoch": 1.3911275415896487, "grad_norm": 5.556185722351074, "learning_rate": 3.124391720480406e-06, "loss": 0.04980615898966789, "step": 1882 }, { "epoch": 1.3918669131238448, "grad_norm": 1.3712522983551025, "learning_rate": 3.122621207761136e-06, "loss": 0.011700092814862728, "step": 1883 }, { "epoch": 1.3926062846580407, "grad_norm": 4.229989051818848, "learning_rate": 3.120850457775346e-06, "loss": 0.043032512068748474, "step": 1884 }, { "epoch": 1.3933456561922366, "grad_norm": 3.4955520629882812, "learning_rate": 3.119079471650559e-06, "loss": 0.04119601100683212, "step": 1885 }, { "epoch": 1.3940850277264325, "grad_norm": 4.850847244262695, "learning_rate": 3.117308250514452e-06, "loss": 0.018718190491199493, "step": 1886 }, { "epoch": 1.3948243992606284, "grad_norm": 3.635776996612549, "learning_rate": 3.1155367954948483e-06, "loss": 0.016720367595553398, "step": 1887 }, { "epoch": 1.3955637707948245, "grad_norm": 4.675966739654541, "learning_rate": 3.1137651077197246e-06, "loss": 0.05551951378583908, "step": 1888 }, { "epoch": 1.3963031423290202, "grad_norm": 4.3305983543396, "learning_rate": 3.1119931883172e-06, "loss": 0.045645736157894135, "step": 1889 }, { "epoch": 1.3970425138632163, "grad_norm": 5.262744903564453, "learning_rate": 3.1102210384155452e-06, "loss": 0.05222761258482933, "step": 1890 }, { "epoch": 1.3977818853974122, "grad_norm": 4.91766881942749, "learning_rate": 3.1084486591431754e-06, "loss": 0.04161448776721954, "step": 1891 }, { "epoch": 1.3985212569316081, "grad_norm": 3.5759220123291016, "learning_rate": 3.106676051628654e-06, "loss": 0.027166329324245453, "step": 1892 }, { "epoch": 1.399260628465804, "grad_norm": 3.573826313018799, "learning_rate": 3.1049032170006897e-06, "loss": 0.02289755828678608, "step": 1893 }, { "epoch": 1.4, "grad_norm": 2.1415886878967285, "learning_rate": 3.1031301563881335e-06, "loss": 0.010580537840723991, "step": 1894 }, { "epoch": 1.400739371534196, "grad_norm": 2.518436908721924, "learning_rate": 3.101356870919982e-06, "loss": 0.015574861317873001, "step": 1895 }, { "epoch": 1.401478743068392, "grad_norm": 4.427599906921387, "learning_rate": 3.099583361725373e-06, "loss": 0.045829951763153076, "step": 1896 }, { "epoch": 1.4022181146025878, "grad_norm": 4.438584327697754, "learning_rate": 3.0978096299335903e-06, "loss": 0.06223393231630325, "step": 1897 }, { "epoch": 1.4029574861367837, "grad_norm": 4.183803558349609, "learning_rate": 3.096035676674057e-06, "loss": 0.018753942102193832, "step": 1898 }, { "epoch": 1.4036968576709796, "grad_norm": 4.397330284118652, "learning_rate": 3.094261503076337e-06, "loss": 0.03933418542146683, "step": 1899 }, { "epoch": 1.4044362292051755, "grad_norm": 4.450538158416748, "learning_rate": 3.0924871102701367e-06, "loss": 0.051722101867198944, "step": 1900 }, { "epoch": 1.4051756007393714, "grad_norm": 4.3741984367370605, "learning_rate": 3.0907124993852975e-06, "loss": 0.08438555896282196, "step": 1901 }, { "epoch": 1.4059149722735675, "grad_norm": 4.16634464263916, "learning_rate": 3.088937671551806e-06, "loss": 0.03436681628227234, "step": 1902 }, { "epoch": 1.4066543438077634, "grad_norm": 2.9783244132995605, "learning_rate": 3.0871626278997823e-06, "loss": 0.03949471190571785, "step": 1903 }, { "epoch": 1.4073937153419593, "grad_norm": 4.460395812988281, "learning_rate": 3.085387369559486e-06, "loss": 0.031905218958854675, "step": 1904 }, { "epoch": 1.4081330868761552, "grad_norm": 3.9484779834747314, "learning_rate": 3.083611897661311e-06, "loss": 0.03743156045675278, "step": 1905 }, { "epoch": 1.4088724584103511, "grad_norm": 4.431934356689453, "learning_rate": 3.0818362133357904e-06, "loss": 0.06055694445967674, "step": 1906 }, { "epoch": 1.4096118299445473, "grad_norm": 3.752963066101074, "learning_rate": 3.080060317713591e-06, "loss": 0.07107581198215485, "step": 1907 }, { "epoch": 1.410351201478743, "grad_norm": 3.1498541831970215, "learning_rate": 3.078284211925513e-06, "loss": 0.04794593155384064, "step": 1908 }, { "epoch": 1.411090573012939, "grad_norm": 2.4723994731903076, "learning_rate": 3.0765078971024925e-06, "loss": 0.01987462304532528, "step": 1909 }, { "epoch": 1.411829944547135, "grad_norm": 4.813337326049805, "learning_rate": 3.0747313743755973e-06, "loss": 0.03317145258188248, "step": 1910 }, { "epoch": 1.4125693160813309, "grad_norm": 3.1758532524108887, "learning_rate": 3.0729546448760293e-06, "loss": 0.01378234289586544, "step": 1911 }, { "epoch": 1.4133086876155267, "grad_norm": 4.6784491539001465, "learning_rate": 3.07117770973512e-06, "loss": 0.054316986352205276, "step": 1912 }, { "epoch": 1.4140480591497226, "grad_norm": 3.478285551071167, "learning_rate": 3.0694005700843327e-06, "loss": 0.04001902416348457, "step": 1913 }, { "epoch": 1.4147874306839188, "grad_norm": 5.67146110534668, "learning_rate": 3.067623227055261e-06, "loss": 0.05393645912408829, "step": 1914 }, { "epoch": 1.4155268022181147, "grad_norm": 4.423172473907471, "learning_rate": 3.0658456817796256e-06, "loss": 0.05114195495843887, "step": 1915 }, { "epoch": 1.4162661737523106, "grad_norm": 4.588649272918701, "learning_rate": 3.064067935389283e-06, "loss": 0.037807755172252655, "step": 1916 }, { "epoch": 1.4170055452865065, "grad_norm": 5.403221130371094, "learning_rate": 3.06228998901621e-06, "loss": 0.03561970591545105, "step": 1917 }, { "epoch": 1.4177449168207024, "grad_norm": 4.011436462402344, "learning_rate": 3.060511843792514e-06, "loss": 0.01869531348347664, "step": 1918 }, { "epoch": 1.4184842883548983, "grad_norm": 5.535740852355957, "learning_rate": 3.05873350085043e-06, "loss": 0.06682844460010529, "step": 1919 }, { "epoch": 1.4192236598890942, "grad_norm": 4.154177665710449, "learning_rate": 3.056954961322316e-06, "loss": 0.03377040475606918, "step": 1920 }, { "epoch": 1.4199630314232903, "grad_norm": 3.3513994216918945, "learning_rate": 3.0551762263406575e-06, "loss": 0.01812131144106388, "step": 1921 }, { "epoch": 1.4207024029574862, "grad_norm": 5.028448581695557, "learning_rate": 3.0533972970380653e-06, "loss": 0.05955328792333603, "step": 1922 }, { "epoch": 1.421441774491682, "grad_norm": 3.9964683055877686, "learning_rate": 3.0516181745472707e-06, "loss": 0.032064199447631836, "step": 1923 }, { "epoch": 1.422181146025878, "grad_norm": 5.631239414215088, "learning_rate": 3.04983886000113e-06, "loss": 0.06987383961677551, "step": 1924 }, { "epoch": 1.4229205175600739, "grad_norm": 4.356898307800293, "learning_rate": 3.0480593545326226e-06, "loss": 0.06826986372470856, "step": 1925 }, { "epoch": 1.42365988909427, "grad_norm": 3.1936182975769043, "learning_rate": 3.0462796592748478e-06, "loss": 0.02428516373038292, "step": 1926 }, { "epoch": 1.4243992606284657, "grad_norm": 4.1430864334106445, "learning_rate": 3.0444997753610257e-06, "loss": 0.03017709217965603, "step": 1927 }, { "epoch": 1.4251386321626618, "grad_norm": 5.438726902008057, "learning_rate": 3.042719703924498e-06, "loss": 0.03655475750565529, "step": 1928 }, { "epoch": 1.4258780036968577, "grad_norm": 4.126948356628418, "learning_rate": 3.040939446098725e-06, "loss": 0.039134785532951355, "step": 1929 }, { "epoch": 1.4266173752310536, "grad_norm": 4.339874744415283, "learning_rate": 3.0391590030172868e-06, "loss": 0.01949901133775711, "step": 1930 }, { "epoch": 1.4273567467652495, "grad_norm": 4.7106781005859375, "learning_rate": 3.037378375813878e-06, "loss": 0.06251044571399689, "step": 1931 }, { "epoch": 1.4280961182994454, "grad_norm": 4.462235450744629, "learning_rate": 3.035597565622315e-06, "loss": 0.02730577439069748, "step": 1932 }, { "epoch": 1.4288354898336415, "grad_norm": 2.796142816543579, "learning_rate": 3.033816573576527e-06, "loss": 0.02980552613735199, "step": 1933 }, { "epoch": 1.4295748613678374, "grad_norm": 3.284698009490967, "learning_rate": 3.0320354008105606e-06, "loss": 0.02414001151919365, "step": 1934 }, { "epoch": 1.4303142329020333, "grad_norm": 3.899141788482666, "learning_rate": 3.0302540484585783e-06, "loss": 0.060081157833337784, "step": 1935 }, { "epoch": 1.4310536044362292, "grad_norm": 5.080705642700195, "learning_rate": 3.0284725176548565e-06, "loss": 0.05282039940357208, "step": 1936 }, { "epoch": 1.431792975970425, "grad_norm": 3.7708253860473633, "learning_rate": 3.0266908095337834e-06, "loss": 0.04168086498975754, "step": 1937 }, { "epoch": 1.432532347504621, "grad_norm": 3.9044582843780518, "learning_rate": 3.024908925229861e-06, "loss": 0.031972289085388184, "step": 1938 }, { "epoch": 1.433271719038817, "grad_norm": 3.280564069747925, "learning_rate": 3.023126865877706e-06, "loss": 0.027527371421456337, "step": 1939 }, { "epoch": 1.434011090573013, "grad_norm": 2.3171629905700684, "learning_rate": 3.0213446326120435e-06, "loss": 0.01939113810658455, "step": 1940 }, { "epoch": 1.434750462107209, "grad_norm": 4.6972479820251465, "learning_rate": 3.0195622265677105e-06, "loss": 0.042007870972156525, "step": 1941 }, { "epoch": 1.4354898336414048, "grad_norm": 4.2968926429748535, "learning_rate": 3.0177796488796535e-06, "loss": 0.02053312212228775, "step": 1942 }, { "epoch": 1.4362292051756007, "grad_norm": 4.3146653175354, "learning_rate": 3.0159969006829286e-06, "loss": 0.034748759120702744, "step": 1943 }, { "epoch": 1.4369685767097966, "grad_norm": 3.1640820503234863, "learning_rate": 3.014213983112701e-06, "loss": 0.03281209617853165, "step": 1944 }, { "epoch": 1.4377079482439927, "grad_norm": 5.356256484985352, "learning_rate": 3.012430897304244e-06, "loss": 0.025032149627804756, "step": 1945 }, { "epoch": 1.4384473197781884, "grad_norm": 3.6457676887512207, "learning_rate": 3.010647644392936e-06, "loss": 0.04706260561943054, "step": 1946 }, { "epoch": 1.4391866913123845, "grad_norm": 3.9497146606445312, "learning_rate": 3.008864225514264e-06, "loss": 0.044065963476896286, "step": 1947 }, { "epoch": 1.4399260628465804, "grad_norm": 3.0572385787963867, "learning_rate": 3.0070806418038194e-06, "loss": 0.010934509336948395, "step": 1948 }, { "epoch": 1.4406654343807763, "grad_norm": 3.7027416229248047, "learning_rate": 3.0052968943972993e-06, "loss": 0.041076742112636566, "step": 1949 }, { "epoch": 1.4414048059149722, "grad_norm": 4.296900272369385, "learning_rate": 3.003512984430504e-06, "loss": 0.04887661337852478, "step": 1950 }, { "epoch": 1.4421441774491681, "grad_norm": 3.6956729888916016, "learning_rate": 3.0017289130393385e-06, "loss": 0.035271137952804565, "step": 1951 }, { "epoch": 1.4428835489833642, "grad_norm": 2.2722318172454834, "learning_rate": 2.999944681359811e-06, "loss": 0.012405750341713428, "step": 1952 }, { "epoch": 1.4436229205175601, "grad_norm": 3.4936554431915283, "learning_rate": 2.998160290528029e-06, "loss": 0.009482761844992638, "step": 1953 }, { "epoch": 1.444362292051756, "grad_norm": 4.956447601318359, "learning_rate": 2.996375741680204e-06, "loss": 0.008802505210042, "step": 1954 }, { "epoch": 1.445101663585952, "grad_norm": 4.503058910369873, "learning_rate": 2.9945910359526485e-06, "loss": 0.01809684932231903, "step": 1955 }, { "epoch": 1.4458410351201478, "grad_norm": 4.61729097366333, "learning_rate": 2.9928061744817717e-06, "loss": 0.02368323877453804, "step": 1956 }, { "epoch": 1.4465804066543437, "grad_norm": 4.379559516906738, "learning_rate": 2.9910211584040857e-06, "loss": 0.03568928316235542, "step": 1957 }, { "epoch": 1.4473197781885396, "grad_norm": 5.424335479736328, "learning_rate": 2.9892359888561976e-06, "loss": 0.04101748764514923, "step": 1958 }, { "epoch": 1.4480591497227358, "grad_norm": 4.951371669769287, "learning_rate": 2.987450666974816e-06, "loss": 0.01939738169312477, "step": 1959 }, { "epoch": 1.4487985212569316, "grad_norm": 4.687196254730225, "learning_rate": 2.985665193896743e-06, "loss": 0.0521998405456543, "step": 1960 }, { "epoch": 1.4495378927911275, "grad_norm": 4.78123664855957, "learning_rate": 2.983879570758879e-06, "loss": 0.035285238176584244, "step": 1961 }, { "epoch": 1.4502772643253234, "grad_norm": 3.155235767364502, "learning_rate": 2.9820937986982185e-06, "loss": 0.018956756219267845, "step": 1962 }, { "epoch": 1.4510166358595193, "grad_norm": 5.1655120849609375, "learning_rate": 2.9803078788518526e-06, "loss": 0.04628627747297287, "step": 1963 }, { "epoch": 1.4517560073937155, "grad_norm": 4.0101447105407715, "learning_rate": 2.978521812356966e-06, "loss": 0.0597144216299057, "step": 1964 }, { "epoch": 1.4524953789279111, "grad_norm": 2.838726282119751, "learning_rate": 2.9767356003508353e-06, "loss": 0.03362023085355759, "step": 1965 }, { "epoch": 1.4532347504621073, "grad_norm": 4.166595458984375, "learning_rate": 2.974949243970832e-06, "loss": 0.0373288094997406, "step": 1966 }, { "epoch": 1.4539741219963032, "grad_norm": 3.828723907470703, "learning_rate": 2.9731627443544165e-06, "loss": 0.05495942384004593, "step": 1967 }, { "epoch": 1.454713493530499, "grad_norm": 4.748375415802002, "learning_rate": 2.971376102639144e-06, "loss": 0.02360156551003456, "step": 1968 }, { "epoch": 1.455452865064695, "grad_norm": 4.467069625854492, "learning_rate": 2.969589319962658e-06, "loss": 0.04608454555273056, "step": 1969 }, { "epoch": 1.4561922365988909, "grad_norm": 3.7180891036987305, "learning_rate": 2.967802397462693e-06, "loss": 0.02188033051788807, "step": 1970 }, { "epoch": 1.456931608133087, "grad_norm": 3.46868896484375, "learning_rate": 2.9660153362770698e-06, "loss": 0.026311460882425308, "step": 1971 }, { "epoch": 1.4576709796672829, "grad_norm": 4.437051773071289, "learning_rate": 2.9642281375437005e-06, "loss": 0.045419301837682724, "step": 1972 }, { "epoch": 1.4584103512014788, "grad_norm": 2.8504109382629395, "learning_rate": 2.962440802400584e-06, "loss": 0.025064650923013687, "step": 1973 }, { "epoch": 1.4591497227356747, "grad_norm": 2.6710546016693115, "learning_rate": 2.960653331985805e-06, "loss": 0.033057764172554016, "step": 1974 }, { "epoch": 1.4598890942698706, "grad_norm": 4.60660457611084, "learning_rate": 2.9588657274375366e-06, "loss": 0.04836399108171463, "step": 1975 }, { "epoch": 1.4606284658040665, "grad_norm": 3.165508985519409, "learning_rate": 2.9570779898940345e-06, "loss": 0.02919403277337551, "step": 1976 }, { "epoch": 1.4613678373382624, "grad_norm": 2.9095776081085205, "learning_rate": 2.9552901204936403e-06, "loss": 0.030656524002552032, "step": 1977 }, { "epoch": 1.4621072088724585, "grad_norm": 4.066230297088623, "learning_rate": 2.953502120374781e-06, "loss": 0.07390530407428741, "step": 1978 }, { "epoch": 1.4628465804066544, "grad_norm": 4.957443714141846, "learning_rate": 2.951713990675964e-06, "loss": 0.04374857619404793, "step": 1979 }, { "epoch": 1.4635859519408503, "grad_norm": 6.572922229766846, "learning_rate": 2.9499257325357815e-06, "loss": 0.054719965904951096, "step": 1980 }, { "epoch": 1.4643253234750462, "grad_norm": 2.0339181423187256, "learning_rate": 2.9481373470929075e-06, "loss": 0.009567564353346825, "step": 1981 }, { "epoch": 1.465064695009242, "grad_norm": 4.234466552734375, "learning_rate": 2.9463488354860943e-06, "loss": 0.03652196377515793, "step": 1982 }, { "epoch": 1.4658040665434382, "grad_norm": 3.2013840675354004, "learning_rate": 2.9445601988541785e-06, "loss": 0.018079619854688644, "step": 1983 }, { "epoch": 1.4665434380776339, "grad_norm": 4.31234884262085, "learning_rate": 2.942771438336073e-06, "loss": 0.05607249215245247, "step": 1984 }, { "epoch": 1.46728280961183, "grad_norm": 4.678189754486084, "learning_rate": 2.940982555070772e-06, "loss": 0.02735525742173195, "step": 1985 }, { "epoch": 1.468022181146026, "grad_norm": 3.11767578125, "learning_rate": 2.939193550197345e-06, "loss": 0.02394535019993782, "step": 1986 }, { "epoch": 1.4687615526802218, "grad_norm": 2.9768691062927246, "learning_rate": 2.937404424854942e-06, "loss": 0.01656603068113327, "step": 1987 }, { "epoch": 1.4695009242144177, "grad_norm": 3.9024970531463623, "learning_rate": 2.935615180182789e-06, "loss": 0.047953907400369644, "step": 1988 }, { "epoch": 1.4702402957486136, "grad_norm": 3.979780435562134, "learning_rate": 2.933825817320186e-06, "loss": 0.05220657214522362, "step": 1989 }, { "epoch": 1.4709796672828097, "grad_norm": 4.127331256866455, "learning_rate": 2.9320363374065106e-06, "loss": 0.04954760521650314, "step": 1990 }, { "epoch": 1.4717190388170056, "grad_norm": 3.533123254776001, "learning_rate": 2.9302467415812136e-06, "loss": 0.016224976629018784, "step": 1991 }, { "epoch": 1.4724584103512015, "grad_norm": 4.689507484436035, "learning_rate": 2.928457030983819e-06, "loss": 0.026616564020514488, "step": 1992 }, { "epoch": 1.4731977818853974, "grad_norm": 4.347389221191406, "learning_rate": 2.9266672067539264e-06, "loss": 0.030398380011320114, "step": 1993 }, { "epoch": 1.4739371534195933, "grad_norm": 3.1971590518951416, "learning_rate": 2.924877270031206e-06, "loss": 0.01920868270099163, "step": 1994 }, { "epoch": 1.4746765249537892, "grad_norm": 4.1846795082092285, "learning_rate": 2.9230872219553995e-06, "loss": 0.03561996668577194, "step": 1995 }, { "epoch": 1.475415896487985, "grad_norm": 2.921715497970581, "learning_rate": 2.9212970636663197e-06, "loss": 0.02092120051383972, "step": 1996 }, { "epoch": 1.4761552680221812, "grad_norm": 3.3783652782440186, "learning_rate": 2.9195067963038505e-06, "loss": 0.03714446723461151, "step": 1997 }, { "epoch": 1.4768946395563771, "grad_norm": 4.1574554443359375, "learning_rate": 2.917716421007943e-06, "loss": 0.06127569079399109, "step": 1998 }, { "epoch": 1.477634011090573, "grad_norm": 3.4705255031585693, "learning_rate": 2.9159259389186197e-06, "loss": 0.023862740024924278, "step": 1999 }, { "epoch": 1.478373382624769, "grad_norm": 4.3726396560668945, "learning_rate": 2.9141353511759696e-06, "loss": 0.01638149656355381, "step": 2000 }, { "epoch": 1.4791127541589648, "grad_norm": 4.74641227722168, "learning_rate": 2.9123446589201493e-06, "loss": 0.04053298756480217, "step": 2001 }, { "epoch": 1.479852125693161, "grad_norm": 4.625601291656494, "learning_rate": 2.910553863291382e-06, "loss": 0.019385619089007378, "step": 2002 }, { "epoch": 1.4805914972273566, "grad_norm": 2.835895299911499, "learning_rate": 2.9087629654299567e-06, "loss": 0.02050464227795601, "step": 2003 }, { "epoch": 1.4813308687615527, "grad_norm": 3.9065144062042236, "learning_rate": 2.906971966476227e-06, "loss": 0.04043756052851677, "step": 2004 }, { "epoch": 1.4820702402957486, "grad_norm": 3.093296527862549, "learning_rate": 2.905180867570612e-06, "loss": 0.01694577746093273, "step": 2005 }, { "epoch": 1.4828096118299445, "grad_norm": 2.679844379425049, "learning_rate": 2.903389669853593e-06, "loss": 0.012654848396778107, "step": 2006 }, { "epoch": 1.4835489833641404, "grad_norm": 3.102372169494629, "learning_rate": 2.9015983744657158e-06, "loss": 0.01731681451201439, "step": 2007 }, { "epoch": 1.4842883548983363, "grad_norm": 3.8433010578155518, "learning_rate": 2.899806982547587e-06, "loss": 0.03735943138599396, "step": 2008 }, { "epoch": 1.4850277264325324, "grad_norm": 5.290180683135986, "learning_rate": 2.8980154952398755e-06, "loss": 0.026810456067323685, "step": 2009 }, { "epoch": 1.4857670979667283, "grad_norm": 5.400670051574707, "learning_rate": 2.8962239136833105e-06, "loss": 0.020539231598377228, "step": 2010 }, { "epoch": 1.4865064695009242, "grad_norm": 5.061708450317383, "learning_rate": 2.8944322390186812e-06, "loss": 0.048390068113803864, "step": 2011 }, { "epoch": 1.4872458410351201, "grad_norm": 3.566866636276245, "learning_rate": 2.8926404723868373e-06, "loss": 0.029389746487140656, "step": 2012 }, { "epoch": 1.487985212569316, "grad_norm": 2.374054431915283, "learning_rate": 2.890848614928685e-06, "loss": 0.009436866268515587, "step": 2013 }, { "epoch": 1.488724584103512, "grad_norm": 3.874850034713745, "learning_rate": 2.88905666778519e-06, "loss": 0.02221044898033142, "step": 2014 }, { "epoch": 1.4894639556377078, "grad_norm": 3.4533746242523193, "learning_rate": 2.8872646320973735e-06, "loss": 0.012443449348211288, "step": 2015 }, { "epoch": 1.490203327171904, "grad_norm": 4.833604335784912, "learning_rate": 2.8854725090063153e-06, "loss": 0.05120781064033508, "step": 2016 }, { "epoch": 1.4909426987060999, "grad_norm": 3.399918556213379, "learning_rate": 2.8836802996531495e-06, "loss": 0.030378170311450958, "step": 2017 }, { "epoch": 1.4916820702402958, "grad_norm": 4.523649215698242, "learning_rate": 2.8818880051790643e-06, "loss": 0.032201651483774185, "step": 2018 }, { "epoch": 1.4924214417744917, "grad_norm": 4.829171180725098, "learning_rate": 2.8800956267253035e-06, "loss": 0.0514591708779335, "step": 2019 }, { "epoch": 1.4931608133086876, "grad_norm": 4.261626720428467, "learning_rate": 2.8783031654331633e-06, "loss": 0.030687250196933746, "step": 2020 }, { "epoch": 1.4939001848428837, "grad_norm": 3.0632104873657227, "learning_rate": 2.876510622443995e-06, "loss": 0.015314146876335144, "step": 2021 }, { "epoch": 1.4946395563770793, "grad_norm": 4.713589191436768, "learning_rate": 2.874717998899199e-06, "loss": 0.03299334645271301, "step": 2022 }, { "epoch": 1.4953789279112755, "grad_norm": 4.621907711029053, "learning_rate": 2.872925295940228e-06, "loss": 0.031340278685092926, "step": 2023 }, { "epoch": 1.4961182994454714, "grad_norm": 3.627938985824585, "learning_rate": 2.871132514708586e-06, "loss": 0.03544803708791733, "step": 2024 }, { "epoch": 1.4968576709796673, "grad_norm": 3.1356260776519775, "learning_rate": 2.869339656345826e-06, "loss": 0.019008416682481766, "step": 2025 }, { "epoch": 1.4975970425138632, "grad_norm": 4.774219036102295, "learning_rate": 2.867546721993551e-06, "loss": 0.04632198065519333, "step": 2026 }, { "epoch": 1.498336414048059, "grad_norm": 4.746023178100586, "learning_rate": 2.8657537127934102e-06, "loss": 0.0404304638504982, "step": 2027 }, { "epoch": 1.4990757855822552, "grad_norm": 5.051352500915527, "learning_rate": 2.863960629887104e-06, "loss": 0.04757878929376602, "step": 2028 }, { "epoch": 1.499815157116451, "grad_norm": 3.700261116027832, "learning_rate": 2.8621674744163767e-06, "loss": 0.022965427488088608, "step": 2029 }, { "epoch": 1.500554528650647, "grad_norm": 4.598008155822754, "learning_rate": 2.8603742475230196e-06, "loss": 0.05535631254315376, "step": 2030 }, { "epoch": 1.5012939001848429, "grad_norm": 3.47391414642334, "learning_rate": 2.858580950348871e-06, "loss": 0.03701433539390564, "step": 2031 }, { "epoch": 1.5020332717190388, "grad_norm": 2.8423891067504883, "learning_rate": 2.856787584035812e-06, "loss": 0.013549918308854103, "step": 2032 }, { "epoch": 1.502772643253235, "grad_norm": 4.121523380279541, "learning_rate": 2.854994149725767e-06, "loss": 0.019229376688599586, "step": 2033 }, { "epoch": 1.5035120147874306, "grad_norm": 3.6260507106781006, "learning_rate": 2.853200648560708e-06, "loss": 0.058182474225759506, "step": 2034 }, { "epoch": 1.5042513863216267, "grad_norm": 3.0703163146972656, "learning_rate": 2.8514070816826443e-06, "loss": 0.016425736248493195, "step": 2035 }, { "epoch": 1.5049907578558226, "grad_norm": 4.1456122398376465, "learning_rate": 2.8496134502336313e-06, "loss": 0.05755992978811264, "step": 2036 }, { "epoch": 1.5057301293900185, "grad_norm": 3.497128486633301, "learning_rate": 2.8478197553557626e-06, "loss": 0.01892842724919319, "step": 2037 }, { "epoch": 1.5064695009242144, "grad_norm": 4.123014450073242, "learning_rate": 2.8460259981911732e-06, "loss": 0.06802278757095337, "step": 2038 }, { "epoch": 1.5072088724584103, "grad_norm": 4.324033737182617, "learning_rate": 2.8442321798820376e-06, "loss": 0.03936494141817093, "step": 2039 }, { "epoch": 1.5079482439926064, "grad_norm": 4.283459663391113, "learning_rate": 2.84243830157057e-06, "loss": 0.04364185035228729, "step": 2040 }, { "epoch": 1.508687615526802, "grad_norm": 3.772080659866333, "learning_rate": 2.8406443643990224e-06, "loss": 0.03716706484556198, "step": 2041 }, { "epoch": 1.5094269870609982, "grad_norm": 3.7228705883026123, "learning_rate": 2.8388503695096838e-06, "loss": 0.03792129456996918, "step": 2042 }, { "epoch": 1.510166358595194, "grad_norm": 4.561285972595215, "learning_rate": 2.83705631804488e-06, "loss": 0.016660090535879135, "step": 2043 }, { "epoch": 1.51090573012939, "grad_norm": 3.1530659198760986, "learning_rate": 2.835262211146974e-06, "loss": 0.02759481780230999, "step": 2044 }, { "epoch": 1.511645101663586, "grad_norm": 3.6655917167663574, "learning_rate": 2.8334680499583615e-06, "loss": 0.029526086524128914, "step": 2045 }, { "epoch": 1.5123844731977818, "grad_norm": 3.2881879806518555, "learning_rate": 2.8316738356214774e-06, "loss": 0.02021661400794983, "step": 2046 }, { "epoch": 1.513123844731978, "grad_norm": 3.5069432258605957, "learning_rate": 2.829879569278785e-06, "loss": 0.023030750453472137, "step": 2047 }, { "epoch": 1.5138632162661736, "grad_norm": 2.9639201164245605, "learning_rate": 2.8280852520727842e-06, "loss": 0.025047965347766876, "step": 2048 }, { "epoch": 1.5146025878003697, "grad_norm": 5.137871742248535, "learning_rate": 2.8262908851460054e-06, "loss": 0.03333357349038124, "step": 2049 }, { "epoch": 1.5153419593345656, "grad_norm": 3.9330801963806152, "learning_rate": 2.8244964696410134e-06, "loss": 0.05514500290155411, "step": 2050 }, { "epoch": 1.5160813308687615, "grad_norm": 2.7428390979766846, "learning_rate": 2.8227020067003996e-06, "loss": 0.021601248532533646, "step": 2051 }, { "epoch": 1.5168207024029576, "grad_norm": 2.899491310119629, "learning_rate": 2.8209074974667905e-06, "loss": 0.021875465288758278, "step": 2052 }, { "epoch": 1.5175600739371533, "grad_norm": 3.8033227920532227, "learning_rate": 2.8191129430828384e-06, "loss": 0.023257708176970482, "step": 2053 }, { "epoch": 1.5182994454713494, "grad_norm": 2.2259414196014404, "learning_rate": 2.817318344691226e-06, "loss": 0.01596982590854168, "step": 2054 }, { "epoch": 1.5190388170055453, "grad_norm": 3.2375943660736084, "learning_rate": 2.815523703434664e-06, "loss": 0.023083221167325974, "step": 2055 }, { "epoch": 1.5197781885397412, "grad_norm": 3.6020965576171875, "learning_rate": 2.8137290204558893e-06, "loss": 0.022473309189081192, "step": 2056 }, { "epoch": 1.5205175600739371, "grad_norm": 3.68684983253479, "learning_rate": 2.811934296897666e-06, "loss": 0.02806912176311016, "step": 2057 }, { "epoch": 1.521256931608133, "grad_norm": 3.205139636993408, "learning_rate": 2.810139533902786e-06, "loss": 0.011577796190977097, "step": 2058 }, { "epoch": 1.5219963031423291, "grad_norm": 2.7315115928649902, "learning_rate": 2.8083447326140615e-06, "loss": 0.019208308309316635, "step": 2059 }, { "epoch": 1.5227356746765248, "grad_norm": 4.154613018035889, "learning_rate": 2.8065498941743347e-06, "loss": 0.02690846100449562, "step": 2060 }, { "epoch": 1.523475046210721, "grad_norm": 4.386315822601318, "learning_rate": 2.804755019726468e-06, "loss": 0.025380276143550873, "step": 2061 }, { "epoch": 1.5242144177449168, "grad_norm": 4.119882583618164, "learning_rate": 2.802960110413347e-06, "loss": 0.02213118225336075, "step": 2062 }, { "epoch": 1.5249537892791127, "grad_norm": 5.167129993438721, "learning_rate": 2.8011651673778804e-06, "loss": 0.07452896982431412, "step": 2063 }, { "epoch": 1.5256931608133086, "grad_norm": 2.0310020446777344, "learning_rate": 2.7993701917629978e-06, "loss": 0.005626820493489504, "step": 2064 }, { "epoch": 1.5264325323475045, "grad_norm": 3.996419906616211, "learning_rate": 2.7975751847116507e-06, "loss": 0.038185350596904755, "step": 2065 }, { "epoch": 1.5271719038817007, "grad_norm": 4.0300445556640625, "learning_rate": 2.7957801473668094e-06, "loss": 0.03378629684448242, "step": 2066 }, { "epoch": 1.5279112754158963, "grad_norm": 3.009652614593506, "learning_rate": 2.7939850808714626e-06, "loss": 0.032043732702732086, "step": 2067 }, { "epoch": 1.5286506469500925, "grad_norm": 3.4882190227508545, "learning_rate": 2.792189986368619e-06, "loss": 0.01419014111161232, "step": 2068 }, { "epoch": 1.5293900184842883, "grad_norm": 2.478475570678711, "learning_rate": 2.790394865001305e-06, "loss": 0.016036417335271835, "step": 2069 }, { "epoch": 1.5301293900184842, "grad_norm": 4.0760979652404785, "learning_rate": 2.788599717912566e-06, "loss": 0.036771565675735474, "step": 2070 }, { "epoch": 1.5308687615526804, "grad_norm": 3.926156520843506, "learning_rate": 2.786804546245459e-06, "loss": 0.024396264925599098, "step": 2071 }, { "epoch": 1.531608133086876, "grad_norm": 3.8534932136535645, "learning_rate": 2.785009351143061e-06, "loss": 0.03918939828872681, "step": 2072 }, { "epoch": 1.5323475046210722, "grad_norm": 3.0898096561431885, "learning_rate": 2.783214133748461e-06, "loss": 0.014491518959403038, "step": 2073 }, { "epoch": 1.533086876155268, "grad_norm": 3.9059953689575195, "learning_rate": 2.7814188952047645e-06, "loss": 0.017025884240865707, "step": 2074 }, { "epoch": 1.533826247689464, "grad_norm": 3.588890790939331, "learning_rate": 2.77962363665509e-06, "loss": 0.008258352987468243, "step": 2075 }, { "epoch": 1.5345656192236599, "grad_norm": 1.4009593725204468, "learning_rate": 2.777828359242567e-06, "loss": 0.006819440051913261, "step": 2076 }, { "epoch": 1.5353049907578558, "grad_norm": 4.525960922241211, "learning_rate": 2.776033064110339e-06, "loss": 0.04391441494226456, "step": 2077 }, { "epoch": 1.5360443622920519, "grad_norm": 2.1202144622802734, "learning_rate": 2.7742377524015595e-06, "loss": 0.01208614930510521, "step": 2078 }, { "epoch": 1.5367837338262476, "grad_norm": 3.6044423580169678, "learning_rate": 2.7724424252593936e-06, "loss": 0.039229802787303925, "step": 2079 }, { "epoch": 1.5375231053604437, "grad_norm": 3.1814258098602295, "learning_rate": 2.7706470838270146e-06, "loss": 0.011209400370717049, "step": 2080 }, { "epoch": 1.5382624768946396, "grad_norm": 5.191589832305908, "learning_rate": 2.7688517292476076e-06, "loss": 0.03669383376836777, "step": 2081 }, { "epoch": 1.5390018484288355, "grad_norm": 4.711766719818115, "learning_rate": 2.7670563626643636e-06, "loss": 0.024888433516025543, "step": 2082 }, { "epoch": 1.5397412199630314, "grad_norm": 3.8805360794067383, "learning_rate": 2.7652609852204817e-06, "loss": 0.044097527861595154, "step": 2083 }, { "epoch": 1.5404805914972273, "grad_norm": 2.486405611038208, "learning_rate": 2.763465598059169e-06, "loss": 0.018522845581173897, "step": 2084 }, { "epoch": 1.5412199630314234, "grad_norm": 4.510000705718994, "learning_rate": 2.7616702023236384e-06, "loss": 0.03989826515316963, "step": 2085 }, { "epoch": 1.541959334565619, "grad_norm": 5.408608913421631, "learning_rate": 2.7598747991571075e-06, "loss": 0.01182643510401249, "step": 2086 }, { "epoch": 1.5426987060998152, "grad_norm": 3.347766160964966, "learning_rate": 2.758079389702798e-06, "loss": 0.021033184602856636, "step": 2087 }, { "epoch": 1.543438077634011, "grad_norm": 5.049282550811768, "learning_rate": 2.7562839751039388e-06, "loss": 0.04167461395263672, "step": 2088 }, { "epoch": 1.544177449168207, "grad_norm": 3.7371773719787598, "learning_rate": 2.7544885565037593e-06, "loss": 0.05849415063858032, "step": 2089 }, { "epoch": 1.544916820702403, "grad_norm": 3.5941901206970215, "learning_rate": 2.7526931350454917e-06, "loss": 0.03376667946577072, "step": 2090 }, { "epoch": 1.5456561922365988, "grad_norm": 3.7044341564178467, "learning_rate": 2.7508977118723703e-06, "loss": 0.033603571355342865, "step": 2091 }, { "epoch": 1.546395563770795, "grad_norm": 4.677684783935547, "learning_rate": 2.749102288127631e-06, "loss": 0.03668828308582306, "step": 2092 }, { "epoch": 1.5471349353049908, "grad_norm": 3.785496473312378, "learning_rate": 2.7473068649545094e-06, "loss": 0.028063688427209854, "step": 2093 }, { "epoch": 1.5478743068391867, "grad_norm": 3.8934106826782227, "learning_rate": 2.7455114434962422e-06, "loss": 0.049689825624227524, "step": 2094 }, { "epoch": 1.5486136783733826, "grad_norm": 3.230536460876465, "learning_rate": 2.743716024896062e-06, "loss": 0.028741229325532913, "step": 2095 }, { "epoch": 1.5493530499075785, "grad_norm": 2.9363694190979004, "learning_rate": 2.7419206102972022e-06, "loss": 0.023149844259023666, "step": 2096 }, { "epoch": 1.5500924214417746, "grad_norm": 3.2253479957580566, "learning_rate": 2.7401252008428937e-06, "loss": 0.025828128680586815, "step": 2097 }, { "epoch": 1.5508317929759703, "grad_norm": 2.1383814811706543, "learning_rate": 2.7383297976763623e-06, "loss": 0.007463611196726561, "step": 2098 }, { "epoch": 1.5515711645101664, "grad_norm": 3.5577774047851562, "learning_rate": 2.736534401940832e-06, "loss": 0.03688427433371544, "step": 2099 }, { "epoch": 1.5523105360443623, "grad_norm": 3.211240768432617, "learning_rate": 2.7347390147795195e-06, "loss": 0.0285252146422863, "step": 2100 }, { "epoch": 1.5530499075785582, "grad_norm": 3.776933193206787, "learning_rate": 2.732943637335637e-06, "loss": 0.04669594764709473, "step": 2101 }, { "epoch": 1.553789279112754, "grad_norm": 2.8506882190704346, "learning_rate": 2.7311482707523935e-06, "loss": 0.030647631734609604, "step": 2102 }, { "epoch": 1.55452865064695, "grad_norm": 2.9740841388702393, "learning_rate": 2.729352916172986e-06, "loss": 0.01953272521495819, "step": 2103 }, { "epoch": 1.5552680221811461, "grad_norm": 3.682354688644409, "learning_rate": 2.727557574740608e-06, "loss": 0.03786110877990723, "step": 2104 }, { "epoch": 1.5560073937153418, "grad_norm": 4.262057304382324, "learning_rate": 2.725762247598442e-06, "loss": 0.052747078239917755, "step": 2105 }, { "epoch": 1.556746765249538, "grad_norm": 2.7701940536499023, "learning_rate": 2.7239669358896625e-06, "loss": 0.03745228797197342, "step": 2106 }, { "epoch": 1.5574861367837338, "grad_norm": 3.9966375827789307, "learning_rate": 2.722171640757434e-06, "loss": 0.02716778591275215, "step": 2107 }, { "epoch": 1.5582255083179297, "grad_norm": 1.1587659120559692, "learning_rate": 2.7203763633449103e-06, "loss": 0.012118863873183727, "step": 2108 }, { "epoch": 1.5589648798521258, "grad_norm": 2.510188579559326, "learning_rate": 2.7185811047952366e-06, "loss": 0.01869913563132286, "step": 2109 }, { "epoch": 1.5597042513863215, "grad_norm": 3.2689011096954346, "learning_rate": 2.7167858662515393e-06, "loss": 0.031162556260824203, "step": 2110 }, { "epoch": 1.5604436229205176, "grad_norm": 3.176523447036743, "learning_rate": 2.7149906488569405e-06, "loss": 0.016916656866669655, "step": 2111 }, { "epoch": 1.5611829944547135, "grad_norm": 1.741491436958313, "learning_rate": 2.7131954537545417e-06, "loss": 0.01680500991642475, "step": 2112 }, { "epoch": 1.5619223659889094, "grad_norm": 3.6840906143188477, "learning_rate": 2.7114002820874345e-06, "loss": 0.03472576290369034, "step": 2113 }, { "epoch": 1.5626617375231053, "grad_norm": 3.86639666557312, "learning_rate": 2.709605134998695e-06, "loss": 0.018041040748357773, "step": 2114 }, { "epoch": 1.5634011090573012, "grad_norm": 2.0357658863067627, "learning_rate": 2.7078100136313817e-06, "loss": 0.014188794419169426, "step": 2115 }, { "epoch": 1.5641404805914974, "grad_norm": 4.162820339202881, "learning_rate": 2.706014919128539e-06, "loss": 0.04228239133954048, "step": 2116 }, { "epoch": 1.564879852125693, "grad_norm": 2.8970718383789062, "learning_rate": 2.704219852633192e-06, "loss": 0.02513071335852146, "step": 2117 }, { "epoch": 1.5656192236598891, "grad_norm": 3.0276734828948975, "learning_rate": 2.7024248152883504e-06, "loss": 0.01136883907020092, "step": 2118 }, { "epoch": 1.566358595194085, "grad_norm": 3.3653106689453125, "learning_rate": 2.7006298082370025e-06, "loss": 0.028043339028954506, "step": 2119 }, { "epoch": 1.567097966728281, "grad_norm": 4.334807395935059, "learning_rate": 2.6988348326221203e-06, "loss": 0.08606678247451782, "step": 2120 }, { "epoch": 1.5678373382624768, "grad_norm": 4.894867897033691, "learning_rate": 2.697039889586654e-06, "loss": 0.05167853832244873, "step": 2121 }, { "epoch": 1.5685767097966727, "grad_norm": 6.088501930236816, "learning_rate": 2.6952449802735325e-06, "loss": 0.02040935680270195, "step": 2122 }, { "epoch": 1.5693160813308689, "grad_norm": 1.9824745655059814, "learning_rate": 2.6934501058256664e-06, "loss": 0.005175154190510511, "step": 2123 }, { "epoch": 1.5700554528650645, "grad_norm": 2.689012289047241, "learning_rate": 2.691655267385939e-06, "loss": 0.017547961324453354, "step": 2124 }, { "epoch": 1.5707948243992607, "grad_norm": 4.688594341278076, "learning_rate": 2.6898604660972155e-06, "loss": 0.03597487509250641, "step": 2125 }, { "epoch": 1.5715341959334566, "grad_norm": 3.141841173171997, "learning_rate": 2.6880657031023342e-06, "loss": 0.022324126213788986, "step": 2126 }, { "epoch": 1.5722735674676525, "grad_norm": 5.54721212387085, "learning_rate": 2.686270979544112e-06, "loss": 0.024538546800613403, "step": 2127 }, { "epoch": 1.5730129390018486, "grad_norm": 2.99066162109375, "learning_rate": 2.6844762965653376e-06, "loss": 0.010682877153158188, "step": 2128 }, { "epoch": 1.5737523105360443, "grad_norm": 3.8980603218078613, "learning_rate": 2.6826816553087746e-06, "loss": 0.020283734425902367, "step": 2129 }, { "epoch": 1.5744916820702404, "grad_norm": 4.501880168914795, "learning_rate": 2.6808870569171623e-06, "loss": 0.03531963378190994, "step": 2130 }, { "epoch": 1.5752310536044363, "grad_norm": 2.8840644359588623, "learning_rate": 2.6790925025332102e-06, "loss": 0.013772494159638882, "step": 2131 }, { "epoch": 1.5759704251386322, "grad_norm": 4.940227508544922, "learning_rate": 2.6772979932996006e-06, "loss": 0.059365931898355484, "step": 2132 }, { "epoch": 1.576709796672828, "grad_norm": 4.54931640625, "learning_rate": 2.675503530358988e-06, "loss": 0.036045536398887634, "step": 2133 }, { "epoch": 1.577449168207024, "grad_norm": 4.276342391967773, "learning_rate": 2.6737091148539953e-06, "loss": 0.030805479735136032, "step": 2134 }, { "epoch": 1.57818853974122, "grad_norm": 5.482246398925781, "learning_rate": 2.6719147479272173e-06, "loss": 0.02705136500298977, "step": 2135 }, { "epoch": 1.5789279112754158, "grad_norm": 3.1551201343536377, "learning_rate": 2.670120430721216e-06, "loss": 0.021236602216959, "step": 2136 }, { "epoch": 1.5796672828096119, "grad_norm": 4.498534679412842, "learning_rate": 2.668326164378523e-06, "loss": 0.03559138998389244, "step": 2137 }, { "epoch": 1.5804066543438078, "grad_norm": 2.3637356758117676, "learning_rate": 2.6665319500416387e-06, "loss": 0.012057129293680191, "step": 2138 }, { "epoch": 1.5811460258780037, "grad_norm": 3.9778454303741455, "learning_rate": 2.664737788853027e-06, "loss": 0.04884276166558266, "step": 2139 }, { "epoch": 1.5818853974121996, "grad_norm": 1.9043936729431152, "learning_rate": 2.6629436819551206e-06, "loss": 0.007099098525941372, "step": 2140 }, { "epoch": 1.5826247689463955, "grad_norm": 4.020323276519775, "learning_rate": 2.6611496304903174e-06, "loss": 0.04351722076535225, "step": 2141 }, { "epoch": 1.5833641404805916, "grad_norm": 4.101926326751709, "learning_rate": 2.659355635600978e-06, "loss": 0.045565731823444366, "step": 2142 }, { "epoch": 1.5841035120147873, "grad_norm": 2.989345073699951, "learning_rate": 2.6575616984294313e-06, "loss": 0.011856739409267902, "step": 2143 }, { "epoch": 1.5848428835489834, "grad_norm": 2.7851152420043945, "learning_rate": 2.655767820117963e-06, "loss": 0.019566120579838753, "step": 2144 }, { "epoch": 1.5855822550831793, "grad_norm": 4.878220081329346, "learning_rate": 2.6539740018088283e-06, "loss": 0.04520207270979881, "step": 2145 }, { "epoch": 1.5863216266173752, "grad_norm": 4.458696365356445, "learning_rate": 2.6521802446442386e-06, "loss": 0.021732255816459656, "step": 2146 }, { "epoch": 1.5870609981515713, "grad_norm": 2.910449981689453, "learning_rate": 2.65038654976637e-06, "loss": 0.011614876799285412, "step": 2147 }, { "epoch": 1.587800369685767, "grad_norm": 3.601928949356079, "learning_rate": 2.6485929183173564e-06, "loss": 0.018376320600509644, "step": 2148 }, { "epoch": 1.588539741219963, "grad_norm": 4.866628646850586, "learning_rate": 2.6467993514392926e-06, "loss": 0.027372542768716812, "step": 2149 }, { "epoch": 1.589279112754159, "grad_norm": 2.883096218109131, "learning_rate": 2.645005850274233e-06, "loss": 0.020169749855995178, "step": 2150 }, { "epoch": 1.590018484288355, "grad_norm": 3.463554859161377, "learning_rate": 2.6432124159641892e-06, "loss": 0.034402016550302505, "step": 2151 }, { "epoch": 1.5907578558225508, "grad_norm": 3.308056592941284, "learning_rate": 2.6414190496511305e-06, "loss": 0.03108818084001541, "step": 2152 }, { "epoch": 1.5914972273567467, "grad_norm": 4.172164440155029, "learning_rate": 2.6396257524769807e-06, "loss": 0.03844154253602028, "step": 2153 }, { "epoch": 1.5922365988909428, "grad_norm": 2.765817880630493, "learning_rate": 2.637832525583624e-06, "loss": 0.012461339123547077, "step": 2154 }, { "epoch": 1.5929759704251385, "grad_norm": 3.266382932662964, "learning_rate": 2.6360393701128965e-06, "loss": 0.041773125529289246, "step": 2155 }, { "epoch": 1.5937153419593346, "grad_norm": 4.147752285003662, "learning_rate": 2.63424628720659e-06, "loss": 0.03238639235496521, "step": 2156 }, { "epoch": 1.5944547134935305, "grad_norm": 4.477581977844238, "learning_rate": 2.6324532780064505e-06, "loss": 0.032760266214609146, "step": 2157 }, { "epoch": 1.5951940850277264, "grad_norm": 4.564809322357178, "learning_rate": 2.6306603436541745e-06, "loss": 0.026150669902563095, "step": 2158 }, { "epoch": 1.5959334565619223, "grad_norm": 4.001355171203613, "learning_rate": 2.628867485291415e-06, "loss": 0.01581651158630848, "step": 2159 }, { "epoch": 1.5966728280961182, "grad_norm": 3.81207275390625, "learning_rate": 2.6270747040597725e-06, "loss": 0.02881780080497265, "step": 2160 }, { "epoch": 1.5974121996303143, "grad_norm": 3.8498263359069824, "learning_rate": 2.625282001100802e-06, "loss": 0.03315620869398117, "step": 2161 }, { "epoch": 1.59815157116451, "grad_norm": 3.6695315837860107, "learning_rate": 2.6234893775560063e-06, "loss": 0.04374134913086891, "step": 2162 }, { "epoch": 1.5988909426987061, "grad_norm": 5.583329200744629, "learning_rate": 2.621696834566837e-06, "loss": 0.03885158896446228, "step": 2163 }, { "epoch": 1.599630314232902, "grad_norm": 3.36185359954834, "learning_rate": 2.619904373274698e-06, "loss": 0.029017649590969086, "step": 2164 }, { "epoch": 1.600369685767098, "grad_norm": 2.9094603061676025, "learning_rate": 2.618111994820937e-06, "loss": 0.018361955881118774, "step": 2165 }, { "epoch": 1.601109057301294, "grad_norm": 3.191549777984619, "learning_rate": 2.6163197003468517e-06, "loss": 0.04414293169975281, "step": 2166 }, { "epoch": 1.6018484288354897, "grad_norm": 4.820195198059082, "learning_rate": 2.614527490993686e-06, "loss": 0.013723446056246758, "step": 2167 }, { "epoch": 1.6025878003696858, "grad_norm": 2.4858298301696777, "learning_rate": 2.6127353679026276e-06, "loss": 0.007553636096417904, "step": 2168 }, { "epoch": 1.6033271719038817, "grad_norm": 3.119143009185791, "learning_rate": 2.6109433322148113e-06, "loss": 0.016562264412641525, "step": 2169 }, { "epoch": 1.6040665434380776, "grad_norm": 1.8144575357437134, "learning_rate": 2.6091513850713156e-06, "loss": 0.004831458441913128, "step": 2170 }, { "epoch": 1.6048059149722735, "grad_norm": 4.801860332489014, "learning_rate": 2.6073595276131642e-06, "loss": 0.04993980750441551, "step": 2171 }, { "epoch": 1.6055452865064694, "grad_norm": 4.738921165466309, "learning_rate": 2.6055677609813195e-06, "loss": 0.03718870133161545, "step": 2172 }, { "epoch": 1.6062846580406656, "grad_norm": 3.413713216781616, "learning_rate": 2.6037760863166906e-06, "loss": 0.029007839038968086, "step": 2173 }, { "epoch": 1.6070240295748612, "grad_norm": 2.7808196544647217, "learning_rate": 2.6019845047601256e-06, "loss": 0.023691486567258835, "step": 2174 }, { "epoch": 1.6077634011090574, "grad_norm": 3.0971434116363525, "learning_rate": 2.6001930174524135e-06, "loss": 0.02784145250916481, "step": 2175 }, { "epoch": 1.6085027726432533, "grad_norm": 4.084527492523193, "learning_rate": 2.5984016255342858e-06, "loss": 0.029221873730421066, "step": 2176 }, { "epoch": 1.6092421441774492, "grad_norm": 5.215152263641357, "learning_rate": 2.5966103301464078e-06, "loss": 0.014876202680170536, "step": 2177 }, { "epoch": 1.609981515711645, "grad_norm": 5.564481735229492, "learning_rate": 2.594819132429389e-06, "loss": 0.03646313399076462, "step": 2178 }, { "epoch": 1.610720887245841, "grad_norm": 4.755812644958496, "learning_rate": 2.5930280335237738e-06, "loss": 0.03172755613923073, "step": 2179 }, { "epoch": 1.611460258780037, "grad_norm": 3.44962739944458, "learning_rate": 2.591237034570044e-06, "loss": 0.04748120903968811, "step": 2180 }, { "epoch": 1.6121996303142327, "grad_norm": 1.8907721042633057, "learning_rate": 2.5894461367086195e-06, "loss": 0.007649576291441917, "step": 2181 }, { "epoch": 1.6129390018484289, "grad_norm": 3.214712142944336, "learning_rate": 2.587655341079852e-06, "loss": 0.06442754715681076, "step": 2182 }, { "epoch": 1.6136783733826248, "grad_norm": 4.872313499450684, "learning_rate": 2.5858646488240315e-06, "loss": 0.049008361995220184, "step": 2183 }, { "epoch": 1.6144177449168207, "grad_norm": 2.9912755489349365, "learning_rate": 2.584074061081381e-06, "loss": 0.01814059168100357, "step": 2184 }, { "epoch": 1.6151571164510168, "grad_norm": 3.498853921890259, "learning_rate": 2.5822835789920574e-06, "loss": 0.021352656185626984, "step": 2185 }, { "epoch": 1.6158964879852125, "grad_norm": 3.266171932220459, "learning_rate": 2.580493203696151e-06, "loss": 0.009127631783485413, "step": 2186 }, { "epoch": 1.6166358595194086, "grad_norm": 4.30954122543335, "learning_rate": 2.5787029363336806e-06, "loss": 0.040453337132930756, "step": 2187 }, { "epoch": 1.6173752310536045, "grad_norm": 3.8558108806610107, "learning_rate": 2.576912778044601e-06, "loss": 0.027418097481131554, "step": 2188 }, { "epoch": 1.6181146025878004, "grad_norm": 3.0457189083099365, "learning_rate": 2.5751227299687942e-06, "loss": 0.017810750752687454, "step": 2189 }, { "epoch": 1.6188539741219963, "grad_norm": 4.5243964195251465, "learning_rate": 2.573332793246074e-06, "loss": 0.015516798011958599, "step": 2190 }, { "epoch": 1.6195933456561922, "grad_norm": 5.077071189880371, "learning_rate": 2.571542969016182e-06, "loss": 0.04133109003305435, "step": 2191 }, { "epoch": 1.6203327171903883, "grad_norm": 4.165071964263916, "learning_rate": 2.569753258418788e-06, "loss": 0.03890908136963844, "step": 2192 }, { "epoch": 1.621072088724584, "grad_norm": 4.60573148727417, "learning_rate": 2.5679636625934905e-06, "loss": 0.01965135708451271, "step": 2193 }, { "epoch": 1.62181146025878, "grad_norm": 3.4938788414001465, "learning_rate": 2.566174182679815e-06, "loss": 0.020129134878516197, "step": 2194 }, { "epoch": 1.622550831792976, "grad_norm": 3.897976875305176, "learning_rate": 2.5643848198172115e-06, "loss": 0.02463747188448906, "step": 2195 }, { "epoch": 1.6232902033271719, "grad_norm": 3.8112595081329346, "learning_rate": 2.5625955751450586e-06, "loss": 0.025300007313489914, "step": 2196 }, { "epoch": 1.6240295748613678, "grad_norm": 2.6549859046936035, "learning_rate": 2.5608064498026557e-06, "loss": 0.013521452434360981, "step": 2197 }, { "epoch": 1.6247689463955637, "grad_norm": 3.480973958969116, "learning_rate": 2.55901744492923e-06, "loss": 0.01464029960334301, "step": 2198 }, { "epoch": 1.6255083179297598, "grad_norm": 4.319767951965332, "learning_rate": 2.5572285616639277e-06, "loss": 0.016592172905802727, "step": 2199 }, { "epoch": 1.6262476894639555, "grad_norm": 3.048246383666992, "learning_rate": 2.555439801145823e-06, "loss": 0.013032794930040836, "step": 2200 }, { "epoch": 1.6269870609981516, "grad_norm": 4.247215270996094, "learning_rate": 2.553651164513907e-06, "loss": 0.029224667698144913, "step": 2201 }, { "epoch": 1.6277264325323475, "grad_norm": 4.225883483886719, "learning_rate": 2.551862652907094e-06, "loss": 0.02637692540884018, "step": 2202 }, { "epoch": 1.6284658040665434, "grad_norm": 4.639348030090332, "learning_rate": 2.5500742674642192e-06, "loss": 0.0200647059828043, "step": 2203 }, { "epoch": 1.6292051756007395, "grad_norm": 4.9393792152404785, "learning_rate": 2.5482860093240365e-06, "loss": 0.025173060595989227, "step": 2204 }, { "epoch": 1.6299445471349352, "grad_norm": 3.9269611835479736, "learning_rate": 2.546497879625221e-06, "loss": 0.018214207142591476, "step": 2205 }, { "epoch": 1.6306839186691313, "grad_norm": 4.741585731506348, "learning_rate": 2.544709879506361e-06, "loss": 0.028485411778092384, "step": 2206 }, { "epoch": 1.6314232902033272, "grad_norm": 3.6371333599090576, "learning_rate": 2.5429220101059666e-06, "loss": 0.016602972522377968, "step": 2207 }, { "epoch": 1.6321626617375231, "grad_norm": 4.908306121826172, "learning_rate": 2.541134272562465e-06, "loss": 0.042321331799030304, "step": 2208 }, { "epoch": 1.632902033271719, "grad_norm": 4.113441467285156, "learning_rate": 2.5393466680141953e-06, "loss": 0.012899376451969147, "step": 2209 }, { "epoch": 1.633641404805915, "grad_norm": 5.288241386413574, "learning_rate": 2.5375591975994174e-06, "loss": 0.020937282592058182, "step": 2210 }, { "epoch": 1.634380776340111, "grad_norm": 2.3809564113616943, "learning_rate": 2.5357718624563006e-06, "loss": 0.011501360684633255, "step": 2211 }, { "epoch": 1.6351201478743067, "grad_norm": 4.177746772766113, "learning_rate": 2.5339846637229314e-06, "loss": 0.03166777640581131, "step": 2212 }, { "epoch": 1.6358595194085028, "grad_norm": 3.514893054962158, "learning_rate": 2.532197602537308e-06, "loss": 0.020802542567253113, "step": 2213 }, { "epoch": 1.6365988909426987, "grad_norm": 3.165635108947754, "learning_rate": 2.5304106800373416e-06, "loss": 0.0324312224984169, "step": 2214 }, { "epoch": 1.6373382624768946, "grad_norm": 2.68092679977417, "learning_rate": 2.5286238973608568e-06, "loss": 0.014812100678682327, "step": 2215 }, { "epoch": 1.6380776340110905, "grad_norm": 3.579054117202759, "learning_rate": 2.5268372556455838e-06, "loss": 0.01912534423172474, "step": 2216 }, { "epoch": 1.6388170055452864, "grad_norm": 2.832308053970337, "learning_rate": 2.525050756029169e-06, "loss": 0.02274181693792343, "step": 2217 }, { "epoch": 1.6395563770794825, "grad_norm": 3.378969669342041, "learning_rate": 2.523264399649165e-06, "loss": 0.02201029099524021, "step": 2218 }, { "epoch": 1.6402957486136782, "grad_norm": 3.7921054363250732, "learning_rate": 2.5214781876430345e-06, "loss": 0.026608718559145927, "step": 2219 }, { "epoch": 1.6410351201478743, "grad_norm": 2.002547264099121, "learning_rate": 2.519692121148148e-06, "loss": 0.012135550379753113, "step": 2220 }, { "epoch": 1.6417744916820702, "grad_norm": 2.861755609512329, "learning_rate": 2.5179062013017818e-06, "loss": 0.015366124920547009, "step": 2221 }, { "epoch": 1.6425138632162661, "grad_norm": 3.1139700412750244, "learning_rate": 2.516120429241122e-06, "loss": 0.015888875350356102, "step": 2222 }, { "epoch": 1.6432532347504623, "grad_norm": 3.5239531993865967, "learning_rate": 2.514334806103258e-06, "loss": 0.03841772675514221, "step": 2223 }, { "epoch": 1.643992606284658, "grad_norm": 4.263808250427246, "learning_rate": 2.5125493330251855e-06, "loss": 0.04863758012652397, "step": 2224 }, { "epoch": 1.644731977818854, "grad_norm": 4.346721649169922, "learning_rate": 2.510764011143803e-06, "loss": 0.020504526793956757, "step": 2225 }, { "epoch": 1.64547134935305, "grad_norm": 4.134269714355469, "learning_rate": 2.5089788415959155e-06, "loss": 0.022496284916996956, "step": 2226 }, { "epoch": 1.6462107208872458, "grad_norm": 5.9540886878967285, "learning_rate": 2.5071938255182294e-06, "loss": 0.018665645271539688, "step": 2227 }, { "epoch": 1.6469500924214417, "grad_norm": 5.121501922607422, "learning_rate": 2.505408964047352e-06, "loss": 0.03920939937233925, "step": 2228 }, { "epoch": 1.6476894639556376, "grad_norm": 3.9685704708099365, "learning_rate": 2.5036242583197966e-06, "loss": 0.014643660746514797, "step": 2229 }, { "epoch": 1.6484288354898338, "grad_norm": 4.25244665145874, "learning_rate": 2.501839709471972e-06, "loss": 0.016370058059692383, "step": 2230 }, { "epoch": 1.6491682070240294, "grad_norm": 2.9850857257843018, "learning_rate": 2.50005531864019e-06, "loss": 0.014751624315977097, "step": 2231 }, { "epoch": 1.6499075785582256, "grad_norm": 2.7486250400543213, "learning_rate": 2.498271086960662e-06, "loss": 0.013475966639816761, "step": 2232 }, { "epoch": 1.6506469500924215, "grad_norm": 3.6529593467712402, "learning_rate": 2.4964870155694964e-06, "loss": 0.0254351869225502, "step": 2233 }, { "epoch": 1.6513863216266174, "grad_norm": 4.225690841674805, "learning_rate": 2.4947031056027022e-06, "loss": 0.018910758197307587, "step": 2234 }, { "epoch": 1.6521256931608133, "grad_norm": 2.9093236923217773, "learning_rate": 2.4929193581961813e-06, "loss": 0.005399193149060011, "step": 2235 }, { "epoch": 1.6528650646950092, "grad_norm": 4.248211860656738, "learning_rate": 2.4911357744857367e-06, "loss": 0.019361088052392006, "step": 2236 }, { "epoch": 1.6536044362292053, "grad_norm": 4.932898998260498, "learning_rate": 2.489352355607064e-06, "loss": 0.028588971123099327, "step": 2237 }, { "epoch": 1.654343807763401, "grad_norm": 3.3856441974639893, "learning_rate": 2.487569102695756e-06, "loss": 0.026681894436478615, "step": 2238 }, { "epoch": 1.655083179297597, "grad_norm": 3.9853861331939697, "learning_rate": 2.4857860168872993e-06, "loss": 0.028581559658050537, "step": 2239 }, { "epoch": 1.655822550831793, "grad_norm": 3.08154034614563, "learning_rate": 2.484003099317072e-06, "loss": 0.014523951336741447, "step": 2240 }, { "epoch": 1.6565619223659889, "grad_norm": 3.663041591644287, "learning_rate": 2.482220351120348e-06, "loss": 0.02577362023293972, "step": 2241 }, { "epoch": 1.657301293900185, "grad_norm": 3.79464054107666, "learning_rate": 2.4804377734322906e-06, "loss": 0.017186908051371574, "step": 2242 }, { "epoch": 1.6580406654343807, "grad_norm": 2.3651371002197266, "learning_rate": 2.4786553673879573e-06, "loss": 0.009456261061131954, "step": 2243 }, { "epoch": 1.6587800369685768, "grad_norm": 4.780938148498535, "learning_rate": 2.476873134122295e-06, "loss": 0.037169717252254486, "step": 2244 }, { "epoch": 1.6595194085027727, "grad_norm": 4.106514930725098, "learning_rate": 2.4750910747701396e-06, "loss": 0.01656350865960121, "step": 2245 }, { "epoch": 1.6602587800369686, "grad_norm": 3.8311965465545654, "learning_rate": 2.473309190466218e-06, "loss": 0.025148963555693626, "step": 2246 }, { "epoch": 1.6609981515711645, "grad_norm": 3.832392454147339, "learning_rate": 2.4715274823451447e-06, "loss": 0.02080075815320015, "step": 2247 }, { "epoch": 1.6617375231053604, "grad_norm": 3.748318910598755, "learning_rate": 2.469745951541422e-06, "loss": 0.03160683065652847, "step": 2248 }, { "epoch": 1.6624768946395565, "grad_norm": 3.7800207138061523, "learning_rate": 2.46796459918944e-06, "loss": 0.026637867093086243, "step": 2249 }, { "epoch": 1.6632162661737522, "grad_norm": 5.39107608795166, "learning_rate": 2.466183426423474e-06, "loss": 0.0466284416615963, "step": 2250 }, { "epoch": 1.6639556377079483, "grad_norm": 3.34173846244812, "learning_rate": 2.4644024343776863e-06, "loss": 0.017087183892726898, "step": 2251 }, { "epoch": 1.6646950092421442, "grad_norm": 3.4112863540649414, "learning_rate": 2.4626216241861224e-06, "loss": 0.021670622751116753, "step": 2252 }, { "epoch": 1.66543438077634, "grad_norm": 5.093894004821777, "learning_rate": 2.4608409969827148e-06, "loss": 0.04217101261019707, "step": 2253 }, { "epoch": 1.666173752310536, "grad_norm": 3.880866289138794, "learning_rate": 2.459060553901275e-06, "loss": 0.027715053409337997, "step": 2254 }, { "epoch": 1.666913123844732, "grad_norm": 3.52370285987854, "learning_rate": 2.4572802960755014e-06, "loss": 0.013565486297011375, "step": 2255 }, { "epoch": 1.667652495378928, "grad_norm": 3.3263986110687256, "learning_rate": 2.4555002246389754e-06, "loss": 0.02409905381500721, "step": 2256 }, { "epoch": 1.6683918669131237, "grad_norm": 3.389965772628784, "learning_rate": 2.453720340725153e-06, "loss": 0.028815191239118576, "step": 2257 }, { "epoch": 1.6691312384473198, "grad_norm": 3.718405246734619, "learning_rate": 2.451940645467379e-06, "loss": 0.04453178867697716, "step": 2258 }, { "epoch": 1.6698706099815157, "grad_norm": 2.0728919506073, "learning_rate": 2.450161139998871e-06, "loss": 0.010756094008684158, "step": 2259 }, { "epoch": 1.6706099815157116, "grad_norm": 4.418429374694824, "learning_rate": 2.4483818254527295e-06, "loss": 0.029085299000144005, "step": 2260 }, { "epoch": 1.6713493530499077, "grad_norm": 5.034226417541504, "learning_rate": 2.4466027029619354e-06, "loss": 0.02200089395046234, "step": 2261 }, { "epoch": 1.6720887245841034, "grad_norm": 4.845980644226074, "learning_rate": 2.4448237736593423e-06, "loss": 0.03161676600575447, "step": 2262 }, { "epoch": 1.6728280961182995, "grad_norm": 3.374791383743286, "learning_rate": 2.4430450386776855e-06, "loss": 0.019193846732378006, "step": 2263 }, { "epoch": 1.6735674676524954, "grad_norm": 4.3976240158081055, "learning_rate": 2.441266499149571e-06, "loss": 0.04156515374779701, "step": 2264 }, { "epoch": 1.6743068391866913, "grad_norm": 2.8635363578796387, "learning_rate": 2.439488156207487e-06, "loss": 0.024174604564905167, "step": 2265 }, { "epoch": 1.6750462107208872, "grad_norm": 3.3264057636260986, "learning_rate": 2.4377100109837907e-06, "loss": 0.03350567817687988, "step": 2266 }, { "epoch": 1.6757855822550831, "grad_norm": 2.721006393432617, "learning_rate": 2.4359320646107176e-06, "loss": 0.026980740949511528, "step": 2267 }, { "epoch": 1.6765249537892792, "grad_norm": 3.710674285888672, "learning_rate": 2.4341543182203747e-06, "loss": 0.03295290842652321, "step": 2268 }, { "epoch": 1.677264325323475, "grad_norm": 3.8217387199401855, "learning_rate": 2.43237677294474e-06, "loss": 0.016600975766777992, "step": 2269 }, { "epoch": 1.678003696857671, "grad_norm": 3.984178066253662, "learning_rate": 2.430599429915669e-06, "loss": 0.017303481698036194, "step": 2270 }, { "epoch": 1.678743068391867, "grad_norm": 3.898719549179077, "learning_rate": 2.4288222902648813e-06, "loss": 0.019104868173599243, "step": 2271 }, { "epoch": 1.6794824399260628, "grad_norm": 1.9477897882461548, "learning_rate": 2.427045355123971e-06, "loss": 0.005281868390738964, "step": 2272 }, { "epoch": 1.6802218114602587, "grad_norm": 4.3093647956848145, "learning_rate": 2.4252686256244034e-06, "loss": 0.030468560755252838, "step": 2273 }, { "epoch": 1.6809611829944546, "grad_norm": 3.545565128326416, "learning_rate": 2.4234921028975078e-06, "loss": 0.028422603383660316, "step": 2274 }, { "epoch": 1.6817005545286507, "grad_norm": 2.6478583812713623, "learning_rate": 2.421715788074488e-06, "loss": 0.016461249440908432, "step": 2275 }, { "epoch": 1.6824399260628464, "grad_norm": 3.4093358516693115, "learning_rate": 2.4199396822864103e-06, "loss": 0.025759069249033928, "step": 2276 }, { "epoch": 1.6831792975970425, "grad_norm": 3.2077932357788086, "learning_rate": 2.418163786664211e-06, "loss": 0.03547879680991173, "step": 2277 }, { "epoch": 1.6839186691312384, "grad_norm": 3.526939868927002, "learning_rate": 2.416388102338689e-06, "loss": 0.024577828124165535, "step": 2278 }, { "epoch": 1.6846580406654343, "grad_norm": 4.07809591293335, "learning_rate": 2.4146126304405147e-06, "loss": 0.034622445702552795, "step": 2279 }, { "epoch": 1.6853974121996305, "grad_norm": 3.8468434810638428, "learning_rate": 2.4128373721002184e-06, "loss": 0.026196900755167007, "step": 2280 }, { "epoch": 1.6861367837338261, "grad_norm": 5.038276195526123, "learning_rate": 2.411062328448194e-06, "loss": 0.0374327152967453, "step": 2281 }, { "epoch": 1.6868761552680223, "grad_norm": 4.311193466186523, "learning_rate": 2.409287500614703e-06, "loss": 0.026636317372322083, "step": 2282 }, { "epoch": 1.6876155268022182, "grad_norm": 2.436724901199341, "learning_rate": 2.4075128897298645e-06, "loss": 0.01207474060356617, "step": 2283 }, { "epoch": 1.688354898336414, "grad_norm": 4.4227213859558105, "learning_rate": 2.405738496923663e-06, "loss": 0.024747345596551895, "step": 2284 }, { "epoch": 1.68909426987061, "grad_norm": 2.555541753768921, "learning_rate": 2.4039643233259443e-06, "loss": 0.011917375028133392, "step": 2285 }, { "epoch": 1.6898336414048059, "grad_norm": 2.410705327987671, "learning_rate": 2.402190370066411e-06, "loss": 0.025761984288692474, "step": 2286 }, { "epoch": 1.690573012939002, "grad_norm": 3.5048508644104004, "learning_rate": 2.4004166382746275e-06, "loss": 0.03313254192471504, "step": 2287 }, { "epoch": 1.6913123844731976, "grad_norm": 3.0678086280822754, "learning_rate": 2.3986431290800195e-06, "loss": 0.015357041731476784, "step": 2288 }, { "epoch": 1.6920517560073938, "grad_norm": 3.681043863296509, "learning_rate": 2.3968698436118677e-06, "loss": 0.028955236077308655, "step": 2289 }, { "epoch": 1.6927911275415897, "grad_norm": 3.0472118854522705, "learning_rate": 2.395096782999311e-06, "loss": 0.016783947125077248, "step": 2290 }, { "epoch": 1.6935304990757856, "grad_norm": 3.17226505279541, "learning_rate": 2.3933239483713457e-06, "loss": 0.013800324872136116, "step": 2291 }, { "epoch": 1.6942698706099815, "grad_norm": 3.5595765113830566, "learning_rate": 2.3915513408568253e-06, "loss": 0.014303334057331085, "step": 2292 }, { "epoch": 1.6950092421441774, "grad_norm": 3.72831654548645, "learning_rate": 2.389778961584456e-06, "loss": 0.020840570330619812, "step": 2293 }, { "epoch": 1.6957486136783735, "grad_norm": 3.4091687202453613, "learning_rate": 2.3880068116828014e-06, "loss": 0.017661523073911667, "step": 2294 }, { "epoch": 1.6964879852125692, "grad_norm": 4.0046162605285645, "learning_rate": 2.3862348922802765e-06, "loss": 0.042785536497831345, "step": 2295 }, { "epoch": 1.6972273567467653, "grad_norm": 3.073822021484375, "learning_rate": 2.3844632045051515e-06, "loss": 0.01740070804953575, "step": 2296 }, { "epoch": 1.6979667282809612, "grad_norm": 4.057745456695557, "learning_rate": 2.382691749485549e-06, "loss": 0.013399899937212467, "step": 2297 }, { "epoch": 1.698706099815157, "grad_norm": 4.370275497436523, "learning_rate": 2.380920528349441e-06, "loss": 0.03711778670549393, "step": 2298 }, { "epoch": 1.6994454713493532, "grad_norm": 3.042551279067993, "learning_rate": 2.3791495422246553e-06, "loss": 0.017035529017448425, "step": 2299 }, { "epoch": 1.7001848428835489, "grad_norm": 2.4010462760925293, "learning_rate": 2.3773787922388646e-06, "loss": 0.013651757501065731, "step": 2300 }, { "epoch": 1.700924214417745, "grad_norm": 3.81661319732666, "learning_rate": 2.3756082795195952e-06, "loss": 0.03946610540151596, "step": 2301 }, { "epoch": 1.701663585951941, "grad_norm": 5.024046897888184, "learning_rate": 2.3738380051942193e-06, "loss": 0.045808132737874985, "step": 2302 }, { "epoch": 1.7024029574861368, "grad_norm": 3.885521411895752, "learning_rate": 2.37206797038996e-06, "loss": 0.028600921854376793, "step": 2303 }, { "epoch": 1.7031423290203327, "grad_norm": 3.274385690689087, "learning_rate": 2.370298176233889e-06, "loss": 0.021771393716335297, "step": 2304 }, { "epoch": 1.7038817005545286, "grad_norm": 3.368257522583008, "learning_rate": 2.368528623852918e-06, "loss": 0.030949480831623077, "step": 2305 }, { "epoch": 1.7046210720887247, "grad_norm": 3.2495460510253906, "learning_rate": 2.3667593143738134e-06, "loss": 0.03356751799583435, "step": 2306 }, { "epoch": 1.7053604436229204, "grad_norm": 3.072146415710449, "learning_rate": 2.3649902489231795e-06, "loss": 0.016124894842505455, "step": 2307 }, { "epoch": 1.7060998151571165, "grad_norm": 2.0861802101135254, "learning_rate": 2.3632214286274713e-06, "loss": 0.01562560722231865, "step": 2308 }, { "epoch": 1.7068391866913124, "grad_norm": 3.117454767227173, "learning_rate": 2.3614528546129847e-06, "loss": 0.017299918457865715, "step": 2309 }, { "epoch": 1.7075785582255083, "grad_norm": 2.657308340072632, "learning_rate": 2.3596845280058566e-06, "loss": 0.018315879628062248, "step": 2310 }, { "epoch": 1.7083179297597042, "grad_norm": 2.8641223907470703, "learning_rate": 2.357916449932071e-06, "loss": 0.016267186030745506, "step": 2311 }, { "epoch": 1.7090573012939, "grad_norm": 3.103268623352051, "learning_rate": 2.356148621517451e-06, "loss": 0.029242921620607376, "step": 2312 }, { "epoch": 1.7097966728280962, "grad_norm": 3.146583080291748, "learning_rate": 2.3543810438876607e-06, "loss": 0.01746893674135208, "step": 2313 }, { "epoch": 1.710536044362292, "grad_norm": 5.043806552886963, "learning_rate": 2.352613718168207e-06, "loss": 0.021496806293725967, "step": 2314 }, { "epoch": 1.711275415896488, "grad_norm": 3.1149189472198486, "learning_rate": 2.350846645484431e-06, "loss": 0.02595711499452591, "step": 2315 }, { "epoch": 1.712014787430684, "grad_norm": 2.9144346714019775, "learning_rate": 2.3490798269615183e-06, "loss": 0.006921641528606415, "step": 2316 }, { "epoch": 1.7127541589648798, "grad_norm": 6.384328842163086, "learning_rate": 2.34731326372449e-06, "loss": 0.0363117940723896, "step": 2317 }, { "epoch": 1.713493530499076, "grad_norm": 3.24922776222229, "learning_rate": 2.3455469568982063e-06, "loss": 0.019358929246664047, "step": 2318 }, { "epoch": 1.7142329020332716, "grad_norm": 4.106117248535156, "learning_rate": 2.3437809076073605e-06, "loss": 0.017269687727093697, "step": 2319 }, { "epoch": 1.7149722735674677, "grad_norm": 2.91957688331604, "learning_rate": 2.3420151169764854e-06, "loss": 0.01759764924645424, "step": 2320 }, { "epoch": 1.7157116451016636, "grad_norm": 4.355286598205566, "learning_rate": 2.3402495861299486e-06, "loss": 0.026261962950229645, "step": 2321 }, { "epoch": 1.7164510166358595, "grad_norm": 4.0125813484191895, "learning_rate": 2.338484316191951e-06, "loss": 0.012858608737587929, "step": 2322 }, { "epoch": 1.7171903881700554, "grad_norm": 5.501418590545654, "learning_rate": 2.3367193082865285e-06, "loss": 0.028274845331907272, "step": 2323 }, { "epoch": 1.7179297597042513, "grad_norm": 3.6245384216308594, "learning_rate": 2.3349545635375497e-06, "loss": 0.02034178003668785, "step": 2324 }, { "epoch": 1.7186691312384474, "grad_norm": 5.436835765838623, "learning_rate": 2.3331900830687148e-06, "loss": 0.02428307756781578, "step": 2325 }, { "epoch": 1.7194085027726431, "grad_norm": 3.8148443698883057, "learning_rate": 2.3314258680035584e-06, "loss": 0.02715928480029106, "step": 2326 }, { "epoch": 1.7201478743068392, "grad_norm": 3.8209753036499023, "learning_rate": 2.3296619194654424e-06, "loss": 0.028351858258247375, "step": 2327 }, { "epoch": 1.7208872458410351, "grad_norm": 1.9815727472305298, "learning_rate": 2.327898238577562e-06, "loss": 0.0051626223139464855, "step": 2328 }, { "epoch": 1.721626617375231, "grad_norm": 3.5638468265533447, "learning_rate": 2.3261348264629407e-06, "loss": 0.023758698254823685, "step": 2329 }, { "epoch": 1.722365988909427, "grad_norm": 3.5325796604156494, "learning_rate": 2.3243716842444308e-06, "loss": 0.01920846477150917, "step": 2330 }, { "epoch": 1.7231053604436228, "grad_norm": 2.4635813236236572, "learning_rate": 2.322608813044712e-06, "loss": 0.0049208784475922585, "step": 2331 }, { "epoch": 1.723844731977819, "grad_norm": 3.90047287940979, "learning_rate": 2.320846213986294e-06, "loss": 0.018197979778051376, "step": 2332 }, { "epoch": 1.7245841035120146, "grad_norm": 1.885125994682312, "learning_rate": 2.3190838881915116e-06, "loss": 0.011636778712272644, "step": 2333 }, { "epoch": 1.7253234750462108, "grad_norm": 5.258152961730957, "learning_rate": 2.317321836782524e-06, "loss": 0.028128089383244514, "step": 2334 }, { "epoch": 1.7260628465804067, "grad_norm": 3.323159694671631, "learning_rate": 2.315560060881319e-06, "loss": 0.01786346174776554, "step": 2335 }, { "epoch": 1.7268022181146025, "grad_norm": 3.4466965198516846, "learning_rate": 2.3137985616097053e-06, "loss": 0.020815221592783928, "step": 2336 }, { "epoch": 1.7275415896487987, "grad_norm": 1.6195108890533447, "learning_rate": 2.312037340089319e-06, "loss": 0.00544291315600276, "step": 2337 }, { "epoch": 1.7282809611829943, "grad_norm": 4.730764865875244, "learning_rate": 2.310276397441618e-06, "loss": 0.018662355840206146, "step": 2338 }, { "epoch": 1.7290203327171905, "grad_norm": 3.856997489929199, "learning_rate": 2.308515734787882e-06, "loss": 0.030451664701104164, "step": 2339 }, { "epoch": 1.7297597042513864, "grad_norm": 6.942995548248291, "learning_rate": 2.3067553532492117e-06, "loss": 0.040018439292907715, "step": 2340 }, { "epoch": 1.7304990757855823, "grad_norm": 4.377633571624756, "learning_rate": 2.3049952539465297e-06, "loss": 0.02344861440360546, "step": 2341 }, { "epoch": 1.7312384473197782, "grad_norm": 3.987415075302124, "learning_rate": 2.303235438000582e-06, "loss": 0.016029536724090576, "step": 2342 }, { "epoch": 1.731977818853974, "grad_norm": 2.9169933795928955, "learning_rate": 2.301475906531928e-06, "loss": 0.014199856668710709, "step": 2343 }, { "epoch": 1.7327171903881702, "grad_norm": 2.4609878063201904, "learning_rate": 2.2997166606609505e-06, "loss": 0.013977241702377796, "step": 2344 }, { "epoch": 1.7334565619223659, "grad_norm": 3.6024022102355957, "learning_rate": 2.2979577015078497e-06, "loss": 0.023516887798905373, "step": 2345 }, { "epoch": 1.734195933456562, "grad_norm": 4.365606784820557, "learning_rate": 2.2961990301926418e-06, "loss": 0.025876987725496292, "step": 2346 }, { "epoch": 1.7349353049907579, "grad_norm": 4.00601863861084, "learning_rate": 2.2944406478351615e-06, "loss": 0.023928221315145493, "step": 2347 }, { "epoch": 1.7356746765249538, "grad_norm": 2.578449010848999, "learning_rate": 2.292682555555057e-06, "loss": 0.014228319749236107, "step": 2348 }, { "epoch": 1.7364140480591497, "grad_norm": 3.812013626098633, "learning_rate": 2.290924754471795e-06, "loss": 0.01216413639485836, "step": 2349 }, { "epoch": 1.7371534195933456, "grad_norm": 3.365164041519165, "learning_rate": 2.2891672457046545e-06, "loss": 0.017062250524759293, "step": 2350 }, { "epoch": 1.7378927911275417, "grad_norm": 3.7219555377960205, "learning_rate": 2.287410030372729e-06, "loss": 0.016293887048959732, "step": 2351 }, { "epoch": 1.7386321626617374, "grad_norm": 4.310502052307129, "learning_rate": 2.2856531095949263e-06, "loss": 0.05316997319459915, "step": 2352 }, { "epoch": 1.7393715341959335, "grad_norm": 4.159209728240967, "learning_rate": 2.2838964844899635e-06, "loss": 0.009063941426575184, "step": 2353 }, { "epoch": 1.7401109057301294, "grad_norm": 1.5583134889602661, "learning_rate": 2.2821401561763736e-06, "loss": 0.0025081171188503504, "step": 2354 }, { "epoch": 1.7408502772643253, "grad_norm": 1.9604299068450928, "learning_rate": 2.2803841257724963e-06, "loss": 0.007108820136636496, "step": 2355 }, { "epoch": 1.7415896487985214, "grad_norm": 2.818726062774658, "learning_rate": 2.2786283943964855e-06, "loss": 0.010156258940696716, "step": 2356 }, { "epoch": 1.742329020332717, "grad_norm": 4.225902557373047, "learning_rate": 2.2768729631663027e-06, "loss": 0.01753763109445572, "step": 2357 }, { "epoch": 1.7430683918669132, "grad_norm": 3.7842750549316406, "learning_rate": 2.2751178331997188e-06, "loss": 0.026155179366469383, "step": 2358 }, { "epoch": 1.743807763401109, "grad_norm": 3.193389654159546, "learning_rate": 2.273363005614312e-06, "loss": 0.014789875596761703, "step": 2359 }, { "epoch": 1.744547134935305, "grad_norm": 4.330848217010498, "learning_rate": 2.271608481527468e-06, "loss": 0.019258223474025726, "step": 2360 }, { "epoch": 1.745286506469501, "grad_norm": 1.5740916728973389, "learning_rate": 2.2698542620563806e-06, "loss": 0.002356966957449913, "step": 2361 }, { "epoch": 1.7460258780036968, "grad_norm": 2.9876863956451416, "learning_rate": 2.2681003483180506e-06, "loss": 0.011176135390996933, "step": 2362 }, { "epoch": 1.746765249537893, "grad_norm": 3.4525959491729736, "learning_rate": 2.2663467414292795e-06, "loss": 0.015460671856999397, "step": 2363 }, { "epoch": 1.7475046210720886, "grad_norm": 4.355672359466553, "learning_rate": 2.2645934425066778e-06, "loss": 0.027341730892658234, "step": 2364 }, { "epoch": 1.7482439926062847, "grad_norm": 1.201949954032898, "learning_rate": 2.2628404526666585e-06, "loss": 0.0016113186720758677, "step": 2365 }, { "epoch": 1.7489833641404806, "grad_norm": 3.258082628250122, "learning_rate": 2.261087773025437e-06, "loss": 0.020061161369085312, "step": 2366 }, { "epoch": 1.7497227356746765, "grad_norm": 5.092148780822754, "learning_rate": 2.2593354046990336e-06, "loss": 0.03500501811504364, "step": 2367 }, { "epoch": 1.7504621072088724, "grad_norm": 3.5656025409698486, "learning_rate": 2.2575833488032667e-06, "loss": 0.027850154787302017, "step": 2368 }, { "epoch": 1.7512014787430683, "grad_norm": 4.061437129974365, "learning_rate": 2.2558316064537582e-06, "loss": 0.01219043880701065, "step": 2369 }, { "epoch": 1.7519408502772644, "grad_norm": 3.4214112758636475, "learning_rate": 2.2540801787659304e-06, "loss": 0.013342330232262611, "step": 2370 }, { "epoch": 1.75268022181146, "grad_norm": 3.888435125350952, "learning_rate": 2.252329066855005e-06, "loss": 0.030443964526057243, "step": 2371 }, { "epoch": 1.7534195933456562, "grad_norm": 2.215669631958008, "learning_rate": 2.2505782718360014e-06, "loss": 0.008965929038822651, "step": 2372 }, { "epoch": 1.7541589648798521, "grad_norm": 2.6345059871673584, "learning_rate": 2.248827794823738e-06, "loss": 0.01995950937271118, "step": 2373 }, { "epoch": 1.754898336414048, "grad_norm": 2.8176822662353516, "learning_rate": 2.247077636932833e-06, "loss": 0.013815215788781643, "step": 2374 }, { "epoch": 1.7556377079482441, "grad_norm": 4.155664920806885, "learning_rate": 2.245327799277696e-06, "loss": 0.036745890974998474, "step": 2375 }, { "epoch": 1.7563770794824398, "grad_norm": 2.999608039855957, "learning_rate": 2.24357828297254e-06, "loss": 0.012460715137422085, "step": 2376 }, { "epoch": 1.757116451016636, "grad_norm": 3.0163888931274414, "learning_rate": 2.2418290891313653e-06, "loss": 0.011357758194208145, "step": 2377 }, { "epoch": 1.7578558225508318, "grad_norm": 4.066549777984619, "learning_rate": 2.2400802188679733e-06, "loss": 0.02111804112792015, "step": 2378 }, { "epoch": 1.7585951940850277, "grad_norm": 3.4116640090942383, "learning_rate": 2.2383316732959565e-06, "loss": 0.026694023981690407, "step": 2379 }, { "epoch": 1.7593345656192236, "grad_norm": 2.7412784099578857, "learning_rate": 2.2365834535287005e-06, "loss": 0.014590496197342873, "step": 2380 }, { "epoch": 1.7600739371534195, "grad_norm": 4.303763389587402, "learning_rate": 2.234835560679386e-06, "loss": 0.015246809460222721, "step": 2381 }, { "epoch": 1.7608133086876157, "grad_norm": 3.6770005226135254, "learning_rate": 2.2330879958609806e-06, "loss": 0.015654275193810463, "step": 2382 }, { "epoch": 1.7615526802218113, "grad_norm": 3.9509403705596924, "learning_rate": 2.231340760186249e-06, "loss": 0.02215556800365448, "step": 2383 }, { "epoch": 1.7622920517560074, "grad_norm": 2.264047861099243, "learning_rate": 2.2295938547677415e-06, "loss": 0.02183609828352928, "step": 2384 }, { "epoch": 1.7630314232902033, "grad_norm": 4.359470367431641, "learning_rate": 2.2278472807178004e-06, "loss": 0.020583637058734894, "step": 2385 }, { "epoch": 1.7637707948243992, "grad_norm": 2.7489120960235596, "learning_rate": 2.226101039148557e-06, "loss": 0.012214656919240952, "step": 2386 }, { "epoch": 1.7645101663585951, "grad_norm": 2.773632049560547, "learning_rate": 2.22435513117193e-06, "loss": 0.013554391451179981, "step": 2387 }, { "epoch": 1.765249537892791, "grad_norm": 2.2618608474731445, "learning_rate": 2.2226095578996264e-06, "loss": 0.01822330430150032, "step": 2388 }, { "epoch": 1.7659889094269872, "grad_norm": 1.106543779373169, "learning_rate": 2.2208643204431396e-06, "loss": 0.007582572754472494, "step": 2389 }, { "epoch": 1.7667282809611828, "grad_norm": 3.143386125564575, "learning_rate": 2.2191194199137494e-06, "loss": 0.017471645027399063, "step": 2390 }, { "epoch": 1.767467652495379, "grad_norm": 4.399631977081299, "learning_rate": 2.2173748574225224e-06, "loss": 0.036028847098350525, "step": 2391 }, { "epoch": 1.7682070240295749, "grad_norm": 3.790013074874878, "learning_rate": 2.215630634080306e-06, "loss": 0.03649940714240074, "step": 2392 }, { "epoch": 1.7689463955637708, "grad_norm": 2.6048028469085693, "learning_rate": 2.2138867509977363e-06, "loss": 0.02308293804526329, "step": 2393 }, { "epoch": 1.7696857670979669, "grad_norm": 2.881810188293457, "learning_rate": 2.2121432092852298e-06, "loss": 0.01218391302973032, "step": 2394 }, { "epoch": 1.7704251386321626, "grad_norm": 3.995593786239624, "learning_rate": 2.2104000100529873e-06, "loss": 0.02590947225689888, "step": 2395 }, { "epoch": 1.7711645101663587, "grad_norm": 1.7587412595748901, "learning_rate": 2.208657154410989e-06, "loss": 0.009115811437368393, "step": 2396 }, { "epoch": 1.7719038817005546, "grad_norm": 3.7189877033233643, "learning_rate": 2.2069146434690003e-06, "loss": 0.016010522842407227, "step": 2397 }, { "epoch": 1.7726432532347505, "grad_norm": 4.09150505065918, "learning_rate": 2.2051724783365634e-06, "loss": 0.02673748880624771, "step": 2398 }, { "epoch": 1.7733826247689464, "grad_norm": 3.077813148498535, "learning_rate": 2.2034306601230025e-06, "loss": 0.0058304728008806705, "step": 2399 }, { "epoch": 1.7741219963031423, "grad_norm": 2.6268601417541504, "learning_rate": 2.2016891899374208e-06, "loss": 0.004101171158254147, "step": 2400 }, { "epoch": 1.7748613678373384, "grad_norm": 3.1871044635772705, "learning_rate": 2.1999480688886974e-06, "loss": 0.02683151513338089, "step": 2401 }, { "epoch": 1.775600739371534, "grad_norm": 1.2588895559310913, "learning_rate": 2.198207298085492e-06, "loss": 0.006219550967216492, "step": 2402 }, { "epoch": 1.7763401109057302, "grad_norm": 1.4101195335388184, "learning_rate": 2.1964668786362407e-06, "loss": 0.002860821085050702, "step": 2403 }, { "epoch": 1.777079482439926, "grad_norm": 5.059312343597412, "learning_rate": 2.194726811649155e-06, "loss": 0.023620054125785828, "step": 2404 }, { "epoch": 1.777818853974122, "grad_norm": 5.706701278686523, "learning_rate": 2.1929870982322234e-06, "loss": 0.0258839912712574, "step": 2405 }, { "epoch": 1.7785582255083179, "grad_norm": 4.169139862060547, "learning_rate": 2.1912477394932056e-06, "loss": 0.035459261387586594, "step": 2406 }, { "epoch": 1.7792975970425138, "grad_norm": 4.4004950523376465, "learning_rate": 2.1895087365396408e-06, "loss": 0.018698304891586304, "step": 2407 }, { "epoch": 1.78003696857671, "grad_norm": 5.06837797164917, "learning_rate": 2.1877700904788372e-06, "loss": 0.027783848345279694, "step": 2408 }, { "epoch": 1.7807763401109056, "grad_norm": 3.9696176052093506, "learning_rate": 2.186031802417879e-06, "loss": 0.018638407811522484, "step": 2409 }, { "epoch": 1.7815157116451017, "grad_norm": 1.6636396646499634, "learning_rate": 2.1842938734636206e-06, "loss": 0.002615541685372591, "step": 2410 }, { "epoch": 1.7822550831792976, "grad_norm": 3.197265386581421, "learning_rate": 2.182556304722687e-06, "loss": 0.021015968173742294, "step": 2411 }, { "epoch": 1.7829944547134935, "grad_norm": 3.716972589492798, "learning_rate": 2.180819097301476e-06, "loss": 0.01034094113856554, "step": 2412 }, { "epoch": 1.7837338262476896, "grad_norm": 3.6617815494537354, "learning_rate": 2.1790822523061537e-06, "loss": 0.020482582971453667, "step": 2413 }, { "epoch": 1.7844731977818853, "grad_norm": 2.8087587356567383, "learning_rate": 2.1773457708426564e-06, "loss": 0.018295619636774063, "step": 2414 }, { "epoch": 1.7852125693160814, "grad_norm": 3.8992536067962646, "learning_rate": 2.175609654016689e-06, "loss": 0.018436912447214127, "step": 2415 }, { "epoch": 1.7859519408502773, "grad_norm": 3.7694220542907715, "learning_rate": 2.1738739029337226e-06, "loss": 0.02034917287528515, "step": 2416 }, { "epoch": 1.7866913123844732, "grad_norm": 3.6766788959503174, "learning_rate": 2.172138518698998e-06, "loss": 0.017549803480505943, "step": 2417 }, { "epoch": 1.787430683918669, "grad_norm": 4.310575485229492, "learning_rate": 2.1704035024175184e-06, "loss": 0.021727636456489563, "step": 2418 }, { "epoch": 1.788170055452865, "grad_norm": 4.417879581451416, "learning_rate": 2.168668855194058e-06, "loss": 0.030745845288038254, "step": 2419 }, { "epoch": 1.7889094269870611, "grad_norm": 2.2890520095825195, "learning_rate": 2.1669345781331525e-06, "loss": 0.010833010077476501, "step": 2420 }, { "epoch": 1.7896487985212568, "grad_norm": 3.402646064758301, "learning_rate": 2.165200672339101e-06, "loss": 0.022983580827713013, "step": 2421 }, { "epoch": 1.790388170055453, "grad_norm": 3.3609817028045654, "learning_rate": 2.16346713891597e-06, "loss": 0.010589862242341042, "step": 2422 }, { "epoch": 1.7911275415896488, "grad_norm": 4.724049091339111, "learning_rate": 2.1617339789675863e-06, "loss": 0.02751976251602173, "step": 2423 }, { "epoch": 1.7918669131238447, "grad_norm": 5.206048965454102, "learning_rate": 2.160001193597539e-06, "loss": 0.00958852656185627, "step": 2424 }, { "epoch": 1.7926062846580406, "grad_norm": 1.880710244178772, "learning_rate": 2.158268783909178e-06, "loss": 0.01415637694299221, "step": 2425 }, { "epoch": 1.7933456561922365, "grad_norm": 4.771102428436279, "learning_rate": 2.1565367510056167e-06, "loss": 0.02817826345562935, "step": 2426 }, { "epoch": 1.7940850277264326, "grad_norm": 3.0026168823242188, "learning_rate": 2.1548050959897267e-06, "loss": 0.015185296535491943, "step": 2427 }, { "epoch": 1.7948243992606283, "grad_norm": 1.9592009782791138, "learning_rate": 2.1530738199641376e-06, "loss": 0.008963466621935368, "step": 2428 }, { "epoch": 1.7955637707948244, "grad_norm": 2.3036251068115234, "learning_rate": 2.151342924031242e-06, "loss": 0.025929508730769157, "step": 2429 }, { "epoch": 1.7963031423290203, "grad_norm": 3.2752437591552734, "learning_rate": 2.149612409293184e-06, "loss": 0.015727248042821884, "step": 2430 }, { "epoch": 1.7970425138632162, "grad_norm": 3.7569661140441895, "learning_rate": 2.147882276851872e-06, "loss": 0.028564980253577232, "step": 2431 }, { "epoch": 1.7977818853974123, "grad_norm": 2.1932222843170166, "learning_rate": 2.146152527808966e-06, "loss": 0.008280735462903976, "step": 2432 }, { "epoch": 1.798521256931608, "grad_norm": 4.589033126831055, "learning_rate": 2.1444231632658845e-06, "loss": 0.0251372791826725, "step": 2433 }, { "epoch": 1.7992606284658041, "grad_norm": 2.3982486724853516, "learning_rate": 2.1426941843238003e-06, "loss": 0.005969917867332697, "step": 2434 }, { "epoch": 1.8, "grad_norm": 1.9761250019073486, "learning_rate": 2.140965592083639e-06, "loss": 0.00776891503483057, "step": 2435 }, { "epoch": 1.800739371534196, "grad_norm": 3.564882278442383, "learning_rate": 2.1392373876460833e-06, "loss": 0.023195520043373108, "step": 2436 }, { "epoch": 1.8014787430683918, "grad_norm": 3.762488603591919, "learning_rate": 2.1375095721115655e-06, "loss": 0.023297926411032677, "step": 2437 }, { "epoch": 1.8022181146025877, "grad_norm": 2.0197479724884033, "learning_rate": 2.1357821465802737e-06, "loss": 0.008520135655999184, "step": 2438 }, { "epoch": 1.8029574861367839, "grad_norm": 4.5516767501831055, "learning_rate": 2.134055112152145e-06, "loss": 0.01757330633699894, "step": 2439 }, { "epoch": 1.8036968576709795, "grad_norm": 4.177767753601074, "learning_rate": 2.1323284699268693e-06, "loss": 0.01855301670730114, "step": 2440 }, { "epoch": 1.8044362292051757, "grad_norm": 3.7651093006134033, "learning_rate": 2.1306022210038844e-06, "loss": 0.012847572565078735, "step": 2441 }, { "epoch": 1.8051756007393716, "grad_norm": 4.661920547485352, "learning_rate": 2.1288763664823803e-06, "loss": 0.016130706295371056, "step": 2442 }, { "epoch": 1.8059149722735675, "grad_norm": 0.9086077213287354, "learning_rate": 2.1271509074612938e-06, "loss": 0.007401757873594761, "step": 2443 }, { "epoch": 1.8066543438077634, "grad_norm": 4.686894416809082, "learning_rate": 2.1254258450393134e-06, "loss": 0.021152470260858536, "step": 2444 }, { "epoch": 1.8073937153419593, "grad_norm": 2.91092586517334, "learning_rate": 2.123701180314868e-06, "loss": 0.016943659633398056, "step": 2445 }, { "epoch": 1.8081330868761554, "grad_norm": 2.7581264972686768, "learning_rate": 2.1219769143861416e-06, "loss": 0.014594347216188908, "step": 2446 }, { "epoch": 1.808872458410351, "grad_norm": 3.3905560970306396, "learning_rate": 2.1202530483510585e-06, "loss": 0.019680093973875046, "step": 2447 }, { "epoch": 1.8096118299445472, "grad_norm": 3.151979923248291, "learning_rate": 2.1185295833072917e-06, "loss": 0.01591341756284237, "step": 2448 }, { "epoch": 1.810351201478743, "grad_norm": 2.038966655731201, "learning_rate": 2.116806520352255e-06, "loss": 0.008122787810862064, "step": 2449 }, { "epoch": 1.811090573012939, "grad_norm": 3.5547854900360107, "learning_rate": 2.11508386058311e-06, "loss": 0.012817684561014175, "step": 2450 }, { "epoch": 1.811829944547135, "grad_norm": 4.88417387008667, "learning_rate": 2.1133616050967607e-06, "loss": 0.02470996230840683, "step": 2451 }, { "epoch": 1.8125693160813308, "grad_norm": 4.236621379852295, "learning_rate": 2.1116397549898514e-06, "loss": 0.010361631400883198, "step": 2452 }, { "epoch": 1.8133086876155269, "grad_norm": 4.106393814086914, "learning_rate": 2.109918311358773e-06, "loss": 0.0161250289529562, "step": 2453 }, { "epoch": 1.8140480591497228, "grad_norm": 2.6988303661346436, "learning_rate": 2.1081972752996514e-06, "loss": 0.011856719851493835, "step": 2454 }, { "epoch": 1.8147874306839187, "grad_norm": 3.6200151443481445, "learning_rate": 2.1064766479083575e-06, "loss": 0.01935610920190811, "step": 2455 }, { "epoch": 1.8155268022181146, "grad_norm": 2.731750011444092, "learning_rate": 2.104756430280501e-06, "loss": 0.013435539789497852, "step": 2456 }, { "epoch": 1.8162661737523105, "grad_norm": 1.7642066478729248, "learning_rate": 2.1030366235114296e-06, "loss": 0.003195903031155467, "step": 2457 }, { "epoch": 1.8170055452865066, "grad_norm": 2.7028400897979736, "learning_rate": 2.1013172286962314e-06, "loss": 0.006337325554341078, "step": 2458 }, { "epoch": 1.8177449168207023, "grad_norm": 5.267459392547607, "learning_rate": 2.0995982469297287e-06, "loss": 0.04332877695560455, "step": 2459 }, { "epoch": 1.8184842883548984, "grad_norm": 1.5918246507644653, "learning_rate": 2.0978796793064853e-06, "loss": 0.012254368513822556, "step": 2460 }, { "epoch": 1.8192236598890943, "grad_norm": 3.011883497238159, "learning_rate": 2.0961615269207967e-06, "loss": 0.006468385923653841, "step": 2461 }, { "epoch": 1.8199630314232902, "grad_norm": 3.677309274673462, "learning_rate": 2.094443790866698e-06, "loss": 0.027356212958693504, "step": 2462 }, { "epoch": 1.820702402957486, "grad_norm": 3.639669895172119, "learning_rate": 2.0927264722379576e-06, "loss": 0.02099538967013359, "step": 2463 }, { "epoch": 1.821441774491682, "grad_norm": 5.0151286125183105, "learning_rate": 2.0910095721280764e-06, "loss": 0.04083260893821716, "step": 2464 }, { "epoch": 1.822181146025878, "grad_norm": 2.4294204711914062, "learning_rate": 2.089293091630292e-06, "loss": 0.006082998588681221, "step": 2465 }, { "epoch": 1.8229205175600738, "grad_norm": 3.1307010650634766, "learning_rate": 2.087577031837572e-06, "loss": 0.025459136813879013, "step": 2466 }, { "epoch": 1.82365988909427, "grad_norm": 2.08546781539917, "learning_rate": 2.085861393842618e-06, "loss": 0.007463822606950998, "step": 2467 }, { "epoch": 1.8243992606284658, "grad_norm": 3.943913459777832, "learning_rate": 2.084146178737863e-06, "loss": 0.022983737289905548, "step": 2468 }, { "epoch": 1.8251386321626617, "grad_norm": 2.0746405124664307, "learning_rate": 2.0824313876154683e-06, "loss": 0.012308754026889801, "step": 2469 }, { "epoch": 1.8258780036968578, "grad_norm": 3.640599250793457, "learning_rate": 2.0807170215673283e-06, "loss": 0.01906050369143486, "step": 2470 }, { "epoch": 1.8266173752310535, "grad_norm": 2.2639191150665283, "learning_rate": 2.079003081685064e-06, "loss": 0.007766381371766329, "step": 2471 }, { "epoch": 1.8273567467652496, "grad_norm": 2.157378911972046, "learning_rate": 2.07728956906003e-06, "loss": 0.012388783507049084, "step": 2472 }, { "epoch": 1.8280961182994455, "grad_norm": 4.338565826416016, "learning_rate": 2.075576484783301e-06, "loss": 0.014728139154613018, "step": 2473 }, { "epoch": 1.8288354898336414, "grad_norm": 3.462287425994873, "learning_rate": 2.0738638299456853e-06, "loss": 0.011560076847672462, "step": 2474 }, { "epoch": 1.8295748613678373, "grad_norm": 3.6259727478027344, "learning_rate": 2.0721516056377162e-06, "loss": 0.012597319670021534, "step": 2475 }, { "epoch": 1.8303142329020332, "grad_norm": 4.314856052398682, "learning_rate": 2.0704398129496507e-06, "loss": 0.021551139652729034, "step": 2476 }, { "epoch": 1.8310536044362293, "grad_norm": 3.8893754482269287, "learning_rate": 2.0687284529714753e-06, "loss": 0.024551209062337875, "step": 2477 }, { "epoch": 1.831792975970425, "grad_norm": 4.30712366104126, "learning_rate": 2.0670175267928954e-06, "loss": 0.011040225625038147, "step": 2478 }, { "epoch": 1.8325323475046211, "grad_norm": 4.189786434173584, "learning_rate": 2.065307035503344e-06, "loss": 0.02088003046810627, "step": 2479 }, { "epoch": 1.833271719038817, "grad_norm": 1.0229260921478271, "learning_rate": 2.0635969801919776e-06, "loss": 0.005271219182759523, "step": 2480 }, { "epoch": 1.834011090573013, "grad_norm": 4.460366249084473, "learning_rate": 2.0618873619476722e-06, "loss": 0.016601696610450745, "step": 2481 }, { "epoch": 1.8347504621072088, "grad_norm": 3.5722532272338867, "learning_rate": 2.0601781818590283e-06, "loss": 0.016410227864980698, "step": 2482 }, { "epoch": 1.8354898336414047, "grad_norm": 3.586667537689209, "learning_rate": 2.0584694410143647e-06, "loss": 0.009934614412486553, "step": 2483 }, { "epoch": 1.8362292051756008, "grad_norm": 3.696007013320923, "learning_rate": 2.0567611405017235e-06, "loss": 0.010011517442762852, "step": 2484 }, { "epoch": 1.8369685767097965, "grad_norm": 3.4734537601470947, "learning_rate": 2.0550532814088635e-06, "loss": 0.009066412225365639, "step": 2485 }, { "epoch": 1.8377079482439926, "grad_norm": 3.1104042530059814, "learning_rate": 2.0533458648232646e-06, "loss": 0.029655221849679947, "step": 2486 }, { "epoch": 1.8384473197781885, "grad_norm": 5.1450042724609375, "learning_rate": 2.0516388918321257e-06, "loss": 0.0356924906373024, "step": 2487 }, { "epoch": 1.8391866913123844, "grad_norm": 3.7162723541259766, "learning_rate": 2.0499323635223583e-06, "loss": 0.015303842723369598, "step": 2488 }, { "epoch": 1.8399260628465806, "grad_norm": 3.033881425857544, "learning_rate": 2.048226280980597e-06, "loss": 0.01909620314836502, "step": 2489 }, { "epoch": 1.8406654343807762, "grad_norm": 2.5080204010009766, "learning_rate": 2.0465206452931874e-06, "loss": 0.019254133105278015, "step": 2490 }, { "epoch": 1.8414048059149724, "grad_norm": 3.6668081283569336, "learning_rate": 2.0448154575461947e-06, "loss": 0.01791715994477272, "step": 2491 }, { "epoch": 1.8421441774491683, "grad_norm": 3.8087096214294434, "learning_rate": 2.0431107188253973e-06, "loss": 0.018676679581403732, "step": 2492 }, { "epoch": 1.8428835489833642, "grad_norm": 4.05318546295166, "learning_rate": 2.0414064302162855e-06, "loss": 0.018575243651866913, "step": 2493 }, { "epoch": 1.84362292051756, "grad_norm": 3.0981976985931396, "learning_rate": 2.039702592804067e-06, "loss": 0.021734917536377907, "step": 2494 }, { "epoch": 1.844362292051756, "grad_norm": 2.508127212524414, "learning_rate": 2.0379992076736584e-06, "loss": 0.01809045299887657, "step": 2495 }, { "epoch": 1.845101663585952, "grad_norm": 3.6174707412719727, "learning_rate": 2.036296275909691e-06, "loss": 0.017042947933077812, "step": 2496 }, { "epoch": 1.8458410351201477, "grad_norm": 2.838428020477295, "learning_rate": 2.0345937985965066e-06, "loss": 0.02286199852824211, "step": 2497 }, { "epoch": 1.8465804066543439, "grad_norm": 3.6460001468658447, "learning_rate": 2.032891776818156e-06, "loss": 0.013638337142765522, "step": 2498 }, { "epoch": 1.8473197781885398, "grad_norm": 2.7401509284973145, "learning_rate": 2.0311902116584037e-06, "loss": 0.013244428671896458, "step": 2499 }, { "epoch": 1.8480591497227357, "grad_norm": 2.5100958347320557, "learning_rate": 2.0294891042007194e-06, "loss": 0.009471779689192772, "step": 2500 }, { "epoch": 1.8487985212569316, "grad_norm": 2.5522286891937256, "learning_rate": 2.027788455528284e-06, "loss": 0.01214597374200821, "step": 2501 }, { "epoch": 1.8495378927911275, "grad_norm": 3.489625930786133, "learning_rate": 2.0260882667239844e-06, "loss": 0.029335670173168182, "step": 2502 }, { "epoch": 1.8502772643253236, "grad_norm": 2.523085117340088, "learning_rate": 2.024388538870417e-06, "loss": 0.014586862176656723, "step": 2503 }, { "epoch": 1.8510166358595193, "grad_norm": 1.6969993114471436, "learning_rate": 2.0226892730498827e-06, "loss": 0.005520989187061787, "step": 2504 }, { "epoch": 1.8517560073937154, "grad_norm": 3.25911808013916, "learning_rate": 2.0209904703443885e-06, "loss": 0.020212309435009956, "step": 2505 }, { "epoch": 1.8524953789279113, "grad_norm": 2.927299737930298, "learning_rate": 2.019292131835648e-06, "loss": 0.0038320699241012335, "step": 2506 }, { "epoch": 1.8532347504621072, "grad_norm": 4.089873790740967, "learning_rate": 2.017594258605077e-06, "loss": 0.0307316817343235, "step": 2507 }, { "epoch": 1.8539741219963033, "grad_norm": 1.7807055711746216, "learning_rate": 2.015896851733796e-06, "loss": 0.004515481181442738, "step": 2508 }, { "epoch": 1.854713493530499, "grad_norm": 5.494533061981201, "learning_rate": 2.014199912302629e-06, "loss": 0.03761644288897514, "step": 2509 }, { "epoch": 1.855452865064695, "grad_norm": 3.418219566345215, "learning_rate": 2.0125034413921024e-06, "loss": 0.009491167031228542, "step": 2510 }, { "epoch": 1.856192236598891, "grad_norm": 4.715724945068359, "learning_rate": 2.0108074400824444e-06, "loss": 0.025438595563173294, "step": 2511 }, { "epoch": 1.8569316081330869, "grad_norm": 3.7262821197509766, "learning_rate": 2.009111909453582e-06, "loss": 0.007098742760717869, "step": 2512 }, { "epoch": 1.8576709796672828, "grad_norm": 3.4748125076293945, "learning_rate": 2.007416850585146e-06, "loss": 0.009745716117322445, "step": 2513 }, { "epoch": 1.8584103512014787, "grad_norm": 3.705963611602783, "learning_rate": 2.0057222645564634e-06, "loss": 0.04493022710084915, "step": 2514 }, { "epoch": 1.8591497227356748, "grad_norm": 3.9773659706115723, "learning_rate": 2.004028152446562e-06, "loss": 0.015622500330209732, "step": 2515 }, { "epoch": 1.8598890942698705, "grad_norm": 5.611099720001221, "learning_rate": 2.0023345153341696e-06, "loss": 0.019458729773759842, "step": 2516 }, { "epoch": 1.8606284658040666, "grad_norm": 3.8900043964385986, "learning_rate": 2.0006413542977065e-06, "loss": 0.028137579560279846, "step": 2517 }, { "epoch": 1.8613678373382625, "grad_norm": 5.129050254821777, "learning_rate": 1.998948670415295e-06, "loss": 0.032516371458768845, "step": 2518 }, { "epoch": 1.8621072088724584, "grad_norm": 3.3075625896453857, "learning_rate": 1.9972564647647496e-06, "loss": 0.014562076888978481, "step": 2519 }, { "epoch": 1.8628465804066543, "grad_norm": 2.6361660957336426, "learning_rate": 1.995564738423583e-06, "loss": 0.009443633258342743, "step": 2520 }, { "epoch": 1.8635859519408502, "grad_norm": 3.488633871078491, "learning_rate": 1.9938734924690035e-06, "loss": 0.020351143553853035, "step": 2521 }, { "epoch": 1.8643253234750463, "grad_norm": 2.6377978324890137, "learning_rate": 1.9921827279779087e-06, "loss": 0.010878841392695904, "step": 2522 }, { "epoch": 1.865064695009242, "grad_norm": 3.2880125045776367, "learning_rate": 1.9904924460268947e-06, "loss": 0.01522357389330864, "step": 2523 }, { "epoch": 1.8658040665434381, "grad_norm": 3.2098374366760254, "learning_rate": 1.9888026476922477e-06, "loss": 0.015666039660573006, "step": 2524 }, { "epoch": 1.866543438077634, "grad_norm": 3.1475136280059814, "learning_rate": 1.9871133340499477e-06, "loss": 0.010549923405051231, "step": 2525 }, { "epoch": 1.86728280961183, "grad_norm": 3.7437243461608887, "learning_rate": 1.9854245061756635e-06, "loss": 0.008770049549639225, "step": 2526 }, { "epoch": 1.868022181146026, "grad_norm": 3.764759063720703, "learning_rate": 1.983736165144756e-06, "loss": 0.01911577582359314, "step": 2527 }, { "epoch": 1.8687615526802217, "grad_norm": 2.913541793823242, "learning_rate": 1.982048312032278e-06, "loss": 0.010813649743795395, "step": 2528 }, { "epoch": 1.8695009242144178, "grad_norm": 3.192661762237549, "learning_rate": 1.980360947912968e-06, "loss": 0.025112736970186234, "step": 2529 }, { "epoch": 1.8702402957486137, "grad_norm": 3.0348057746887207, "learning_rate": 1.978674073861257e-06, "loss": 0.011066978797316551, "step": 2530 }, { "epoch": 1.8709796672828096, "grad_norm": 3.515044689178467, "learning_rate": 1.97698769095126e-06, "loss": 0.018442511558532715, "step": 2531 }, { "epoch": 1.8717190388170055, "grad_norm": 3.509636402130127, "learning_rate": 1.9753018002567807e-06, "loss": 0.02753857895731926, "step": 2532 }, { "epoch": 1.8724584103512014, "grad_norm": 2.974996328353882, "learning_rate": 1.973616402851312e-06, "loss": 0.017207089811563492, "step": 2533 }, { "epoch": 1.8731977818853975, "grad_norm": 3.2207627296447754, "learning_rate": 1.971931499808029e-06, "loss": 0.025128718465566635, "step": 2534 }, { "epoch": 1.8739371534195932, "grad_norm": 3.9219188690185547, "learning_rate": 1.970247092199795e-06, "loss": 0.025501323863863945, "step": 2535 }, { "epoch": 1.8746765249537893, "grad_norm": 1.757284164428711, "learning_rate": 1.968563181099155e-06, "loss": 0.0053723216988146305, "step": 2536 }, { "epoch": 1.8754158964879852, "grad_norm": 3.7875301837921143, "learning_rate": 1.96687976757834e-06, "loss": 0.024803148582577705, "step": 2537 }, { "epoch": 1.8761552680221811, "grad_norm": 2.813736915588379, "learning_rate": 1.965196852709262e-06, "loss": 0.010527173057198524, "step": 2538 }, { "epoch": 1.876894639556377, "grad_norm": 5.087325572967529, "learning_rate": 1.9635144375635184e-06, "loss": 0.013853006064891815, "step": 2539 }, { "epoch": 1.877634011090573, "grad_norm": 3.4073686599731445, "learning_rate": 1.961832523212387e-06, "loss": 0.014763442799448967, "step": 2540 }, { "epoch": 1.878373382624769, "grad_norm": 2.977489471435547, "learning_rate": 1.9601511107268256e-06, "loss": 0.01405268907546997, "step": 2541 }, { "epoch": 1.8791127541589647, "grad_norm": 2.912625551223755, "learning_rate": 1.9584702011774748e-06, "loss": 0.012350047938525677, "step": 2542 }, { "epoch": 1.8798521256931608, "grad_norm": 2.7538230419158936, "learning_rate": 1.9567897956346517e-06, "loss": 0.022199474275112152, "step": 2543 }, { "epoch": 1.8805914972273567, "grad_norm": 2.570904493331909, "learning_rate": 1.9551098951683556e-06, "loss": 0.013901055790483952, "step": 2544 }, { "epoch": 1.8813308687615526, "grad_norm": 2.2610507011413574, "learning_rate": 1.9534305008482636e-06, "loss": 0.012707959860563278, "step": 2545 }, { "epoch": 1.8820702402957488, "grad_norm": 2.7905526161193848, "learning_rate": 1.951751613743728e-06, "loss": 0.018538974225521088, "step": 2546 }, { "epoch": 1.8828096118299444, "grad_norm": 3.186262369155884, "learning_rate": 1.950073234923782e-06, "loss": 0.013871033675968647, "step": 2547 }, { "epoch": 1.8835489833641406, "grad_norm": 4.2799224853515625, "learning_rate": 1.948395365457132e-06, "loss": 0.01344123762100935, "step": 2548 }, { "epoch": 1.8842883548983365, "grad_norm": 3.4807639122009277, "learning_rate": 1.946718006412161e-06, "loss": 0.013025575317442417, "step": 2549 }, { "epoch": 1.8850277264325324, "grad_norm": 3.0026752948760986, "learning_rate": 1.945041158856929e-06, "loss": 0.013906736858189106, "step": 2550 }, { "epoch": 1.8857670979667283, "grad_norm": 2.737553119659424, "learning_rate": 1.9433648238591657e-06, "loss": 0.01715557835996151, "step": 2551 }, { "epoch": 1.8865064695009242, "grad_norm": 3.115222454071045, "learning_rate": 1.94168900248628e-06, "loss": 0.011402231641113758, "step": 2552 }, { "epoch": 1.8872458410351203, "grad_norm": 4.862003326416016, "learning_rate": 1.940013695805349e-06, "loss": 0.028259025886654854, "step": 2553 }, { "epoch": 1.887985212569316, "grad_norm": 4.419849395751953, "learning_rate": 1.9383389048831257e-06, "loss": 0.014222763478755951, "step": 2554 }, { "epoch": 1.888724584103512, "grad_norm": 5.574449062347412, "learning_rate": 1.9366646307860316e-06, "loss": 0.020874667912721634, "step": 2555 }, { "epoch": 1.889463955637708, "grad_norm": 2.9645118713378906, "learning_rate": 1.9349908745801615e-06, "loss": 0.017527000978589058, "step": 2556 }, { "epoch": 1.8902033271719039, "grad_norm": 3.160423755645752, "learning_rate": 1.9333176373312794e-06, "loss": 0.02151082083582878, "step": 2557 }, { "epoch": 1.8909426987060998, "grad_norm": 2.594607353210449, "learning_rate": 1.931644920104819e-06, "loss": 0.011080844327807426, "step": 2558 }, { "epoch": 1.8916820702402957, "grad_norm": 2.818969249725342, "learning_rate": 1.9299727239658827e-06, "loss": 0.02957616001367569, "step": 2559 }, { "epoch": 1.8924214417744918, "grad_norm": 2.285057544708252, "learning_rate": 1.9283010499792405e-06, "loss": 0.017289716750383377, "step": 2560 }, { "epoch": 1.8931608133086875, "grad_norm": 3.59547758102417, "learning_rate": 1.926629899209331e-06, "loss": 0.01630765199661255, "step": 2561 }, { "epoch": 1.8939001848428836, "grad_norm": 4.411393642425537, "learning_rate": 1.9249592727202608e-06, "loss": 0.019423849880695343, "step": 2562 }, { "epoch": 1.8946395563770795, "grad_norm": 3.359018564224243, "learning_rate": 1.9232891715757995e-06, "loss": 0.020574504509568214, "step": 2563 }, { "epoch": 1.8953789279112754, "grad_norm": 2.451221466064453, "learning_rate": 1.921619596839386e-06, "loss": 0.013680260628461838, "step": 2564 }, { "epoch": 1.8961182994454715, "grad_norm": 1.5320148468017578, "learning_rate": 1.9199505495741184e-06, "loss": 0.01122130174189806, "step": 2565 }, { "epoch": 1.8968576709796672, "grad_norm": 2.1561384201049805, "learning_rate": 1.9182820308427653e-06, "loss": 0.006632218137383461, "step": 2566 }, { "epoch": 1.8975970425138633, "grad_norm": 3.2377803325653076, "learning_rate": 1.916614041707754e-06, "loss": 0.01781860552728176, "step": 2567 }, { "epoch": 1.8983364140480592, "grad_norm": 3.128405809402466, "learning_rate": 1.914946583231177e-06, "loss": 0.02400435507297516, "step": 2568 }, { "epoch": 1.899075785582255, "grad_norm": 3.438673973083496, "learning_rate": 1.91327965647479e-06, "loss": 0.02096051722764969, "step": 2569 }, { "epoch": 1.899815157116451, "grad_norm": 2.6885745525360107, "learning_rate": 1.911613262500006e-06, "loss": 0.014881380833685398, "step": 2570 }, { "epoch": 1.900554528650647, "grad_norm": 1.5891108512878418, "learning_rate": 1.9099474023679028e-06, "loss": 0.004002361558377743, "step": 2571 }, { "epoch": 1.901293900184843, "grad_norm": 3.1028757095336914, "learning_rate": 1.908282077139216e-06, "loss": 0.011091664433479309, "step": 2572 }, { "epoch": 1.9020332717190387, "grad_norm": 3.5601718425750732, "learning_rate": 1.906617287874341e-06, "loss": 0.021292012184858322, "step": 2573 }, { "epoch": 1.9027726432532348, "grad_norm": 3.3160924911499023, "learning_rate": 1.9049530356333335e-06, "loss": 0.015741800889372826, "step": 2574 }, { "epoch": 1.9035120147874307, "grad_norm": 4.4338860511779785, "learning_rate": 1.9032893214759046e-06, "loss": 0.012256084010004997, "step": 2575 }, { "epoch": 1.9042513863216266, "grad_norm": 3.109445333480835, "learning_rate": 1.901626146461425e-06, "loss": 0.04004257917404175, "step": 2576 }, { "epoch": 1.9049907578558225, "grad_norm": 5.1455230712890625, "learning_rate": 1.89996351164892e-06, "loss": 0.027885276824235916, "step": 2577 }, { "epoch": 1.9057301293900184, "grad_norm": 2.409987211227417, "learning_rate": 1.898301418097075e-06, "loss": 0.015603492967784405, "step": 2578 }, { "epoch": 1.9064695009242145, "grad_norm": 3.4041199684143066, "learning_rate": 1.8966398668642245e-06, "loss": 0.007124275900423527, "step": 2579 }, { "epoch": 1.9072088724584102, "grad_norm": 1.644808053970337, "learning_rate": 1.894978859008363e-06, "loss": 0.01091003604233265, "step": 2580 }, { "epoch": 1.9079482439926063, "grad_norm": 3.31950044631958, "learning_rate": 1.8933183955871364e-06, "loss": 0.022927802056074142, "step": 2581 }, { "epoch": 1.9086876155268022, "grad_norm": 1.8364616632461548, "learning_rate": 1.8916584776578444e-06, "loss": 0.009941250085830688, "step": 2582 }, { "epoch": 1.9094269870609981, "grad_norm": 3.7455315589904785, "learning_rate": 1.889999106277442e-06, "loss": 0.020394597202539444, "step": 2583 }, { "epoch": 1.9101663585951942, "grad_norm": 2.5581207275390625, "learning_rate": 1.8883402825025305e-06, "loss": 0.01320491824299097, "step": 2584 }, { "epoch": 1.91090573012939, "grad_norm": 3.4753708839416504, "learning_rate": 1.8866820073893676e-06, "loss": 0.005066542886197567, "step": 2585 }, { "epoch": 1.911645101663586, "grad_norm": 1.678920865058899, "learning_rate": 1.8850242819938596e-06, "loss": 0.013212006539106369, "step": 2586 }, { "epoch": 1.912384473197782, "grad_norm": 2.520986795425415, "learning_rate": 1.8833671073715625e-06, "loss": 0.00488297687843442, "step": 2587 }, { "epoch": 1.9131238447319778, "grad_norm": 2.311912775039673, "learning_rate": 1.8817104845776829e-06, "loss": 0.007029008585959673, "step": 2588 }, { "epoch": 1.9138632162661737, "grad_norm": 3.8761696815490723, "learning_rate": 1.8800544146670747e-06, "loss": 0.02363251894712448, "step": 2589 }, { "epoch": 1.9146025878003696, "grad_norm": 5.21687126159668, "learning_rate": 1.8783988986942404e-06, "loss": 0.037312015891075134, "step": 2590 }, { "epoch": 1.9153419593345657, "grad_norm": 3.8292601108551025, "learning_rate": 1.8767439377133284e-06, "loss": 0.01551235280930996, "step": 2591 }, { "epoch": 1.9160813308687614, "grad_norm": 4.650663375854492, "learning_rate": 1.8750895327781365e-06, "loss": 0.029318328946828842, "step": 2592 }, { "epoch": 1.9168207024029575, "grad_norm": 3.5815179347991943, "learning_rate": 1.8734356849421075e-06, "loss": 0.02336857095360756, "step": 2593 }, { "epoch": 1.9175600739371534, "grad_norm": 4.407413959503174, "learning_rate": 1.8717823952583275e-06, "loss": 0.0174778550863266, "step": 2594 }, { "epoch": 1.9182994454713493, "grad_norm": 3.5410008430480957, "learning_rate": 1.8701296647795292e-06, "loss": 0.018241610378026962, "step": 2595 }, { "epoch": 1.9190388170055452, "grad_norm": 2.462214708328247, "learning_rate": 1.8684774945580874e-06, "loss": 0.01292932964861393, "step": 2596 }, { "epoch": 1.9197781885397411, "grad_norm": 2.8325414657592773, "learning_rate": 1.8668258856460225e-06, "loss": 0.015230849385261536, "step": 2597 }, { "epoch": 1.9205175600739373, "grad_norm": 0.9402478933334351, "learning_rate": 1.865174839094997e-06, "loss": 0.003564033657312393, "step": 2598 }, { "epoch": 1.921256931608133, "grad_norm": 3.200505018234253, "learning_rate": 1.8635243559563127e-06, "loss": 0.023666441440582275, "step": 2599 }, { "epoch": 1.921996303142329, "grad_norm": 3.145677089691162, "learning_rate": 1.8618744372809166e-06, "loss": 0.022824279963970184, "step": 2600 }, { "epoch": 1.922735674676525, "grad_norm": 2.231393575668335, "learning_rate": 1.8602250841193926e-06, "loss": 0.005737572908401489, "step": 2601 }, { "epoch": 1.9234750462107209, "grad_norm": 2.388536214828491, "learning_rate": 1.8585762975219673e-06, "loss": 0.009186677634716034, "step": 2602 }, { "epoch": 1.924214417744917, "grad_norm": 3.3825747966766357, "learning_rate": 1.8569280785385048e-06, "loss": 0.01676245965063572, "step": 2603 }, { "epoch": 1.9249537892791126, "grad_norm": 1.512338399887085, "learning_rate": 1.8552804282185078e-06, "loss": 0.003917570225894451, "step": 2604 }, { "epoch": 1.9256931608133088, "grad_norm": 2.2300429344177246, "learning_rate": 1.853633347611119e-06, "loss": 0.020230170339345932, "step": 2605 }, { "epoch": 1.9264325323475047, "grad_norm": 2.4728353023529053, "learning_rate": 1.8519868377651162e-06, "loss": 0.01820462755858898, "step": 2606 }, { "epoch": 1.9271719038817006, "grad_norm": 3.7090213298797607, "learning_rate": 1.8503408997289145e-06, "loss": 0.017540760338306427, "step": 2607 }, { "epoch": 1.9279112754158965, "grad_norm": 4.228249549865723, "learning_rate": 1.848695534550564e-06, "loss": 0.036250222474336624, "step": 2608 }, { "epoch": 1.9286506469500924, "grad_norm": 2.7595407962799072, "learning_rate": 1.8470507432777512e-06, "loss": 0.019223205745220184, "step": 2609 }, { "epoch": 1.9293900184842885, "grad_norm": 2.622697591781616, "learning_rate": 1.845406526957798e-06, "loss": 0.006925911642611027, "step": 2610 }, { "epoch": 1.9301293900184842, "grad_norm": 3.6940255165100098, "learning_rate": 1.8437628866376573e-06, "loss": 0.014792063273489475, "step": 2611 }, { "epoch": 1.9308687615526803, "grad_norm": 3.318878412246704, "learning_rate": 1.8421198233639187e-06, "loss": 0.024628572165966034, "step": 2612 }, { "epoch": 1.9316081330868762, "grad_norm": 2.730419158935547, "learning_rate": 1.8404773381828007e-06, "loss": 0.018542081117630005, "step": 2613 }, { "epoch": 1.932347504621072, "grad_norm": 2.448462963104248, "learning_rate": 1.8388354321401559e-06, "loss": 0.007579020690172911, "step": 2614 }, { "epoch": 1.933086876155268, "grad_norm": 3.1847612857818604, "learning_rate": 1.837194106281469e-06, "loss": 0.009917194955050945, "step": 2615 }, { "epoch": 1.9338262476894639, "grad_norm": 1.666242003440857, "learning_rate": 1.8355533616518534e-06, "loss": 0.008449562825262547, "step": 2616 }, { "epoch": 1.93456561922366, "grad_norm": 0.8524744510650635, "learning_rate": 1.833913199296054e-06, "loss": 0.0023125496227294207, "step": 2617 }, { "epoch": 1.9353049907578557, "grad_norm": 1.9056636095046997, "learning_rate": 1.8322736202584423e-06, "loss": 0.010103561915457249, "step": 2618 }, { "epoch": 1.9360443622920518, "grad_norm": 3.9240238666534424, "learning_rate": 1.830634625583021e-06, "loss": 0.02679106406867504, "step": 2619 }, { "epoch": 1.9367837338262477, "grad_norm": 2.6147570610046387, "learning_rate": 1.8289962163134196e-06, "loss": 0.019392989575862885, "step": 2620 }, { "epoch": 1.9375231053604436, "grad_norm": 2.3031179904937744, "learning_rate": 1.827358393492895e-06, "loss": 0.006820867769420147, "step": 2621 }, { "epoch": 1.9382624768946397, "grad_norm": 1.7397563457489014, "learning_rate": 1.8257211581643325e-06, "loss": 0.020943081006407738, "step": 2622 }, { "epoch": 1.9390018484288354, "grad_norm": 3.1057868003845215, "learning_rate": 1.8240845113702382e-06, "loss": 0.008784599602222443, "step": 2623 }, { "epoch": 1.9397412199630315, "grad_norm": 4.64809513092041, "learning_rate": 1.8224484541527504e-06, "loss": 0.02487381361424923, "step": 2624 }, { "epoch": 1.9404805914972274, "grad_norm": 2.85463809967041, "learning_rate": 1.8208129875536256e-06, "loss": 0.010487016290426254, "step": 2625 }, { "epoch": 1.9412199630314233, "grad_norm": 2.50286602973938, "learning_rate": 1.8191781126142476e-06, "loss": 0.010723087005317211, "step": 2626 }, { "epoch": 1.9419593345656192, "grad_norm": 1.4065439701080322, "learning_rate": 1.817543830375625e-06, "loss": 0.013847202993929386, "step": 2627 }, { "epoch": 1.942698706099815, "grad_norm": 3.130000352859497, "learning_rate": 1.815910141878383e-06, "loss": 0.02109493315219879, "step": 2628 }, { "epoch": 1.9434380776340112, "grad_norm": 4.079592227935791, "learning_rate": 1.8142770481627763e-06, "loss": 0.025709927082061768, "step": 2629 }, { "epoch": 1.944177449168207, "grad_norm": 1.6798573732376099, "learning_rate": 1.8126445502686747e-06, "loss": 0.005074533633887768, "step": 2630 }, { "epoch": 1.944916820702403, "grad_norm": 3.3952503204345703, "learning_rate": 1.811012649235574e-06, "loss": 0.020179815590381622, "step": 2631 }, { "epoch": 1.945656192236599, "grad_norm": 2.7161993980407715, "learning_rate": 1.8093813461025828e-06, "loss": 0.006107627414166927, "step": 2632 }, { "epoch": 1.9463955637707948, "grad_norm": 2.8276596069335938, "learning_rate": 1.8077506419084368e-06, "loss": 0.017256850376725197, "step": 2633 }, { "epoch": 1.9471349353049907, "grad_norm": 3.510815143585205, "learning_rate": 1.8061205376914853e-06, "loss": 0.016296714544296265, "step": 2634 }, { "epoch": 1.9478743068391866, "grad_norm": 5.087317943572998, "learning_rate": 1.804491034489696e-06, "loss": 0.027957256883382797, "step": 2635 }, { "epoch": 1.9486136783733827, "grad_norm": 2.5234603881835938, "learning_rate": 1.8028621333406588e-06, "loss": 0.012986913323402405, "step": 2636 }, { "epoch": 1.9493530499075784, "grad_norm": 2.9210612773895264, "learning_rate": 1.8012338352815711e-06, "loss": 0.013944745995104313, "step": 2637 }, { "epoch": 1.9500924214417745, "grad_norm": 3.9143760204315186, "learning_rate": 1.799606141349255e-06, "loss": 0.018283186480402946, "step": 2638 }, { "epoch": 1.9508317929759704, "grad_norm": 3.1477761268615723, "learning_rate": 1.7979790525801439e-06, "loss": 0.012649117037653923, "step": 2639 }, { "epoch": 1.9515711645101663, "grad_norm": 3.9837450981140137, "learning_rate": 1.7963525700102846e-06, "loss": 0.027188453823328018, "step": 2640 }, { "epoch": 1.9523105360443624, "grad_norm": 2.6227259635925293, "learning_rate": 1.7947266946753434e-06, "loss": 0.011824147775769234, "step": 2641 }, { "epoch": 1.9530499075785581, "grad_norm": 2.241793155670166, "learning_rate": 1.7931014276105924e-06, "loss": 0.008016071282327175, "step": 2642 }, { "epoch": 1.9537892791127542, "grad_norm": 2.446063756942749, "learning_rate": 1.7914767698509238e-06, "loss": 0.011099990457296371, "step": 2643 }, { "epoch": 1.9545286506469501, "grad_norm": 2.684382438659668, "learning_rate": 1.7898527224308349e-06, "loss": 0.019460022449493408, "step": 2644 }, { "epoch": 1.955268022181146, "grad_norm": 1.766122817993164, "learning_rate": 1.7882292863844386e-06, "loss": 0.010015372186899185, "step": 2645 }, { "epoch": 1.956007393715342, "grad_norm": 3.180995464324951, "learning_rate": 1.7866064627454611e-06, "loss": 0.011759483255445957, "step": 2646 }, { "epoch": 1.9567467652495378, "grad_norm": 2.0905404090881348, "learning_rate": 1.78498425254723e-06, "loss": 0.004458367358893156, "step": 2647 }, { "epoch": 1.957486136783734, "grad_norm": 4.313170433044434, "learning_rate": 1.783362656822692e-06, "loss": 0.029495110735297203, "step": 2648 }, { "epoch": 1.9582255083179296, "grad_norm": 3.8106210231781006, "learning_rate": 1.7817416766043938e-06, "loss": 0.019497502595186234, "step": 2649 }, { "epoch": 1.9589648798521258, "grad_norm": 2.677617311477661, "learning_rate": 1.780121312924497e-06, "loss": 0.014122757129371166, "step": 2650 }, { "epoch": 1.9597042513863216, "grad_norm": 4.005002021789551, "learning_rate": 1.7785015668147692e-06, "loss": 0.014262248761951923, "step": 2651 }, { "epoch": 1.9604436229205175, "grad_norm": 3.240267038345337, "learning_rate": 1.7768824393065804e-06, "loss": 0.017366668209433556, "step": 2652 }, { "epoch": 1.9611829944547134, "grad_norm": 2.880427837371826, "learning_rate": 1.7752639314309122e-06, "loss": 0.005995279178023338, "step": 2653 }, { "epoch": 1.9619223659889093, "grad_norm": 2.901531457901001, "learning_rate": 1.773646044218349e-06, "loss": 0.005575364455580711, "step": 2654 }, { "epoch": 1.9626617375231055, "grad_norm": 3.9463367462158203, "learning_rate": 1.7720287786990797e-06, "loss": 0.0189892016351223, "step": 2655 }, { "epoch": 1.9634011090573011, "grad_norm": 3.4969000816345215, "learning_rate": 1.7704121359028987e-06, "loss": 0.011559023521840572, "step": 2656 }, { "epoch": 1.9641404805914973, "grad_norm": 1.6983001232147217, "learning_rate": 1.768796116859201e-06, "loss": 0.002961203921586275, "step": 2657 }, { "epoch": 1.9648798521256932, "grad_norm": 2.136058807373047, "learning_rate": 1.76718072259699e-06, "loss": 0.005912424065172672, "step": 2658 }, { "epoch": 1.965619223659889, "grad_norm": 2.163344621658325, "learning_rate": 1.7655659541448662e-06, "loss": 0.01646912842988968, "step": 2659 }, { "epoch": 1.9663585951940852, "grad_norm": 1.4081248044967651, "learning_rate": 1.7639518125310328e-06, "loss": 0.011507082730531693, "step": 2660 }, { "epoch": 1.9670979667282809, "grad_norm": 5.016354560852051, "learning_rate": 1.7623382987832948e-06, "loss": 0.018794864416122437, "step": 2661 }, { "epoch": 1.967837338262477, "grad_norm": 2.4368085861206055, "learning_rate": 1.7607254139290563e-06, "loss": 0.0132742989808321, "step": 2662 }, { "epoch": 1.9685767097966729, "grad_norm": 5.208108425140381, "learning_rate": 1.7591131589953231e-06, "loss": 0.008535366505384445, "step": 2663 }, { "epoch": 1.9693160813308688, "grad_norm": 1.9396418333053589, "learning_rate": 1.7575015350086972e-06, "loss": 0.00645950622856617, "step": 2664 }, { "epoch": 1.9700554528650647, "grad_norm": 4.335019588470459, "learning_rate": 1.7558905429953804e-06, "loss": 0.015437587164342403, "step": 2665 }, { "epoch": 1.9707948243992606, "grad_norm": 4.528462886810303, "learning_rate": 1.754280183981172e-06, "loss": 0.007513603661209345, "step": 2666 }, { "epoch": 1.9715341959334567, "grad_norm": 1.5137859582901, "learning_rate": 1.7526704589914656e-06, "loss": 0.006913037970662117, "step": 2667 }, { "epoch": 1.9722735674676524, "grad_norm": 3.2464523315429688, "learning_rate": 1.7510613690512574e-06, "loss": 0.01357488427311182, "step": 2668 }, { "epoch": 1.9730129390018485, "grad_norm": 3.569657564163208, "learning_rate": 1.749452915185133e-06, "loss": 0.017631448805332184, "step": 2669 }, { "epoch": 1.9737523105360444, "grad_norm": 2.130852222442627, "learning_rate": 1.747845098417275e-06, "loss": 0.008079858496785164, "step": 2670 }, { "epoch": 1.9744916820702403, "grad_norm": 3.1221351623535156, "learning_rate": 1.7462379197714619e-06, "loss": 0.00626937672495842, "step": 2671 }, { "epoch": 1.9752310536044362, "grad_norm": 2.992461681365967, "learning_rate": 1.7446313802710636e-06, "loss": 0.015037234872579575, "step": 2672 }, { "epoch": 1.975970425138632, "grad_norm": 4.610738277435303, "learning_rate": 1.7430254809390431e-06, "loss": 0.008459974080324173, "step": 2673 }, { "epoch": 1.9767097966728282, "grad_norm": 2.3413705825805664, "learning_rate": 1.7414202227979587e-06, "loss": 0.01615959405899048, "step": 2674 }, { "epoch": 1.9774491682070239, "grad_norm": 3.299849510192871, "learning_rate": 1.7398156068699574e-06, "loss": 0.010568017140030861, "step": 2675 }, { "epoch": 1.97818853974122, "grad_norm": 2.7215888500213623, "learning_rate": 1.7382116341767785e-06, "loss": 0.013309899717569351, "step": 2676 }, { "epoch": 1.978927911275416, "grad_norm": 1.9905391931533813, "learning_rate": 1.736608305739751e-06, "loss": 0.009206774644553661, "step": 2677 }, { "epoch": 1.9796672828096118, "grad_norm": 1.5988456010818481, "learning_rate": 1.7350056225797938e-06, "loss": 0.0059140753000974655, "step": 2678 }, { "epoch": 1.980406654343808, "grad_norm": 3.0836243629455566, "learning_rate": 1.7334035857174163e-06, "loss": 0.012724051252007484, "step": 2679 }, { "epoch": 1.9811460258780036, "grad_norm": 3.473362445831299, "learning_rate": 1.731802196172715e-06, "loss": 0.025437787175178528, "step": 2680 }, { "epoch": 1.9818853974121997, "grad_norm": 2.5470004081726074, "learning_rate": 1.7302014549653744e-06, "loss": 0.015046168118715286, "step": 2681 }, { "epoch": 1.9826247689463956, "grad_norm": 2.9508867263793945, "learning_rate": 1.7286013631146665e-06, "loss": 0.0037522276397794485, "step": 2682 }, { "epoch": 1.9833641404805915, "grad_norm": 2.9569928646087646, "learning_rate": 1.7270019216394485e-06, "loss": 0.010462714359164238, "step": 2683 }, { "epoch": 1.9841035120147874, "grad_norm": 1.9917893409729004, "learning_rate": 1.725403131558167e-06, "loss": 0.005201437510550022, "step": 2684 }, { "epoch": 1.9848428835489833, "grad_norm": 3.4123775959014893, "learning_rate": 1.723804993888849e-06, "loss": 0.014068081974983215, "step": 2685 }, { "epoch": 1.9855822550831794, "grad_norm": 4.135313034057617, "learning_rate": 1.72220750964911e-06, "loss": 0.02219175174832344, "step": 2686 }, { "epoch": 1.986321626617375, "grad_norm": 4.596225738525391, "learning_rate": 1.7206106798561478e-06, "loss": 0.012380598112940788, "step": 2687 }, { "epoch": 1.9870609981515712, "grad_norm": 0.8384724855422974, "learning_rate": 1.719014505526742e-06, "loss": 0.004578684456646442, "step": 2688 }, { "epoch": 1.9878003696857671, "grad_norm": 3.4580955505371094, "learning_rate": 1.7174189876772596e-06, "loss": 0.01539861224591732, "step": 2689 }, { "epoch": 1.988539741219963, "grad_norm": 3.8417258262634277, "learning_rate": 1.7158241273236431e-06, "loss": 0.019083812832832336, "step": 2690 }, { "epoch": 1.989279112754159, "grad_norm": 2.449845790863037, "learning_rate": 1.714229925481422e-06, "loss": 0.008634298108518124, "step": 2691 }, { "epoch": 1.9900184842883548, "grad_norm": 3.8344812393188477, "learning_rate": 1.7126363831657039e-06, "loss": 0.013049611821770668, "step": 2692 }, { "epoch": 1.990757855822551, "grad_norm": 2.9693827629089355, "learning_rate": 1.7110435013911748e-06, "loss": 0.013275526463985443, "step": 2693 }, { "epoch": 1.9914972273567466, "grad_norm": 2.2328994274139404, "learning_rate": 1.7094512811721053e-06, "loss": 0.018960390239953995, "step": 2694 }, { "epoch": 1.9922365988909427, "grad_norm": 2.1726434230804443, "learning_rate": 1.7078597235223376e-06, "loss": 0.010748470202088356, "step": 2695 }, { "epoch": 1.9929759704251386, "grad_norm": 2.2021596431732178, "learning_rate": 1.7062688294552992e-06, "loss": 0.005071976222097874, "step": 2696 }, { "epoch": 1.9937153419593345, "grad_norm": 2.792102098464966, "learning_rate": 1.7046785999839905e-06, "loss": 0.007577487267553806, "step": 2697 }, { "epoch": 1.9944547134935307, "grad_norm": 2.782825469970703, "learning_rate": 1.703089036120989e-06, "loss": 0.014980247244238853, "step": 2698 }, { "epoch": 1.9951940850277263, "grad_norm": 3.6940629482269287, "learning_rate": 1.701500138878452e-06, "loss": 0.01821412518620491, "step": 2699 }, { "epoch": 1.9959334565619224, "grad_norm": 0.679865300655365, "learning_rate": 1.6999119092681066e-06, "loss": 0.004912871867418289, "step": 2700 }, { "epoch": 1.9966728280961183, "grad_norm": 2.463540554046631, "learning_rate": 1.6983243483012597e-06, "loss": 0.005265563726425171, "step": 2701 }, { "epoch": 1.9974121996303142, "grad_norm": 2.8141775131225586, "learning_rate": 1.6967374569887904e-06, "loss": 0.01703951507806778, "step": 2702 }, { "epoch": 1.9981515711645101, "grad_norm": 2.4945425987243652, "learning_rate": 1.6951512363411502e-06, "loss": 0.009591843001544476, "step": 2703 }, { "epoch": 1.998890942698706, "grad_norm": 3.2061121463775635, "learning_rate": 1.6935656873683682e-06, "loss": 0.010070151649415493, "step": 2704 }, { "epoch": 1.9996303142329022, "grad_norm": 1.5787009000778198, "learning_rate": 1.6919808110800385e-06, "loss": 0.005644980818033218, "step": 2705 }, { "epoch": 2.0, "grad_norm": 3.217663526535034, "learning_rate": 1.6903966084853341e-06, "loss": 0.02034686878323555, "step": 2706 } ], "logging_steps": 1, "max_steps": 4059, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }