CodeV-SVA-VL-9B / trainer_state.json
wyt2000's picture
Upload folder using huggingface_hub
1905999 verified
Raw History Blame Contribute Delete
324 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 1734,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0011534025374855825,
"grad_norm": 4.495698014593769,
"learning_rate": 0.0,
"loss": 0.4020993113517761,
"step": 1
},
{
"epoch": 0.002306805074971165,
"grad_norm": 4.084672663171,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.40232712030410767,
"step": 2
},
{
"epoch": 0.0034602076124567475,
"grad_norm": 3.9339748411420934,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.3950231671333313,
"step": 3
},
{
"epoch": 0.00461361014994233,
"grad_norm": 3.553755784599115,
"learning_rate": 3e-06,
"loss": 0.39159896969795227,
"step": 4
},
{
"epoch": 0.0057670126874279125,
"grad_norm": 5.716295608907302,
"learning_rate": 4.000000000000001e-06,
"loss": 0.3793337941169739,
"step": 5
},
{
"epoch": 0.006920415224913495,
"grad_norm": 4.600659201094378,
"learning_rate": 5e-06,
"loss": 0.36095085740089417,
"step": 6
},
{
"epoch": 0.008073817762399077,
"grad_norm": 3.5658854597859655,
"learning_rate": 6e-06,
"loss": 0.3388616144657135,
"step": 7
},
{
"epoch": 0.00922722029988466,
"grad_norm": 3.1573873425693404,
"learning_rate": 7e-06,
"loss": 0.348488986492157,
"step": 8
},
{
"epoch": 0.010380622837370242,
"grad_norm": 2.2570596175668314,
"learning_rate": 8.000000000000001e-06,
"loss": 0.3143322467803955,
"step": 9
},
{
"epoch": 0.011534025374855825,
"grad_norm": 1.9509618277564946,
"learning_rate": 9e-06,
"loss": 0.30949121713638306,
"step": 10
},
{
"epoch": 0.012687427912341407,
"grad_norm": 1.7347006387664359,
"learning_rate": 1e-05,
"loss": 0.30733388662338257,
"step": 11
},
{
"epoch": 0.01384083044982699,
"grad_norm": 1.4402349166700867,
"learning_rate": 1.1000000000000001e-05,
"loss": 0.2608252763748169,
"step": 12
},
{
"epoch": 0.014994232987312572,
"grad_norm": 1.349096234750999,
"learning_rate": 1.2e-05,
"loss": 0.2670484781265259,
"step": 13
},
{
"epoch": 0.016147635524798153,
"grad_norm": 1.2083977806132666,
"learning_rate": 1.3000000000000001e-05,
"loss": 0.2627828121185303,
"step": 14
},
{
"epoch": 0.01730103806228374,
"grad_norm": 1.3310000767131982,
"learning_rate": 1.4e-05,
"loss": 0.22677546739578247,
"step": 15
},
{
"epoch": 0.01845444059976932,
"grad_norm": 1.2421061646995548,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.22651028633117676,
"step": 16
},
{
"epoch": 0.0196078431372549,
"grad_norm": 1.2938288598662275,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.22234304249286652,
"step": 17
},
{
"epoch": 0.020761245674740483,
"grad_norm": 1.4364937197493592,
"learning_rate": 1.7e-05,
"loss": 0.2238374948501587,
"step": 18
},
{
"epoch": 0.02191464821222607,
"grad_norm": 1.5193452893165287,
"learning_rate": 1.8e-05,
"loss": 0.25704097747802734,
"step": 19
},
{
"epoch": 0.02306805074971165,
"grad_norm": 0.9577160786216784,
"learning_rate": 1.9e-05,
"loss": 0.22603656351566315,
"step": 20
},
{
"epoch": 0.02422145328719723,
"grad_norm": 1.7250123161659838,
"learning_rate": 2e-05,
"loss": 0.22263137996196747,
"step": 21
},
{
"epoch": 0.025374855824682813,
"grad_norm": 1.1170685594102856,
"learning_rate": 1.9999983202370694e-05,
"loss": 0.2646973431110382,
"step": 22
},
{
"epoch": 0.0265282583621684,
"grad_norm": 1.2653792374896884,
"learning_rate": 1.9999932809539202e-05,
"loss": 0.2285732626914978,
"step": 23
},
{
"epoch": 0.02768166089965398,
"grad_norm": 1.017015308345842,
"learning_rate": 1.9999848821674823e-05,
"loss": 0.20624320209026337,
"step": 24
},
{
"epoch": 0.02883506343713956,
"grad_norm": 1.0392422567255213,
"learning_rate": 1.9999731239059717e-05,
"loss": 0.24688449501991272,
"step": 25
},
{
"epoch": 0.029988465974625143,
"grad_norm": 1.0278844728857768,
"learning_rate": 1.9999580062088904e-05,
"loss": 0.24513927102088928,
"step": 26
},
{
"epoch": 0.031141868512110725,
"grad_norm": 0.9258272623475648,
"learning_rate": 1.9999395291270266e-05,
"loss": 0.2528718113899231,
"step": 27
},
{
"epoch": 0.03229527104959631,
"grad_norm": 0.9962179968337926,
"learning_rate": 1.999917692722455e-05,
"loss": 0.1915189027786255,
"step": 28
},
{
"epoch": 0.03344867358708189,
"grad_norm": 1.0416038318520384,
"learning_rate": 1.999892497068535e-05,
"loss": 0.230655238032341,
"step": 29
},
{
"epoch": 0.03460207612456748,
"grad_norm": 0.9524263031083638,
"learning_rate": 1.9998639422499125e-05,
"loss": 0.20174655318260193,
"step": 30
},
{
"epoch": 0.03575547866205306,
"grad_norm": 1.1708413473303303,
"learning_rate": 1.999832028362518e-05,
"loss": 0.25079044699668884,
"step": 31
},
{
"epoch": 0.03690888119953864,
"grad_norm": 1.0592418422823295,
"learning_rate": 1.999796755513567e-05,
"loss": 0.21050980687141418,
"step": 32
},
{
"epoch": 0.03806228373702422,
"grad_norm": 1.3306260548202338,
"learning_rate": 1.9997581238215593e-05,
"loss": 0.23325186967849731,
"step": 33
},
{
"epoch": 0.0392156862745098,
"grad_norm": 1.118931113649939,
"learning_rate": 1.9997161334162794e-05,
"loss": 0.23002584278583527,
"step": 34
},
{
"epoch": 0.040369088811995385,
"grad_norm": 1.0513011647651154,
"learning_rate": 1.999670784438795e-05,
"loss": 0.2421700656414032,
"step": 35
},
{
"epoch": 0.04152249134948097,
"grad_norm": 0.9907875006476332,
"learning_rate": 1.9996220770414573e-05,
"loss": 0.2103942334651947,
"step": 36
},
{
"epoch": 0.04267589388696655,
"grad_norm": 0.8246052733310042,
"learning_rate": 1.9995700113879e-05,
"loss": 0.18875542283058167,
"step": 37
},
{
"epoch": 0.04382929642445214,
"grad_norm": 0.8906929778764381,
"learning_rate": 1.9995145876530393e-05,
"loss": 0.22451245784759521,
"step": 38
},
{
"epoch": 0.04498269896193772,
"grad_norm": 0.7776447844140424,
"learning_rate": 1.9994558060230722e-05,
"loss": 0.18571923673152924,
"step": 39
},
{
"epoch": 0.0461361014994233,
"grad_norm": 0.8432808597551156,
"learning_rate": 1.9993936666954773e-05,
"loss": 0.18852660059928894,
"step": 40
},
{
"epoch": 0.04728950403690888,
"grad_norm": 0.933765604151769,
"learning_rate": 1.9993281698790132e-05,
"loss": 0.2217416763305664,
"step": 41
},
{
"epoch": 0.04844290657439446,
"grad_norm": 0.806489358792373,
"learning_rate": 1.999259315793718e-05,
"loss": 0.1801045835018158,
"step": 42
},
{
"epoch": 0.049596309111880045,
"grad_norm": 0.7585462098085481,
"learning_rate": 1.999187104670909e-05,
"loss": 0.18967005610466003,
"step": 43
},
{
"epoch": 0.05074971164936563,
"grad_norm": 0.8000978397100376,
"learning_rate": 1.9991115367531814e-05,
"loss": 0.1850614845752716,
"step": 44
},
{
"epoch": 0.05190311418685121,
"grad_norm": 0.7830743786002725,
"learning_rate": 1.9990326122944076e-05,
"loss": 0.1842389553785324,
"step": 45
},
{
"epoch": 0.0530565167243368,
"grad_norm": 0.7700207911397281,
"learning_rate": 1.998950331559736e-05,
"loss": 0.1886480748653412,
"step": 46
},
{
"epoch": 0.05420991926182238,
"grad_norm": 0.8785778067202274,
"learning_rate": 1.9988646948255915e-05,
"loss": 0.18026632070541382,
"step": 47
},
{
"epoch": 0.05536332179930796,
"grad_norm": 0.778431585548906,
"learning_rate": 1.9987757023796724e-05,
"loss": 0.20572340488433838,
"step": 48
},
{
"epoch": 0.05651672433679354,
"grad_norm": 0.7730621952434611,
"learning_rate": 1.998683354520951e-05,
"loss": 0.19894349575042725,
"step": 49
},
{
"epoch": 0.05767012687427912,
"grad_norm": 0.7779776845442864,
"learning_rate": 1.9985876515596733e-05,
"loss": 0.2092895805835724,
"step": 50
},
{
"epoch": 0.058823529411764705,
"grad_norm": 0.78028219904086,
"learning_rate": 1.9984885938173545e-05,
"loss": 0.1944088637828827,
"step": 51
},
{
"epoch": 0.05997693194925029,
"grad_norm": 0.7709406369325121,
"learning_rate": 1.9983861816267823e-05,
"loss": 0.2166958749294281,
"step": 52
},
{
"epoch": 0.06113033448673587,
"grad_norm": 0.7713648350442719,
"learning_rate": 1.998280415332013e-05,
"loss": 0.22369879484176636,
"step": 53
},
{
"epoch": 0.06228373702422145,
"grad_norm": 0.8086620934710609,
"learning_rate": 1.998171295288372e-05,
"loss": 0.20533880591392517,
"step": 54
},
{
"epoch": 0.06343713956170703,
"grad_norm": 0.7110082026885636,
"learning_rate": 1.9980588218624496e-05,
"loss": 0.2042270004749298,
"step": 55
},
{
"epoch": 0.06459054209919261,
"grad_norm": 0.7188180580991053,
"learning_rate": 1.997942995432104e-05,
"loss": 0.17431491613388062,
"step": 56
},
{
"epoch": 0.0657439446366782,
"grad_norm": 0.7695273259939246,
"learning_rate": 1.9978238163864567e-05,
"loss": 0.16922461986541748,
"step": 57
},
{
"epoch": 0.06689734717416378,
"grad_norm": 0.7151789780212761,
"learning_rate": 1.997701285125893e-05,
"loss": 0.18981802463531494,
"step": 58
},
{
"epoch": 0.06805074971164937,
"grad_norm": 0.7457963164048023,
"learning_rate": 1.99757540206206e-05,
"loss": 0.20055337250232697,
"step": 59
},
{
"epoch": 0.06920415224913495,
"grad_norm": 0.7106633142827684,
"learning_rate": 1.9974461676178653e-05,
"loss": 0.19016343355178833,
"step": 60
},
{
"epoch": 0.07035755478662054,
"grad_norm": 0.6787700054010403,
"learning_rate": 1.997313582227474e-05,
"loss": 0.21186664700508118,
"step": 61
},
{
"epoch": 0.07151095732410612,
"grad_norm": 0.7235077621040789,
"learning_rate": 1.997177646336312e-05,
"loss": 0.1914021074771881,
"step": 62
},
{
"epoch": 0.0726643598615917,
"grad_norm": 0.7691174918121025,
"learning_rate": 1.997038360401058e-05,
"loss": 0.2368081510066986,
"step": 63
},
{
"epoch": 0.07381776239907728,
"grad_norm": 0.747998060141863,
"learning_rate": 1.9968957248896475e-05,
"loss": 0.21126055717468262,
"step": 64
},
{
"epoch": 0.07497116493656286,
"grad_norm": 0.774816228484377,
"learning_rate": 1.9967497402812678e-05,
"loss": 0.19711562991142273,
"step": 65
},
{
"epoch": 0.07612456747404844,
"grad_norm": 0.741331599142382,
"learning_rate": 1.9966004070663584e-05,
"loss": 0.16449815034866333,
"step": 66
},
{
"epoch": 0.07727797001153403,
"grad_norm": 0.7795432887912381,
"learning_rate": 1.9964477257466075e-05,
"loss": 0.22083425521850586,
"step": 67
},
{
"epoch": 0.0784313725490196,
"grad_norm": 0.7334006490725261,
"learning_rate": 1.9962916968349524e-05,
"loss": 0.17930898070335388,
"step": 68
},
{
"epoch": 0.07958477508650519,
"grad_norm": 1.0376744193292347,
"learning_rate": 1.9961323208555762e-05,
"loss": 0.19702863693237305,
"step": 69
},
{
"epoch": 0.08073817762399077,
"grad_norm": 0.7674160561790067,
"learning_rate": 1.9959695983439066e-05,
"loss": 0.18416546285152435,
"step": 70
},
{
"epoch": 0.08189158016147635,
"grad_norm": 0.6628770875250466,
"learning_rate": 1.995803529846614e-05,
"loss": 0.19464576244354248,
"step": 71
},
{
"epoch": 0.08304498269896193,
"grad_norm": 0.6504951058097224,
"learning_rate": 1.9956341159216095e-05,
"loss": 0.1906358301639557,
"step": 72
},
{
"epoch": 0.08419838523644751,
"grad_norm": 0.6517011356063094,
"learning_rate": 1.9954613571380443e-05,
"loss": 0.19010229408740997,
"step": 73
},
{
"epoch": 0.0853517877739331,
"grad_norm": 0.6327669758706768,
"learning_rate": 1.9952852540763054e-05,
"loss": 0.18857397139072418,
"step": 74
},
{
"epoch": 0.08650519031141868,
"grad_norm": 0.704108495110942,
"learning_rate": 1.995105807328016e-05,
"loss": 0.15688985586166382,
"step": 75
},
{
"epoch": 0.08765859284890427,
"grad_norm": 0.7912809895398992,
"learning_rate": 1.9949230174960314e-05,
"loss": 0.19355568289756775,
"step": 76
},
{
"epoch": 0.08881199538638986,
"grad_norm": 0.8067266915157705,
"learning_rate": 1.9947368851944397e-05,
"loss": 0.1635342389345169,
"step": 77
},
{
"epoch": 0.08996539792387544,
"grad_norm": 0.8528294200498603,
"learning_rate": 1.9945474110485568e-05,
"loss": 0.19267328083515167,
"step": 78
},
{
"epoch": 0.09111880046136102,
"grad_norm": 0.7301623791872695,
"learning_rate": 1.9943545956949258e-05,
"loss": 0.20365720987319946,
"step": 79
},
{
"epoch": 0.0922722029988466,
"grad_norm": 0.7425915083838776,
"learning_rate": 1.9941584397813152e-05,
"loss": 0.20796117186546326,
"step": 80
},
{
"epoch": 0.09342560553633218,
"grad_norm": 0.7370443525445284,
"learning_rate": 1.9939589439667154e-05,
"loss": 0.160355344414711,
"step": 81
},
{
"epoch": 0.09457900807381776,
"grad_norm": 0.6879209582205407,
"learning_rate": 1.993756108921338e-05,
"loss": 0.18066641688346863,
"step": 82
},
{
"epoch": 0.09573241061130335,
"grad_norm": 0.7501986474920204,
"learning_rate": 1.9935499353266126e-05,
"loss": 0.17343878746032715,
"step": 83
},
{
"epoch": 0.09688581314878893,
"grad_norm": 0.8453546646184825,
"learning_rate": 1.9933404238751848e-05,
"loss": 0.18964549899101257,
"step": 84
},
{
"epoch": 0.09803921568627451,
"grad_norm": 0.7660225096070951,
"learning_rate": 1.9931275752709135e-05,
"loss": 0.18064546585083008,
"step": 85
},
{
"epoch": 0.09919261822376009,
"grad_norm": 0.7128608927357949,
"learning_rate": 1.99291139022887e-05,
"loss": 0.1978132128715515,
"step": 86
},
{
"epoch": 0.10034602076124567,
"grad_norm": 0.7102978074464146,
"learning_rate": 1.9926918694753316e-05,
"loss": 0.1548643708229065,
"step": 87
},
{
"epoch": 0.10149942329873125,
"grad_norm": 0.7684238299725789,
"learning_rate": 1.992469013747786e-05,
"loss": 0.16184785962104797,
"step": 88
},
{
"epoch": 0.10265282583621683,
"grad_norm": 0.7461469883116325,
"learning_rate": 1.9922428237949215e-05,
"loss": 0.22292672097682953,
"step": 89
},
{
"epoch": 0.10380622837370242,
"grad_norm": 0.8059321331245797,
"learning_rate": 1.9920133003766298e-05,
"loss": 0.16984286904335022,
"step": 90
},
{
"epoch": 0.104959630911188,
"grad_norm": 0.8724177524199226,
"learning_rate": 1.9917804442640003e-05,
"loss": 0.1581200659275055,
"step": 91
},
{
"epoch": 0.1061130334486736,
"grad_norm": 0.8185655485997323,
"learning_rate": 1.9915442562393194e-05,
"loss": 0.16904079914093018,
"step": 92
},
{
"epoch": 0.10726643598615918,
"grad_norm": 0.6790832179688941,
"learning_rate": 1.9913047370960667e-05,
"loss": 0.18937250971794128,
"step": 93
},
{
"epoch": 0.10841983852364476,
"grad_norm": 0.7208039642626578,
"learning_rate": 1.991061887638913e-05,
"loss": 0.16727524995803833,
"step": 94
},
{
"epoch": 0.10957324106113034,
"grad_norm": 0.7260609493742487,
"learning_rate": 1.9908157086837173e-05,
"loss": 0.1600365936756134,
"step": 95
},
{
"epoch": 0.11072664359861592,
"grad_norm": 0.7599022223322853,
"learning_rate": 1.9905662010575245e-05,
"loss": 0.1855992078781128,
"step": 96
},
{
"epoch": 0.1118800461361015,
"grad_norm": 0.7203578332312915,
"learning_rate": 1.9903133655985614e-05,
"loss": 0.1814870834350586,
"step": 97
},
{
"epoch": 0.11303344867358708,
"grad_norm": 0.8400490804196428,
"learning_rate": 1.9900572031562356e-05,
"loss": 0.1677895188331604,
"step": 98
},
{
"epoch": 0.11418685121107267,
"grad_norm": 0.8299924486926631,
"learning_rate": 1.9897977145911317e-05,
"loss": 0.20058929920196533,
"step": 99
},
{
"epoch": 0.11534025374855825,
"grad_norm": 0.7166475597710489,
"learning_rate": 1.9895349007750077e-05,
"loss": 0.16462388634681702,
"step": 100
},
{
"epoch": 0.11649365628604383,
"grad_norm": 0.7898404461210468,
"learning_rate": 1.9892687625907933e-05,
"loss": 0.18888401985168457,
"step": 101
},
{
"epoch": 0.11764705882352941,
"grad_norm": 0.7606486446667211,
"learning_rate": 1.9889993009325872e-05,
"loss": 0.15768247842788696,
"step": 102
},
{
"epoch": 0.11880046136101499,
"grad_norm": 0.7518137858099312,
"learning_rate": 1.9887265167056526e-05,
"loss": 0.16646791994571686,
"step": 103
},
{
"epoch": 0.11995386389850057,
"grad_norm": 0.7456858354803523,
"learning_rate": 1.988450410826415e-05,
"loss": 0.17068099975585938,
"step": 104
},
{
"epoch": 0.12110726643598616,
"grad_norm": 0.7672698184819925,
"learning_rate": 1.9881709842224593e-05,
"loss": 0.15105988085269928,
"step": 105
},
{
"epoch": 0.12226066897347174,
"grad_norm": 0.7260136347547786,
"learning_rate": 1.987888237832527e-05,
"loss": 0.1487008035182953,
"step": 106
},
{
"epoch": 0.12341407151095732,
"grad_norm": 0.8104592728214355,
"learning_rate": 1.9876021726065107e-05,
"loss": 0.1582726687192917,
"step": 107
},
{
"epoch": 0.1245674740484429,
"grad_norm": 0.7015588724722166,
"learning_rate": 1.987312789505455e-05,
"loss": 0.19047990441322327,
"step": 108
},
{
"epoch": 0.12572087658592848,
"grad_norm": 0.7717800038845771,
"learning_rate": 1.9870200895015492e-05,
"loss": 0.1607568860054016,
"step": 109
},
{
"epoch": 0.12687427912341406,
"grad_norm": 0.6508959505175478,
"learning_rate": 1.986724073578127e-05,
"loss": 0.17037968337535858,
"step": 110
},
{
"epoch": 0.12802768166089964,
"grad_norm": 0.7509745088194525,
"learning_rate": 1.9864247427296616e-05,
"loss": 0.18344631791114807,
"step": 111
},
{
"epoch": 0.12918108419838523,
"grad_norm": 0.723638889751479,
"learning_rate": 1.9861220979617622e-05,
"loss": 0.17803481221199036,
"step": 112
},
{
"epoch": 0.1303344867358708,
"grad_norm": 0.6690700869937638,
"learning_rate": 1.9858161402911722e-05,
"loss": 0.1759791225194931,
"step": 113
},
{
"epoch": 0.1314878892733564,
"grad_norm": 0.6999734177843382,
"learning_rate": 1.985506870745764e-05,
"loss": 0.16697293519973755,
"step": 114
},
{
"epoch": 0.13264129181084197,
"grad_norm": 0.7155964614863718,
"learning_rate": 1.985194290364537e-05,
"loss": 0.17260026931762695,
"step": 115
},
{
"epoch": 0.13379469434832755,
"grad_norm": 0.6733810919072469,
"learning_rate": 1.9848784001976126e-05,
"loss": 0.17716319859027863,
"step": 116
},
{
"epoch": 0.13494809688581316,
"grad_norm": 0.7676515297694675,
"learning_rate": 1.9845592013062325e-05,
"loss": 0.15381976962089539,
"step": 117
},
{
"epoch": 0.13610149942329874,
"grad_norm": 0.6711250063380779,
"learning_rate": 1.9842366947627534e-05,
"loss": 0.1781279742717743,
"step": 118
},
{
"epoch": 0.13725490196078433,
"grad_norm": 0.6937278250371831,
"learning_rate": 1.9839108816506442e-05,
"loss": 0.16928637027740479,
"step": 119
},
{
"epoch": 0.1384083044982699,
"grad_norm": 0.6978264528874867,
"learning_rate": 1.9835817630644826e-05,
"loss": 0.1834072470664978,
"step": 120
},
{
"epoch": 0.1395617070357555,
"grad_norm": 0.7117828148098732,
"learning_rate": 1.9832493401099507e-05,
"loss": 0.16920702159404755,
"step": 121
},
{
"epoch": 0.14071510957324107,
"grad_norm": 0.6815724729197252,
"learning_rate": 1.982913613903833e-05,
"loss": 0.16225093603134155,
"step": 122
},
{
"epoch": 0.14186851211072665,
"grad_norm": 0.6060510951668272,
"learning_rate": 1.9825745855740096e-05,
"loss": 0.1567312330007553,
"step": 123
},
{
"epoch": 0.14302191464821223,
"grad_norm": 0.6335609668418263,
"learning_rate": 1.9822322562594545e-05,
"loss": 0.1627032458782196,
"step": 124
},
{
"epoch": 0.14417531718569782,
"grad_norm": 0.6980426831015336,
"learning_rate": 1.9818866271102332e-05,
"loss": 0.16960784792900085,
"step": 125
},
{
"epoch": 0.1453287197231834,
"grad_norm": 0.7331889657583718,
"learning_rate": 1.9815376992874946e-05,
"loss": 0.16772682964801788,
"step": 126
},
{
"epoch": 0.14648212226066898,
"grad_norm": 0.7769961061707333,
"learning_rate": 1.981185473963471e-05,
"loss": 0.19278088212013245,
"step": 127
},
{
"epoch": 0.14763552479815456,
"grad_norm": 0.6489913647322587,
"learning_rate": 1.9808299523214727e-05,
"loss": 0.17111808061599731,
"step": 128
},
{
"epoch": 0.14878892733564014,
"grad_norm": 0.712038022990334,
"learning_rate": 1.9804711355558838e-05,
"loss": 0.1926974058151245,
"step": 129
},
{
"epoch": 0.14994232987312572,
"grad_norm": 0.6281774229260446,
"learning_rate": 1.9801090248721588e-05,
"loss": 0.15971839427947998,
"step": 130
},
{
"epoch": 0.1510957324106113,
"grad_norm": 0.698907132063945,
"learning_rate": 1.9797436214868172e-05,
"loss": 0.1802392601966858,
"step": 131
},
{
"epoch": 0.1522491349480969,
"grad_norm": 0.7166423387908352,
"learning_rate": 1.9793749266274416e-05,
"loss": 0.1831294447183609,
"step": 132
},
{
"epoch": 0.15340253748558247,
"grad_norm": 0.6809858400425628,
"learning_rate": 1.979002941532672e-05,
"loss": 0.19361042976379395,
"step": 133
},
{
"epoch": 0.15455594002306805,
"grad_norm": 0.5868163764208815,
"learning_rate": 1.9786276674522014e-05,
"loss": 0.16919901967048645,
"step": 134
},
{
"epoch": 0.15570934256055363,
"grad_norm": 0.682534775920136,
"learning_rate": 1.9782491056467737e-05,
"loss": 0.15192270278930664,
"step": 135
},
{
"epoch": 0.1568627450980392,
"grad_norm": 0.6168695809239229,
"learning_rate": 1.977867257388176e-05,
"loss": 0.17174585163593292,
"step": 136
},
{
"epoch": 0.1580161476355248,
"grad_norm": 0.6767685304844478,
"learning_rate": 1.9774821239592383e-05,
"loss": 0.15824061632156372,
"step": 137
},
{
"epoch": 0.15916955017301038,
"grad_norm": 0.6569001976163854,
"learning_rate": 1.977093706653826e-05,
"loss": 0.1818104088306427,
"step": 138
},
{
"epoch": 0.16032295271049596,
"grad_norm": 0.6746854765941305,
"learning_rate": 1.9767020067768365e-05,
"loss": 0.1537202149629593,
"step": 139
},
{
"epoch": 0.16147635524798154,
"grad_norm": 0.8613823360709193,
"learning_rate": 1.9763070256441968e-05,
"loss": 0.1882435381412506,
"step": 140
},
{
"epoch": 0.16262975778546712,
"grad_norm": 0.6590904456495016,
"learning_rate": 1.9759087645828554e-05,
"loss": 0.19184091687202454,
"step": 141
},
{
"epoch": 0.1637831603229527,
"grad_norm": 0.8131211006058583,
"learning_rate": 1.9755072249307807e-05,
"loss": 0.1724315583705902,
"step": 142
},
{
"epoch": 0.16493656286043828,
"grad_norm": 0.7530893941193774,
"learning_rate": 1.975102408036956e-05,
"loss": 0.18615314364433289,
"step": 143
},
{
"epoch": 0.16608996539792387,
"grad_norm": 0.6822828843062119,
"learning_rate": 1.9746943152613734e-05,
"loss": 0.1724388599395752,
"step": 144
},
{
"epoch": 0.16724336793540945,
"grad_norm": 0.7283527535241309,
"learning_rate": 1.974282947975032e-05,
"loss": 0.20811283588409424,
"step": 145
},
{
"epoch": 0.16839677047289503,
"grad_norm": 0.6879638461141416,
"learning_rate": 1.97386830755993e-05,
"loss": 0.1503899097442627,
"step": 146
},
{
"epoch": 0.1695501730103806,
"grad_norm": 0.6797044328715199,
"learning_rate": 1.9734503954090634e-05,
"loss": 0.17163418233394623,
"step": 147
},
{
"epoch": 0.1707035755478662,
"grad_norm": 0.7396679853833975,
"learning_rate": 1.9730292129264185e-05,
"loss": 0.15767835080623627,
"step": 148
},
{
"epoch": 0.17185697808535177,
"grad_norm": 0.6517491453600455,
"learning_rate": 1.9726047615269682e-05,
"loss": 0.15945325791835785,
"step": 149
},
{
"epoch": 0.17301038062283736,
"grad_norm": 0.6496308304955081,
"learning_rate": 1.9721770426366688e-05,
"loss": 0.17854586243629456,
"step": 150
},
{
"epoch": 0.17416378316032297,
"grad_norm": 0.8471540101303121,
"learning_rate": 1.9717460576924528e-05,
"loss": 0.16005681455135345,
"step": 151
},
{
"epoch": 0.17531718569780855,
"grad_norm": 0.7158572380122824,
"learning_rate": 1.971311808142225e-05,
"loss": 0.1603868156671524,
"step": 152
},
{
"epoch": 0.17647058823529413,
"grad_norm": 0.6613145618152824,
"learning_rate": 1.9708742954448578e-05,
"loss": 0.16552971303462982,
"step": 153
},
{
"epoch": 0.1776239907727797,
"grad_norm": 0.750817110907634,
"learning_rate": 1.970433521070187e-05,
"loss": 0.18519628047943115,
"step": 154
},
{
"epoch": 0.1787773933102653,
"grad_norm": 0.6786596208094904,
"learning_rate": 1.9699894864990054e-05,
"loss": 0.15193766355514526,
"step": 155
},
{
"epoch": 0.17993079584775087,
"grad_norm": 0.7086543849182976,
"learning_rate": 1.9695421932230583e-05,
"loss": 0.14914155006408691,
"step": 156
},
{
"epoch": 0.18108419838523646,
"grad_norm": 0.6699914820868181,
"learning_rate": 1.969091642745039e-05,
"loss": 0.18154610693454742,
"step": 157
},
{
"epoch": 0.18223760092272204,
"grad_norm": 0.8045405493430372,
"learning_rate": 1.968637836578584e-05,
"loss": 0.17390501499176025,
"step": 158
},
{
"epoch": 0.18339100346020762,
"grad_norm": 0.7840064563130626,
"learning_rate": 1.9681807762482665e-05,
"loss": 0.1906685084104538,
"step": 159
},
{
"epoch": 0.1845444059976932,
"grad_norm": 0.7059232786382783,
"learning_rate": 1.9677204632895926e-05,
"loss": 0.18064354360103607,
"step": 160
},
{
"epoch": 0.18569780853517878,
"grad_norm": 0.748103668927526,
"learning_rate": 1.9672568992489952e-05,
"loss": 0.15183711051940918,
"step": 161
},
{
"epoch": 0.18685121107266436,
"grad_norm": 0.7260783559031867,
"learning_rate": 1.96679008568383e-05,
"loss": 0.2089284062385559,
"step": 162
},
{
"epoch": 0.18800461361014995,
"grad_norm": 0.6703069824231739,
"learning_rate": 1.9663200241623695e-05,
"loss": 0.17618629336357117,
"step": 163
},
{
"epoch": 0.18915801614763553,
"grad_norm": 0.6989469017222434,
"learning_rate": 1.965846716263797e-05,
"loss": 0.13952553272247314,
"step": 164
},
{
"epoch": 0.1903114186851211,
"grad_norm": 0.6615823083965273,
"learning_rate": 1.9653701635782033e-05,
"loss": 0.15998944640159607,
"step": 165
},
{
"epoch": 0.1914648212226067,
"grad_norm": 0.6581045996895691,
"learning_rate": 1.9648903677065787e-05,
"loss": 0.18140161037445068,
"step": 166
},
{
"epoch": 0.19261822376009227,
"grad_norm": 0.6844581947788555,
"learning_rate": 1.9644073302608104e-05,
"loss": 0.18786802887916565,
"step": 167
},
{
"epoch": 0.19377162629757785,
"grad_norm": 0.8978506977929855,
"learning_rate": 1.963921052863675e-05,
"loss": 0.18338137865066528,
"step": 168
},
{
"epoch": 0.19492502883506344,
"grad_norm": 0.6118803449893772,
"learning_rate": 1.963431537148834e-05,
"loss": 0.14717982709407806,
"step": 169
},
{
"epoch": 0.19607843137254902,
"grad_norm": 0.6921321365674334,
"learning_rate": 1.9629387847608282e-05,
"loss": 0.17424622178077698,
"step": 170
},
{
"epoch": 0.1972318339100346,
"grad_norm": 0.635519426557306,
"learning_rate": 1.9624427973550717e-05,
"loss": 0.20706820487976074,
"step": 171
},
{
"epoch": 0.19838523644752018,
"grad_norm": 0.691081999633674,
"learning_rate": 1.961943576597847e-05,
"loss": 0.14305469393730164,
"step": 172
},
{
"epoch": 0.19953863898500576,
"grad_norm": 0.6569901436690089,
"learning_rate": 1.9614411241663e-05,
"loss": 0.19898062944412231,
"step": 173
},
{
"epoch": 0.20069204152249134,
"grad_norm": 0.7938813433918841,
"learning_rate": 1.9609354417484315e-05,
"loss": 0.15062376856803894,
"step": 174
},
{
"epoch": 0.20184544405997693,
"grad_norm": 0.8784657842893214,
"learning_rate": 1.9604265310430954e-05,
"loss": 0.21294081211090088,
"step": 175
},
{
"epoch": 0.2029988465974625,
"grad_norm": 0.800950438271067,
"learning_rate": 1.95991439375999e-05,
"loss": 0.15397536754608154,
"step": 176
},
{
"epoch": 0.2041522491349481,
"grad_norm": 0.8506015048962129,
"learning_rate": 1.959399031619654e-05,
"loss": 0.1569744348526001,
"step": 177
},
{
"epoch": 0.20530565167243367,
"grad_norm": 0.7550605504194676,
"learning_rate": 1.9588804463534595e-05,
"loss": 0.15508490800857544,
"step": 178
},
{
"epoch": 0.20645905420991925,
"grad_norm": 0.7069978699221183,
"learning_rate": 1.9583586397036077e-05,
"loss": 0.17187631130218506,
"step": 179
},
{
"epoch": 0.20761245674740483,
"grad_norm": 0.7606744058347206,
"learning_rate": 1.957833613423121e-05,
"loss": 0.16045120358467102,
"step": 180
},
{
"epoch": 0.20876585928489041,
"grad_norm": 0.6488865372501639,
"learning_rate": 1.9573053692758387e-05,
"loss": 0.15017543733119965,
"step": 181
},
{
"epoch": 0.209919261822376,
"grad_norm": 0.7516938225173669,
"learning_rate": 1.956773909036411e-05,
"loss": 0.15792176127433777,
"step": 182
},
{
"epoch": 0.21107266435986158,
"grad_norm": 0.7195436431694959,
"learning_rate": 1.956239234490292e-05,
"loss": 0.1756700873374939,
"step": 183
},
{
"epoch": 0.2122260668973472,
"grad_norm": 0.7539290249596812,
"learning_rate": 1.9557013474337358e-05,
"loss": 0.14884887635707855,
"step": 184
},
{
"epoch": 0.21337946943483277,
"grad_norm": 0.6115891853558258,
"learning_rate": 1.955160249673786e-05,
"loss": 0.1404731273651123,
"step": 185
},
{
"epoch": 0.21453287197231835,
"grad_norm": 0.6833264921900011,
"learning_rate": 1.954615943028276e-05,
"loss": 0.18250282108783722,
"step": 186
},
{
"epoch": 0.21568627450980393,
"grad_norm": 0.9608735278788254,
"learning_rate": 1.9540684293258172e-05,
"loss": 0.1494218409061432,
"step": 187
},
{
"epoch": 0.21683967704728951,
"grad_norm": 0.782019885961227,
"learning_rate": 1.9535177104057966e-05,
"loss": 0.14098602533340454,
"step": 188
},
{
"epoch": 0.2179930795847751,
"grad_norm": 0.575355295999276,
"learning_rate": 1.9529637881183684e-05,
"loss": 0.14377009868621826,
"step": 189
},
{
"epoch": 0.21914648212226068,
"grad_norm": 0.7174136965676335,
"learning_rate": 1.952406664324449e-05,
"loss": 0.15598389506340027,
"step": 190
},
{
"epoch": 0.22029988465974626,
"grad_norm": 0.5578953127982875,
"learning_rate": 1.9518463408957097e-05,
"loss": 0.1604020595550537,
"step": 191
},
{
"epoch": 0.22145328719723184,
"grad_norm": 0.6454965792404662,
"learning_rate": 1.9512828197145723e-05,
"loss": 0.1589151918888092,
"step": 192
},
{
"epoch": 0.22260668973471742,
"grad_norm": 0.6324358589514775,
"learning_rate": 1.9507161026742004e-05,
"loss": 0.1622220277786255,
"step": 193
},
{
"epoch": 0.223760092272203,
"grad_norm": 0.7017861436507348,
"learning_rate": 1.9501461916784944e-05,
"loss": 0.14456167817115784,
"step": 194
},
{
"epoch": 0.22491349480968859,
"grad_norm": 0.6058184902619975,
"learning_rate": 1.9495730886420853e-05,
"loss": 0.1365896314382553,
"step": 195
},
{
"epoch": 0.22606689734717417,
"grad_norm": 0.7500588260908763,
"learning_rate": 1.9489967954903272e-05,
"loss": 0.17968465387821198,
"step": 196
},
{
"epoch": 0.22722029988465975,
"grad_norm": 0.7089570342281032,
"learning_rate": 1.9484173141592926e-05,
"loss": 0.18182900547981262,
"step": 197
},
{
"epoch": 0.22837370242214533,
"grad_norm": 0.663156769202086,
"learning_rate": 1.947834646595763e-05,
"loss": 0.14773045480251312,
"step": 198
},
{
"epoch": 0.2295271049596309,
"grad_norm": 0.6238639161576802,
"learning_rate": 1.947248794757226e-05,
"loss": 0.14862895011901855,
"step": 199
},
{
"epoch": 0.2306805074971165,
"grad_norm": 0.712591434122004,
"learning_rate": 1.9466597606118657e-05,
"loss": 0.14908409118652344,
"step": 200
},
{
"epoch": 0.23183391003460208,
"grad_norm": 0.7857855428388536,
"learning_rate": 1.9460675461385575e-05,
"loss": 0.14093071222305298,
"step": 201
},
{
"epoch": 0.23298731257208766,
"grad_norm": 0.6834540289794465,
"learning_rate": 1.9454721533268612e-05,
"loss": 0.20165839791297913,
"step": 202
},
{
"epoch": 0.23414071510957324,
"grad_norm": 0.6681437803511335,
"learning_rate": 1.9448735841770147e-05,
"loss": 0.17805591225624084,
"step": 203
},
{
"epoch": 0.23529411764705882,
"grad_norm": 0.6964491767880527,
"learning_rate": 1.944271840699926e-05,
"loss": 0.18638867139816284,
"step": 204
},
{
"epoch": 0.2364475201845444,
"grad_norm": 0.6309107103768836,
"learning_rate": 1.9436669249171677e-05,
"loss": 0.16904030740261078,
"step": 205
},
{
"epoch": 0.23760092272202998,
"grad_norm": 0.6161482732232728,
"learning_rate": 1.943058838860971e-05,
"loss": 0.16694962978363037,
"step": 206
},
{
"epoch": 0.23875432525951557,
"grad_norm": 0.6809688946377181,
"learning_rate": 1.942447584574216e-05,
"loss": 0.1818813979625702,
"step": 207
},
{
"epoch": 0.23990772779700115,
"grad_norm": 0.6148694115943677,
"learning_rate": 1.9418331641104274e-05,
"loss": 0.15761923789978027,
"step": 208
},
{
"epoch": 0.24106113033448673,
"grad_norm": 0.7133630688557838,
"learning_rate": 1.941215579533767e-05,
"loss": 0.17556048929691315,
"step": 209
},
{
"epoch": 0.2422145328719723,
"grad_norm": 0.6154658066060851,
"learning_rate": 1.9405948329190252e-05,
"loss": 0.15637770295143127,
"step": 210
},
{
"epoch": 0.2433679354094579,
"grad_norm": 0.6148371473971329,
"learning_rate": 1.9399709263516176e-05,
"loss": 0.1426171362400055,
"step": 211
},
{
"epoch": 0.24452133794694347,
"grad_norm": 0.6367721888296183,
"learning_rate": 1.939343861927574e-05,
"loss": 0.13779184222221375,
"step": 212
},
{
"epoch": 0.24567474048442905,
"grad_norm": 0.5661006058580291,
"learning_rate": 1.9387136417535325e-05,
"loss": 0.1561303734779358,
"step": 213
},
{
"epoch": 0.24682814302191464,
"grad_norm": 0.66923061174313,
"learning_rate": 1.9380802679467355e-05,
"loss": 0.14185982942581177,
"step": 214
},
{
"epoch": 0.24798154555940022,
"grad_norm": 0.5951887363506578,
"learning_rate": 1.937443742635018e-05,
"loss": 0.16747041046619415,
"step": 215
},
{
"epoch": 0.2491349480968858,
"grad_norm": 0.6398408955945288,
"learning_rate": 1.9368040679568033e-05,
"loss": 0.1757366955280304,
"step": 216
},
{
"epoch": 0.2502883506343714,
"grad_norm": 0.6794820007813638,
"learning_rate": 1.936161246061095e-05,
"loss": 0.15195152163505554,
"step": 217
},
{
"epoch": 0.25144175317185696,
"grad_norm": 0.6320585340854773,
"learning_rate": 1.9355152791074704e-05,
"loss": 0.1725984513759613,
"step": 218
},
{
"epoch": 0.25259515570934254,
"grad_norm": 0.6586738289496019,
"learning_rate": 1.934866169266071e-05,
"loss": 0.17071212828159332,
"step": 219
},
{
"epoch": 0.2537485582468281,
"grad_norm": 0.620559540975561,
"learning_rate": 1.9342139187175998e-05,
"loss": 0.1379246562719345,
"step": 220
},
{
"epoch": 0.2549019607843137,
"grad_norm": 0.6318904398066311,
"learning_rate": 1.9335585296533076e-05,
"loss": 0.14164702594280243,
"step": 221
},
{
"epoch": 0.2560553633217993,
"grad_norm": 0.6689395869931279,
"learning_rate": 1.9329000042749923e-05,
"loss": 0.15543320775032043,
"step": 222
},
{
"epoch": 0.25720876585928487,
"grad_norm": 0.6258515919683159,
"learning_rate": 1.932238344794986e-05,
"loss": 0.16550573706626892,
"step": 223
},
{
"epoch": 0.25836216839677045,
"grad_norm": 0.6521453735419664,
"learning_rate": 1.9315735534361514e-05,
"loss": 0.19588306546211243,
"step": 224
},
{
"epoch": 0.25951557093425603,
"grad_norm": 0.5966485650652636,
"learning_rate": 1.930905632431872e-05,
"loss": 0.16947650909423828,
"step": 225
},
{
"epoch": 0.2606689734717416,
"grad_norm": 0.7024615389998373,
"learning_rate": 1.930234584026046e-05,
"loss": 0.13438433408737183,
"step": 226
},
{
"epoch": 0.2618223760092272,
"grad_norm": 0.6194418377694445,
"learning_rate": 1.9295604104730775e-05,
"loss": 0.15439243614673615,
"step": 227
},
{
"epoch": 0.2629757785467128,
"grad_norm": 0.6403842214670022,
"learning_rate": 1.9288831140378705e-05,
"loss": 0.1744384467601776,
"step": 228
},
{
"epoch": 0.26412918108419836,
"grad_norm": 0.6048581296616878,
"learning_rate": 1.9282026969958187e-05,
"loss": 0.1487625241279602,
"step": 229
},
{
"epoch": 0.26528258362168394,
"grad_norm": 0.6475299786983004,
"learning_rate": 1.927519161632802e-05,
"loss": 0.14852109551429749,
"step": 230
},
{
"epoch": 0.2664359861591695,
"grad_norm": 0.6322761841306727,
"learning_rate": 1.9268325102451747e-05,
"loss": 0.19121116399765015,
"step": 231
},
{
"epoch": 0.2675893886966551,
"grad_norm": 0.6164528535008689,
"learning_rate": 1.9261427451397602e-05,
"loss": 0.1762842833995819,
"step": 232
},
{
"epoch": 0.2687427912341407,
"grad_norm": 0.5552926123427286,
"learning_rate": 1.9254498686338415e-05,
"loss": 0.17892390489578247,
"step": 233
},
{
"epoch": 0.2698961937716263,
"grad_norm": 0.6245195793553985,
"learning_rate": 1.924753883055156e-05,
"loss": 0.18121552467346191,
"step": 234
},
{
"epoch": 0.2710495963091119,
"grad_norm": 0.57463942535133,
"learning_rate": 1.9240547907418843e-05,
"loss": 0.14027521014213562,
"step": 235
},
{
"epoch": 0.2722029988465975,
"grad_norm": 0.5591351052884336,
"learning_rate": 1.9233525940426458e-05,
"loss": 0.15384751558303833,
"step": 236
},
{
"epoch": 0.27335640138408307,
"grad_norm": 0.6852180111968864,
"learning_rate": 1.9226472953164884e-05,
"loss": 0.14967069029808044,
"step": 237
},
{
"epoch": 0.27450980392156865,
"grad_norm": 0.638661328119356,
"learning_rate": 1.921938896932881e-05,
"loss": 0.15989956259727478,
"step": 238
},
{
"epoch": 0.27566320645905423,
"grad_norm": 0.6160720333630387,
"learning_rate": 1.9212274012717066e-05,
"loss": 0.18037046492099762,
"step": 239
},
{
"epoch": 0.2768166089965398,
"grad_norm": 0.726438823280282,
"learning_rate": 1.9205128107232535e-05,
"loss": 0.1522429883480072,
"step": 240
},
{
"epoch": 0.2779700115340254,
"grad_norm": 0.5696132233247583,
"learning_rate": 1.9197951276882068e-05,
"loss": 0.15229398012161255,
"step": 241
},
{
"epoch": 0.279123414071511,
"grad_norm": 0.6168875314316694,
"learning_rate": 1.919074354577641e-05,
"loss": 0.1699984073638916,
"step": 242
},
{
"epoch": 0.28027681660899656,
"grad_norm": 0.6396885409258831,
"learning_rate": 1.9183504938130126e-05,
"loss": 0.1515495777130127,
"step": 243
},
{
"epoch": 0.28143021914648214,
"grad_norm": 0.6469346040841166,
"learning_rate": 1.91762354782615e-05,
"loss": 0.151616171002388,
"step": 244
},
{
"epoch": 0.2825836216839677,
"grad_norm": 0.6182297639408516,
"learning_rate": 1.9168935190592477e-05,
"loss": 0.15752270817756653,
"step": 245
},
{
"epoch": 0.2837370242214533,
"grad_norm": 0.767772964658083,
"learning_rate": 1.9161604099648555e-05,
"loss": 0.15428704023361206,
"step": 246
},
{
"epoch": 0.2848904267589389,
"grad_norm": 0.6207761455425719,
"learning_rate": 1.9154242230058728e-05,
"loss": 0.18557901680469513,
"step": 247
},
{
"epoch": 0.28604382929642447,
"grad_norm": 0.7291914618534315,
"learning_rate": 1.914684960655539e-05,
"loss": 0.14835432171821594,
"step": 248
},
{
"epoch": 0.28719723183391005,
"grad_norm": 0.634126281523298,
"learning_rate": 1.9139426253974243e-05,
"loss": 0.16896021366119385,
"step": 249
},
{
"epoch": 0.28835063437139563,
"grad_norm": 0.6278587313110084,
"learning_rate": 1.9131972197254233e-05,
"loss": 0.16670472919940948,
"step": 250
},
{
"epoch": 0.2895040369088812,
"grad_norm": 0.587673568906245,
"learning_rate": 1.9124487461437465e-05,
"loss": 0.17196440696716309,
"step": 251
},
{
"epoch": 0.2906574394463668,
"grad_norm": 0.6040499957923918,
"learning_rate": 1.9116972071669094e-05,
"loss": 0.15291623771190643,
"step": 252
},
{
"epoch": 0.2918108419838524,
"grad_norm": 0.5799600479514249,
"learning_rate": 1.910942605319727e-05,
"loss": 0.1717730313539505,
"step": 253
},
{
"epoch": 0.29296424452133796,
"grad_norm": 0.6692640528967084,
"learning_rate": 1.9101849431373034e-05,
"loss": 0.1386372447013855,
"step": 254
},
{
"epoch": 0.29411764705882354,
"grad_norm": 0.5954575440073485,
"learning_rate": 1.9094242231650246e-05,
"loss": 0.14453719556331635,
"step": 255
},
{
"epoch": 0.2952710495963091,
"grad_norm": 0.5888129238621411,
"learning_rate": 1.908660447958549e-05,
"loss": 0.17573729157447815,
"step": 256
},
{
"epoch": 0.2964244521337947,
"grad_norm": 0.5944268766683269,
"learning_rate": 1.9078936200837993e-05,
"loss": 0.15886855125427246,
"step": 257
},
{
"epoch": 0.2975778546712803,
"grad_norm": 0.589067341194725,
"learning_rate": 1.9071237421169534e-05,
"loss": 0.15152607858181,
"step": 258
},
{
"epoch": 0.29873125720876587,
"grad_norm": 0.6024693762250394,
"learning_rate": 1.906350816644436e-05,
"loss": 0.15554502606391907,
"step": 259
},
{
"epoch": 0.29988465974625145,
"grad_norm": 0.6183723883775224,
"learning_rate": 1.9055748462629103e-05,
"loss": 0.1618952751159668,
"step": 260
},
{
"epoch": 0.30103806228373703,
"grad_norm": 0.5344994423355516,
"learning_rate": 1.9047958335792693e-05,
"loss": 0.1390984207391739,
"step": 261
},
{
"epoch": 0.3021914648212226,
"grad_norm": 0.570623910998358,
"learning_rate": 1.9040137812106257e-05,
"loss": 0.1390601396560669,
"step": 262
},
{
"epoch": 0.3033448673587082,
"grad_norm": 0.650639034377059,
"learning_rate": 1.903228691784305e-05,
"loss": 0.18048053979873657,
"step": 263
},
{
"epoch": 0.3044982698961938,
"grad_norm": 0.5969481077584313,
"learning_rate": 1.9024405679378355e-05,
"loss": 0.16585248708724976,
"step": 264
},
{
"epoch": 0.30565167243367936,
"grad_norm": 0.6010680653192075,
"learning_rate": 1.9016494123189396e-05,
"loss": 0.14491364359855652,
"step": 265
},
{
"epoch": 0.30680507497116494,
"grad_norm": 0.6515267414342089,
"learning_rate": 1.9008552275855244e-05,
"loss": 0.14044110476970673,
"step": 266
},
{
"epoch": 0.3079584775086505,
"grad_norm": 0.6463289692154163,
"learning_rate": 1.900058016405675e-05,
"loss": 0.16240356862545013,
"step": 267
},
{
"epoch": 0.3091118800461361,
"grad_norm": 0.64799743412743,
"learning_rate": 1.8992577814576426e-05,
"loss": 0.1671900451183319,
"step": 268
},
{
"epoch": 0.3102652825836217,
"grad_norm": 0.6294774268467922,
"learning_rate": 1.8984545254298373e-05,
"loss": 0.15457934141159058,
"step": 269
},
{
"epoch": 0.31141868512110726,
"grad_norm": 0.692013678499771,
"learning_rate": 1.897648251020818e-05,
"loss": 0.1681273877620697,
"step": 270
},
{
"epoch": 0.31257208765859285,
"grad_norm": 0.5785043343835079,
"learning_rate": 1.896838960939285e-05,
"loss": 0.1647273749113083,
"step": 271
},
{
"epoch": 0.3137254901960784,
"grad_norm": 0.7091233777971023,
"learning_rate": 1.896026657904069e-05,
"loss": 0.15869927406311035,
"step": 272
},
{
"epoch": 0.314878892733564,
"grad_norm": 1.6348366189592263,
"learning_rate": 1.8952113446441234e-05,
"loss": 0.17177990078926086,
"step": 273
},
{
"epoch": 0.3160322952710496,
"grad_norm": 0.759979619764839,
"learning_rate": 1.8943930238985134e-05,
"loss": 0.1421811729669571,
"step": 274
},
{
"epoch": 0.31718569780853517,
"grad_norm": 0.6968387202104807,
"learning_rate": 1.8935716984164098e-05,
"loss": 0.19615796208381653,
"step": 275
},
{
"epoch": 0.31833910034602075,
"grad_norm": 0.6938605270404232,
"learning_rate": 1.892747370957076e-05,
"loss": 0.20023885369300842,
"step": 276
},
{
"epoch": 0.31949250288350634,
"grad_norm": 0.6499190809111743,
"learning_rate": 1.8919200442898617e-05,
"loss": 0.14965467154979706,
"step": 277
},
{
"epoch": 0.3206459054209919,
"grad_norm": 0.7955725027718614,
"learning_rate": 1.8910897211941915e-05,
"loss": 0.18283946812152863,
"step": 278
},
{
"epoch": 0.3217993079584775,
"grad_norm": 0.6731809381793898,
"learning_rate": 1.890256404459559e-05,
"loss": 0.18026655912399292,
"step": 279
},
{
"epoch": 0.3229527104959631,
"grad_norm": 0.6398474909892362,
"learning_rate": 1.8894200968855116e-05,
"loss": 0.15962138772010803,
"step": 280
},
{
"epoch": 0.32410611303344866,
"grad_norm": 0.6471145778176509,
"learning_rate": 1.8885808012816472e-05,
"loss": 0.15755227208137512,
"step": 281
},
{
"epoch": 0.32525951557093424,
"grad_norm": 0.5884607536713726,
"learning_rate": 1.8877385204676007e-05,
"loss": 0.15950888395309448,
"step": 282
},
{
"epoch": 0.3264129181084198,
"grad_norm": 0.6114916621132576,
"learning_rate": 1.8868932572730363e-05,
"loss": 0.15973015129566193,
"step": 283
},
{
"epoch": 0.3275663206459054,
"grad_norm": 0.6627184316801389,
"learning_rate": 1.8860450145376377e-05,
"loss": 0.1577485203742981,
"step": 284
},
{
"epoch": 0.328719723183391,
"grad_norm": 0.5679564525841323,
"learning_rate": 1.8851937951110985e-05,
"loss": 0.16121065616607666,
"step": 285
},
{
"epoch": 0.32987312572087657,
"grad_norm": 0.5592178238528515,
"learning_rate": 1.8843396018531118e-05,
"loss": 0.16865059733390808,
"step": 286
},
{
"epoch": 0.33102652825836215,
"grad_norm": 0.5714687993980185,
"learning_rate": 1.8834824376333625e-05,
"loss": 0.18550503253936768,
"step": 287
},
{
"epoch": 0.33217993079584773,
"grad_norm": 0.5928080155462283,
"learning_rate": 1.8826223053315156e-05,
"loss": 0.1783304214477539,
"step": 288
},
{
"epoch": 0.3333333333333333,
"grad_norm": 0.5122798425732935,
"learning_rate": 1.881759207837208e-05,
"loss": 0.13825464248657227,
"step": 289
},
{
"epoch": 0.3344867358708189,
"grad_norm": 0.5769949357016659,
"learning_rate": 1.880893148050038e-05,
"loss": 0.15711499750614166,
"step": 290
},
{
"epoch": 0.3356401384083045,
"grad_norm": 0.5534893673696315,
"learning_rate": 1.8800241288795557e-05,
"loss": 0.170371413230896,
"step": 291
},
{
"epoch": 0.33679354094579006,
"grad_norm": 0.6110534669629217,
"learning_rate": 1.879152153245254e-05,
"loss": 0.167488232254982,
"step": 292
},
{
"epoch": 0.33794694348327564,
"grad_norm": 0.5323763957072871,
"learning_rate": 1.878277224076557e-05,
"loss": 0.1558186262845993,
"step": 293
},
{
"epoch": 0.3391003460207612,
"grad_norm": 0.6565228435339502,
"learning_rate": 1.8773993443128123e-05,
"loss": 0.22588524222373962,
"step": 294
},
{
"epoch": 0.3402537485582468,
"grad_norm": 0.6144431120807249,
"learning_rate": 1.876518516903279e-05,
"loss": 0.13824054598808289,
"step": 295
},
{
"epoch": 0.3414071510957324,
"grad_norm": 0.5860828779571663,
"learning_rate": 1.8756347448071204e-05,
"loss": 0.13891658186912537,
"step": 296
},
{
"epoch": 0.34256055363321797,
"grad_norm": 0.682527634088077,
"learning_rate": 1.874748030993391e-05,
"loss": 0.20061641931533813,
"step": 297
},
{
"epoch": 0.34371395617070355,
"grad_norm": 0.6322914587786336,
"learning_rate": 1.8738583784410294e-05,
"loss": 0.1826009452342987,
"step": 298
},
{
"epoch": 0.34486735870818913,
"grad_norm": 0.6154520777371884,
"learning_rate": 1.872965790138846e-05,
"loss": 0.21193917095661163,
"step": 299
},
{
"epoch": 0.3460207612456747,
"grad_norm": 0.5837519929746886,
"learning_rate": 1.872070269085515e-05,
"loss": 0.1457318365573883,
"step": 300
},
{
"epoch": 0.34717416378316035,
"grad_norm": 0.7250716077101558,
"learning_rate": 1.871171818289561e-05,
"loss": 0.16374683380126953,
"step": 301
},
{
"epoch": 0.34832756632064593,
"grad_norm": 0.5872771951030308,
"learning_rate": 1.870270440769354e-05,
"loss": 0.1658208966255188,
"step": 302
},
{
"epoch": 0.3494809688581315,
"grad_norm": 0.6262927143341512,
"learning_rate": 1.8693661395530946e-05,
"loss": 0.14369769394397736,
"step": 303
},
{
"epoch": 0.3506343713956171,
"grad_norm": 0.6466712820338771,
"learning_rate": 1.8684589176788062e-05,
"loss": 0.19318701326847076,
"step": 304
},
{
"epoch": 0.3517877739331027,
"grad_norm": 0.6905293969420009,
"learning_rate": 1.867548778194324e-05,
"loss": 0.1600923240184784,
"step": 305
},
{
"epoch": 0.35294117647058826,
"grad_norm": 0.5311838943926463,
"learning_rate": 1.8666357241572855e-05,
"loss": 0.14665910601615906,
"step": 306
},
{
"epoch": 0.35409457900807384,
"grad_norm": 0.737452564875159,
"learning_rate": 1.8657197586351192e-05,
"loss": 0.2010802924633026,
"step": 307
},
{
"epoch": 0.3552479815455594,
"grad_norm": 0.6368820129055907,
"learning_rate": 1.864800884705035e-05,
"loss": 0.1569179892539978,
"step": 308
},
{
"epoch": 0.356401384083045,
"grad_norm": 0.609848048693859,
"learning_rate": 1.863879105454013e-05,
"loss": 0.17925408482551575,
"step": 309
},
{
"epoch": 0.3575547866205306,
"grad_norm": 0.6611473960727973,
"learning_rate": 1.8629544239787955e-05,
"loss": 0.20669366419315338,
"step": 310
},
{
"epoch": 0.35870818915801617,
"grad_norm": 0.5978202995521881,
"learning_rate": 1.862026843385873e-05,
"loss": 0.1488860696554184,
"step": 311
},
{
"epoch": 0.35986159169550175,
"grad_norm": 0.668068187753409,
"learning_rate": 1.861096366791477e-05,
"loss": 0.18036362528800964,
"step": 312
},
{
"epoch": 0.36101499423298733,
"grad_norm": 0.5453948029949389,
"learning_rate": 1.8601629973215673e-05,
"loss": 0.13395975530147552,
"step": 313
},
{
"epoch": 0.3621683967704729,
"grad_norm": 0.5580942182143247,
"learning_rate": 1.859226738111823e-05,
"loss": 0.1414610594511032,
"step": 314
},
{
"epoch": 0.3633217993079585,
"grad_norm": 0.8125715426313076,
"learning_rate": 1.8582875923076312e-05,
"loss": 0.15479719638824463,
"step": 315
},
{
"epoch": 0.3644752018454441,
"grad_norm": 1.0610266776964536,
"learning_rate": 1.857345563064076e-05,
"loss": 0.14872941374778748,
"step": 316
},
{
"epoch": 0.36562860438292966,
"grad_norm": 0.7065889230123819,
"learning_rate": 1.8564006535459298e-05,
"loss": 0.1397535502910614,
"step": 317
},
{
"epoch": 0.36678200692041524,
"grad_norm": 0.6441135190208248,
"learning_rate": 1.85545286692764e-05,
"loss": 0.16468822956085205,
"step": 318
},
{
"epoch": 0.3679354094579008,
"grad_norm": 0.6977198279005254,
"learning_rate": 1.8545022063933206e-05,
"loss": 0.1629921793937683,
"step": 319
},
{
"epoch": 0.3690888119953864,
"grad_norm": 0.6263724735262808,
"learning_rate": 1.8535486751367397e-05,
"loss": 0.19503770768642426,
"step": 320
},
{
"epoch": 0.370242214532872,
"grad_norm": 0.6494272992699572,
"learning_rate": 1.8525922763613106e-05,
"loss": 0.14250437915325165,
"step": 321
},
{
"epoch": 0.37139561707035756,
"grad_norm": 0.6514052160472804,
"learning_rate": 1.85163301328008e-05,
"loss": 0.16930492222309113,
"step": 322
},
{
"epoch": 0.37254901960784315,
"grad_norm": 0.6391733192871718,
"learning_rate": 1.8506708891157162e-05,
"loss": 0.17864246666431427,
"step": 323
},
{
"epoch": 0.3737024221453287,
"grad_norm": 0.6909009148124259,
"learning_rate": 1.8497059071005014e-05,
"loss": 0.13533613085746765,
"step": 324
},
{
"epoch": 0.3748558246828143,
"grad_norm": 0.6090850311140993,
"learning_rate": 1.8487380704763166e-05,
"loss": 0.16766279935836792,
"step": 325
},
{
"epoch": 0.3760092272202999,
"grad_norm": 0.7075638436579788,
"learning_rate": 1.8477673824946343e-05,
"loss": 0.1420484334230423,
"step": 326
},
{
"epoch": 0.3771626297577855,
"grad_norm": 0.5149551166058534,
"learning_rate": 1.8467938464165063e-05,
"loss": 0.14712229371070862,
"step": 327
},
{
"epoch": 0.37831603229527105,
"grad_norm": 0.7038759359157707,
"learning_rate": 1.8458174655125515e-05,
"loss": 0.1626979410648346,
"step": 328
},
{
"epoch": 0.37946943483275664,
"grad_norm": 0.5357980684471451,
"learning_rate": 1.8448382430629475e-05,
"loss": 0.1347598135471344,
"step": 329
},
{
"epoch": 0.3806228373702422,
"grad_norm": 0.6578490546518286,
"learning_rate": 1.8438561823574166e-05,
"loss": 0.1374199092388153,
"step": 330
},
{
"epoch": 0.3817762399077278,
"grad_norm": 0.6108853275109521,
"learning_rate": 1.8428712866952182e-05,
"loss": 0.14258575439453125,
"step": 331
},
{
"epoch": 0.3829296424452134,
"grad_norm": 0.5312639647319896,
"learning_rate": 1.8418835593851336e-05,
"loss": 0.13776928186416626,
"step": 332
},
{
"epoch": 0.38408304498269896,
"grad_norm": 0.5717372763551319,
"learning_rate": 1.8408930037454592e-05,
"loss": 0.1634255349636078,
"step": 333
},
{
"epoch": 0.38523644752018454,
"grad_norm": 0.6518585322816427,
"learning_rate": 1.8398996231039916e-05,
"loss": 0.16223083436489105,
"step": 334
},
{
"epoch": 0.3863898500576701,
"grad_norm": 0.6423659119475283,
"learning_rate": 1.8389034207980195e-05,
"loss": 0.16551171243190765,
"step": 335
},
{
"epoch": 0.3875432525951557,
"grad_norm": 0.6448976487869127,
"learning_rate": 1.8379044001743094e-05,
"loss": 0.15386125445365906,
"step": 336
},
{
"epoch": 0.3886966551326413,
"grad_norm": 0.6316838041901517,
"learning_rate": 1.8369025645890976e-05,
"loss": 0.17140522599220276,
"step": 337
},
{
"epoch": 0.38985005767012687,
"grad_norm": 0.6567665761506373,
"learning_rate": 1.8358979174080765e-05,
"loss": 0.1711106300354004,
"step": 338
},
{
"epoch": 0.39100346020761245,
"grad_norm": 0.509547386008856,
"learning_rate": 1.8348904620063843e-05,
"loss": 0.16814981400966644,
"step": 339
},
{
"epoch": 0.39215686274509803,
"grad_norm": 0.5649473931155227,
"learning_rate": 1.8338802017685933e-05,
"loss": 0.1602228581905365,
"step": 340
},
{
"epoch": 0.3933102652825836,
"grad_norm": 0.5549737006835778,
"learning_rate": 1.8328671400886993e-05,
"loss": 0.1781570464372635,
"step": 341
},
{
"epoch": 0.3944636678200692,
"grad_norm": 0.6323738856472507,
"learning_rate": 1.8318512803701087e-05,
"loss": 0.13819089531898499,
"step": 342
},
{
"epoch": 0.3956170703575548,
"grad_norm": 0.5808729876117982,
"learning_rate": 1.8308326260256287e-05,
"loss": 0.19749534130096436,
"step": 343
},
{
"epoch": 0.39677047289504036,
"grad_norm": 0.5704876025053673,
"learning_rate": 1.829811180477455e-05,
"loss": 0.1375598907470703,
"step": 344
},
{
"epoch": 0.39792387543252594,
"grad_norm": 0.5664136613446569,
"learning_rate": 1.82878694715716e-05,
"loss": 0.16664184629917145,
"step": 345
},
{
"epoch": 0.3990772779700115,
"grad_norm": 0.5415734462200937,
"learning_rate": 1.8277599295056826e-05,
"loss": 0.1529524028301239,
"step": 346
},
{
"epoch": 0.4002306805074971,
"grad_norm": 0.5233547182141234,
"learning_rate": 1.826730130973315e-05,
"loss": 0.16251197457313538,
"step": 347
},
{
"epoch": 0.4013840830449827,
"grad_norm": 0.6001653603365329,
"learning_rate": 1.8256975550196915e-05,
"loss": 0.17979922890663147,
"step": 348
},
{
"epoch": 0.40253748558246827,
"grad_norm": 0.5507780005481694,
"learning_rate": 1.824662205113778e-05,
"loss": 0.16273164749145508,
"step": 349
},
{
"epoch": 0.40369088811995385,
"grad_norm": 0.484714637634931,
"learning_rate": 1.8236240847338597e-05,
"loss": 0.13890139758586884,
"step": 350
},
{
"epoch": 0.40484429065743943,
"grad_norm": 0.6127094837879373,
"learning_rate": 1.8225831973675285e-05,
"loss": 0.16631168127059937,
"step": 351
},
{
"epoch": 0.405997693194925,
"grad_norm": 0.532735252248148,
"learning_rate": 1.8215395465116726e-05,
"loss": 0.1492110788822174,
"step": 352
},
{
"epoch": 0.4071510957324106,
"grad_norm": 0.6272429244921338,
"learning_rate": 1.8204931356724636e-05,
"loss": 0.15365289151668549,
"step": 353
},
{
"epoch": 0.4083044982698962,
"grad_norm": 0.5427028642278637,
"learning_rate": 1.819443968365346e-05,
"loss": 0.16043487191200256,
"step": 354
},
{
"epoch": 0.40945790080738176,
"grad_norm": 0.5915601581739729,
"learning_rate": 1.8183920481150247e-05,
"loss": 0.20037689805030823,
"step": 355
},
{
"epoch": 0.41061130334486734,
"grad_norm": 0.6474743675745677,
"learning_rate": 1.817337378455453e-05,
"loss": 0.14570726454257965,
"step": 356
},
{
"epoch": 0.4117647058823529,
"grad_norm": 0.5768676971896113,
"learning_rate": 1.8162799629298207e-05,
"loss": 0.16605910658836365,
"step": 357
},
{
"epoch": 0.4129181084198385,
"grad_norm": 0.6868191738360956,
"learning_rate": 1.8152198050905426e-05,
"loss": 0.17490999400615692,
"step": 358
},
{
"epoch": 0.4140715109573241,
"grad_norm": 0.6147909751635406,
"learning_rate": 1.8141569084992466e-05,
"loss": 0.16756336390972137,
"step": 359
},
{
"epoch": 0.41522491349480967,
"grad_norm": 0.6252677134906619,
"learning_rate": 1.8130912767267613e-05,
"loss": 0.1820719689130783,
"step": 360
},
{
"epoch": 0.41637831603229525,
"grad_norm": 0.6786390548801235,
"learning_rate": 1.8120229133531037e-05,
"loss": 0.14838874340057373,
"step": 361
},
{
"epoch": 0.41753171856978083,
"grad_norm": 0.6077230227246183,
"learning_rate": 1.810951821967468e-05,
"loss": 0.1385819911956787,
"step": 362
},
{
"epoch": 0.4186851211072664,
"grad_norm": 0.5131208699071214,
"learning_rate": 1.8098780061682148e-05,
"loss": 0.12897606194019318,
"step": 363
},
{
"epoch": 0.419838523644752,
"grad_norm": 0.6534967117809036,
"learning_rate": 1.8088014695628547e-05,
"loss": 0.14001844823360443,
"step": 364
},
{
"epoch": 0.4209919261822376,
"grad_norm": 0.5147387519995673,
"learning_rate": 1.8077222157680408e-05,
"loss": 0.14925754070281982,
"step": 365
},
{
"epoch": 0.42214532871972316,
"grad_norm": 0.6170624591352436,
"learning_rate": 1.8066402484095544e-05,
"loss": 0.1398134082555771,
"step": 366
},
{
"epoch": 0.4232987312572088,
"grad_norm": 0.561778445999366,
"learning_rate": 1.8055555711222924e-05,
"loss": 0.14288640022277832,
"step": 367
},
{
"epoch": 0.4244521337946944,
"grad_norm": 0.5672430429172519,
"learning_rate": 1.804468187550256e-05,
"loss": 0.14533036947250366,
"step": 368
},
{
"epoch": 0.42560553633217996,
"grad_norm": 0.5218767822241767,
"learning_rate": 1.803378101346539e-05,
"loss": 0.15439116954803467,
"step": 369
},
{
"epoch": 0.42675893886966554,
"grad_norm": 0.5491970357850229,
"learning_rate": 1.8022853161733136e-05,
"loss": 0.16101045906543732,
"step": 370
},
{
"epoch": 0.4279123414071511,
"grad_norm": 0.5487132286130078,
"learning_rate": 1.80118983570182e-05,
"loss": 0.14062638580799103,
"step": 371
},
{
"epoch": 0.4290657439446367,
"grad_norm": 0.5512738712431802,
"learning_rate": 1.800091663612354e-05,
"loss": 0.1546398252248764,
"step": 372
},
{
"epoch": 0.4302191464821223,
"grad_norm": 0.5904216150903602,
"learning_rate": 1.798990803594252e-05,
"loss": 0.13248053193092346,
"step": 373
},
{
"epoch": 0.43137254901960786,
"grad_norm": 0.4870722542687213,
"learning_rate": 1.7978872593458823e-05,
"loss": 0.12665371596813202,
"step": 374
},
{
"epoch": 0.43252595155709345,
"grad_norm": 0.5560981805406187,
"learning_rate": 1.7967810345746305e-05,
"loss": 0.1362338811159134,
"step": 375
},
{
"epoch": 0.43367935409457903,
"grad_norm": 0.5066381656967647,
"learning_rate": 1.795672132996887e-05,
"loss": 0.15236693620681763,
"step": 376
},
{
"epoch": 0.4348327566320646,
"grad_norm": 0.5532836507064487,
"learning_rate": 1.794560558338035e-05,
"loss": 0.19554941356182098,
"step": 377
},
{
"epoch": 0.4359861591695502,
"grad_norm": 0.9071793188245438,
"learning_rate": 1.7934463143324386e-05,
"loss": 0.15714818239212036,
"step": 378
},
{
"epoch": 0.4371395617070358,
"grad_norm": 0.6259103899151703,
"learning_rate": 1.7923294047234296e-05,
"loss": 0.13867619633674622,
"step": 379
},
{
"epoch": 0.43829296424452135,
"grad_norm": 0.6323637236022777,
"learning_rate": 1.7912098332632947e-05,
"loss": 0.15434211492538452,
"step": 380
},
{
"epoch": 0.43944636678200694,
"grad_norm": 0.6666879339469248,
"learning_rate": 1.790087603713263e-05,
"loss": 0.13556981086730957,
"step": 381
},
{
"epoch": 0.4405997693194925,
"grad_norm": 0.5255720231855027,
"learning_rate": 1.7889627198434937e-05,
"loss": 0.17278407514095306,
"step": 382
},
{
"epoch": 0.4417531718569781,
"grad_norm": 0.6127939213794201,
"learning_rate": 1.7878351854330635e-05,
"loss": 0.2017221450805664,
"step": 383
},
{
"epoch": 0.4429065743944637,
"grad_norm": 0.6426642184845965,
"learning_rate": 1.7867050042699527e-05,
"loss": 0.1491549015045166,
"step": 384
},
{
"epoch": 0.44405997693194926,
"grad_norm": 0.5735648612882913,
"learning_rate": 1.7855721801510347e-05,
"loss": 0.14504510164260864,
"step": 385
},
{
"epoch": 0.44521337946943484,
"grad_norm": 0.59670769589012,
"learning_rate": 1.7844367168820617e-05,
"loss": 0.20007307827472687,
"step": 386
},
{
"epoch": 0.4463667820069204,
"grad_norm": 0.6722822530407403,
"learning_rate": 1.783298618277652e-05,
"loss": 0.13457295298576355,
"step": 387
},
{
"epoch": 0.447520184544406,
"grad_norm": 0.5723154595955655,
"learning_rate": 1.782157888161276e-05,
"loss": 0.1361749768257141,
"step": 388
},
{
"epoch": 0.4486735870818916,
"grad_norm": 0.5933765975571021,
"learning_rate": 1.781014530365247e-05,
"loss": 0.13172583281993866,
"step": 389
},
{
"epoch": 0.44982698961937717,
"grad_norm": 0.5595725805936463,
"learning_rate": 1.7798685487307055e-05,
"loss": 0.1408577859401703,
"step": 390
},
{
"epoch": 0.45098039215686275,
"grad_norm": 0.5386623717885152,
"learning_rate": 1.7787199471076054e-05,
"loss": 0.13466668128967285,
"step": 391
},
{
"epoch": 0.45213379469434833,
"grad_norm": 0.616876225588993,
"learning_rate": 1.7775687293547045e-05,
"loss": 0.17032642662525177,
"step": 392
},
{
"epoch": 0.4532871972318339,
"grad_norm": 0.5060959550264947,
"learning_rate": 1.776414899339548e-05,
"loss": 0.14022529125213623,
"step": 393
},
{
"epoch": 0.4544405997693195,
"grad_norm": 0.6187093775149125,
"learning_rate": 1.775258460938458e-05,
"loss": 0.19888664782047272,
"step": 394
},
{
"epoch": 0.4555940023068051,
"grad_norm": 0.5469243152898912,
"learning_rate": 1.7740994180365185e-05,
"loss": 0.1494068205356598,
"step": 395
},
{
"epoch": 0.45674740484429066,
"grad_norm": 0.5549945925107316,
"learning_rate": 1.772937774527565e-05,
"loss": 0.17779028415679932,
"step": 396
},
{
"epoch": 0.45790080738177624,
"grad_norm": 0.5356630783087206,
"learning_rate": 1.7717735343141687e-05,
"loss": 0.14874984323978424,
"step": 397
},
{
"epoch": 0.4590542099192618,
"grad_norm": 0.5784401696611337,
"learning_rate": 1.770606701307625e-05,
"loss": 0.19778814911842346,
"step": 398
},
{
"epoch": 0.4602076124567474,
"grad_norm": 0.5749195806630031,
"learning_rate": 1.7694372794279388e-05,
"loss": 0.16131365299224854,
"step": 399
},
{
"epoch": 0.461361014994233,
"grad_norm": 0.5166305494211104,
"learning_rate": 1.7682652726038134e-05,
"loss": 0.1501253843307495,
"step": 400
},
{
"epoch": 0.46251441753171857,
"grad_norm": 0.590451513359013,
"learning_rate": 1.7670906847726363e-05,
"loss": 0.13478678464889526,
"step": 401
},
{
"epoch": 0.46366782006920415,
"grad_norm": 0.5223232636743267,
"learning_rate": 1.7659135198804654e-05,
"loss": 0.1519884467124939,
"step": 402
},
{
"epoch": 0.46482122260668973,
"grad_norm": 0.6456534120492474,
"learning_rate": 1.7647337818820168e-05,
"loss": 0.1633578985929489,
"step": 403
},
{
"epoch": 0.4659746251441753,
"grad_norm": 0.5707995506974195,
"learning_rate": 1.7635514747406507e-05,
"loss": 0.16134953498840332,
"step": 404
},
{
"epoch": 0.4671280276816609,
"grad_norm": 0.5627966144257297,
"learning_rate": 1.7623666024283587e-05,
"loss": 0.1460132598876953,
"step": 405
},
{
"epoch": 0.4682814302191465,
"grad_norm": 0.5537969912885383,
"learning_rate": 1.7611791689257497e-05,
"loss": 0.12994131445884705,
"step": 406
},
{
"epoch": 0.46943483275663206,
"grad_norm": 0.5689783457344849,
"learning_rate": 1.7599891782220378e-05,
"loss": 0.14559157192707062,
"step": 407
},
{
"epoch": 0.47058823529411764,
"grad_norm": 0.6303674894841822,
"learning_rate": 1.7587966343150267e-05,
"loss": 0.12237685918807983,
"step": 408
},
{
"epoch": 0.4717416378316032,
"grad_norm": 0.4950439887286743,
"learning_rate": 1.7576015412110992e-05,
"loss": 0.14725524187088013,
"step": 409
},
{
"epoch": 0.4728950403690888,
"grad_norm": 0.5793020374820799,
"learning_rate": 1.756403902925201e-05,
"loss": 0.15199531614780426,
"step": 410
},
{
"epoch": 0.4740484429065744,
"grad_norm": 0.5300021678901531,
"learning_rate": 1.7552037234808296e-05,
"loss": 0.1341474950313568,
"step": 411
},
{
"epoch": 0.47520184544405997,
"grad_norm": 0.5243138259238604,
"learning_rate": 1.754001006910018e-05,
"loss": 0.1351647973060608,
"step": 412
},
{
"epoch": 0.47635524798154555,
"grad_norm": 0.6132285217078636,
"learning_rate": 1.7527957572533243e-05,
"loss": 0.19658911228179932,
"step": 413
},
{
"epoch": 0.47750865051903113,
"grad_norm": 0.6096593878290506,
"learning_rate": 1.7515879785598154e-05,
"loss": 0.1873995065689087,
"step": 414
},
{
"epoch": 0.4786620530565167,
"grad_norm": 0.5404931372810174,
"learning_rate": 1.7503776748870554e-05,
"loss": 0.1355239450931549,
"step": 415
},
{
"epoch": 0.4798154555940023,
"grad_norm": 0.5231049640469617,
"learning_rate": 1.749164850301091e-05,
"loss": 0.15796911716461182,
"step": 416
},
{
"epoch": 0.4809688581314879,
"grad_norm": 0.7403124896998432,
"learning_rate": 1.7479495088764368e-05,
"loss": 0.13449671864509583,
"step": 417
},
{
"epoch": 0.48212226066897346,
"grad_norm": 0.5115671865218216,
"learning_rate": 1.7467316546960648e-05,
"loss": 0.1528121829032898,
"step": 418
},
{
"epoch": 0.48327566320645904,
"grad_norm": 0.5912951041291811,
"learning_rate": 1.7455112918513872e-05,
"loss": 0.13427142798900604,
"step": 419
},
{
"epoch": 0.4844290657439446,
"grad_norm": 0.5469561442758858,
"learning_rate": 1.7442884244422448e-05,
"loss": 0.17089733481407166,
"step": 420
},
{
"epoch": 0.4855824682814302,
"grad_norm": 0.5201342254379955,
"learning_rate": 1.7430630565768916e-05,
"loss": 0.1518099009990692,
"step": 421
},
{
"epoch": 0.4867358708189158,
"grad_norm": 0.5000689377955542,
"learning_rate": 1.741835192371983e-05,
"loss": 0.13482986390590668,
"step": 422
},
{
"epoch": 0.48788927335640137,
"grad_norm": 0.49241332366094487,
"learning_rate": 1.7406048359525608e-05,
"loss": 0.16338984668254852,
"step": 423
},
{
"epoch": 0.48904267589388695,
"grad_norm": 0.509700585188127,
"learning_rate": 1.739371991452039e-05,
"loss": 0.15633586049079895,
"step": 424
},
{
"epoch": 0.49019607843137253,
"grad_norm": 0.49838307329547027,
"learning_rate": 1.7381366630121904e-05,
"loss": 0.1475672423839569,
"step": 425
},
{
"epoch": 0.4913494809688581,
"grad_norm": 0.5495990152846041,
"learning_rate": 1.7368988547831335e-05,
"loss": 0.17451348900794983,
"step": 426
},
{
"epoch": 0.4925028835063437,
"grad_norm": 0.5010970042568553,
"learning_rate": 1.7356585709233163e-05,
"loss": 0.12543466687202454,
"step": 427
},
{
"epoch": 0.4936562860438293,
"grad_norm": 0.5125572744867564,
"learning_rate": 1.7344158155995047e-05,
"loss": 0.13997386395931244,
"step": 428
},
{
"epoch": 0.49480968858131485,
"grad_norm": 0.5123389498965938,
"learning_rate": 1.7331705929867676e-05,
"loss": 0.14170202612876892,
"step": 429
},
{
"epoch": 0.49596309111880044,
"grad_norm": 0.6146771224585806,
"learning_rate": 1.7319229072684623e-05,
"loss": 0.1580689251422882,
"step": 430
},
{
"epoch": 0.497116493656286,
"grad_norm": 0.5136609821480289,
"learning_rate": 1.7306727626362215e-05,
"loss": 0.15322911739349365,
"step": 431
},
{
"epoch": 0.4982698961937716,
"grad_norm": 0.5396210984688472,
"learning_rate": 1.7294201632899383e-05,
"loss": 0.13010656833648682,
"step": 432
},
{
"epoch": 0.4994232987312572,
"grad_norm": 0.5385182617726646,
"learning_rate": 1.7281651134377525e-05,
"loss": 0.15869855880737305,
"step": 433
},
{
"epoch": 0.5005767012687428,
"grad_norm": 0.5229599653725078,
"learning_rate": 1.726907617296037e-05,
"loss": 0.1403420865535736,
"step": 434
},
{
"epoch": 0.5017301038062284,
"grad_norm": 0.609385079413866,
"learning_rate": 1.7256476790893817e-05,
"loss": 0.13169164955615997,
"step": 435
},
{
"epoch": 0.5028835063437139,
"grad_norm": 0.5005388120496388,
"learning_rate": 1.724385303050582e-05,
"loss": 0.13182707130908966,
"step": 436
},
{
"epoch": 0.5040369088811996,
"grad_norm": 0.7508475702578633,
"learning_rate": 1.7231204934206234e-05,
"loss": 0.1548827886581421,
"step": 437
},
{
"epoch": 0.5051903114186851,
"grad_norm": 0.5972607914123017,
"learning_rate": 1.7218532544486662e-05,
"loss": 0.14812692999839783,
"step": 438
},
{
"epoch": 0.5063437139561707,
"grad_norm": 0.6209667109581565,
"learning_rate": 1.7205835903920323e-05,
"loss": 0.16896986961364746,
"step": 439
},
{
"epoch": 0.5074971164936563,
"grad_norm": 0.5824894858833444,
"learning_rate": 1.719311505516191e-05,
"loss": 0.1277136206626892,
"step": 440
},
{
"epoch": 0.5086505190311419,
"grad_norm": 0.56479919677272,
"learning_rate": 1.7180370040947443e-05,
"loss": 0.14347553253173828,
"step": 441
},
{
"epoch": 0.5098039215686274,
"grad_norm": 0.6516735076984296,
"learning_rate": 1.716760090409413e-05,
"loss": 0.16942009329795837,
"step": 442
},
{
"epoch": 0.510957324106113,
"grad_norm": 0.5601824858382677,
"learning_rate": 1.715480768750021e-05,
"loss": 0.14430741965770721,
"step": 443
},
{
"epoch": 0.5121107266435986,
"grad_norm": 0.5066565852392919,
"learning_rate": 1.7141990434144835e-05,
"loss": 0.1629716008901596,
"step": 444
},
{
"epoch": 0.5132641291810842,
"grad_norm": 0.5177132030337044,
"learning_rate": 1.712914918708789e-05,
"loss": 0.169794961810112,
"step": 445
},
{
"epoch": 0.5144175317185697,
"grad_norm": 0.5919431526672617,
"learning_rate": 1.7116283989469883e-05,
"loss": 0.16447743773460388,
"step": 446
},
{
"epoch": 0.5155709342560554,
"grad_norm": 0.5119353412151587,
"learning_rate": 1.7103394884511773e-05,
"loss": 0.1600750982761383,
"step": 447
},
{
"epoch": 0.5167243367935409,
"grad_norm": 0.5257803628068568,
"learning_rate": 1.7090481915514843e-05,
"loss": 0.16211356222629547,
"step": 448
},
{
"epoch": 0.5178777393310265,
"grad_norm": 0.5134868045361936,
"learning_rate": 1.7077545125860548e-05,
"loss": 0.16266633570194244,
"step": 449
},
{
"epoch": 0.5190311418685121,
"grad_norm": 0.5703141457148184,
"learning_rate": 1.7064584559010366e-05,
"loss": 0.17013393342494965,
"step": 450
},
{
"epoch": 0.5201845444059977,
"grad_norm": 0.5080239431931232,
"learning_rate": 1.7051600258505658e-05,
"loss": 0.15469172596931458,
"step": 451
},
{
"epoch": 0.5213379469434832,
"grad_norm": 0.5562279210986152,
"learning_rate": 1.7038592267967514e-05,
"loss": 0.17757435142993927,
"step": 452
},
{
"epoch": 0.5224913494809689,
"grad_norm": 0.563969328201003,
"learning_rate": 1.7025560631096614e-05,
"loss": 0.14963868260383606,
"step": 453
},
{
"epoch": 0.5236447520184544,
"grad_norm": 0.5164609552755912,
"learning_rate": 1.7012505391673087e-05,
"loss": 0.13683462142944336,
"step": 454
},
{
"epoch": 0.52479815455594,
"grad_norm": 0.5235506604885183,
"learning_rate": 1.6999426593556342e-05,
"loss": 0.14717784523963928,
"step": 455
},
{
"epoch": 0.5259515570934256,
"grad_norm": 0.49301637174294705,
"learning_rate": 1.698632428068494e-05,
"loss": 0.15806545317173004,
"step": 456
},
{
"epoch": 0.5271049596309112,
"grad_norm": 0.5046932747874421,
"learning_rate": 1.6973198497076437e-05,
"loss": 0.15590038895606995,
"step": 457
},
{
"epoch": 0.5282583621683967,
"grad_norm": 0.47235048046147915,
"learning_rate": 1.6960049286827248e-05,
"loss": 0.1356411576271057,
"step": 458
},
{
"epoch": 0.5294117647058824,
"grad_norm": 0.4994423465676713,
"learning_rate": 1.6946876694112478e-05,
"loss": 0.15324820578098297,
"step": 459
},
{
"epoch": 0.5305651672433679,
"grad_norm": 0.4815505791552659,
"learning_rate": 1.6933680763185802e-05,
"loss": 0.12883895635604858,
"step": 460
},
{
"epoch": 0.5317185697808535,
"grad_norm": 0.49420061779042523,
"learning_rate": 1.6920461538379286e-05,
"loss": 0.14862468838691711,
"step": 461
},
{
"epoch": 0.532871972318339,
"grad_norm": 0.49939245756691797,
"learning_rate": 1.6907219064103255e-05,
"loss": 0.15597042441368103,
"step": 462
},
{
"epoch": 0.5340253748558247,
"grad_norm": 0.5083477543161224,
"learning_rate": 1.689395338484615e-05,
"loss": 0.13210950791835785,
"step": 463
},
{
"epoch": 0.5351787773933102,
"grad_norm": 0.5064737720590998,
"learning_rate": 1.6880664545174356e-05,
"loss": 0.134365051984787,
"step": 464
},
{
"epoch": 0.5363321799307958,
"grad_norm": 0.5251521042595475,
"learning_rate": 1.686735258973208e-05,
"loss": 0.1622808873653412,
"step": 465
},
{
"epoch": 0.5374855824682814,
"grad_norm": 0.5322571868962346,
"learning_rate": 1.685401756324118e-05,
"loss": 0.14772941172122955,
"step": 466
},
{
"epoch": 0.538638985005767,
"grad_norm": 0.5251645980020719,
"learning_rate": 1.684065951050102e-05,
"loss": 0.16897282004356384,
"step": 467
},
{
"epoch": 0.5397923875432526,
"grad_norm": 0.5680238369995724,
"learning_rate": 1.682727847638832e-05,
"loss": 0.1572577804327011,
"step": 468
},
{
"epoch": 0.5409457900807382,
"grad_norm": 0.5059706981011328,
"learning_rate": 1.6813874505857018e-05,
"loss": 0.1252511441707611,
"step": 469
},
{
"epoch": 0.5420991926182238,
"grad_norm": 0.5671929804845061,
"learning_rate": 1.6800447643938097e-05,
"loss": 0.16724829375743866,
"step": 470
},
{
"epoch": 0.5432525951557093,
"grad_norm": 0.6185247911645135,
"learning_rate": 1.6786997935739444e-05,
"loss": 0.12754085659980774,
"step": 471
},
{
"epoch": 0.544405997693195,
"grad_norm": 0.5792159310806154,
"learning_rate": 1.6773525426445706e-05,
"loss": 0.144723042845726,
"step": 472
},
{
"epoch": 0.5455594002306805,
"grad_norm": 0.6034427081641701,
"learning_rate": 1.676003016131812e-05,
"loss": 0.12894895672798157,
"step": 473
},
{
"epoch": 0.5467128027681661,
"grad_norm": 0.5881141794695095,
"learning_rate": 1.6746512185694385e-05,
"loss": 0.18118900060653687,
"step": 474
},
{
"epoch": 0.5478662053056517,
"grad_norm": 0.6886850857698014,
"learning_rate": 1.6732971544988487e-05,
"loss": 0.15676993131637573,
"step": 475
},
{
"epoch": 0.5490196078431373,
"grad_norm": 0.5492751063089862,
"learning_rate": 1.671940828469056e-05,
"loss": 0.16350331902503967,
"step": 476
},
{
"epoch": 0.5501730103806228,
"grad_norm": 0.7861294255441532,
"learning_rate": 1.6705822450366726e-05,
"loss": 0.17945683002471924,
"step": 477
},
{
"epoch": 0.5513264129181085,
"grad_norm": 0.6157782510220919,
"learning_rate": 1.669221408765894e-05,
"loss": 0.13672518730163574,
"step": 478
},
{
"epoch": 0.552479815455594,
"grad_norm": 0.6117788102482531,
"learning_rate": 1.6678583242284864e-05,
"loss": 0.16273924708366394,
"step": 479
},
{
"epoch": 0.5536332179930796,
"grad_norm": 0.6404019486139891,
"learning_rate": 1.6664929960037662e-05,
"loss": 0.15442579984664917,
"step": 480
},
{
"epoch": 0.5547866205305652,
"grad_norm": 0.5141432568034691,
"learning_rate": 1.6651254286785893e-05,
"loss": 0.139705628156662,
"step": 481
},
{
"epoch": 0.5559400230680508,
"grad_norm": 0.6589879011157578,
"learning_rate": 1.6637556268473338e-05,
"loss": 0.14485619962215424,
"step": 482
},
{
"epoch": 0.5570934256055363,
"grad_norm": 0.8764652457597896,
"learning_rate": 1.6623835951118843e-05,
"loss": 0.13132694363594055,
"step": 483
},
{
"epoch": 0.558246828143022,
"grad_norm": 0.6438072562698549,
"learning_rate": 1.6610093380816167e-05,
"loss": 0.18907666206359863,
"step": 484
},
{
"epoch": 0.5594002306805075,
"grad_norm": 0.6955214423181818,
"learning_rate": 1.6596328603733828e-05,
"loss": 0.14733359217643738,
"step": 485
},
{
"epoch": 0.5605536332179931,
"grad_norm": 0.5491181307064507,
"learning_rate": 1.6582541666114954e-05,
"loss": 0.1575850546360016,
"step": 486
},
{
"epoch": 0.5617070357554786,
"grad_norm": 0.6228834673953163,
"learning_rate": 1.656873261427712e-05,
"loss": 0.15319159626960754,
"step": 487
},
{
"epoch": 0.5628604382929643,
"grad_norm": 0.5709353561488977,
"learning_rate": 1.655490149461219e-05,
"loss": 0.13675610721111298,
"step": 488
},
{
"epoch": 0.5640138408304498,
"grad_norm": 0.631645594863702,
"learning_rate": 1.654104835358617e-05,
"loss": 0.14342458546161652,
"step": 489
},
{
"epoch": 0.5651672433679354,
"grad_norm": 0.5964646801877486,
"learning_rate": 1.652717323773904e-05,
"loss": 0.1375924050807953,
"step": 490
},
{
"epoch": 0.566320645905421,
"grad_norm": 0.6047339189305303,
"learning_rate": 1.6513276193684618e-05,
"loss": 0.13377784192562103,
"step": 491
},
{
"epoch": 0.5674740484429066,
"grad_norm": 0.5404104591242939,
"learning_rate": 1.6499357268110382e-05,
"loss": 0.15617550909519196,
"step": 492
},
{
"epoch": 0.5686274509803921,
"grad_norm": 0.6292652961591134,
"learning_rate": 1.6485416507777316e-05,
"loss": 0.1749780923128128,
"step": 493
},
{
"epoch": 0.5697808535178778,
"grad_norm": 0.5304462340817008,
"learning_rate": 1.647145395951977e-05,
"loss": 0.13451187312602997,
"step": 494
},
{
"epoch": 0.5709342560553633,
"grad_norm": 0.6730638976392467,
"learning_rate": 1.6457469670245284e-05,
"loss": 0.13611406087875366,
"step": 495
},
{
"epoch": 0.5720876585928489,
"grad_norm": 0.501762486687127,
"learning_rate": 1.6443463686934445e-05,
"loss": 0.14352793991565704,
"step": 496
},
{
"epoch": 0.5732410611303345,
"grad_norm": 0.5499176462212256,
"learning_rate": 1.6429436056640708e-05,
"loss": 0.12907668948173523,
"step": 497
},
{
"epoch": 0.5743944636678201,
"grad_norm": 0.5623427319529885,
"learning_rate": 1.6415386826490264e-05,
"loss": 0.19299966096878052,
"step": 498
},
{
"epoch": 0.5755478662053056,
"grad_norm": 0.5157042408548885,
"learning_rate": 1.6401316043681867e-05,
"loss": 0.13448208570480347,
"step": 499
},
{
"epoch": 0.5767012687427913,
"grad_norm": 0.4752457867177338,
"learning_rate": 1.6387223755486674e-05,
"loss": 0.11987307667732239,
"step": 500
},
{
"epoch": 0.5778546712802768,
"grad_norm": 0.5120938696100806,
"learning_rate": 1.6373110009248087e-05,
"loss": 0.12943559885025024,
"step": 501
},
{
"epoch": 0.5790080738177624,
"grad_norm": 0.532690338033902,
"learning_rate": 1.6358974852381607e-05,
"loss": 0.17673994600772858,
"step": 502
},
{
"epoch": 0.580161476355248,
"grad_norm": 0.5623065663191178,
"learning_rate": 1.634481833237466e-05,
"loss": 0.13277395069599152,
"step": 503
},
{
"epoch": 0.5813148788927336,
"grad_norm": 0.4784148765658445,
"learning_rate": 1.633064049678643e-05,
"loss": 0.12811553478240967,
"step": 504
},
{
"epoch": 0.5824682814302191,
"grad_norm": 0.5131296691651874,
"learning_rate": 1.631644139324774e-05,
"loss": 0.13314424455165863,
"step": 505
},
{
"epoch": 0.5836216839677048,
"grad_norm": 0.5681965057057857,
"learning_rate": 1.630222106946083e-05,
"loss": 0.19336840510368347,
"step": 506
},
{
"epoch": 0.5847750865051903,
"grad_norm": 0.5137386434712415,
"learning_rate": 1.6287979573199257e-05,
"loss": 0.14501304924488068,
"step": 507
},
{
"epoch": 0.5859284890426759,
"grad_norm": 0.5605418587245659,
"learning_rate": 1.6273716952307687e-05,
"loss": 0.19734391570091248,
"step": 508
},
{
"epoch": 0.5870818915801614,
"grad_norm": 0.5265417141890952,
"learning_rate": 1.625943325470177e-05,
"loss": 0.12686699628829956,
"step": 509
},
{
"epoch": 0.5882352941176471,
"grad_norm": 0.4907316977168637,
"learning_rate": 1.6245128528367956e-05,
"loss": 0.14856919646263123,
"step": 510
},
{
"epoch": 0.5893886966551326,
"grad_norm": 0.524445087967102,
"learning_rate": 1.623080282136334e-05,
"loss": 0.13928894698619843,
"step": 511
},
{
"epoch": 0.5905420991926182,
"grad_norm": 0.580455383806871,
"learning_rate": 1.6216456181815506e-05,
"loss": 0.1829647719860077,
"step": 512
},
{
"epoch": 0.5916955017301038,
"grad_norm": 0.5953375438678165,
"learning_rate": 1.6202088657922365e-05,
"loss": 0.12916627526283264,
"step": 513
},
{
"epoch": 0.5928489042675894,
"grad_norm": 0.5088706116353038,
"learning_rate": 1.6187700297951982e-05,
"loss": 0.16603748500347137,
"step": 514
},
{
"epoch": 0.5940023068050749,
"grad_norm": 0.6486324524371444,
"learning_rate": 1.617329115024242e-05,
"loss": 0.1245475485920906,
"step": 515
},
{
"epoch": 0.5951557093425606,
"grad_norm": 0.5839872502168708,
"learning_rate": 1.6158861263201592e-05,
"loss": 0.13916662335395813,
"step": 516
},
{
"epoch": 0.5963091118800461,
"grad_norm": 0.5413976854346556,
"learning_rate": 1.614441068530707e-05,
"loss": 0.13750016689300537,
"step": 517
},
{
"epoch": 0.5974625144175317,
"grad_norm": 0.5945849086859157,
"learning_rate": 1.612993946510595e-05,
"loss": 0.1695195585489273,
"step": 518
},
{
"epoch": 0.5986159169550173,
"grad_norm": 0.5388970557414748,
"learning_rate": 1.6115447651214668e-05,
"loss": 0.12012791633605957,
"step": 519
},
{
"epoch": 0.5997693194925029,
"grad_norm": 0.5783037091235467,
"learning_rate": 1.6100935292318842e-05,
"loss": 0.12007131427526474,
"step": 520
},
{
"epoch": 0.6009227220299884,
"grad_norm": 0.5046815230724989,
"learning_rate": 1.6086402437173126e-05,
"loss": 0.15104244649410248,
"step": 521
},
{
"epoch": 0.6020761245674741,
"grad_norm": 0.6108327119155138,
"learning_rate": 1.607184913460101e-05,
"loss": 0.13571733236312866,
"step": 522
},
{
"epoch": 0.6032295271049596,
"grad_norm": 0.4891322589149242,
"learning_rate": 1.605727543349471e-05,
"loss": 0.13314920663833618,
"step": 523
},
{
"epoch": 0.6043829296424452,
"grad_norm": 0.6480602703804762,
"learning_rate": 1.6042681382814934e-05,
"loss": 0.1488885134458542,
"step": 524
},
{
"epoch": 0.6055363321799307,
"grad_norm": 0.49515559782504487,
"learning_rate": 1.6028067031590785e-05,
"loss": 0.1305495798587799,
"step": 525
},
{
"epoch": 0.6066897347174164,
"grad_norm": 0.5613758571333316,
"learning_rate": 1.6013432428919543e-05,
"loss": 0.15761324763298035,
"step": 526
},
{
"epoch": 0.6078431372549019,
"grad_norm": 0.719440728106589,
"learning_rate": 1.5998777623966536e-05,
"loss": 0.13021963834762573,
"step": 527
},
{
"epoch": 0.6089965397923875,
"grad_norm": 0.6448633202339258,
"learning_rate": 1.5984102665964968e-05,
"loss": 0.13886308670043945,
"step": 528
},
{
"epoch": 0.6101499423298731,
"grad_norm": 0.6856492575522682,
"learning_rate": 1.596940760421574e-05,
"loss": 0.14724163711071014,
"step": 529
},
{
"epoch": 0.6113033448673587,
"grad_norm": 0.5439850780926365,
"learning_rate": 1.595469248808728e-05,
"loss": 0.1788213551044464,
"step": 530
},
{
"epoch": 0.6124567474048442,
"grad_norm": 0.6216048283187898,
"learning_rate": 1.5939957367015414e-05,
"loss": 0.1568169891834259,
"step": 531
},
{
"epoch": 0.6136101499423299,
"grad_norm": 0.5335354607655224,
"learning_rate": 1.5925202290503153e-05,
"loss": 0.14316213130950928,
"step": 532
},
{
"epoch": 0.6147635524798154,
"grad_norm": 0.5306633865907578,
"learning_rate": 1.5910427308120564e-05,
"loss": 0.15020892024040222,
"step": 533
},
{
"epoch": 0.615916955017301,
"grad_norm": 0.6808119627634571,
"learning_rate": 1.589563246950458e-05,
"loss": 0.15737614035606384,
"step": 534
},
{
"epoch": 0.6170703575547867,
"grad_norm": 0.510290728587089,
"learning_rate": 1.5880817824358846e-05,
"loss": 0.11930564045906067,
"step": 535
},
{
"epoch": 0.6182237600922722,
"grad_norm": 0.5256930577377047,
"learning_rate": 1.586598342245354e-05,
"loss": 0.15343284606933594,
"step": 536
},
{
"epoch": 0.6193771626297578,
"grad_norm": 0.640843053670881,
"learning_rate": 1.5851129313625225e-05,
"loss": 0.13962233066558838,
"step": 537
},
{
"epoch": 0.6205305651672434,
"grad_norm": 0.48512946575182725,
"learning_rate": 1.583625554777666e-05,
"loss": 0.1539442539215088,
"step": 538
},
{
"epoch": 0.621683967704729,
"grad_norm": 0.9091375250132737,
"learning_rate": 1.582136217487665e-05,
"loss": 0.16035687923431396,
"step": 539
},
{
"epoch": 0.6228373702422145,
"grad_norm": 0.6361248214551573,
"learning_rate": 1.5806449244959864e-05,
"loss": 0.15062326192855835,
"step": 540
},
{
"epoch": 0.6239907727797002,
"grad_norm": 0.5795341332890258,
"learning_rate": 1.5791516808126674e-05,
"loss": 0.14406441152095795,
"step": 541
},
{
"epoch": 0.6251441753171857,
"grad_norm": 0.5158258241285123,
"learning_rate": 1.577656491454299e-05,
"loss": 0.13476890325546265,
"step": 542
},
{
"epoch": 0.6262975778546713,
"grad_norm": 0.49358926793391183,
"learning_rate": 1.5761593614440088e-05,
"loss": 0.14457282423973083,
"step": 543
},
{
"epoch": 0.6274509803921569,
"grad_norm": 0.5079736151909183,
"learning_rate": 1.5746602958114428e-05,
"loss": 0.13668256998062134,
"step": 544
},
{
"epoch": 0.6286043829296425,
"grad_norm": 0.4835605771669892,
"learning_rate": 1.573159299592752e-05,
"loss": 0.1446947157382965,
"step": 545
},
{
"epoch": 0.629757785467128,
"grad_norm": 0.5260157360973866,
"learning_rate": 1.571656377830571e-05,
"loss": 0.12934987246990204,
"step": 546
},
{
"epoch": 0.6309111880046137,
"grad_norm": 0.5581197651221572,
"learning_rate": 1.5701515355740048e-05,
"loss": 0.18074943125247955,
"step": 547
},
{
"epoch": 0.6320645905420992,
"grad_norm": 0.6025254040494772,
"learning_rate": 1.56864477787861e-05,
"loss": 0.12877404689788818,
"step": 548
},
{
"epoch": 0.6332179930795848,
"grad_norm": 0.4976275614372827,
"learning_rate": 1.567136109806378e-05,
"loss": 0.13635510206222534,
"step": 549
},
{
"epoch": 0.6343713956170703,
"grad_norm": 0.5231621751083734,
"learning_rate": 1.565625536425718e-05,
"loss": 0.14098894596099854,
"step": 550
},
{
"epoch": 0.635524798154556,
"grad_norm": 0.6388145368349771,
"learning_rate": 1.5641130628114403e-05,
"loss": 0.13117524981498718,
"step": 551
},
{
"epoch": 0.6366782006920415,
"grad_norm": 0.4749665988022428,
"learning_rate": 1.5625986940447394e-05,
"loss": 0.1493483930826187,
"step": 552
},
{
"epoch": 0.6378316032295271,
"grad_norm": 0.6239135620354836,
"learning_rate": 1.5610824352131758e-05,
"loss": 0.1407947540283203,
"step": 553
},
{
"epoch": 0.6389850057670127,
"grad_norm": 0.5211356683059194,
"learning_rate": 1.559564291410661e-05,
"loss": 0.15622186660766602,
"step": 554
},
{
"epoch": 0.6401384083044983,
"grad_norm": 0.5439645588974373,
"learning_rate": 1.5580442677374378e-05,
"loss": 0.16329573094844818,
"step": 555
},
{
"epoch": 0.6412918108419838,
"grad_norm": 0.5466176515472843,
"learning_rate": 1.5565223693000654e-05,
"loss": 0.1400141417980194,
"step": 556
},
{
"epoch": 0.6424452133794695,
"grad_norm": 0.5600380279978392,
"learning_rate": 1.5549986012114008e-05,
"loss": 0.15253818035125732,
"step": 557
},
{
"epoch": 0.643598615916955,
"grad_norm": 0.6590375057714741,
"learning_rate": 1.5534729685905822e-05,
"loss": 0.15048329532146454,
"step": 558
},
{
"epoch": 0.6447520184544406,
"grad_norm": 0.5676088491896859,
"learning_rate": 1.5519454765630123e-05,
"loss": 0.16173312067985535,
"step": 559
},
{
"epoch": 0.6459054209919262,
"grad_norm": 0.608170871148904,
"learning_rate": 1.550416130260339e-05,
"loss": 0.14939799904823303,
"step": 560
},
{
"epoch": 0.6470588235294118,
"grad_norm": 0.5618013541990615,
"learning_rate": 1.5488849348204417e-05,
"loss": 0.171311616897583,
"step": 561
},
{
"epoch": 0.6482122260668973,
"grad_norm": 0.5350643953554979,
"learning_rate": 1.5473518953874115e-05,
"loss": 0.15433278679847717,
"step": 562
},
{
"epoch": 0.649365628604383,
"grad_norm": 0.5461538358486903,
"learning_rate": 1.5458170171115328e-05,
"loss": 0.1339482069015503,
"step": 563
},
{
"epoch": 0.6505190311418685,
"grad_norm": 0.528031450301865,
"learning_rate": 1.544280305149269e-05,
"loss": 0.1293746829032898,
"step": 564
},
{
"epoch": 0.6516724336793541,
"grad_norm": 0.5471415737400961,
"learning_rate": 1.5427417646632447e-05,
"loss": 0.15067921578884125,
"step": 565
},
{
"epoch": 0.6528258362168397,
"grad_norm": 0.522070439832573,
"learning_rate": 1.5412014008222263e-05,
"loss": 0.16512134671211243,
"step": 566
},
{
"epoch": 0.6539792387543253,
"grad_norm": 0.49701358265268786,
"learning_rate": 1.5396592188011047e-05,
"loss": 0.17120397090911865,
"step": 567
},
{
"epoch": 0.6551326412918108,
"grad_norm": 0.5459954927277397,
"learning_rate": 1.5381152237808816e-05,
"loss": 0.1849687099456787,
"step": 568
},
{
"epoch": 0.6562860438292965,
"grad_norm": 0.5278338530829496,
"learning_rate": 1.5365694209486475e-05,
"loss": 0.14413192868232727,
"step": 569
},
{
"epoch": 0.657439446366782,
"grad_norm": 0.501584227039677,
"learning_rate": 1.5350218154975672e-05,
"loss": 0.15337622165679932,
"step": 570
},
{
"epoch": 0.6585928489042676,
"grad_norm": 0.5408968247163672,
"learning_rate": 1.533472412626861e-05,
"loss": 0.14485733211040497,
"step": 571
},
{
"epoch": 0.6597462514417531,
"grad_norm": 0.572354753695507,
"learning_rate": 1.5319212175417884e-05,
"loss": 0.1300359070301056,
"step": 572
},
{
"epoch": 0.6608996539792388,
"grad_norm": 0.4659276220217205,
"learning_rate": 1.5303682354536294e-05,
"loss": 0.14252139627933502,
"step": 573
},
{
"epoch": 0.6620530565167243,
"grad_norm": 0.49419884513805645,
"learning_rate": 1.528813471579667e-05,
"loss": 0.13386285305023193,
"step": 574
},
{
"epoch": 0.6632064590542099,
"grad_norm": 0.5147789880783188,
"learning_rate": 1.527256931143171e-05,
"loss": 0.1558796465396881,
"step": 575
},
{
"epoch": 0.6643598615916955,
"grad_norm": 0.5275232297542298,
"learning_rate": 1.5256986193733792e-05,
"loss": 0.1288929283618927,
"step": 576
},
{
"epoch": 0.6655132641291811,
"grad_norm": 0.5312240439822914,
"learning_rate": 1.52413854150548e-05,
"loss": 0.13220518827438354,
"step": 577
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.5110414826155746,
"learning_rate": 1.5225767027805956e-05,
"loss": 0.15769299864768982,
"step": 578
},
{
"epoch": 0.6678200692041523,
"grad_norm": 0.498460810716016,
"learning_rate": 1.5210131084457635e-05,
"loss": 0.12131652235984802,
"step": 579
},
{
"epoch": 0.6689734717416378,
"grad_norm": 0.49390710871071464,
"learning_rate": 1.5194477637539194e-05,
"loss": 0.11926689743995667,
"step": 580
},
{
"epoch": 0.6701268742791234,
"grad_norm": 0.5431560538943355,
"learning_rate": 1.5178806739638794e-05,
"loss": 0.12431739270687103,
"step": 581
},
{
"epoch": 0.671280276816609,
"grad_norm": 0.4969474380077296,
"learning_rate": 1.5163118443403218e-05,
"loss": 0.1454714834690094,
"step": 582
},
{
"epoch": 0.6724336793540946,
"grad_norm": 0.5446278468019289,
"learning_rate": 1.514741280153771e-05,
"loss": 0.12180251628160477,
"step": 583
},
{
"epoch": 0.6735870818915801,
"grad_norm": 0.5213849520042092,
"learning_rate": 1.513168986680577e-05,
"loss": 0.13198694586753845,
"step": 584
},
{
"epoch": 0.6747404844290658,
"grad_norm": 0.482705856545049,
"learning_rate": 1.511594969202901e-05,
"loss": 0.12537050247192383,
"step": 585
},
{
"epoch": 0.6758938869665513,
"grad_norm": 0.5460449425240769,
"learning_rate": 1.5100192330086954e-05,
"loss": 0.13353708386421204,
"step": 586
},
{
"epoch": 0.6770472895040369,
"grad_norm": 0.489975776760947,
"learning_rate": 1.5084417833916866e-05,
"loss": 0.14451898634433746,
"step": 587
},
{
"epoch": 0.6782006920415224,
"grad_norm": 0.5712097699919916,
"learning_rate": 1.5068626256513575e-05,
"loss": 0.1839696615934372,
"step": 588
},
{
"epoch": 0.6793540945790081,
"grad_norm": 0.5078368873465547,
"learning_rate": 1.5052817650929288e-05,
"loss": 0.12648722529411316,
"step": 589
},
{
"epoch": 0.6805074971164936,
"grad_norm": 0.5796413401669295,
"learning_rate": 1.5036992070273431e-05,
"loss": 0.1492060422897339,
"step": 590
},
{
"epoch": 0.6816608996539792,
"grad_norm": 0.5353727705310136,
"learning_rate": 1.5021149567712452e-05,
"loss": 0.11778081953525543,
"step": 591
},
{
"epoch": 0.6828143021914648,
"grad_norm": 0.5507435950551299,
"learning_rate": 1.5005290196469644e-05,
"loss": 0.16449977457523346,
"step": 592
},
{
"epoch": 0.6839677047289504,
"grad_norm": 0.6154857462055485,
"learning_rate": 1.4989414009824978e-05,
"loss": 0.16693055629730225,
"step": 593
},
{
"epoch": 0.6851211072664359,
"grad_norm": 0.5363066452784009,
"learning_rate": 1.497352106111491e-05,
"loss": 0.1481233388185501,
"step": 594
},
{
"epoch": 0.6862745098039216,
"grad_norm": 0.5351326577132005,
"learning_rate": 1.4957611403732216e-05,
"loss": 0.15320725739002228,
"step": 595
},
{
"epoch": 0.6874279123414071,
"grad_norm": 0.4927935287201684,
"learning_rate": 1.4941685091125797e-05,
"loss": 0.12467750161886215,
"step": 596
},
{
"epoch": 0.6885813148788927,
"grad_norm": 0.5339869900670627,
"learning_rate": 1.4925742176800518e-05,
"loss": 0.1490432322025299,
"step": 597
},
{
"epoch": 0.6897347174163783,
"grad_norm": 0.5319021282760147,
"learning_rate": 1.4909782714317003e-05,
"loss": 0.13420483469963074,
"step": 598
},
{
"epoch": 0.6908881199538639,
"grad_norm": 0.5472528748138635,
"learning_rate": 1.489380675729149e-05,
"loss": 0.15919135510921478,
"step": 599
},
{
"epoch": 0.6920415224913494,
"grad_norm": 0.5655058091517462,
"learning_rate": 1.4877814359395612e-05,
"loss": 0.1313835084438324,
"step": 600
},
{
"epoch": 0.6931949250288351,
"grad_norm": 0.5272225341921626,
"learning_rate": 1.4861805574356247e-05,
"loss": 0.1361112743616104,
"step": 601
},
{
"epoch": 0.6943483275663207,
"grad_norm": 0.46316532791751275,
"learning_rate": 1.4845780455955321e-05,
"loss": 0.12278570234775543,
"step": 602
},
{
"epoch": 0.6955017301038062,
"grad_norm": 0.5157405803709645,
"learning_rate": 1.4829739058029632e-05,
"loss": 0.1393338441848755,
"step": 603
},
{
"epoch": 0.6966551326412919,
"grad_norm": 0.4660425967458831,
"learning_rate": 1.4813681434470675e-05,
"loss": 0.1265096664428711,
"step": 604
},
{
"epoch": 0.6978085351787774,
"grad_norm": 0.607169193452165,
"learning_rate": 1.479760763922445e-05,
"loss": 0.19993136823177338,
"step": 605
},
{
"epoch": 0.698961937716263,
"grad_norm": 0.5561024362942076,
"learning_rate": 1.4781517726291283e-05,
"loss": 0.134174183011055,
"step": 606
},
{
"epoch": 0.7001153402537486,
"grad_norm": 0.5568699135101743,
"learning_rate": 1.476541174972566e-05,
"loss": 0.17245635390281677,
"step": 607
},
{
"epoch": 0.7012687427912342,
"grad_norm": 0.5574218585856895,
"learning_rate": 1.4749289763636026e-05,
"loss": 0.12473335862159729,
"step": 608
},
{
"epoch": 0.7024221453287197,
"grad_norm": 0.49991404161903524,
"learning_rate": 1.4733151822184603e-05,
"loss": 0.14183908700942993,
"step": 609
},
{
"epoch": 0.7035755478662054,
"grad_norm": 0.5784015824100245,
"learning_rate": 1.4716997979587229e-05,
"loss": 0.16254518926143646,
"step": 610
},
{
"epoch": 0.7047289504036909,
"grad_norm": 0.47502758559634706,
"learning_rate": 1.4700828290113152e-05,
"loss": 0.13118690252304077,
"step": 611
},
{
"epoch": 0.7058823529411765,
"grad_norm": 0.5756515677752417,
"learning_rate": 1.4684642808084863e-05,
"loss": 0.12527240812778473,
"step": 612
},
{
"epoch": 0.707035755478662,
"grad_norm": 0.5944669507429753,
"learning_rate": 1.4668441587877908e-05,
"loss": 0.12315365672111511,
"step": 613
},
{
"epoch": 0.7081891580161477,
"grad_norm": 0.5820491361474616,
"learning_rate": 1.4652224683920708e-05,
"loss": 0.152072012424469,
"step": 614
},
{
"epoch": 0.7093425605536332,
"grad_norm": 0.6936615898796944,
"learning_rate": 1.4635992150694364e-05,
"loss": 0.11985116451978683,
"step": 615
},
{
"epoch": 0.7104959630911188,
"grad_norm": 0.5862175497048799,
"learning_rate": 1.4619744042732503e-05,
"loss": 0.14260166883468628,
"step": 616
},
{
"epoch": 0.7116493656286044,
"grad_norm": 0.5696355599687706,
"learning_rate": 1.4603480414621049e-05,
"loss": 0.1583096981048584,
"step": 617
},
{
"epoch": 0.71280276816609,
"grad_norm": 0.5536064300076412,
"learning_rate": 1.4587201320998095e-05,
"loss": 0.12779922783374786,
"step": 618
},
{
"epoch": 0.7139561707035755,
"grad_norm": 0.5683725441563136,
"learning_rate": 1.4570906816553667e-05,
"loss": 0.15696051716804504,
"step": 619
},
{
"epoch": 0.7151095732410612,
"grad_norm": 0.5816795170443614,
"learning_rate": 1.455459695602958e-05,
"loss": 0.15622369945049286,
"step": 620
},
{
"epoch": 0.7162629757785467,
"grad_norm": 0.5246602210023453,
"learning_rate": 1.4538271794219227e-05,
"loss": 0.11773015558719635,
"step": 621
},
{
"epoch": 0.7174163783160323,
"grad_norm": 0.5838056472598705,
"learning_rate": 1.4521931385967419e-05,
"loss": 0.11321108043193817,
"step": 622
},
{
"epoch": 0.7185697808535179,
"grad_norm": 0.48813990213399494,
"learning_rate": 1.4505575786170171e-05,
"loss": 0.1298905313014984,
"step": 623
},
{
"epoch": 0.7197231833910035,
"grad_norm": 0.6037055905884313,
"learning_rate": 1.448920504977455e-05,
"loss": 0.1801263988018036,
"step": 624
},
{
"epoch": 0.720876585928489,
"grad_norm": 0.5647072451963943,
"learning_rate": 1.447281923177847e-05,
"loss": 0.132421612739563,
"step": 625
},
{
"epoch": 0.7220299884659747,
"grad_norm": 0.48239759073931937,
"learning_rate": 1.4456418387230504e-05,
"loss": 0.12824435532093048,
"step": 626
},
{
"epoch": 0.7231833910034602,
"grad_norm": 0.5480487766867851,
"learning_rate": 1.4440002571229714e-05,
"loss": 0.14619705080986023,
"step": 627
},
{
"epoch": 0.7243367935409458,
"grad_norm": 0.5577460487290515,
"learning_rate": 1.4423571838925462e-05,
"loss": 0.1666150689125061,
"step": 628
},
{
"epoch": 0.7254901960784313,
"grad_norm": 0.5033289115787662,
"learning_rate": 1.4407126245517219e-05,
"loss": 0.13386300206184387,
"step": 629
},
{
"epoch": 0.726643598615917,
"grad_norm": 0.5079288376531816,
"learning_rate": 1.4390665846254377e-05,
"loss": 0.1228678971529007,
"step": 630
},
{
"epoch": 0.7277970011534025,
"grad_norm": 0.5359222589213934,
"learning_rate": 1.4374190696436074e-05,
"loss": 0.1296224743127823,
"step": 631
},
{
"epoch": 0.7289504036908881,
"grad_norm": 0.5455294698369951,
"learning_rate": 1.4357700851411005e-05,
"loss": 0.14228416979312897,
"step": 632
},
{
"epoch": 0.7301038062283737,
"grad_norm": 0.5072219903444594,
"learning_rate": 1.4341196366577228e-05,
"loss": 0.12831538915634155,
"step": 633
},
{
"epoch": 0.7312572087658593,
"grad_norm": 0.49469433441892874,
"learning_rate": 1.4324677297381988e-05,
"loss": 0.13411608338356018,
"step": 634
},
{
"epoch": 0.7324106113033448,
"grad_norm": 0.5292023886526898,
"learning_rate": 1.4308143699321523e-05,
"loss": 0.18858088552951813,
"step": 635
},
{
"epoch": 0.7335640138408305,
"grad_norm": 0.6069208658754568,
"learning_rate": 1.4291595627940886e-05,
"loss": 0.12956708669662476,
"step": 636
},
{
"epoch": 0.734717416378316,
"grad_norm": 0.5166295778006793,
"learning_rate": 1.427503313883375e-05,
"loss": 0.17079350352287292,
"step": 637
},
{
"epoch": 0.7358708189158016,
"grad_norm": 0.6147879103137568,
"learning_rate": 1.4258456287642223e-05,
"loss": 0.12353013455867767,
"step": 638
},
{
"epoch": 0.7370242214532872,
"grad_norm": 0.5253661800075267,
"learning_rate": 1.4241865130056669e-05,
"loss": 0.1629999279975891,
"step": 639
},
{
"epoch": 0.7381776239907728,
"grad_norm": 0.5314516080253134,
"learning_rate": 1.422525972181551e-05,
"loss": 0.1264171600341797,
"step": 640
},
{
"epoch": 0.7393310265282583,
"grad_norm": 0.5374944603876363,
"learning_rate": 1.4208640118705041e-05,
"loss": 0.14531268179416656,
"step": 641
},
{
"epoch": 0.740484429065744,
"grad_norm": 0.49710057604718516,
"learning_rate": 1.4192006376559253e-05,
"loss": 0.11916156113147736,
"step": 642
},
{
"epoch": 0.7416378316032295,
"grad_norm": 0.4957107233791663,
"learning_rate": 1.4175358551259633e-05,
"loss": 0.13886961340904236,
"step": 643
},
{
"epoch": 0.7427912341407151,
"grad_norm": 0.5016765856347394,
"learning_rate": 1.4158696698734976e-05,
"loss": 0.15586960315704346,
"step": 644
},
{
"epoch": 0.7439446366782007,
"grad_norm": 0.5902947704403175,
"learning_rate": 1.4142020874961212e-05,
"loss": 0.11920519173145294,
"step": 645
},
{
"epoch": 0.7450980392156863,
"grad_norm": 0.4676806601416915,
"learning_rate": 1.4125331135961195e-05,
"loss": 0.14290127158164978,
"step": 646
},
{
"epoch": 0.7462514417531718,
"grad_norm": 0.5813314894842488,
"learning_rate": 1.4108627537804542e-05,
"loss": 0.1420619636774063,
"step": 647
},
{
"epoch": 0.7474048442906575,
"grad_norm": 0.49411599897321457,
"learning_rate": 1.409191013660742e-05,
"loss": 0.1233484297990799,
"step": 648
},
{
"epoch": 0.748558246828143,
"grad_norm": 0.49482834529055764,
"learning_rate": 1.407517898853237e-05,
"loss": 0.1243283599615097,
"step": 649
},
{
"epoch": 0.7497116493656286,
"grad_norm": 0.5083680203441916,
"learning_rate": 1.4058434149788118e-05,
"loss": 0.14333392679691315,
"step": 650
},
{
"epoch": 0.7508650519031141,
"grad_norm": 0.5027502862293466,
"learning_rate": 1.4041675676629383e-05,
"loss": 0.12547186017036438,
"step": 651
},
{
"epoch": 0.7520184544405998,
"grad_norm": 0.5010900988152549,
"learning_rate": 1.402490362535669e-05,
"loss": 0.12978696823120117,
"step": 652
},
{
"epoch": 0.7531718569780853,
"grad_norm": 0.5957236302434203,
"learning_rate": 1.4008118052316173e-05,
"loss": 0.16652148962020874,
"step": 653
},
{
"epoch": 0.754325259515571,
"grad_norm": 0.4934641176466029,
"learning_rate": 1.3991319013899407e-05,
"loss": 0.14946141839027405,
"step": 654
},
{
"epoch": 0.7554786620530565,
"grad_norm": 0.5155200648020438,
"learning_rate": 1.3974506566543189e-05,
"loss": 0.16961389780044556,
"step": 655
},
{
"epoch": 0.7566320645905421,
"grad_norm": 0.5233906629651089,
"learning_rate": 1.3957680766729376e-05,
"loss": 0.18178865313529968,
"step": 656
},
{
"epoch": 0.7577854671280276,
"grad_norm": 0.5542014402977351,
"learning_rate": 1.3940841670984675e-05,
"loss": 0.12632358074188232,
"step": 657
},
{
"epoch": 0.7589388696655133,
"grad_norm": 0.5029068883970547,
"learning_rate": 1.3923989335880464e-05,
"loss": 0.130338653922081,
"step": 658
},
{
"epoch": 0.7600922722029988,
"grad_norm": 0.4723947009840195,
"learning_rate": 1.3907123818032598e-05,
"loss": 0.12922826409339905,
"step": 659
},
{
"epoch": 0.7612456747404844,
"grad_norm": 0.5251476496178913,
"learning_rate": 1.389024517410122e-05,
"loss": 0.1291044056415558,
"step": 660
},
{
"epoch": 0.76239907727797,
"grad_norm": 0.461583231258406,
"learning_rate": 1.3873353460790573e-05,
"loss": 0.11793453991413116,
"step": 661
},
{
"epoch": 0.7635524798154556,
"grad_norm": 0.47553283146240927,
"learning_rate": 1.3856448734848805e-05,
"loss": 0.1147337406873703,
"step": 662
},
{
"epoch": 0.7647058823529411,
"grad_norm": 0.5157238281093511,
"learning_rate": 1.3839531053067776e-05,
"loss": 0.1541939228773117,
"step": 663
},
{
"epoch": 0.7658592848904268,
"grad_norm": 0.5482609980951465,
"learning_rate": 1.3822600472282878e-05,
"loss": 0.130991131067276,
"step": 664
},
{
"epoch": 0.7670126874279123,
"grad_norm": 0.5818730812946653,
"learning_rate": 1.3805657049372835e-05,
"loss": 0.1250099539756775,
"step": 665
},
{
"epoch": 0.7681660899653979,
"grad_norm": 0.45706447647693627,
"learning_rate": 1.3788700841259517e-05,
"loss": 0.12014703452587128,
"step": 666
},
{
"epoch": 0.7693194925028836,
"grad_norm": 0.6300276674529146,
"learning_rate": 1.377173190490774e-05,
"loss": 0.15254420042037964,
"step": 667
},
{
"epoch": 0.7704728950403691,
"grad_norm": 0.48293225028464276,
"learning_rate": 1.3754750297325084e-05,
"loss": 0.1490272879600525,
"step": 668
},
{
"epoch": 0.7716262975778547,
"grad_norm": 0.4852686704522429,
"learning_rate": 1.37377560755617e-05,
"loss": 0.12159977108240128,
"step": 669
},
{
"epoch": 0.7727797001153403,
"grad_norm": 0.5425438852383447,
"learning_rate": 1.3720749296710118e-05,
"loss": 0.13024558126926422,
"step": 670
},
{
"epoch": 0.7739331026528259,
"grad_norm": 0.47477802830012056,
"learning_rate": 1.3703730017905051e-05,
"loss": 0.12879452109336853,
"step": 671
},
{
"epoch": 0.7750865051903114,
"grad_norm": 0.4720908147870451,
"learning_rate": 1.36866982963232e-05,
"loss": 0.13055327534675598,
"step": 672
},
{
"epoch": 0.776239907727797,
"grad_norm": 0.4853023559384227,
"learning_rate": 1.366965418918308e-05,
"loss": 0.14996758103370667,
"step": 673
},
{
"epoch": 0.7773933102652826,
"grad_norm": 0.5275656228404062,
"learning_rate": 1.3652597753744811e-05,
"loss": 0.1717807650566101,
"step": 674
},
{
"epoch": 0.7785467128027682,
"grad_norm": 0.48909206558574764,
"learning_rate": 1.3635529047309925e-05,
"loss": 0.11478856950998306,
"step": 675
},
{
"epoch": 0.7797001153402537,
"grad_norm": 0.5121037563682057,
"learning_rate": 1.3618448127221184e-05,
"loss": 0.16400906443595886,
"step": 676
},
{
"epoch": 0.7808535178777394,
"grad_norm": 0.5118924397674435,
"learning_rate": 1.3601355050862382e-05,
"loss": 0.1637585312128067,
"step": 677
},
{
"epoch": 0.7820069204152249,
"grad_norm": 0.5114643706722616,
"learning_rate": 1.3584249875658148e-05,
"loss": 0.1955689638853073,
"step": 678
},
{
"epoch": 0.7831603229527105,
"grad_norm": 0.5692216930220083,
"learning_rate": 1.3567132659073763e-05,
"loss": 0.15024235844612122,
"step": 679
},
{
"epoch": 0.7843137254901961,
"grad_norm": 0.4662502367486114,
"learning_rate": 1.355000345861496e-05,
"loss": 0.12984469532966614,
"step": 680
},
{
"epoch": 0.7854671280276817,
"grad_norm": 0.5475875327663269,
"learning_rate": 1.3532862331827725e-05,
"loss": 0.11616875976324081,
"step": 681
},
{
"epoch": 0.7866205305651672,
"grad_norm": 0.559572917336331,
"learning_rate": 1.3515709336298124e-05,
"loss": 0.167021706700325,
"step": 682
},
{
"epoch": 0.7877739331026529,
"grad_norm": 0.4954824697557678,
"learning_rate": 1.3498544529652087e-05,
"loss": 0.13411027193069458,
"step": 683
},
{
"epoch": 0.7889273356401384,
"grad_norm": 0.47520662256460044,
"learning_rate": 1.3481367969555224e-05,
"loss": 0.13749516010284424,
"step": 684
},
{
"epoch": 0.790080738177624,
"grad_norm": 0.5050923211392107,
"learning_rate": 1.3464179713712633e-05,
"loss": 0.13095010817050934,
"step": 685
},
{
"epoch": 0.7912341407151096,
"grad_norm": 0.43013110731458276,
"learning_rate": 1.3446979819868703e-05,
"loss": 0.11840403079986572,
"step": 686
},
{
"epoch": 0.7923875432525952,
"grad_norm": 0.5142941001435608,
"learning_rate": 1.3429768345806926e-05,
"loss": 0.12289038300514221,
"step": 687
},
{
"epoch": 0.7935409457900807,
"grad_norm": 0.5666346535066954,
"learning_rate": 1.341254534934969e-05,
"loss": 0.19119775295257568,
"step": 688
},
{
"epoch": 0.7946943483275664,
"grad_norm": 0.5308711716675759,
"learning_rate": 1.3395310888358102e-05,
"loss": 0.12390265613794327,
"step": 689
},
{
"epoch": 0.7958477508650519,
"grad_norm": 0.45967933586708387,
"learning_rate": 1.3378065020731774e-05,
"loss": 0.1265391856431961,
"step": 690
},
{
"epoch": 0.7970011534025375,
"grad_norm": 0.48912421779338106,
"learning_rate": 1.336080780440865e-05,
"loss": 0.1423664093017578,
"step": 691
},
{
"epoch": 0.798154555940023,
"grad_norm": 0.4577029116117999,
"learning_rate": 1.3343539297364789e-05,
"loss": 0.12507611513137817,
"step": 692
},
{
"epoch": 0.7993079584775087,
"grad_norm": 0.4565066360212717,
"learning_rate": 1.3326259557614189e-05,
"loss": 0.13531243801116943,
"step": 693
},
{
"epoch": 0.8004613610149942,
"grad_norm": 0.47172263440090917,
"learning_rate": 1.3308968643208583e-05,
"loss": 0.13763310015201569,
"step": 694
},
{
"epoch": 0.8016147635524798,
"grad_norm": 0.5005317166303402,
"learning_rate": 1.3291666612237246e-05,
"loss": 0.11014001816511154,
"step": 695
},
{
"epoch": 0.8027681660899654,
"grad_norm": 0.5079854536706654,
"learning_rate": 1.32743535228268e-05,
"loss": 0.12492197006940842,
"step": 696
},
{
"epoch": 0.803921568627451,
"grad_norm": 0.5313593682158041,
"learning_rate": 1.325702943314101e-05,
"loss": 0.13400518894195557,
"step": 697
},
{
"epoch": 0.8050749711649365,
"grad_norm": 0.48081325126213154,
"learning_rate": 1.3239694401380609e-05,
"loss": 0.11496663093566895,
"step": 698
},
{
"epoch": 0.8062283737024222,
"grad_norm": 0.5637182884624731,
"learning_rate": 1.3222348485783084e-05,
"loss": 0.12158745527267456,
"step": 699
},
{
"epoch": 0.8073817762399077,
"grad_norm": 0.47283709211937475,
"learning_rate": 1.3204991744622485e-05,
"loss": 0.1463463455438614,
"step": 700
},
{
"epoch": 0.8085351787773933,
"grad_norm": 0.49726695222908784,
"learning_rate": 1.3187624236209233e-05,
"loss": 0.1618843674659729,
"step": 701
},
{
"epoch": 0.8096885813148789,
"grad_norm": 0.5467791988771391,
"learning_rate": 1.3170246018889925e-05,
"loss": 0.11895115673542023,
"step": 702
},
{
"epoch": 0.8108419838523645,
"grad_norm": 0.5039661326746293,
"learning_rate": 1.3152857151047126e-05,
"loss": 0.12421370297670364,
"step": 703
},
{
"epoch": 0.81199538638985,
"grad_norm": 0.7001030107430828,
"learning_rate": 1.3135457691099192e-05,
"loss": 0.158072829246521,
"step": 704
},
{
"epoch": 0.8131487889273357,
"grad_norm": 0.5760067674552856,
"learning_rate": 1.3118047697500055e-05,
"loss": 0.17845988273620605,
"step": 705
},
{
"epoch": 0.8143021914648212,
"grad_norm": 0.639366877251007,
"learning_rate": 1.3100627228739045e-05,
"loss": 0.13280446827411652,
"step": 706
},
{
"epoch": 0.8154555940023068,
"grad_norm": 0.5948266085567002,
"learning_rate": 1.308319634334067e-05,
"loss": 0.15607911348342896,
"step": 707
},
{
"epoch": 0.8166089965397924,
"grad_norm": 0.5658609958504864,
"learning_rate": 1.3065755099864446e-05,
"loss": 0.13090167939662933,
"step": 708
},
{
"epoch": 0.817762399077278,
"grad_norm": 0.5601525196337492,
"learning_rate": 1.3048303556904674e-05,
"loss": 0.12660911679267883,
"step": 709
},
{
"epoch": 0.8189158016147635,
"grad_norm": 0.5799186801955697,
"learning_rate": 1.303084177309027e-05,
"loss": 0.12165719270706177,
"step": 710
},
{
"epoch": 0.8200692041522492,
"grad_norm": 0.5429121468307906,
"learning_rate": 1.3013369807084547e-05,
"loss": 0.141845703125,
"step": 711
},
{
"epoch": 0.8212226066897347,
"grad_norm": 0.5674836097345141,
"learning_rate": 1.299588771758503e-05,
"loss": 0.1135028749704361,
"step": 712
},
{
"epoch": 0.8223760092272203,
"grad_norm": 0.5649176625917081,
"learning_rate": 1.2978395563323245e-05,
"loss": 0.12597960233688354,
"step": 713
},
{
"epoch": 0.8235294117647058,
"grad_norm": 0.5606121031681092,
"learning_rate": 1.2960893403064539e-05,
"loss": 0.14138728380203247,
"step": 714
},
{
"epoch": 0.8246828143021915,
"grad_norm": 0.5047229087018015,
"learning_rate": 1.2943381295607872e-05,
"loss": 0.11548430472612381,
"step": 715
},
{
"epoch": 0.825836216839677,
"grad_norm": 0.7033121691981485,
"learning_rate": 1.2925859299785623e-05,
"loss": 0.14183145761489868,
"step": 716
},
{
"epoch": 0.8269896193771626,
"grad_norm": 0.5195038420873457,
"learning_rate": 1.290832747446339e-05,
"loss": 0.1535172015428543,
"step": 717
},
{
"epoch": 0.8281430219146482,
"grad_norm": 0.6971509554815981,
"learning_rate": 1.2890785878539789e-05,
"loss": 0.13211797177791595,
"step": 718
},
{
"epoch": 0.8292964244521338,
"grad_norm": 0.5234097153970256,
"learning_rate": 1.2873234570946268e-05,
"loss": 0.1272798776626587,
"step": 719
},
{
"epoch": 0.8304498269896193,
"grad_norm": 0.5817526320975626,
"learning_rate": 1.2855673610646903e-05,
"loss": 0.12169615924358368,
"step": 720
},
{
"epoch": 0.831603229527105,
"grad_norm": 0.5923207387610938,
"learning_rate": 1.2838103056638191e-05,
"loss": 0.16157564520835876,
"step": 721
},
{
"epoch": 0.8327566320645905,
"grad_norm": 0.5324122703906454,
"learning_rate": 1.2820522967948863e-05,
"loss": 0.12075008451938629,
"step": 722
},
{
"epoch": 0.8339100346020761,
"grad_norm": 0.5056685320354494,
"learning_rate": 1.280293340363968e-05,
"loss": 0.11899920552968979,
"step": 723
},
{
"epoch": 0.8350634371395617,
"grad_norm": 0.4779186151257459,
"learning_rate": 1.2785334422803239e-05,
"loss": 0.12023557722568512,
"step": 724
},
{
"epoch": 0.8362168396770473,
"grad_norm": 0.5017649554594961,
"learning_rate": 1.2767726084563772e-05,
"loss": 0.11929436028003693,
"step": 725
},
{
"epoch": 0.8373702422145328,
"grad_norm": 0.49647551405511614,
"learning_rate": 1.2750108448076945e-05,
"loss": 0.12515047192573547,
"step": 726
},
{
"epoch": 0.8385236447520185,
"grad_norm": 0.48602643661153844,
"learning_rate": 1.2732481572529664e-05,
"loss": 0.11817323416471481,
"step": 727
},
{
"epoch": 0.839677047289504,
"grad_norm": 0.4991749854519916,
"learning_rate": 1.2714845517139876e-05,
"loss": 0.1402069330215454,
"step": 728
},
{
"epoch": 0.8408304498269896,
"grad_norm": 0.47012932228311793,
"learning_rate": 1.2697200341156361e-05,
"loss": 0.12989844381809235,
"step": 729
},
{
"epoch": 0.8419838523644751,
"grad_norm": 0.48849362054317597,
"learning_rate": 1.2679546103858549e-05,
"loss": 0.14675335586071014,
"step": 730
},
{
"epoch": 0.8431372549019608,
"grad_norm": 0.4659596907765367,
"learning_rate": 1.2661882864556301e-05,
"loss": 0.12276650965213776,
"step": 731
},
{
"epoch": 0.8442906574394463,
"grad_norm": 0.5085407038574923,
"learning_rate": 1.2644210682589731e-05,
"loss": 0.11952273547649384,
"step": 732
},
{
"epoch": 0.845444059976932,
"grad_norm": 0.44694377522138,
"learning_rate": 1.2626529617328989e-05,
"loss": 0.11922066658735275,
"step": 733
},
{
"epoch": 0.8465974625144176,
"grad_norm": 0.48154144938877586,
"learning_rate": 1.260883972817407e-05,
"loss": 0.11177340149879456,
"step": 734
},
{
"epoch": 0.8477508650519031,
"grad_norm": 0.8402337994085274,
"learning_rate": 1.2591141074554616e-05,
"loss": 0.18846005201339722,
"step": 735
},
{
"epoch": 0.8489042675893888,
"grad_norm": 0.5578115521535476,
"learning_rate": 1.2573433715929712e-05,
"loss": 0.13192817568778992,
"step": 736
},
{
"epoch": 0.8500576701268743,
"grad_norm": 0.5160747665464651,
"learning_rate": 1.2555717711787687e-05,
"loss": 0.1245187446475029,
"step": 737
},
{
"epoch": 0.8512110726643599,
"grad_norm": 0.5245387806778021,
"learning_rate": 1.2537993121645913e-05,
"loss": 0.14407718181610107,
"step": 738
},
{
"epoch": 0.8523644752018454,
"grad_norm": 0.5426156168104281,
"learning_rate": 1.2520260005050608e-05,
"loss": 0.1544804573059082,
"step": 739
},
{
"epoch": 0.8535178777393311,
"grad_norm": 0.5375909541096028,
"learning_rate": 1.2502518421576641e-05,
"loss": 0.1352873146533966,
"step": 740
},
{
"epoch": 0.8546712802768166,
"grad_norm": 0.5343735161722114,
"learning_rate": 1.248476843082732e-05,
"loss": 0.11682512611150742,
"step": 741
},
{
"epoch": 0.8558246828143022,
"grad_norm": 0.6182163975726941,
"learning_rate": 1.2467010092434195e-05,
"loss": 0.120292529463768,
"step": 742
},
{
"epoch": 0.8569780853517878,
"grad_norm": 0.5204246308057422,
"learning_rate": 1.2449243466056862e-05,
"loss": 0.14841249585151672,
"step": 743
},
{
"epoch": 0.8581314878892734,
"grad_norm": 0.6462650873945603,
"learning_rate": 1.2431468611382759e-05,
"loss": 0.1314428448677063,
"step": 744
},
{
"epoch": 0.8592848904267589,
"grad_norm": 0.5099572777574999,
"learning_rate": 1.241368558812698e-05,
"loss": 0.13948190212249756,
"step": 745
},
{
"epoch": 0.8604382929642446,
"grad_norm": 0.5910089500066701,
"learning_rate": 1.2395894456032045e-05,
"loss": 0.11840378493070602,
"step": 746
},
{
"epoch": 0.8615916955017301,
"grad_norm": 0.5532853790146433,
"learning_rate": 1.2378095274867725e-05,
"loss": 0.18207389116287231,
"step": 747
},
{
"epoch": 0.8627450980392157,
"grad_norm": 0.5676825868416598,
"learning_rate": 1.2360288104430824e-05,
"loss": 0.1385842263698578,
"step": 748
},
{
"epoch": 0.8638985005767013,
"grad_norm": 0.4903953788517216,
"learning_rate": 1.2342473004545e-05,
"loss": 0.13495568931102753,
"step": 749
},
{
"epoch": 0.8650519031141869,
"grad_norm": 0.6213380525686397,
"learning_rate": 1.2324650035060537e-05,
"loss": 0.12037555873394012,
"step": 750
},
{
"epoch": 0.8662053056516724,
"grad_norm": 0.5226311452027257,
"learning_rate": 1.230681925585416e-05,
"loss": 0.1713240146636963,
"step": 751
},
{
"epoch": 0.8673587081891581,
"grad_norm": 0.6148899940090881,
"learning_rate": 1.2288980726828834e-05,
"loss": 0.12939223647117615,
"step": 752
},
{
"epoch": 0.8685121107266436,
"grad_norm": 0.5477841283785141,
"learning_rate": 1.2271134507913557e-05,
"loss": 0.12186455726623535,
"step": 753
},
{
"epoch": 0.8696655132641292,
"grad_norm": 0.5810753088092774,
"learning_rate": 1.2253280659063172e-05,
"loss": 0.11791893094778061,
"step": 754
},
{
"epoch": 0.8708189158016147,
"grad_norm": 0.5519442091504363,
"learning_rate": 1.2235419240258136e-05,
"loss": 0.12686342000961304,
"step": 755
},
{
"epoch": 0.8719723183391004,
"grad_norm": 0.558470086169441,
"learning_rate": 1.2217550311504352e-05,
"loss": 0.13953900337219238,
"step": 756
},
{
"epoch": 0.8731257208765859,
"grad_norm": 0.5503546827858067,
"learning_rate": 1.2199673932832945e-05,
"loss": 0.12774379551410675,
"step": 757
},
{
"epoch": 0.8742791234140715,
"grad_norm": 0.5338232948926647,
"learning_rate": 1.2181790164300079e-05,
"loss": 0.12202088534832001,
"step": 758
},
{
"epoch": 0.8754325259515571,
"grad_norm": 0.5154115293230783,
"learning_rate": 1.2163899065986726e-05,
"loss": 0.1431630551815033,
"step": 759
},
{
"epoch": 0.8765859284890427,
"grad_norm": 0.5308762322612174,
"learning_rate": 1.2146000697998502e-05,
"loss": 0.13109466433525085,
"step": 760
},
{
"epoch": 0.8777393310265282,
"grad_norm": 0.49450423151107537,
"learning_rate": 1.2128095120465434e-05,
"loss": 0.11216701567173004,
"step": 761
},
{
"epoch": 0.8788927335640139,
"grad_norm": 0.663197055960104,
"learning_rate": 1.2110182393541769e-05,
"loss": 0.12343467772006989,
"step": 762
},
{
"epoch": 0.8800461361014994,
"grad_norm": 0.5849625900724198,
"learning_rate": 1.2092262577405784e-05,
"loss": 0.16109780967235565,
"step": 763
},
{
"epoch": 0.881199538638985,
"grad_norm": 0.5376597031775099,
"learning_rate": 1.2074335732259559e-05,
"loss": 0.1572968065738678,
"step": 764
},
{
"epoch": 0.8823529411764706,
"grad_norm": 0.5621965679687541,
"learning_rate": 1.2056401918328798e-05,
"loss": 0.11584329605102539,
"step": 765
},
{
"epoch": 0.8835063437139562,
"grad_norm": 0.5225715902294277,
"learning_rate": 1.2038461195862612e-05,
"loss": 0.1535029262304306,
"step": 766
},
{
"epoch": 0.8846597462514417,
"grad_norm": 0.6107379642678605,
"learning_rate": 1.2020513625133319e-05,
"loss": 0.16522571444511414,
"step": 767
},
{
"epoch": 0.8858131487889274,
"grad_norm": 0.5086866518587267,
"learning_rate": 1.2002559266436249e-05,
"loss": 0.11956141889095306,
"step": 768
},
{
"epoch": 0.8869665513264129,
"grad_norm": 0.5458069357065302,
"learning_rate": 1.1984598180089533e-05,
"loss": 0.14245794713497162,
"step": 769
},
{
"epoch": 0.8881199538638985,
"grad_norm": 0.5759474673309344,
"learning_rate": 1.1966630426433903e-05,
"loss": 0.1315859705209732,
"step": 770
},
{
"epoch": 0.889273356401384,
"grad_norm": 0.4608515690852653,
"learning_rate": 1.19486560658325e-05,
"loss": 0.136341854929924,
"step": 771
},
{
"epoch": 0.8904267589388697,
"grad_norm": 0.6391478819198044,
"learning_rate": 1.1930675158670648e-05,
"loss": 0.12057085335254669,
"step": 772
},
{
"epoch": 0.8915801614763552,
"grad_norm": 0.5430114177335108,
"learning_rate": 1.1912687765355669e-05,
"loss": 0.1715458333492279,
"step": 773
},
{
"epoch": 0.8927335640138409,
"grad_norm": 0.538042525371012,
"learning_rate": 1.189469394631668e-05,
"loss": 0.17306232452392578,
"step": 774
},
{
"epoch": 0.8938869665513264,
"grad_norm": 0.7948781439623782,
"learning_rate": 1.1876693762004377e-05,
"loss": 0.12249374389648438,
"step": 775
},
{
"epoch": 0.895040369088812,
"grad_norm": 0.4687819464856926,
"learning_rate": 1.1858687272890846e-05,
"loss": 0.12969157099723816,
"step": 776
},
{
"epoch": 0.8961937716262975,
"grad_norm": 0.6380709330206916,
"learning_rate": 1.1840674539469352e-05,
"loss": 0.131486713886261,
"step": 777
},
{
"epoch": 0.8973471741637832,
"grad_norm": 0.5468895396981319,
"learning_rate": 1.182265562225414e-05,
"loss": 0.11392475664615631,
"step": 778
},
{
"epoch": 0.8985005767012687,
"grad_norm": 0.5247661289062109,
"learning_rate": 1.1804630581780227e-05,
"loss": 0.12471131980419159,
"step": 779
},
{
"epoch": 0.8996539792387543,
"grad_norm": 0.6131361935312979,
"learning_rate": 1.1786599478603206e-05,
"loss": 0.11894915997982025,
"step": 780
},
{
"epoch": 0.9008073817762399,
"grad_norm": 0.47212138826656463,
"learning_rate": 1.1768562373299033e-05,
"loss": 0.13203996419906616,
"step": 781
},
{
"epoch": 0.9019607843137255,
"grad_norm": 0.5624898939898191,
"learning_rate": 1.1750519326463829e-05,
"loss": 0.153026282787323,
"step": 782
},
{
"epoch": 0.903114186851211,
"grad_norm": 0.511563993314677,
"learning_rate": 1.1732470398713678e-05,
"loss": 0.16525156795978546,
"step": 783
},
{
"epoch": 0.9042675893886967,
"grad_norm": 0.4604873818993575,
"learning_rate": 1.1714415650684417e-05,
"loss": 0.11781368404626846,
"step": 784
},
{
"epoch": 0.9054209919261822,
"grad_norm": 0.48452202279332945,
"learning_rate": 1.169635514303144e-05,
"loss": 0.13027019798755646,
"step": 785
},
{
"epoch": 0.9065743944636678,
"grad_norm": 0.5010849053405766,
"learning_rate": 1.1678288936429487e-05,
"loss": 0.1446325182914734,
"step": 786
},
{
"epoch": 0.9077277970011534,
"grad_norm": 0.5951864573013925,
"learning_rate": 1.1660217091572453e-05,
"loss": 0.17229454219341278,
"step": 787
},
{
"epoch": 0.908881199538639,
"grad_norm": 0.5008925340736188,
"learning_rate": 1.1642139669173159e-05,
"loss": 0.141371950507164,
"step": 788
},
{
"epoch": 0.9100346020761245,
"grad_norm": 0.5157202721233193,
"learning_rate": 1.1624056729963181e-05,
"loss": 0.14333415031433105,
"step": 789
},
{
"epoch": 0.9111880046136102,
"grad_norm": 0.5176798677496323,
"learning_rate": 1.1605968334692617e-05,
"loss": 0.15442237257957458,
"step": 790
},
{
"epoch": 0.9123414071510957,
"grad_norm": 0.48284207722835076,
"learning_rate": 1.1587874544129905e-05,
"loss": 0.11121197044849396,
"step": 791
},
{
"epoch": 0.9134948096885813,
"grad_norm": 0.4768312188348215,
"learning_rate": 1.156977541906159e-05,
"loss": 0.10819806158542633,
"step": 792
},
{
"epoch": 0.9146482122260668,
"grad_norm": 0.5048155762921698,
"learning_rate": 1.1551671020292164e-05,
"loss": 0.1131187379360199,
"step": 793
},
{
"epoch": 0.9158016147635525,
"grad_norm": 0.4956212220796523,
"learning_rate": 1.1533561408643813e-05,
"loss": 0.13360702991485596,
"step": 794
},
{
"epoch": 0.916955017301038,
"grad_norm": 0.5136457608697987,
"learning_rate": 1.1515446644956247e-05,
"loss": 0.11669032275676727,
"step": 795
},
{
"epoch": 0.9181084198385236,
"grad_norm": 0.6017077076485671,
"learning_rate": 1.1497326790086489e-05,
"loss": 0.20555585622787476,
"step": 796
},
{
"epoch": 0.9192618223760092,
"grad_norm": 0.5767280956547872,
"learning_rate": 1.1479201904908655e-05,
"loss": 0.11772092431783676,
"step": 797
},
{
"epoch": 0.9204152249134948,
"grad_norm": 0.5045620081758398,
"learning_rate": 1.1461072050313767e-05,
"loss": 0.11465321481227875,
"step": 798
},
{
"epoch": 0.9215686274509803,
"grad_norm": 0.4933328600377331,
"learning_rate": 1.1442937287209543e-05,
"loss": 0.12301330268383026,
"step": 799
},
{
"epoch": 0.922722029988466,
"grad_norm": 0.49513830348898163,
"learning_rate": 1.1424797676520185e-05,
"loss": 0.13406047224998474,
"step": 800
},
{
"epoch": 0.9238754325259516,
"grad_norm": 0.524227145139979,
"learning_rate": 1.1406653279186185e-05,
"loss": 0.17211738228797913,
"step": 801
},
{
"epoch": 0.9250288350634371,
"grad_norm": 0.5218991759392988,
"learning_rate": 1.1388504156164112e-05,
"loss": 0.14991195499897003,
"step": 802
},
{
"epoch": 0.9261822376009228,
"grad_norm": 0.4898630020070673,
"learning_rate": 1.1370350368426422e-05,
"loss": 0.1146138459444046,
"step": 803
},
{
"epoch": 0.9273356401384083,
"grad_norm": 0.46151887636867944,
"learning_rate": 1.1352191976961228e-05,
"loss": 0.1429271548986435,
"step": 804
},
{
"epoch": 0.9284890426758939,
"grad_norm": 0.4913048169720137,
"learning_rate": 1.1334029042772114e-05,
"loss": 0.12484600394964218,
"step": 805
},
{
"epoch": 0.9296424452133795,
"grad_norm": 0.5984926689951884,
"learning_rate": 1.1315861626877935e-05,
"loss": 0.14442124962806702,
"step": 806
},
{
"epoch": 0.9307958477508651,
"grad_norm": 0.55953177057377,
"learning_rate": 1.1297689790312592e-05,
"loss": 0.1499641239643097,
"step": 807
},
{
"epoch": 0.9319492502883506,
"grad_norm": 0.5514716403719319,
"learning_rate": 1.1279513594124835e-05,
"loss": 0.17451085150241852,
"step": 808
},
{
"epoch": 0.9331026528258363,
"grad_norm": 0.5212972365686701,
"learning_rate": 1.1261333099378071e-05,
"loss": 0.15393301844596863,
"step": 809
},
{
"epoch": 0.9342560553633218,
"grad_norm": 0.5170363626827706,
"learning_rate": 1.1243148367150138e-05,
"loss": 0.14219734072685242,
"step": 810
},
{
"epoch": 0.9354094579008074,
"grad_norm": 0.5013620849473337,
"learning_rate": 1.1224959458533115e-05,
"loss": 0.15779711306095123,
"step": 811
},
{
"epoch": 0.936562860438293,
"grad_norm": 0.5022990485144617,
"learning_rate": 1.1206766434633111e-05,
"loss": 0.12351110577583313,
"step": 812
},
{
"epoch": 0.9377162629757786,
"grad_norm": 0.4926635630144464,
"learning_rate": 1.1188569356570063e-05,
"loss": 0.14798204600811005,
"step": 813
},
{
"epoch": 0.9388696655132641,
"grad_norm": 0.5102851532701447,
"learning_rate": 1.117036828547752e-05,
"loss": 0.15228979289531708,
"step": 814
},
{
"epoch": 0.9400230680507498,
"grad_norm": 0.49598035883621716,
"learning_rate": 1.1152163282502458e-05,
"loss": 0.161825492978096,
"step": 815
},
{
"epoch": 0.9411764705882353,
"grad_norm": 0.475031756473608,
"learning_rate": 1.1133954408805053e-05,
"loss": 0.14184290170669556,
"step": 816
},
{
"epoch": 0.9423298731257209,
"grad_norm": 0.5368002467804762,
"learning_rate": 1.1115741725558482e-05,
"loss": 0.11386921256780624,
"step": 817
},
{
"epoch": 0.9434832756632064,
"grad_norm": 0.5275024260826342,
"learning_rate": 1.1097525293948732e-05,
"loss": 0.12222331762313843,
"step": 818
},
{
"epoch": 0.9446366782006921,
"grad_norm": 0.4659500826631294,
"learning_rate": 1.107930517517437e-05,
"loss": 0.1318773776292801,
"step": 819
},
{
"epoch": 0.9457900807381776,
"grad_norm": 0.5214209389184257,
"learning_rate": 1.1061081430446363e-05,
"loss": 0.1519978642463684,
"step": 820
},
{
"epoch": 0.9469434832756632,
"grad_norm": 0.4962295204200481,
"learning_rate": 1.1042854120987846e-05,
"loss": 0.14240394532680511,
"step": 821
},
{
"epoch": 0.9480968858131488,
"grad_norm": 0.5090614097673495,
"learning_rate": 1.1024623308033942e-05,
"loss": 0.11301392316818237,
"step": 822
},
{
"epoch": 0.9492502883506344,
"grad_norm": 0.4885392340764444,
"learning_rate": 1.1006389052831534e-05,
"loss": 0.15481115877628326,
"step": 823
},
{
"epoch": 0.9504036908881199,
"grad_norm": 0.4971726786499656,
"learning_rate": 1.098815141663908e-05,
"loss": 0.13558566570281982,
"step": 824
},
{
"epoch": 0.9515570934256056,
"grad_norm": 0.4922156516529219,
"learning_rate": 1.0969910460726386e-05,
"loss": 0.13970401883125305,
"step": 825
},
{
"epoch": 0.9527104959630911,
"grad_norm": 0.4747963720175551,
"learning_rate": 1.0951666246374413e-05,
"loss": 0.1722104847431183,
"step": 826
},
{
"epoch": 0.9538638985005767,
"grad_norm": 0.4833703670478662,
"learning_rate": 1.0933418834875074e-05,
"loss": 0.14162388443946838,
"step": 827
},
{
"epoch": 0.9550173010380623,
"grad_norm": 0.48953110031894925,
"learning_rate": 1.0915168287531022e-05,
"loss": 0.12644685804843903,
"step": 828
},
{
"epoch": 0.9561707035755479,
"grad_norm": 0.5021243442579109,
"learning_rate": 1.0896914665655438e-05,
"loss": 0.1451805680990219,
"step": 829
},
{
"epoch": 0.9573241061130334,
"grad_norm": 0.509842379291639,
"learning_rate": 1.087865803057184e-05,
"loss": 0.1587900072336197,
"step": 830
},
{
"epoch": 0.9584775086505191,
"grad_norm": 0.518472294210678,
"learning_rate": 1.0860398443613865e-05,
"loss": 0.1341312825679779,
"step": 831
},
{
"epoch": 0.9596309111880046,
"grad_norm": 0.48690093943597035,
"learning_rate": 1.0842135966125065e-05,
"loss": 0.11342040449380875,
"step": 832
},
{
"epoch": 0.9607843137254902,
"grad_norm": 0.5058721748635181,
"learning_rate": 1.0823870659458713e-05,
"loss": 0.13377347588539124,
"step": 833
},
{
"epoch": 0.9619377162629758,
"grad_norm": 0.4948748300661748,
"learning_rate": 1.080560258497757e-05,
"loss": 0.13863548636436462,
"step": 834
},
{
"epoch": 0.9630911188004614,
"grad_norm": 0.49057320827722234,
"learning_rate": 1.0787331804053708e-05,
"loss": 0.1253422498703003,
"step": 835
},
{
"epoch": 0.9642445213379469,
"grad_norm": 0.4711855787001409,
"learning_rate": 1.076905837806829e-05,
"loss": 0.14167127013206482,
"step": 836
},
{
"epoch": 0.9653979238754326,
"grad_norm": 0.5303181272211027,
"learning_rate": 1.0750782368411362e-05,
"loss": 0.11201933026313782,
"step": 837
},
{
"epoch": 0.9665513264129181,
"grad_norm": 0.4503683240189161,
"learning_rate": 1.0732503836481654e-05,
"loss": 0.13806921243667603,
"step": 838
},
{
"epoch": 0.9677047289504037,
"grad_norm": 0.5409803677136359,
"learning_rate": 1.0714222843686359e-05,
"loss": 0.12202418595552444,
"step": 839
},
{
"epoch": 0.9688581314878892,
"grad_norm": 0.43788579180198056,
"learning_rate": 1.0695939451440951e-05,
"loss": 0.11685791611671448,
"step": 840
},
{
"epoch": 0.9700115340253749,
"grad_norm": 0.4914659401821282,
"learning_rate": 1.067765372116896e-05,
"loss": 0.14914211630821228,
"step": 841
},
{
"epoch": 0.9711649365628604,
"grad_norm": 0.4667636357814174,
"learning_rate": 1.0659365714301766e-05,
"loss": 0.11691362410783768,
"step": 842
},
{
"epoch": 0.972318339100346,
"grad_norm": 0.6125837495674231,
"learning_rate": 1.06410754922784e-05,
"loss": 0.1188657134771347,
"step": 843
},
{
"epoch": 0.9734717416378316,
"grad_norm": 1.0043300934472579,
"learning_rate": 1.0622783116545343e-05,
"loss": 0.11905190348625183,
"step": 844
},
{
"epoch": 0.9746251441753172,
"grad_norm": 0.5330004992155631,
"learning_rate": 1.0604488648556298e-05,
"loss": 0.12654386460781097,
"step": 845
},
{
"epoch": 0.9757785467128027,
"grad_norm": 0.5281034446050751,
"learning_rate": 1.0586192149772003e-05,
"loss": 0.14189785718917847,
"step": 846
},
{
"epoch": 0.9769319492502884,
"grad_norm": 0.5383181149979296,
"learning_rate": 1.0567893681660024e-05,
"loss": 0.119781494140625,
"step": 847
},
{
"epoch": 0.9780853517877739,
"grad_norm": 0.5625076607005653,
"learning_rate": 1.0549593305694533e-05,
"loss": 0.14032383263111115,
"step": 848
},
{
"epoch": 0.9792387543252595,
"grad_norm": 0.47260446152158325,
"learning_rate": 1.0531291083356118e-05,
"loss": 0.12955546379089355,
"step": 849
},
{
"epoch": 0.9803921568627451,
"grad_norm": 0.5783249924870085,
"learning_rate": 1.051298707613157e-05,
"loss": 0.14123381674289703,
"step": 850
},
{
"epoch": 0.9815455594002307,
"grad_norm": 0.4836464061932191,
"learning_rate": 1.0494681345513673e-05,
"loss": 0.15069706737995148,
"step": 851
},
{
"epoch": 0.9826989619377162,
"grad_norm": 0.5329320199710786,
"learning_rate": 1.0476373953001002e-05,
"loss": 0.11646130681037903,
"step": 852
},
{
"epoch": 0.9838523644752019,
"grad_norm": 0.48361674545123523,
"learning_rate": 1.0458064960097716e-05,
"loss": 0.1173904687166214,
"step": 853
},
{
"epoch": 0.9850057670126874,
"grad_norm": 0.5674904773433913,
"learning_rate": 1.043975442831335e-05,
"loss": 0.1230940967798233,
"step": 854
},
{
"epoch": 0.986159169550173,
"grad_norm": 0.49797818150021184,
"learning_rate": 1.042144241916261e-05,
"loss": 0.1367897093296051,
"step": 855
},
{
"epoch": 0.9873125720876585,
"grad_norm": 0.48811809773331,
"learning_rate": 1.0403128994165164e-05,
"loss": 0.12335164844989777,
"step": 856
},
{
"epoch": 0.9884659746251442,
"grad_norm": 0.5035184860546984,
"learning_rate": 1.0384814214845436e-05,
"loss": 0.11108942329883575,
"step": 857
},
{
"epoch": 0.9896193771626297,
"grad_norm": 0.5517137406492241,
"learning_rate": 1.0366498142732401e-05,
"loss": 0.16597163677215576,
"step": 858
},
{
"epoch": 0.9907727797001153,
"grad_norm": 0.5430814888126808,
"learning_rate": 1.0348180839359377e-05,
"loss": 0.12709940969944,
"step": 859
},
{
"epoch": 0.9919261822376009,
"grad_norm": 0.4802162809023051,
"learning_rate": 1.032986236626382e-05,
"loss": 0.11109043657779694,
"step": 860
},
{
"epoch": 0.9930795847750865,
"grad_norm": 0.5185920265987902,
"learning_rate": 1.0311542784987112e-05,
"loss": 0.12917207181453705,
"step": 861
},
{
"epoch": 0.994232987312572,
"grad_norm": 0.5220510997992924,
"learning_rate": 1.029322215707436e-05,
"loss": 0.1368136703968048,
"step": 862
},
{
"epoch": 0.9953863898500577,
"grad_norm": 0.467528762890588,
"learning_rate": 1.0274900544074189e-05,
"loss": 0.11447203159332275,
"step": 863
},
{
"epoch": 0.9965397923875432,
"grad_norm": 0.5539728684505872,
"learning_rate": 1.0256578007538531e-05,
"loss": 0.12644827365875244,
"step": 864
},
{
"epoch": 0.9976931949250288,
"grad_norm": 0.48906135060999084,
"learning_rate": 1.0238254609022419e-05,
"loss": 0.14615187048912048,
"step": 865
},
{
"epoch": 0.9988465974625144,
"grad_norm": 0.5666809447070674,
"learning_rate": 1.0219930410083789e-05,
"loss": 0.1465197205543518,
"step": 866
},
{
"epoch": 1.0,
"grad_norm": 0.5249353638294506,
"learning_rate": 1.0201605472283258e-05,
"loss": 0.12244215607643127,
"step": 867
},
{
"epoch": 1.0011534025374855,
"grad_norm": 0.4763864542193517,
"learning_rate": 1.018327985718393e-05,
"loss": 0.09151667356491089,
"step": 868
},
{
"epoch": 1.0023068050749713,
"grad_norm": 0.49757073346057734,
"learning_rate": 1.0164953626351181e-05,
"loss": 0.08958524465560913,
"step": 869
},
{
"epoch": 1.0034602076124568,
"grad_norm": 0.47091055694380995,
"learning_rate": 1.0146626841352458e-05,
"loss": 0.10383825749158859,
"step": 870
},
{
"epoch": 1.0046136101499423,
"grad_norm": 0.48872949717440484,
"learning_rate": 1.0128299563757069e-05,
"loss": 0.10968658328056335,
"step": 871
},
{
"epoch": 1.0057670126874279,
"grad_norm": 0.5062515150178212,
"learning_rate": 1.0109971855135978e-05,
"loss": 0.09833651781082153,
"step": 872
},
{
"epoch": 1.0069204152249136,
"grad_norm": 0.5165030226618152,
"learning_rate": 1.0091643777061598e-05,
"loss": 0.10747067630290985,
"step": 873
},
{
"epoch": 1.0080738177623991,
"grad_norm": 0.6041854777807889,
"learning_rate": 1.0073315391107575e-05,
"loss": 0.12004461884498596,
"step": 874
},
{
"epoch": 1.0092272202998847,
"grad_norm": 0.6793854230910817,
"learning_rate": 1.0054986758848602e-05,
"loss": 0.10915791988372803,
"step": 875
},
{
"epoch": 1.0103806228373702,
"grad_norm": 0.6167791401170976,
"learning_rate": 1.0036657941860189e-05,
"loss": 0.08258497714996338,
"step": 876
},
{
"epoch": 1.011534025374856,
"grad_norm": 0.5925963768392973,
"learning_rate": 1.0018329001718472e-05,
"loss": 0.09908583015203476,
"step": 877
},
{
"epoch": 1.0126874279123415,
"grad_norm": 0.5507248103884372,
"learning_rate": 1e-05,
"loss": 0.09428858011960983,
"step": 878
},
{
"epoch": 1.013840830449827,
"grad_norm": 0.5523783402770842,
"learning_rate": 9.981670998281533e-06,
"loss": 0.0894874781370163,
"step": 879
},
{
"epoch": 1.0149942329873125,
"grad_norm": 0.5178223700713711,
"learning_rate": 9.963342058139815e-06,
"loss": 0.09048782289028168,
"step": 880
},
{
"epoch": 1.0161476355247983,
"grad_norm": 0.4989814397939169,
"learning_rate": 9.945013241151404e-06,
"loss": 0.09086169302463531,
"step": 881
},
{
"epoch": 1.0173010380622838,
"grad_norm": 0.49202899303808106,
"learning_rate": 9.926684608892428e-06,
"loss": 0.10291647911071777,
"step": 882
},
{
"epoch": 1.0184544405997693,
"grad_norm": 0.5035357514947397,
"learning_rate": 9.908356222938407e-06,
"loss": 0.09086708724498749,
"step": 883
},
{
"epoch": 1.0196078431372548,
"grad_norm": 0.5396255221812811,
"learning_rate": 9.890028144864022e-06,
"loss": 0.09750939905643463,
"step": 884
},
{
"epoch": 1.0207612456747406,
"grad_norm": 0.5572623208685673,
"learning_rate": 9.871700436242934e-06,
"loss": 0.12063789367675781,
"step": 885
},
{
"epoch": 1.021914648212226,
"grad_norm": 0.5337829669899,
"learning_rate": 9.853373158647544e-06,
"loss": 0.08750061690807343,
"step": 886
},
{
"epoch": 1.0230680507497116,
"grad_norm": 0.5862638278677369,
"learning_rate": 9.835046373648822e-06,
"loss": 0.12191341072320938,
"step": 887
},
{
"epoch": 1.0242214532871972,
"grad_norm": 0.5169011300199624,
"learning_rate": 9.816720142816076e-06,
"loss": 0.09341713786125183,
"step": 888
},
{
"epoch": 1.025374855824683,
"grad_norm": 0.5013077051724067,
"learning_rate": 9.798394527716743e-06,
"loss": 0.09151067584753036,
"step": 889
},
{
"epoch": 1.0265282583621684,
"grad_norm": 0.5186026211010281,
"learning_rate": 9.780069589916213e-06,
"loss": 0.11461058259010315,
"step": 890
},
{
"epoch": 1.027681660899654,
"grad_norm": 0.501393304782996,
"learning_rate": 9.761745390977583e-06,
"loss": 0.0893915444612503,
"step": 891
},
{
"epoch": 1.0288350634371395,
"grad_norm": 0.5036806512446272,
"learning_rate": 9.743421992461472e-06,
"loss": 0.10196788609027863,
"step": 892
},
{
"epoch": 1.0299884659746252,
"grad_norm": 0.5357512134896675,
"learning_rate": 9.725099455925811e-06,
"loss": 0.12573838233947754,
"step": 893
},
{
"epoch": 1.0311418685121108,
"grad_norm": 0.7173152417521917,
"learning_rate": 9.706777842925643e-06,
"loss": 0.07852639257907867,
"step": 894
},
{
"epoch": 1.0322952710495963,
"grad_norm": 0.620471585263273,
"learning_rate": 9.688457215012894e-06,
"loss": 0.08761699497699738,
"step": 895
},
{
"epoch": 1.0334486735870818,
"grad_norm": 0.5499344367463908,
"learning_rate": 9.670137633736182e-06,
"loss": 0.09657341986894608,
"step": 896
},
{
"epoch": 1.0346020761245676,
"grad_norm": 0.5940027227855502,
"learning_rate": 9.651819160640627e-06,
"loss": 0.10603274405002594,
"step": 897
},
{
"epoch": 1.035755478662053,
"grad_norm": 0.5959495994604215,
"learning_rate": 9.633501857267602e-06,
"loss": 0.08676831424236298,
"step": 898
},
{
"epoch": 1.0369088811995386,
"grad_norm": 0.5753503042689567,
"learning_rate": 9.615185785154568e-06,
"loss": 0.08439299464225769,
"step": 899
},
{
"epoch": 1.0380622837370241,
"grad_norm": 0.6597637664199886,
"learning_rate": 9.59687100583484e-06,
"loss": 0.08585464209318161,
"step": 900
},
{
"epoch": 1.0392156862745099,
"grad_norm": 0.646651501719703,
"learning_rate": 9.578557580837393e-06,
"loss": 0.10925036668777466,
"step": 901
},
{
"epoch": 1.0403690888119954,
"grad_norm": 0.6554604716956176,
"learning_rate": 9.560245571686651e-06,
"loss": 0.10689708590507507,
"step": 902
},
{
"epoch": 1.041522491349481,
"grad_norm": 0.5599762011096946,
"learning_rate": 9.541935039902287e-06,
"loss": 0.10270076990127563,
"step": 903
},
{
"epoch": 1.0426758938869665,
"grad_norm": 0.5692603663008247,
"learning_rate": 9.523626046999003e-06,
"loss": 0.11272953450679779,
"step": 904
},
{
"epoch": 1.0438292964244522,
"grad_norm": 0.5422715067380376,
"learning_rate": 9.50531865448633e-06,
"loss": 0.0740877240896225,
"step": 905
},
{
"epoch": 1.0449826989619377,
"grad_norm": 0.5253323907683212,
"learning_rate": 9.487012923868434e-06,
"loss": 0.09957445412874222,
"step": 906
},
{
"epoch": 1.0461361014994233,
"grad_norm": 0.6022308155436835,
"learning_rate": 9.468708916643883e-06,
"loss": 0.08926476538181305,
"step": 907
},
{
"epoch": 1.0472895040369088,
"grad_norm": 0.5419414802368134,
"learning_rate": 9.45040669430547e-06,
"loss": 0.10540592670440674,
"step": 908
},
{
"epoch": 1.0484429065743945,
"grad_norm": 0.5194221839415228,
"learning_rate": 9.432106318339978e-06,
"loss": 0.09325025230646133,
"step": 909
},
{
"epoch": 1.04959630911188,
"grad_norm": 0.5392220761001569,
"learning_rate": 9.413807850227998e-06,
"loss": 0.10640532523393631,
"step": 910
},
{
"epoch": 1.0507497116493656,
"grad_norm": 0.5629214433911204,
"learning_rate": 9.395511351443704e-06,
"loss": 0.0883219987154007,
"step": 911
},
{
"epoch": 1.0519031141868511,
"grad_norm": 0.561682874295784,
"learning_rate": 9.37721688345466e-06,
"loss": 0.0960318073630333,
"step": 912
},
{
"epoch": 1.0530565167243369,
"grad_norm": 0.5349921051416878,
"learning_rate": 9.358924507721603e-06,
"loss": 0.10955075919628143,
"step": 913
},
{
"epoch": 1.0542099192618224,
"grad_norm": 0.5087772864323833,
"learning_rate": 9.340634285698238e-06,
"loss": 0.10333843529224396,
"step": 914
},
{
"epoch": 1.055363321799308,
"grad_norm": 0.5550896952638728,
"learning_rate": 9.322346278831045e-06,
"loss": 0.10751191526651382,
"step": 915
},
{
"epoch": 1.0565167243367934,
"grad_norm": 0.5244940724744629,
"learning_rate": 9.30406054855905e-06,
"loss": 0.10530083626508713,
"step": 916
},
{
"epoch": 1.0576701268742792,
"grad_norm": 0.503589252129561,
"learning_rate": 9.285777156313643e-06,
"loss": 0.09009189903736115,
"step": 917
},
{
"epoch": 1.0588235294117647,
"grad_norm": 0.5998164735634413,
"learning_rate": 9.26749616351835e-06,
"loss": 0.08612369000911713,
"step": 918
},
{
"epoch": 1.0599769319492502,
"grad_norm": 0.5015073822304937,
"learning_rate": 9.24921763158864e-06,
"loss": 0.10262610018253326,
"step": 919
},
{
"epoch": 1.0611303344867358,
"grad_norm": 0.5396030977573607,
"learning_rate": 9.230941621931713e-06,
"loss": 0.11248406022787094,
"step": 920
},
{
"epoch": 1.0622837370242215,
"grad_norm": 0.5356091664078048,
"learning_rate": 9.212668195946293e-06,
"loss": 0.08686747401952744,
"step": 921
},
{
"epoch": 1.063437139561707,
"grad_norm": 0.5310748906874857,
"learning_rate": 9.194397415022435e-06,
"loss": 0.10193166881799698,
"step": 922
},
{
"epoch": 1.0645905420991926,
"grad_norm": 0.5793604514733595,
"learning_rate": 9.176129340541292e-06,
"loss": 0.0832764059305191,
"step": 923
},
{
"epoch": 1.065743944636678,
"grad_norm": 0.5347126863377805,
"learning_rate": 9.157864033874939e-06,
"loss": 0.08657500147819519,
"step": 924
},
{
"epoch": 1.0668973471741638,
"grad_norm": 0.5339627707615295,
"learning_rate": 9.139601556386138e-06,
"loss": 0.08516493439674377,
"step": 925
},
{
"epoch": 1.0680507497116494,
"grad_norm": 0.49630926721301755,
"learning_rate": 9.121341969428164e-06,
"loss": 0.09067849069833755,
"step": 926
},
{
"epoch": 1.069204152249135,
"grad_norm": 0.5412175452554143,
"learning_rate": 9.103085334344565e-06,
"loss": 0.08592383563518524,
"step": 927
},
{
"epoch": 1.0703575547866206,
"grad_norm": 0.5500019723761345,
"learning_rate": 9.084831712468981e-06,
"loss": 0.11557412147521973,
"step": 928
},
{
"epoch": 1.0715109573241062,
"grad_norm": 0.49441094014111986,
"learning_rate": 9.06658116512493e-06,
"loss": 0.10171924531459808,
"step": 929
},
{
"epoch": 1.0726643598615917,
"grad_norm": 0.5994457482475518,
"learning_rate": 9.04833375362559e-06,
"loss": 0.10486065596342087,
"step": 930
},
{
"epoch": 1.0738177623990772,
"grad_norm": 0.5164275777290822,
"learning_rate": 9.030089539273621e-06,
"loss": 0.10328168421983719,
"step": 931
},
{
"epoch": 1.0749711649365628,
"grad_norm": 0.5857004060235275,
"learning_rate": 9.011848583360922e-06,
"loss": 0.12410283088684082,
"step": 932
},
{
"epoch": 1.0761245674740485,
"grad_norm": 0.5220635774451153,
"learning_rate": 8.99361094716847e-06,
"loss": 0.09167494624853134,
"step": 933
},
{
"epoch": 1.077277970011534,
"grad_norm": 0.46642348433101044,
"learning_rate": 8.97537669196606e-06,
"loss": 0.09721170365810394,
"step": 934
},
{
"epoch": 1.0784313725490196,
"grad_norm": 0.48142370622363756,
"learning_rate": 8.957145879012155e-06,
"loss": 0.09602795541286469,
"step": 935
},
{
"epoch": 1.0795847750865053,
"grad_norm": 0.5583182497964692,
"learning_rate": 8.93891856955364e-06,
"loss": 0.10774899274110794,
"step": 936
},
{
"epoch": 1.0807381776239908,
"grad_norm": 0.4749468803559607,
"learning_rate": 8.920694824825632e-06,
"loss": 0.08575998246669769,
"step": 937
},
{
"epoch": 1.0818915801614764,
"grad_norm": 0.46575184483844423,
"learning_rate": 8.902474706051275e-06,
"loss": 0.08790190517902374,
"step": 938
},
{
"epoch": 1.0830449826989619,
"grad_norm": 0.5941184223358359,
"learning_rate": 8.88425827444152e-06,
"loss": 0.07909614592790604,
"step": 939
},
{
"epoch": 1.0841983852364474,
"grad_norm": 0.5220520893407326,
"learning_rate": 8.866045591194952e-06,
"loss": 0.09688889980316162,
"step": 940
},
{
"epoch": 1.0853517877739332,
"grad_norm": 0.528094110289125,
"learning_rate": 8.847836717497544e-06,
"loss": 0.09018624573945999,
"step": 941
},
{
"epoch": 1.0865051903114187,
"grad_norm": 0.49440114165182886,
"learning_rate": 8.829631714522482e-06,
"loss": 0.07982371747493744,
"step": 942
},
{
"epoch": 1.0876585928489042,
"grad_norm": 0.5123737857918038,
"learning_rate": 8.811430643429939e-06,
"loss": 0.10689489543437958,
"step": 943
},
{
"epoch": 1.08881199538639,
"grad_norm": 0.5431794892390595,
"learning_rate": 8.79323356536689e-06,
"loss": 0.09043840318918228,
"step": 944
},
{
"epoch": 1.0899653979238755,
"grad_norm": 0.5846231629902161,
"learning_rate": 8.77504054146689e-06,
"loss": 0.130972757935524,
"step": 945
},
{
"epoch": 1.091118800461361,
"grad_norm": 0.5094125433000469,
"learning_rate": 8.756851632849865e-06,
"loss": 0.1053023487329483,
"step": 946
},
{
"epoch": 1.0922722029988465,
"grad_norm": 0.47903712690505734,
"learning_rate": 8.738666900621935e-06,
"loss": 0.10153735429048538,
"step": 947
},
{
"epoch": 1.0934256055363323,
"grad_norm": 0.4743816070033095,
"learning_rate": 8.720486405875167e-06,
"loss": 0.08185769617557526,
"step": 948
},
{
"epoch": 1.0945790080738178,
"grad_norm": 0.5403401487924467,
"learning_rate": 8.702310209687413e-06,
"loss": 0.10509772598743439,
"step": 949
},
{
"epoch": 1.0957324106113033,
"grad_norm": 0.4909468687812748,
"learning_rate": 8.684138373122065e-06,
"loss": 0.08688805997371674,
"step": 950
},
{
"epoch": 1.0968858131487889,
"grad_norm": 0.5561956941410496,
"learning_rate": 8.665970957227888e-06,
"loss": 0.11150307953357697,
"step": 951
},
{
"epoch": 1.0980392156862746,
"grad_norm": 0.5671609426325777,
"learning_rate": 8.647808023038775e-06,
"loss": 0.09001736342906952,
"step": 952
},
{
"epoch": 1.0991926182237601,
"grad_norm": 0.4822590904386884,
"learning_rate": 8.629649631573582e-06,
"loss": 0.09384094178676605,
"step": 953
},
{
"epoch": 1.1003460207612457,
"grad_norm": 0.5054607907453892,
"learning_rate": 8.611495843835891e-06,
"loss": 0.08971919119358063,
"step": 954
},
{
"epoch": 1.1014994232987312,
"grad_norm": 0.5343792624216148,
"learning_rate": 8.593346720813819e-06,
"loss": 0.10019970685243607,
"step": 955
},
{
"epoch": 1.102652825836217,
"grad_norm": 0.5056886784937115,
"learning_rate": 8.57520232347982e-06,
"loss": 0.08857197314500809,
"step": 956
},
{
"epoch": 1.1038062283737025,
"grad_norm": 0.4871272792567786,
"learning_rate": 8.557062712790459e-06,
"loss": 0.10381016135215759,
"step": 957
},
{
"epoch": 1.104959630911188,
"grad_norm": 0.5181715469508408,
"learning_rate": 8.538927949686235e-06,
"loss": 0.1068318784236908,
"step": 958
},
{
"epoch": 1.1061130334486735,
"grad_norm": 0.5009058256000951,
"learning_rate": 8.520798095091347e-06,
"loss": 0.08693573623895645,
"step": 959
},
{
"epoch": 1.1072664359861593,
"grad_norm": 0.4620382808993134,
"learning_rate": 8.502673209913513e-06,
"loss": 0.09147288650274277,
"step": 960
},
{
"epoch": 1.1084198385236448,
"grad_norm": 0.4614981440191872,
"learning_rate": 8.484553355043755e-06,
"loss": 0.08958452194929123,
"step": 961
},
{
"epoch": 1.1095732410611303,
"grad_norm": 0.5042272675553611,
"learning_rate": 8.46643859135619e-06,
"loss": 0.1052074059844017,
"step": 962
},
{
"epoch": 1.1107266435986158,
"grad_norm": 0.46260684767155785,
"learning_rate": 8.448328979707843e-06,
"loss": 0.08317890763282776,
"step": 963
},
{
"epoch": 1.1118800461361016,
"grad_norm": 0.49922249986506223,
"learning_rate": 8.430224580938411e-06,
"loss": 0.10049428045749664,
"step": 964
},
{
"epoch": 1.1130334486735871,
"grad_norm": 0.49779594022515894,
"learning_rate": 8.412125455870102e-06,
"loss": 0.09185273200273514,
"step": 965
},
{
"epoch": 1.1141868512110726,
"grad_norm": 0.5045972240328359,
"learning_rate": 8.394031665307383e-06,
"loss": 0.10500700771808624,
"step": 966
},
{
"epoch": 1.1153402537485582,
"grad_norm": 0.541783147569593,
"learning_rate": 8.375943270036822e-06,
"loss": 0.08786934614181519,
"step": 967
},
{
"epoch": 1.116493656286044,
"grad_norm": 0.5165284647238945,
"learning_rate": 8.357860330826843e-06,
"loss": 0.10999307036399841,
"step": 968
},
{
"epoch": 1.1176470588235294,
"grad_norm": 0.5046220380925249,
"learning_rate": 8.339782908427552e-06,
"loss": 0.08341707289218903,
"step": 969
},
{
"epoch": 1.118800461361015,
"grad_norm": 0.5034106420855234,
"learning_rate": 8.321711063570516e-06,
"loss": 0.10172156989574432,
"step": 970
},
{
"epoch": 1.1199538638985005,
"grad_norm": 0.48950691310522676,
"learning_rate": 8.303644856968564e-06,
"loss": 0.10010553151369095,
"step": 971
},
{
"epoch": 1.1211072664359862,
"grad_norm": 0.5360353446259297,
"learning_rate": 8.285584349315588e-06,
"loss": 0.1211581826210022,
"step": 972
},
{
"epoch": 1.1222606689734718,
"grad_norm": 0.4992868014235251,
"learning_rate": 8.267529601286325e-06,
"loss": 0.0885929986834526,
"step": 973
},
{
"epoch": 1.1234140715109573,
"grad_norm": 0.5025467167187584,
"learning_rate": 8.249480673536174e-06,
"loss": 0.08859886974096298,
"step": 974
},
{
"epoch": 1.1245674740484428,
"grad_norm": 0.47042932527929593,
"learning_rate": 8.231437626700968e-06,
"loss": 0.09510938823223114,
"step": 975
},
{
"epoch": 1.1257208765859286,
"grad_norm": 0.5673495246698553,
"learning_rate": 8.213400521396796e-06,
"loss": 0.10550633072853088,
"step": 976
},
{
"epoch": 1.126874279123414,
"grad_norm": 0.5289746602061379,
"learning_rate": 8.195369418219773e-06,
"loss": 0.08814887702465057,
"step": 977
},
{
"epoch": 1.1280276816608996,
"grad_norm": 0.514623710758184,
"learning_rate": 8.177344377745862e-06,
"loss": 0.0875728502869606,
"step": 978
},
{
"epoch": 1.1291810841983851,
"grad_norm": 0.5175516519684839,
"learning_rate": 8.159325460530653e-06,
"loss": 0.09927032887935638,
"step": 979
},
{
"epoch": 1.130334486735871,
"grad_norm": 0.4952671500138912,
"learning_rate": 8.14131272710916e-06,
"loss": 0.0860392153263092,
"step": 980
},
{
"epoch": 1.1314878892733564,
"grad_norm": 0.49807978613176096,
"learning_rate": 8.123306237995628e-06,
"loss": 0.10014220327138901,
"step": 981
},
{
"epoch": 1.132641291810842,
"grad_norm": 0.5254769505587604,
"learning_rate": 8.105306053683323e-06,
"loss": 0.07768946886062622,
"step": 982
},
{
"epoch": 1.1337946943483275,
"grad_norm": 0.4727711476745625,
"learning_rate": 8.087312234644333e-06,
"loss": 0.08815547823905945,
"step": 983
},
{
"epoch": 1.1349480968858132,
"grad_norm": 0.5470174027355911,
"learning_rate": 8.069324841329354e-06,
"loss": 0.12428925186395645,
"step": 984
},
{
"epoch": 1.1361014994232987,
"grad_norm": 0.5042032258002711,
"learning_rate": 8.051343934167504e-06,
"loss": 0.10129491984844208,
"step": 985
},
{
"epoch": 1.1372549019607843,
"grad_norm": 0.49592670175171577,
"learning_rate": 8.033369573566097e-06,
"loss": 0.10581555217504501,
"step": 986
},
{
"epoch": 1.1384083044982698,
"grad_norm": 0.46619285899849516,
"learning_rate": 8.015401819910472e-06,
"loss": 0.08362239599227905,
"step": 987
},
{
"epoch": 1.1395617070357555,
"grad_norm": 0.482253558941966,
"learning_rate": 7.997440733563756e-06,
"loss": 0.09193193167448044,
"step": 988
},
{
"epoch": 1.140715109573241,
"grad_norm": 0.5388176103172546,
"learning_rate": 7.979486374866684e-06,
"loss": 0.11026648432016373,
"step": 989
},
{
"epoch": 1.1418685121107266,
"grad_norm": 0.5235325255028277,
"learning_rate": 7.961538804137393e-06,
"loss": 0.08698016405105591,
"step": 990
},
{
"epoch": 1.1430219146482123,
"grad_norm": 0.478258529490472,
"learning_rate": 7.943598081671202e-06,
"loss": 0.08453488349914551,
"step": 991
},
{
"epoch": 1.1441753171856979,
"grad_norm": 0.4964560444901134,
"learning_rate": 7.925664267740443e-06,
"loss": 0.09722678363323212,
"step": 992
},
{
"epoch": 1.1453287197231834,
"grad_norm": 0.48042904794507296,
"learning_rate": 7.907737422594217e-06,
"loss": 0.08202920854091644,
"step": 993
},
{
"epoch": 1.146482122260669,
"grad_norm": 0.5267702447553305,
"learning_rate": 7.889817606458235e-06,
"loss": 0.10414986312389374,
"step": 994
},
{
"epoch": 1.1476355247981544,
"grad_norm": 0.5124124766730157,
"learning_rate": 7.871904879534571e-06,
"loss": 0.0977637842297554,
"step": 995
},
{
"epoch": 1.1487889273356402,
"grad_norm": 0.5075323809325216,
"learning_rate": 7.8539993020015e-06,
"loss": 0.10953503847122192,
"step": 996
},
{
"epoch": 1.1499423298731257,
"grad_norm": 0.5105690514695158,
"learning_rate": 7.836100934013278e-06,
"loss": 0.08410903811454773,
"step": 997
},
{
"epoch": 1.1510957324106112,
"grad_norm": 0.5175148821936078,
"learning_rate": 7.818209835699925e-06,
"loss": 0.09517618268728256,
"step": 998
},
{
"epoch": 1.152249134948097,
"grad_norm": 0.49416915918199844,
"learning_rate": 7.800326067167058e-06,
"loss": 0.08611415326595306,
"step": 999
},
{
"epoch": 1.1534025374855825,
"grad_norm": 0.4785789839825236,
"learning_rate": 7.78244968849565e-06,
"loss": 0.09135805070400238,
"step": 1000
},
{
"epoch": 1.154555940023068,
"grad_norm": 0.5365859966547547,
"learning_rate": 7.764580759741867e-06,
"loss": 0.08865796774625778,
"step": 1001
},
{
"epoch": 1.1557093425605536,
"grad_norm": 0.5096719006182749,
"learning_rate": 7.74671934093683e-06,
"loss": 0.10502101480960846,
"step": 1002
},
{
"epoch": 1.156862745098039,
"grad_norm": 0.5364719225161734,
"learning_rate": 7.728865492086444e-06,
"loss": 0.1084267795085907,
"step": 1003
},
{
"epoch": 1.1580161476355249,
"grad_norm": 0.4948778376763926,
"learning_rate": 7.711019273171171e-06,
"loss": 0.0822351723909378,
"step": 1004
},
{
"epoch": 1.1591695501730104,
"grad_norm": 0.5003349450067912,
"learning_rate": 7.693180744145844e-06,
"loss": 0.10146962851285934,
"step": 1005
},
{
"epoch": 1.160322952710496,
"grad_norm": 0.5160635979593039,
"learning_rate": 7.67534996493947e-06,
"loss": 0.09175290912389755,
"step": 1006
},
{
"epoch": 1.1614763552479817,
"grad_norm": 0.494492537943702,
"learning_rate": 7.657526995455002e-06,
"loss": 0.11067642271518707,
"step": 1007
},
{
"epoch": 1.1626297577854672,
"grad_norm": 0.5187626705274467,
"learning_rate": 7.639711895569178e-06,
"loss": 0.0985628068447113,
"step": 1008
},
{
"epoch": 1.1637831603229527,
"grad_norm": 0.581223955735769,
"learning_rate": 7.621904725132277e-06,
"loss": 0.12328938394784927,
"step": 1009
},
{
"epoch": 1.1649365628604382,
"grad_norm": 0.5157738287087184,
"learning_rate": 7.604105543967957e-06,
"loss": 0.09032996743917465,
"step": 1010
},
{
"epoch": 1.1660899653979238,
"grad_norm": 0.4745532894254481,
"learning_rate": 7.58631441187302e-06,
"loss": 0.08756215870380402,
"step": 1011
},
{
"epoch": 1.1672433679354095,
"grad_norm": 0.6507511380647181,
"learning_rate": 7.568531388617242e-06,
"loss": 0.0906791090965271,
"step": 1012
},
{
"epoch": 1.168396770472895,
"grad_norm": 0.49479358223537545,
"learning_rate": 7.550756533943144e-06,
"loss": 0.10055435448884964,
"step": 1013
},
{
"epoch": 1.1695501730103806,
"grad_norm": 0.517109838192588,
"learning_rate": 7.532989907565808e-06,
"loss": 0.09011873602867126,
"step": 1014
},
{
"epoch": 1.1707035755478663,
"grad_norm": 0.49007647609264093,
"learning_rate": 7.515231569172683e-06,
"loss": 0.10140611231327057,
"step": 1015
},
{
"epoch": 1.1718569780853518,
"grad_norm": 0.5532227305348156,
"learning_rate": 7.497481578423358e-06,
"loss": 0.08936945348978043,
"step": 1016
},
{
"epoch": 1.1730103806228374,
"grad_norm": 0.5066836271153341,
"learning_rate": 7.479739994949393e-06,
"loss": 0.08611628413200378,
"step": 1017
},
{
"epoch": 1.1741637831603229,
"grad_norm": 0.4917863333523073,
"learning_rate": 7.462006878354089e-06,
"loss": 0.09299027919769287,
"step": 1018
},
{
"epoch": 1.1753171856978086,
"grad_norm": 0.5460936121970834,
"learning_rate": 7.444282288212317e-06,
"loss": 0.08408843725919724,
"step": 1019
},
{
"epoch": 1.1764705882352942,
"grad_norm": 0.5332533217885734,
"learning_rate": 7.426566284070291e-06,
"loss": 0.08506901562213898,
"step": 1020
},
{
"epoch": 1.1776239907727797,
"grad_norm": 0.5615604926981128,
"learning_rate": 7.4088589254453855e-06,
"loss": 0.09462149441242218,
"step": 1021
},
{
"epoch": 1.1787773933102652,
"grad_norm": 0.5658541185135038,
"learning_rate": 7.3911602718259326e-06,
"loss": 0.10566244274377823,
"step": 1022
},
{
"epoch": 1.179930795847751,
"grad_norm": 0.5057541342839357,
"learning_rate": 7.373470382671013e-06,
"loss": 0.08625955879688263,
"step": 1023
},
{
"epoch": 1.1810841983852365,
"grad_norm": 0.4821897577129044,
"learning_rate": 7.355789317410272e-06,
"loss": 0.10066573321819305,
"step": 1024
},
{
"epoch": 1.182237600922722,
"grad_norm": 0.49701453019519826,
"learning_rate": 7.3381171354437e-06,
"loss": 0.0926642119884491,
"step": 1025
},
{
"epoch": 1.1833910034602075,
"grad_norm": 0.4883069156518409,
"learning_rate": 7.3204538961414554e-06,
"loss": 0.08802008628845215,
"step": 1026
},
{
"epoch": 1.1845444059976933,
"grad_norm": 0.47819254272315365,
"learning_rate": 7.302799658843639e-06,
"loss": 0.08817782998085022,
"step": 1027
},
{
"epoch": 1.1856978085351788,
"grad_norm": 0.4986794081716864,
"learning_rate": 7.285154482860127e-06,
"loss": 0.10408937186002731,
"step": 1028
},
{
"epoch": 1.1868512110726643,
"grad_norm": 0.48374989538272384,
"learning_rate": 7.267518427470338e-06,
"loss": 0.0858943983912468,
"step": 1029
},
{
"epoch": 1.1880046136101499,
"grad_norm": 0.49784392767623137,
"learning_rate": 7.249891551923056e-06,
"loss": 0.11606677621603012,
"step": 1030
},
{
"epoch": 1.1891580161476356,
"grad_norm": 0.490821951992738,
"learning_rate": 7.232273915436232e-06,
"loss": 0.09444928169250488,
"step": 1031
},
{
"epoch": 1.1903114186851211,
"grad_norm": 0.4523414212412621,
"learning_rate": 7.214665577196762e-06,
"loss": 0.0993557721376419,
"step": 1032
},
{
"epoch": 1.1914648212226067,
"grad_norm": 0.5289008812093915,
"learning_rate": 7.197066596360324e-06,
"loss": 0.12773090600967407,
"step": 1033
},
{
"epoch": 1.1926182237600922,
"grad_norm": 0.45354276698756274,
"learning_rate": 7.179477032051139e-06,
"loss": 0.0877922773361206,
"step": 1034
},
{
"epoch": 1.193771626297578,
"grad_norm": 0.5170926265944734,
"learning_rate": 7.161896943361811e-06,
"loss": 0.0849873274564743,
"step": 1035
},
{
"epoch": 1.1949250288350635,
"grad_norm": 0.4493013137482951,
"learning_rate": 7.144326389353095e-06,
"loss": 0.0837244987487793,
"step": 1036
},
{
"epoch": 1.196078431372549,
"grad_norm": 0.49754870490666464,
"learning_rate": 7.126765429053732e-06,
"loss": 0.09449903666973114,
"step": 1037
},
{
"epoch": 1.1972318339100345,
"grad_norm": 0.45948298891910827,
"learning_rate": 7.109214121460215e-06,
"loss": 0.08699238300323486,
"step": 1038
},
{
"epoch": 1.1983852364475203,
"grad_norm": 0.524269158749167,
"learning_rate": 7.0916725255366136e-06,
"loss": 0.11119051277637482,
"step": 1039
},
{
"epoch": 1.1995386389850058,
"grad_norm": 0.4670984408906168,
"learning_rate": 7.07414070021438e-06,
"loss": 0.08967498689889908,
"step": 1040
},
{
"epoch": 1.2006920415224913,
"grad_norm": 0.4975162623942099,
"learning_rate": 7.056618704392128e-06,
"loss": 0.08757388591766357,
"step": 1041
},
{
"epoch": 1.2018454440599768,
"grad_norm": 0.4780706337201267,
"learning_rate": 7.039106596935465e-06,
"loss": 0.10718037188053131,
"step": 1042
},
{
"epoch": 1.2029988465974626,
"grad_norm": 0.5135631607664823,
"learning_rate": 7.021604436676757e-06,
"loss": 0.09813686460256577,
"step": 1043
},
{
"epoch": 1.2041522491349481,
"grad_norm": 0.5065903814466779,
"learning_rate": 7.004112282414974e-06,
"loss": 0.09846250712871552,
"step": 1044
},
{
"epoch": 1.2053056516724336,
"grad_norm": 0.4902689258853185,
"learning_rate": 6.986630192915457e-06,
"loss": 0.10626232624053955,
"step": 1045
},
{
"epoch": 1.2064590542099192,
"grad_norm": 0.504817311681189,
"learning_rate": 6.969158226909732e-06,
"loss": 0.11492729187011719,
"step": 1046
},
{
"epoch": 1.207612456747405,
"grad_norm": 0.4632942056905242,
"learning_rate": 6.951696443095331e-06,
"loss": 0.08765972405672073,
"step": 1047
},
{
"epoch": 1.2087658592848904,
"grad_norm": 0.5313721042777281,
"learning_rate": 6.934244900135559e-06,
"loss": 0.11615556478500366,
"step": 1048
},
{
"epoch": 1.209919261822376,
"grad_norm": 0.559049991836223,
"learning_rate": 6.916803656659335e-06,
"loss": 0.0846204161643982,
"step": 1049
},
{
"epoch": 1.2110726643598615,
"grad_norm": 0.460708954781832,
"learning_rate": 6.899372771260957e-06,
"loss": 0.09682544320821762,
"step": 1050
},
{
"epoch": 1.2122260668973472,
"grad_norm": 0.4535747655253053,
"learning_rate": 6.8819523024999454e-06,
"loss": 0.08187022805213928,
"step": 1051
},
{
"epoch": 1.2133794694348328,
"grad_norm": 0.49447721761842983,
"learning_rate": 6.864542308900809e-06,
"loss": 0.09626784920692444,
"step": 1052
},
{
"epoch": 1.2145328719723183,
"grad_norm": 0.48902845697289193,
"learning_rate": 6.847142848952877e-06,
"loss": 0.10504262149333954,
"step": 1053
},
{
"epoch": 1.215686274509804,
"grad_norm": 0.5010152534031115,
"learning_rate": 6.829753981110079e-06,
"loss": 0.08587254583835602,
"step": 1054
},
{
"epoch": 1.2168396770472896,
"grad_norm": 0.5214192342477043,
"learning_rate": 6.812375763790769e-06,
"loss": 0.09980383515357971,
"step": 1055
},
{
"epoch": 1.217993079584775,
"grad_norm": 0.4842242084454059,
"learning_rate": 6.7950082553775185e-06,
"loss": 0.10281448066234589,
"step": 1056
},
{
"epoch": 1.2191464821222606,
"grad_norm": 0.5175555400455312,
"learning_rate": 6.777651514216917e-06,
"loss": 0.11097170412540436,
"step": 1057
},
{
"epoch": 1.2202998846597461,
"grad_norm": 0.5114483051534833,
"learning_rate": 6.760305598619394e-06,
"loss": 0.10225895792245865,
"step": 1058
},
{
"epoch": 1.221453287197232,
"grad_norm": 0.4860780522020304,
"learning_rate": 6.74297056685899e-06,
"loss": 0.10749275237321854,
"step": 1059
},
{
"epoch": 1.2226066897347174,
"grad_norm": 0.5091761402344968,
"learning_rate": 6.725646477173204e-06,
"loss": 0.08309759944677353,
"step": 1060
},
{
"epoch": 1.223760092272203,
"grad_norm": 0.49154901317590366,
"learning_rate": 6.708333387762753e-06,
"loss": 0.10968708992004395,
"step": 1061
},
{
"epoch": 1.2249134948096887,
"grad_norm": 0.5242418899891917,
"learning_rate": 6.691031356791417e-06,
"loss": 0.08381320536136627,
"step": 1062
},
{
"epoch": 1.2260668973471742,
"grad_norm": 0.5137469263179224,
"learning_rate": 6.6737404423858135e-06,
"loss": 0.11715946346521378,
"step": 1063
},
{
"epoch": 1.2272202998846597,
"grad_norm": 0.47752705871823825,
"learning_rate": 6.656460702635213e-06,
"loss": 0.08981136977672577,
"step": 1064
},
{
"epoch": 1.2283737024221453,
"grad_norm": 0.4759861471766702,
"learning_rate": 6.639192195591353e-06,
"loss": 0.09328687191009521,
"step": 1065
},
{
"epoch": 1.2295271049596308,
"grad_norm": 0.4868380888784573,
"learning_rate": 6.621934979268225e-06,
"loss": 0.08579070121049881,
"step": 1066
},
{
"epoch": 1.2306805074971165,
"grad_norm": 0.5121258904408004,
"learning_rate": 6.6046891116419e-06,
"loss": 0.08853509277105331,
"step": 1067
},
{
"epoch": 1.231833910034602,
"grad_norm": 0.504039744029831,
"learning_rate": 6.587454650650309e-06,
"loss": 0.08403575420379639,
"step": 1068
},
{
"epoch": 1.2329873125720876,
"grad_norm": 0.4941910499355994,
"learning_rate": 6.570231654193078e-06,
"loss": 0.10076552629470825,
"step": 1069
},
{
"epoch": 1.2341407151095733,
"grad_norm": 0.520860295401891,
"learning_rate": 6.553020180131302e-06,
"loss": 0.07828101515769958,
"step": 1070
},
{
"epoch": 1.2352941176470589,
"grad_norm": 0.48995727228242386,
"learning_rate": 6.535820286287371e-06,
"loss": 0.10417379438877106,
"step": 1071
},
{
"epoch": 1.2364475201845444,
"grad_norm": 0.6628954386782753,
"learning_rate": 6.5186320304447805e-06,
"loss": 0.08277340978384018,
"step": 1072
},
{
"epoch": 1.23760092272203,
"grad_norm": 0.4929201555071923,
"learning_rate": 6.501455470347917e-06,
"loss": 0.08730141073465347,
"step": 1073
},
{
"epoch": 1.2387543252595155,
"grad_norm": 0.5358201455668762,
"learning_rate": 6.484290663701877e-06,
"loss": 0.07814203202724457,
"step": 1074
},
{
"epoch": 1.2399077277970012,
"grad_norm": 0.5842554197762848,
"learning_rate": 6.467137668172274e-06,
"loss": 0.12421269714832306,
"step": 1075
},
{
"epoch": 1.2410611303344867,
"grad_norm": 0.5530594585875135,
"learning_rate": 6.449996541385044e-06,
"loss": 0.09736786782741547,
"step": 1076
},
{
"epoch": 1.2422145328719723,
"grad_norm": 0.5366949706571222,
"learning_rate": 6.432867340926238e-06,
"loss": 0.10894203186035156,
"step": 1077
},
{
"epoch": 1.243367935409458,
"grad_norm": 0.533823672142275,
"learning_rate": 6.415750124341855e-06,
"loss": 0.07957334816455841,
"step": 1078
},
{
"epoch": 1.2445213379469435,
"grad_norm": 0.5398533966282152,
"learning_rate": 6.3986449491376246e-06,
"loss": 0.08436086773872375,
"step": 1079
},
{
"epoch": 1.245674740484429,
"grad_norm": 0.557111343500373,
"learning_rate": 6.381551872778819e-06,
"loss": 0.09227782487869263,
"step": 1080
},
{
"epoch": 1.2468281430219146,
"grad_norm": 0.500638630988167,
"learning_rate": 6.364470952690078e-06,
"loss": 0.09067467600107193,
"step": 1081
},
{
"epoch": 1.2479815455594,
"grad_norm": 0.5181104930156516,
"learning_rate": 6.347402246255192e-06,
"loss": 0.08355002105236053,
"step": 1082
},
{
"epoch": 1.2491349480968859,
"grad_norm": 0.5241081163626568,
"learning_rate": 6.33034581081692e-06,
"loss": 0.08575861155986786,
"step": 1083
},
{
"epoch": 1.2502883506343714,
"grad_norm": 0.4883906229893202,
"learning_rate": 6.3133017036768e-06,
"loss": 0.09739427268505096,
"step": 1084
},
{
"epoch": 1.251441753171857,
"grad_norm": 0.5064508188720717,
"learning_rate": 6.296269982094953e-06,
"loss": 0.08717295527458191,
"step": 1085
},
{
"epoch": 1.2525951557093427,
"grad_norm": 0.5407323691523166,
"learning_rate": 6.2792507032898815e-06,
"loss": 0.111013263463974,
"step": 1086
},
{
"epoch": 1.2537485582468282,
"grad_norm": 0.5526712696722236,
"learning_rate": 6.2622439244383e-06,
"loss": 0.08177759498357773,
"step": 1087
},
{
"epoch": 1.2549019607843137,
"grad_norm": 0.49088562718217044,
"learning_rate": 6.24524970267492e-06,
"loss": 0.08803164958953857,
"step": 1088
},
{
"epoch": 1.2560553633217992,
"grad_norm": 0.5314875415844492,
"learning_rate": 6.228268095092263e-06,
"loss": 0.08679408580064774,
"step": 1089
},
{
"epoch": 1.2572087658592848,
"grad_norm": 0.5186180203152905,
"learning_rate": 6.211299158740486e-06,
"loss": 0.09869305044412613,
"step": 1090
},
{
"epoch": 1.2583621683967705,
"grad_norm": 0.4776143750324184,
"learning_rate": 6.1943429506271655e-06,
"loss": 0.08108265697956085,
"step": 1091
},
{
"epoch": 1.259515570934256,
"grad_norm": 0.5158201085971993,
"learning_rate": 6.1773995277171235e-06,
"loss": 0.10209430754184723,
"step": 1092
},
{
"epoch": 1.2606689734717416,
"grad_norm": 0.49351992674337575,
"learning_rate": 6.160468946932225e-06,
"loss": 0.08316084742546082,
"step": 1093
},
{
"epoch": 1.2618223760092273,
"grad_norm": 0.47647124444128075,
"learning_rate": 6.1435512651512e-06,
"loss": 0.10119596123695374,
"step": 1094
},
{
"epoch": 1.2629757785467128,
"grad_norm": 0.6142901971995941,
"learning_rate": 6.126646539209431e-06,
"loss": 0.08824190497398376,
"step": 1095
},
{
"epoch": 1.2641291810841984,
"grad_norm": 0.5186130137117047,
"learning_rate": 6.109754825898783e-06,
"loss": 0.1093616634607315,
"step": 1096
},
{
"epoch": 1.2652825836216839,
"grad_norm": 0.45832295833534004,
"learning_rate": 6.092876181967408e-06,
"loss": 0.09331536293029785,
"step": 1097
},
{
"epoch": 1.2664359861591694,
"grad_norm": 0.46596822595970633,
"learning_rate": 6.0760106641195406e-06,
"loss": 0.08940400183200836,
"step": 1098
},
{
"epoch": 1.2675893886966552,
"grad_norm": 0.4774216502239242,
"learning_rate": 6.059158329015329e-06,
"loss": 0.10458675026893616,
"step": 1099
},
{
"epoch": 1.2687427912341407,
"grad_norm": 0.5099450447776908,
"learning_rate": 6.042319233270627e-06,
"loss": 0.08113732933998108,
"step": 1100
},
{
"epoch": 1.2698961937716264,
"grad_norm": 0.5012341836018229,
"learning_rate": 6.025493433456813e-06,
"loss": 0.08787193149328232,
"step": 1101
},
{
"epoch": 1.271049596309112,
"grad_norm": 0.5272494201023792,
"learning_rate": 6.0086809861005945e-06,
"loss": 0.09474918991327286,
"step": 1102
},
{
"epoch": 1.2722029988465975,
"grad_norm": 0.5496764276942484,
"learning_rate": 5.991881947683828e-06,
"loss": 0.11352242529392242,
"step": 1103
},
{
"epoch": 1.273356401384083,
"grad_norm": 0.508974513764939,
"learning_rate": 5.975096374643317e-06,
"loss": 0.09265436977148056,
"step": 1104
},
{
"epoch": 1.2745098039215685,
"grad_norm": 0.4872379719082291,
"learning_rate": 5.958324323370619e-06,
"loss": 0.0872151181101799,
"step": 1105
},
{
"epoch": 1.2756632064590543,
"grad_norm": 0.5199480133175085,
"learning_rate": 5.941565850211886e-06,
"loss": 0.09725511074066162,
"step": 1106
},
{
"epoch": 1.2768166089965398,
"grad_norm": 0.5102384734937775,
"learning_rate": 5.924821011467632e-06,
"loss": 0.10504769533872604,
"step": 1107
},
{
"epoch": 1.2779700115340253,
"grad_norm": 0.5034609414986727,
"learning_rate": 5.908089863392583e-06,
"loss": 0.09033522754907608,
"step": 1108
},
{
"epoch": 1.279123414071511,
"grad_norm": 0.5291901100071729,
"learning_rate": 5.891372462195459e-06,
"loss": 0.12067270278930664,
"step": 1109
},
{
"epoch": 1.2802768166089966,
"grad_norm": 0.5143690352209813,
"learning_rate": 5.874668864038808e-06,
"loss": 0.08715777099132538,
"step": 1110
},
{
"epoch": 1.2814302191464821,
"grad_norm": 0.5120680745359965,
"learning_rate": 5.857979125038795e-06,
"loss": 0.1045406311750412,
"step": 1111
},
{
"epoch": 1.2825836216839677,
"grad_norm": 0.5220036036709266,
"learning_rate": 5.841303301265027e-06,
"loss": 0.07616324722766876,
"step": 1112
},
{
"epoch": 1.2837370242214532,
"grad_norm": 0.4810294683870358,
"learning_rate": 5.824641448740373e-06,
"loss": 0.08584298193454742,
"step": 1113
},
{
"epoch": 1.284890426758939,
"grad_norm": 0.47580798199708935,
"learning_rate": 5.807993623440746e-06,
"loss": 0.08185971528291702,
"step": 1114
},
{
"epoch": 1.2860438292964245,
"grad_norm": 0.4925385213197463,
"learning_rate": 5.791359881294962e-06,
"loss": 0.09213608503341675,
"step": 1115
},
{
"epoch": 1.28719723183391,
"grad_norm": 0.457164240147007,
"learning_rate": 5.774740278184492e-06,
"loss": 0.08021476864814758,
"step": 1116
},
{
"epoch": 1.2883506343713957,
"grad_norm": 0.4836397565759473,
"learning_rate": 5.758134869943335e-06,
"loss": 0.08997669816017151,
"step": 1117
},
{
"epoch": 1.2895040369088813,
"grad_norm": 0.5363412714110818,
"learning_rate": 5.7415437123577776e-06,
"loss": 0.09681268036365509,
"step": 1118
},
{
"epoch": 1.2906574394463668,
"grad_norm": 0.4985088214699496,
"learning_rate": 5.724966861166256e-06,
"loss": 0.087282195687294,
"step": 1119
},
{
"epoch": 1.2918108419838523,
"grad_norm": 0.5390463250702708,
"learning_rate": 5.708404372059118e-06,
"loss": 0.0947989970445633,
"step": 1120
},
{
"epoch": 1.2929642445213378,
"grad_norm": 0.5020916553222697,
"learning_rate": 5.691856300678478e-06,
"loss": 0.08303868770599365,
"step": 1121
},
{
"epoch": 1.2941176470588236,
"grad_norm": 0.528933364342531,
"learning_rate": 5.675322702618018e-06,
"loss": 0.09628322720527649,
"step": 1122
},
{
"epoch": 1.2952710495963091,
"grad_norm": 0.5195465632648432,
"learning_rate": 5.658803633422773e-06,
"loss": 0.0886673778295517,
"step": 1123
},
{
"epoch": 1.2964244521337946,
"grad_norm": 0.5181463312476864,
"learning_rate": 5.642299148588999e-06,
"loss": 0.11223585903644562,
"step": 1124
},
{
"epoch": 1.2975778546712804,
"grad_norm": 0.5279605605387799,
"learning_rate": 5.625809303563925e-06,
"loss": 0.09991466999053955,
"step": 1125
},
{
"epoch": 1.298731257208766,
"grad_norm": 0.48686887305452325,
"learning_rate": 5.609334153745628e-06,
"loss": 0.0920310914516449,
"step": 1126
},
{
"epoch": 1.2998846597462514,
"grad_norm": 0.48807133536723973,
"learning_rate": 5.592873754482781e-06,
"loss": 0.08726586401462555,
"step": 1127
},
{
"epoch": 1.301038062283737,
"grad_norm": 0.4526244646368317,
"learning_rate": 5.576428161074541e-06,
"loss": 0.08500439673662186,
"step": 1128
},
{
"epoch": 1.3021914648212225,
"grad_norm": 0.49550547512807813,
"learning_rate": 5.55999742877029e-06,
"loss": 0.08454027771949768,
"step": 1129
},
{
"epoch": 1.3033448673587082,
"grad_norm": 0.4838682640144393,
"learning_rate": 5.543581612769498e-06,
"loss": 0.08620169758796692,
"step": 1130
},
{
"epoch": 1.3044982698961938,
"grad_norm": 0.4765713073281102,
"learning_rate": 5.527180768221536e-06,
"loss": 0.0826345831155777,
"step": 1131
},
{
"epoch": 1.3056516724336793,
"grad_norm": 0.4942475015165598,
"learning_rate": 5.510794950225448e-06,
"loss": 0.0953306257724762,
"step": 1132
},
{
"epoch": 1.306805074971165,
"grad_norm": 0.44920895037104636,
"learning_rate": 5.494424213829832e-06,
"loss": 0.08424252271652222,
"step": 1133
},
{
"epoch": 1.3079584775086506,
"grad_norm": 0.4680295307795166,
"learning_rate": 5.478068614032584e-06,
"loss": 0.08729878067970276,
"step": 1134
},
{
"epoch": 1.309111880046136,
"grad_norm": 0.9268281946572282,
"learning_rate": 5.461728205780777e-06,
"loss": 0.08794678747653961,
"step": 1135
},
{
"epoch": 1.3102652825836216,
"grad_norm": 0.5292653690977969,
"learning_rate": 5.445403043970427e-06,
"loss": 0.10001164674758911,
"step": 1136
},
{
"epoch": 1.3114186851211072,
"grad_norm": 0.5943457086043997,
"learning_rate": 5.429093183446339e-06,
"loss": 0.09134934097528458,
"step": 1137
},
{
"epoch": 1.312572087658593,
"grad_norm": 0.5179802790257814,
"learning_rate": 5.412798679001912e-06,
"loss": 0.09967174381017685,
"step": 1138
},
{
"epoch": 1.3137254901960784,
"grad_norm": 0.510163328544469,
"learning_rate": 5.39651958537895e-06,
"loss": 0.08825188875198364,
"step": 1139
},
{
"epoch": 1.314878892733564,
"grad_norm": 0.5520274094430528,
"learning_rate": 5.380255957267504e-06,
"loss": 0.09385664761066437,
"step": 1140
},
{
"epoch": 1.3160322952710497,
"grad_norm": 0.48588560932271924,
"learning_rate": 5.364007849305635e-06,
"loss": 0.08724480867385864,
"step": 1141
},
{
"epoch": 1.3171856978085352,
"grad_norm": 0.5242982939879761,
"learning_rate": 5.347775316079298e-06,
"loss": 0.10368078947067261,
"step": 1142
},
{
"epoch": 1.3183391003460208,
"grad_norm": 0.5465640687689112,
"learning_rate": 5.331558412122092e-06,
"loss": 0.08240247517824173,
"step": 1143
},
{
"epoch": 1.3194925028835063,
"grad_norm": 0.4820788471611164,
"learning_rate": 5.315357191915141e-06,
"loss": 0.08277444541454315,
"step": 1144
},
{
"epoch": 1.3206459054209918,
"grad_norm": 0.4638089933744181,
"learning_rate": 5.2991717098868534e-06,
"loss": 0.08547809720039368,
"step": 1145
},
{
"epoch": 1.3217993079584776,
"grad_norm": 0.5367881982395373,
"learning_rate": 5.283002020412777e-06,
"loss": 0.10128656029701233,
"step": 1146
},
{
"epoch": 1.322952710495963,
"grad_norm": 0.5548074677296512,
"learning_rate": 5.266848177815402e-06,
"loss": 0.09210200607776642,
"step": 1147
},
{
"epoch": 1.3241061130334486,
"grad_norm": 0.46471853395075313,
"learning_rate": 5.250710236363976e-06,
"loss": 0.08336486667394638,
"step": 1148
},
{
"epoch": 1.3252595155709344,
"grad_norm": 0.49365986706679243,
"learning_rate": 5.2345882502743415e-06,
"loss": 0.08845160901546478,
"step": 1149
},
{
"epoch": 1.3264129181084199,
"grad_norm": 0.6049999870630645,
"learning_rate": 5.218482273708716e-06,
"loss": 0.1451820731163025,
"step": 1150
},
{
"epoch": 1.3275663206459054,
"grad_norm": 0.5940508742058602,
"learning_rate": 5.202392360775556e-06,
"loss": 0.09655202925205231,
"step": 1151
},
{
"epoch": 1.328719723183391,
"grad_norm": 0.5107202877112091,
"learning_rate": 5.186318565529325e-06,
"loss": 0.10245556384325027,
"step": 1152
},
{
"epoch": 1.3298731257208765,
"grad_norm": 0.4917746847730811,
"learning_rate": 5.170260941970371e-06,
"loss": 0.08417163044214249,
"step": 1153
},
{
"epoch": 1.3310265282583622,
"grad_norm": 0.53080621009012,
"learning_rate": 5.154219544044684e-06,
"loss": 0.10081610083580017,
"step": 1154
},
{
"epoch": 1.3321799307958477,
"grad_norm": 0.5112307443919099,
"learning_rate": 5.138194425643757e-06,
"loss": 0.10317876189947128,
"step": 1155
},
{
"epoch": 1.3333333333333333,
"grad_norm": 0.4527248745177868,
"learning_rate": 5.122185640604391e-06,
"loss": 0.0830250084400177,
"step": 1156
},
{
"epoch": 1.334486735870819,
"grad_norm": 0.4795790499105225,
"learning_rate": 5.10619324270851e-06,
"loss": 0.09759711474180222,
"step": 1157
},
{
"epoch": 1.3356401384083045,
"grad_norm": 0.48060255331324636,
"learning_rate": 5.090217285682999e-06,
"loss": 0.0819205790758133,
"step": 1158
},
{
"epoch": 1.33679354094579,
"grad_norm": 0.528408061320208,
"learning_rate": 5.074257823199485e-06,
"loss": 0.1273103803396225,
"step": 1159
},
{
"epoch": 1.3379469434832756,
"grad_norm": 0.5165064721911903,
"learning_rate": 5.058314908874206e-06,
"loss": 0.07449788600206375,
"step": 1160
},
{
"epoch": 1.3391003460207611,
"grad_norm": 0.46541544463379503,
"learning_rate": 5.0423885962677885e-06,
"loss": 0.09530683606863022,
"step": 1161
},
{
"epoch": 1.3402537485582469,
"grad_norm": 0.8214019070815849,
"learning_rate": 5.026478938885093e-06,
"loss": 0.0844576433300972,
"step": 1162
},
{
"epoch": 1.3414071510957324,
"grad_norm": 0.517070707727408,
"learning_rate": 5.010585990175027e-06,
"loss": 0.0959005132317543,
"step": 1163
},
{
"epoch": 1.342560553633218,
"grad_norm": 0.544523534510525,
"learning_rate": 4.9947098035303586e-06,
"loss": 0.1044202670454979,
"step": 1164
},
{
"epoch": 1.3437139561707037,
"grad_norm": 0.5949562714957899,
"learning_rate": 4.978850432287551e-06,
"loss": 0.08823059499263763,
"step": 1165
},
{
"epoch": 1.3448673587081892,
"grad_norm": 0.5081981200861693,
"learning_rate": 4.963007929726568e-06,
"loss": 0.08562524616718292,
"step": 1166
},
{
"epoch": 1.3460207612456747,
"grad_norm": 0.49670658811222096,
"learning_rate": 4.947182349070715e-06,
"loss": 0.08057855814695358,
"step": 1167
},
{
"epoch": 1.3471741637831602,
"grad_norm": 0.5490532587924616,
"learning_rate": 4.931373743486427e-06,
"loss": 0.10181853175163269,
"step": 1168
},
{
"epoch": 1.348327566320646,
"grad_norm": 0.47955951911667216,
"learning_rate": 4.9155821660831396e-06,
"loss": 0.09622423350811005,
"step": 1169
},
{
"epoch": 1.3494809688581315,
"grad_norm": 0.4716695202322506,
"learning_rate": 4.899807669913051e-06,
"loss": 0.08078745007514954,
"step": 1170
},
{
"epoch": 1.350634371395617,
"grad_norm": 0.5676071693005171,
"learning_rate": 4.884050307970994e-06,
"loss": 0.0984252542257309,
"step": 1171
},
{
"epoch": 1.3517877739331028,
"grad_norm": 0.463011077766537,
"learning_rate": 4.868310133194235e-06,
"loss": 0.07783740758895874,
"step": 1172
},
{
"epoch": 1.3529411764705883,
"grad_norm": 0.4823346913639178,
"learning_rate": 4.8525871984622965e-06,
"loss": 0.0836503654718399,
"step": 1173
},
{
"epoch": 1.3540945790080738,
"grad_norm": 0.5105201008610478,
"learning_rate": 4.8368815565967845e-06,
"loss": 0.0777088925242424,
"step": 1174
},
{
"epoch": 1.3552479815455594,
"grad_norm": 0.4941952887672344,
"learning_rate": 4.8211932603612056e-06,
"loss": 0.1049129068851471,
"step": 1175
},
{
"epoch": 1.356401384083045,
"grad_norm": 0.47097315944052,
"learning_rate": 4.805522362460809e-06,
"loss": 0.09377576410770416,
"step": 1176
},
{
"epoch": 1.3575547866205306,
"grad_norm": 0.49344652854833276,
"learning_rate": 4.789868915542365e-06,
"loss": 0.08657893538475037,
"step": 1177
},
{
"epoch": 1.3587081891580162,
"grad_norm": 0.5244724005683726,
"learning_rate": 4.774232972194048e-06,
"loss": 0.09125713258981705,
"step": 1178
},
{
"epoch": 1.3598615916955017,
"grad_norm": 0.45064702104131016,
"learning_rate": 4.758614584945205e-06,
"loss": 0.0777345523238182,
"step": 1179
},
{
"epoch": 1.3610149942329874,
"grad_norm": 0.4629886241619075,
"learning_rate": 4.743013806266214e-06,
"loss": 0.07935482263565063,
"step": 1180
},
{
"epoch": 1.362168396770473,
"grad_norm": 0.525314691614035,
"learning_rate": 4.727430688568294e-06,
"loss": 0.11505504697561264,
"step": 1181
},
{
"epoch": 1.3633217993079585,
"grad_norm": 0.4749288663446185,
"learning_rate": 4.71186528420333e-06,
"loss": 0.08527901768684387,
"step": 1182
},
{
"epoch": 1.364475201845444,
"grad_norm": 0.4704118601011978,
"learning_rate": 4.69631764546371e-06,
"loss": 0.08753247559070587,
"step": 1183
},
{
"epoch": 1.3656286043829295,
"grad_norm": 0.481622422010927,
"learning_rate": 4.680787824582113e-06,
"loss": 0.09783068299293518,
"step": 1184
},
{
"epoch": 1.3667820069204153,
"grad_norm": 0.47959709800828304,
"learning_rate": 4.665275873731392e-06,
"loss": 0.09854909777641296,
"step": 1185
},
{
"epoch": 1.3679354094579008,
"grad_norm": 0.447728460027515,
"learning_rate": 4.649781845024333e-06,
"loss": 0.08357270061969757,
"step": 1186
},
{
"epoch": 1.3690888119953863,
"grad_norm": 0.4592438605823468,
"learning_rate": 4.634305790513529e-06,
"loss": 0.08320782333612442,
"step": 1187
},
{
"epoch": 1.370242214532872,
"grad_norm": 0.4864981030060889,
"learning_rate": 4.618847762191189e-06,
"loss": 0.08831924200057983,
"step": 1188
},
{
"epoch": 1.3713956170703576,
"grad_norm": 0.4510100672054657,
"learning_rate": 4.603407811988956e-06,
"loss": 0.09139207750558853,
"step": 1189
},
{
"epoch": 1.3725490196078431,
"grad_norm": 0.5114977204093722,
"learning_rate": 4.587985991777744e-06,
"loss": 0.11378781497478485,
"step": 1190
},
{
"epoch": 1.3737024221453287,
"grad_norm": 0.4944951341432841,
"learning_rate": 4.57258235336755e-06,
"loss": 0.09309037029743195,
"step": 1191
},
{
"epoch": 1.3748558246828142,
"grad_norm": 0.47120187692507726,
"learning_rate": 4.557196948507311e-06,
"loss": 0.0843665823340416,
"step": 1192
},
{
"epoch": 1.3760092272203,
"grad_norm": 0.5127696916400873,
"learning_rate": 4.541829828884674e-06,
"loss": 0.08751867711544037,
"step": 1193
},
{
"epoch": 1.3771626297577855,
"grad_norm": 0.5112634026223221,
"learning_rate": 4.526481046125891e-06,
"loss": 0.08220349252223969,
"step": 1194
},
{
"epoch": 1.378316032295271,
"grad_norm": 0.464705694444387,
"learning_rate": 4.511150651795584e-06,
"loss": 0.08250938355922699,
"step": 1195
},
{
"epoch": 1.3794694348327567,
"grad_norm": 0.5409947247038671,
"learning_rate": 4.4958386973966125e-06,
"loss": 0.10774454474449158,
"step": 1196
},
{
"epoch": 1.3806228373702423,
"grad_norm": 0.541102041183711,
"learning_rate": 4.480545234369884e-06,
"loss": 0.1112159714102745,
"step": 1197
},
{
"epoch": 1.3817762399077278,
"grad_norm": 0.5034326809508693,
"learning_rate": 4.465270314094181e-06,
"loss": 0.07743170857429504,
"step": 1198
},
{
"epoch": 1.3829296424452133,
"grad_norm": 0.49063289154619055,
"learning_rate": 4.450013987885996e-06,
"loss": 0.09016168117523193,
"step": 1199
},
{
"epoch": 1.3840830449826989,
"grad_norm": 0.5536891472289158,
"learning_rate": 4.434776306999345e-06,
"loss": 0.0837908536195755,
"step": 1200
},
{
"epoch": 1.3852364475201846,
"grad_norm": 0.5083962391757554,
"learning_rate": 4.419557322625625e-06,
"loss": 0.08363713324069977,
"step": 1201
},
{
"epoch": 1.3863898500576701,
"grad_norm": 0.4852864372610325,
"learning_rate": 4.404357085893389e-06,
"loss": 0.10005521774291992,
"step": 1202
},
{
"epoch": 1.3875432525951557,
"grad_norm": 0.5568987616818593,
"learning_rate": 4.389175647868245e-06,
"loss": 0.0869610607624054,
"step": 1203
},
{
"epoch": 1.3886966551326414,
"grad_norm": 0.5467006333899488,
"learning_rate": 4.374013059552611e-06,
"loss": 0.11631981283426285,
"step": 1204
},
{
"epoch": 1.389850057670127,
"grad_norm": 0.5004638656697998,
"learning_rate": 4.3588693718856e-06,
"loss": 0.07439540326595306,
"step": 1205
},
{
"epoch": 1.3910034602076125,
"grad_norm": 0.4998381891942109,
"learning_rate": 4.343744635742824e-06,
"loss": 0.08437787741422653,
"step": 1206
},
{
"epoch": 1.392156862745098,
"grad_norm": 0.5155886917153059,
"learning_rate": 4.328638901936223e-06,
"loss": 0.1088452860713005,
"step": 1207
},
{
"epoch": 1.3933102652825835,
"grad_norm": 0.4626559143040623,
"learning_rate": 4.313552221213901e-06,
"loss": 0.07555978000164032,
"step": 1208
},
{
"epoch": 1.3944636678200693,
"grad_norm": 0.5192520301471394,
"learning_rate": 4.2984846442599495e-06,
"loss": 0.1174163967370987,
"step": 1209
},
{
"epoch": 1.3956170703575548,
"grad_norm": 0.5041858564795372,
"learning_rate": 4.283436221694293e-06,
"loss": 0.09407342970371246,
"step": 1210
},
{
"epoch": 1.3967704728950403,
"grad_norm": 0.4643839906580603,
"learning_rate": 4.268407004072485e-06,
"loss": 0.08721020817756653,
"step": 1211
},
{
"epoch": 1.397923875432526,
"grad_norm": 0.48174100727151536,
"learning_rate": 4.253397041885574e-06,
"loss": 0.07627159357070923,
"step": 1212
},
{
"epoch": 1.3990772779700116,
"grad_norm": 0.473580345282926,
"learning_rate": 4.2384063855599164e-06,
"loss": 0.07980696856975555,
"step": 1213
},
{
"epoch": 1.400230680507497,
"grad_norm": 0.4908552611174717,
"learning_rate": 4.223435085457012e-06,
"loss": 0.1050553172826767,
"step": 1214
},
{
"epoch": 1.4013840830449826,
"grad_norm": 0.5183885404992002,
"learning_rate": 4.208483191873328e-06,
"loss": 0.08839884400367737,
"step": 1215
},
{
"epoch": 1.4025374855824682,
"grad_norm": 0.5249247137180852,
"learning_rate": 4.19355075504014e-06,
"loss": 0.08334405720233917,
"step": 1216
},
{
"epoch": 1.403690888119954,
"grad_norm": 0.4726097805920548,
"learning_rate": 4.178637825123353e-06,
"loss": 0.09830236434936523,
"step": 1217
},
{
"epoch": 1.4048442906574394,
"grad_norm": 0.506151120720196,
"learning_rate": 4.163744452223338e-06,
"loss": 0.09475752711296082,
"step": 1218
},
{
"epoch": 1.405997693194925,
"grad_norm": 0.5111565940240601,
"learning_rate": 4.148870686374778e-06,
"loss": 0.08506309241056442,
"step": 1219
},
{
"epoch": 1.4071510957324107,
"grad_norm": 0.468851162026766,
"learning_rate": 4.134016577546463e-06,
"loss": 0.08116403222084045,
"step": 1220
},
{
"epoch": 1.4083044982698962,
"grad_norm": 0.4895444737244898,
"learning_rate": 4.11918217564116e-06,
"loss": 0.08761732280254364,
"step": 1221
},
{
"epoch": 1.4094579008073818,
"grad_norm": 0.48203098333140004,
"learning_rate": 4.104367530495424e-06,
"loss": 0.08589303493499756,
"step": 1222
},
{
"epoch": 1.4106113033448673,
"grad_norm": 0.5089708008190156,
"learning_rate": 4.089572691879441e-06,
"loss": 0.08399146795272827,
"step": 1223
},
{
"epoch": 1.4117647058823528,
"grad_norm": 0.5428912261753253,
"learning_rate": 4.0747977094968505e-06,
"loss": 0.07639371603727341,
"step": 1224
},
{
"epoch": 1.4129181084198386,
"grad_norm": 0.5338443768258041,
"learning_rate": 4.060042632984591e-06,
"loss": 0.08502918481826782,
"step": 1225
},
{
"epoch": 1.414071510957324,
"grad_norm": 0.4847131753051582,
"learning_rate": 4.045307511912723e-06,
"loss": 0.09035171568393707,
"step": 1226
},
{
"epoch": 1.4152249134948096,
"grad_norm": 0.4992345879547538,
"learning_rate": 4.0305923957842615e-06,
"loss": 0.07799342274665833,
"step": 1227
},
{
"epoch": 1.4163783160322954,
"grad_norm": 0.5028127437708988,
"learning_rate": 4.015897334035033e-06,
"loss": 0.08772207796573639,
"step": 1228
},
{
"epoch": 1.4175317185697809,
"grad_norm": 0.48029334493691606,
"learning_rate": 4.001222376033464e-06,
"loss": 0.08191703259944916,
"step": 1229
},
{
"epoch": 1.4186851211072664,
"grad_norm": 0.5405987552099338,
"learning_rate": 3.986567571080462e-06,
"loss": 0.1108025386929512,
"step": 1230
},
{
"epoch": 1.419838523644752,
"grad_norm": 0.5584439486966768,
"learning_rate": 3.971932968409221e-06,
"loss": 0.07912584394216537,
"step": 1231
},
{
"epoch": 1.4209919261822375,
"grad_norm": 0.5205550739552034,
"learning_rate": 3.9573186171850685e-06,
"loss": 0.09308473765850067,
"step": 1232
},
{
"epoch": 1.4221453287197232,
"grad_norm": 0.565873566726639,
"learning_rate": 3.942724566505294e-06,
"loss": 0.09804917871952057,
"step": 1233
},
{
"epoch": 1.4232987312572087,
"grad_norm": 0.4918892254363981,
"learning_rate": 3.928150865398989e-06,
"loss": 0.0782637894153595,
"step": 1234
},
{
"epoch": 1.4244521337946945,
"grad_norm": 0.4706469015944614,
"learning_rate": 3.91359756282688e-06,
"loss": 0.09189221262931824,
"step": 1235
},
{
"epoch": 1.42560553633218,
"grad_norm": 0.4888588472550852,
"learning_rate": 3.8990647076811615e-06,
"loss": 0.08393605798482895,
"step": 1236
},
{
"epoch": 1.4267589388696655,
"grad_norm": 0.5614821291867843,
"learning_rate": 3.884552348785338e-06,
"loss": 0.12941831350326538,
"step": 1237
},
{
"epoch": 1.427912341407151,
"grad_norm": 0.5082338587623986,
"learning_rate": 3.870060534894052e-06,
"loss": 0.07559476792812347,
"step": 1238
},
{
"epoch": 1.4290657439446366,
"grad_norm": 0.4824916833194081,
"learning_rate": 3.855589314692931e-06,
"loss": 0.09814043343067169,
"step": 1239
},
{
"epoch": 1.4302191464821223,
"grad_norm": 0.48706096092282386,
"learning_rate": 3.84113873679841e-06,
"loss": 0.0985657274723053,
"step": 1240
},
{
"epoch": 1.4313725490196079,
"grad_norm": 0.5308778286608933,
"learning_rate": 3.826708849757581e-06,
"loss": 0.077473483979702,
"step": 1241
},
{
"epoch": 1.4325259515570934,
"grad_norm": 0.5227434393434751,
"learning_rate": 3.8122997020480224e-06,
"loss": 0.08808223903179169,
"step": 1242
},
{
"epoch": 1.4336793540945791,
"grad_norm": 0.46283667262817524,
"learning_rate": 3.797911342077638e-06,
"loss": 0.0912768542766571,
"step": 1243
},
{
"epoch": 1.4348327566320647,
"grad_norm": 0.49959968371422087,
"learning_rate": 3.7835438181844963e-06,
"loss": 0.08537976443767548,
"step": 1244
},
{
"epoch": 1.4359861591695502,
"grad_norm": 0.5047588404995199,
"learning_rate": 3.7691971786366643e-06,
"loss": 0.09026628732681274,
"step": 1245
},
{
"epoch": 1.4371395617070357,
"grad_norm": 0.5058758521903348,
"learning_rate": 3.7548714716320477e-06,
"loss": 0.11224260926246643,
"step": 1246
},
{
"epoch": 1.4382929642445212,
"grad_norm": 0.4867962521219152,
"learning_rate": 3.740566745298233e-06,
"loss": 0.08994022011756897,
"step": 1247
},
{
"epoch": 1.439446366782007,
"grad_norm": 0.4496312380317118,
"learning_rate": 3.726283047692315e-06,
"loss": 0.0852772444486618,
"step": 1248
},
{
"epoch": 1.4405997693194925,
"grad_norm": 0.519409377012893,
"learning_rate": 3.7120204268007464e-06,
"loss": 0.1168537437915802,
"step": 1249
},
{
"epoch": 1.441753171856978,
"grad_norm": 0.4655173638999558,
"learning_rate": 3.6977789305391708e-06,
"loss": 0.08203011006116867,
"step": 1250
},
{
"epoch": 1.4429065743944638,
"grad_norm": 0.5523329214276352,
"learning_rate": 3.6835586067522645e-06,
"loss": 0.12155325710773468,
"step": 1251
},
{
"epoch": 1.4440599769319493,
"grad_norm": 0.5007737603899637,
"learning_rate": 3.6693595032135664e-06,
"loss": 0.09166581928730011,
"step": 1252
},
{
"epoch": 1.4452133794694348,
"grad_norm": 0.5289486910279214,
"learning_rate": 3.655181667625346e-06,
"loss": 0.11802002787590027,
"step": 1253
},
{
"epoch": 1.4463667820069204,
"grad_norm": 0.49151304896977055,
"learning_rate": 3.6410251476183965e-06,
"loss": 0.10033787786960602,
"step": 1254
},
{
"epoch": 1.447520184544406,
"grad_norm": 0.48337061070334447,
"learning_rate": 3.626889990751916e-06,
"loss": 0.0916968509554863,
"step": 1255
},
{
"epoch": 1.4486735870818916,
"grad_norm": 0.4844919915917855,
"learning_rate": 3.612776244513331e-06,
"loss": 0.082099549472332,
"step": 1256
},
{
"epoch": 1.4498269896193772,
"grad_norm": 0.4741277293814583,
"learning_rate": 3.598683956318135e-06,
"loss": 0.09391163289546967,
"step": 1257
},
{
"epoch": 1.4509803921568627,
"grad_norm": 0.4629188538325247,
"learning_rate": 3.5846131735097366e-06,
"loss": 0.08973279595375061,
"step": 1258
},
{
"epoch": 1.4521337946943484,
"grad_norm": 0.4466920886821788,
"learning_rate": 3.570563943359293e-06,
"loss": 0.08001997321844101,
"step": 1259
},
{
"epoch": 1.453287197231834,
"grad_norm": 0.4848140420058611,
"learning_rate": 3.556536313065557e-06,
"loss": 0.09635907411575317,
"step": 1260
},
{
"epoch": 1.4544405997693195,
"grad_norm": 0.5011753391060038,
"learning_rate": 3.5425303297547163e-06,
"loss": 0.07817291468381882,
"step": 1261
},
{
"epoch": 1.455594002306805,
"grad_norm": 0.4856361011103989,
"learning_rate": 3.528546040480232e-06,
"loss": 0.10424292832612991,
"step": 1262
},
{
"epoch": 1.4567474048442905,
"grad_norm": 0.4700754724992772,
"learning_rate": 3.5145834922226875e-06,
"loss": 0.09175125509500504,
"step": 1263
},
{
"epoch": 1.4579008073817763,
"grad_norm": 0.5005745117204867,
"learning_rate": 3.500642731889623e-06,
"loss": 0.10491032898426056,
"step": 1264
},
{
"epoch": 1.4590542099192618,
"grad_norm": 0.502898102473839,
"learning_rate": 3.486723806315384e-06,
"loss": 0.09685451537370682,
"step": 1265
},
{
"epoch": 1.4602076124567474,
"grad_norm": 0.559034911954663,
"learning_rate": 3.4728267622609623e-06,
"loss": 0.12815001606941223,
"step": 1266
},
{
"epoch": 1.461361014994233,
"grad_norm": 0.4501763329994075,
"learning_rate": 3.458951646413835e-06,
"loss": 0.07831063121557236,
"step": 1267
},
{
"epoch": 1.4625144175317186,
"grad_norm": 0.4869571766104193,
"learning_rate": 3.445098505387814e-06,
"loss": 0.09639817476272583,
"step": 1268
},
{
"epoch": 1.4636678200692042,
"grad_norm": 0.5216308163274526,
"learning_rate": 3.4312673857228827e-06,
"loss": 0.09379415959119797,
"step": 1269
},
{
"epoch": 1.4648212226066897,
"grad_norm": 0.44323059795411895,
"learning_rate": 3.4174583338850476e-06,
"loss": 0.08638792484998703,
"step": 1270
},
{
"epoch": 1.4659746251441752,
"grad_norm": 0.497851700802119,
"learning_rate": 3.4036713962661747e-06,
"loss": 0.0931597650051117,
"step": 1271
},
{
"epoch": 1.467128027681661,
"grad_norm": 0.4715405170685261,
"learning_rate": 3.3899066191838382e-06,
"loss": 0.08654098957777023,
"step": 1272
},
{
"epoch": 1.4682814302191465,
"grad_norm": 0.43186503000424104,
"learning_rate": 3.3761640488811607e-06,
"loss": 0.0785563737154007,
"step": 1273
},
{
"epoch": 1.469434832756632,
"grad_norm": 0.4562506392990808,
"learning_rate": 3.362443731526663e-06,
"loss": 0.07910420000553131,
"step": 1274
},
{
"epoch": 1.4705882352941178,
"grad_norm": 0.4735535676237498,
"learning_rate": 3.3487457132141087e-06,
"loss": 0.08227407932281494,
"step": 1275
},
{
"epoch": 1.4717416378316033,
"grad_norm": 0.4424422620006913,
"learning_rate": 3.3350700399623416e-06,
"loss": 0.07782106846570969,
"step": 1276
},
{
"epoch": 1.4728950403690888,
"grad_norm": 0.45232969238926535,
"learning_rate": 3.3214167577151412e-06,
"loss": 0.0773819088935852,
"step": 1277
},
{
"epoch": 1.4740484429065743,
"grad_norm": 0.46502112235953913,
"learning_rate": 3.3077859123410617e-06,
"loss": 0.08315959572792053,
"step": 1278
},
{
"epoch": 1.4752018454440599,
"grad_norm": 0.4592918532931937,
"learning_rate": 3.2941775496332807e-06,
"loss": 0.07742109149694443,
"step": 1279
},
{
"epoch": 1.4763552479815456,
"grad_norm": 0.4704113459904759,
"learning_rate": 3.2805917153094445e-06,
"loss": 0.09802589565515518,
"step": 1280
},
{
"epoch": 1.4775086505190311,
"grad_norm": 0.4767455819596402,
"learning_rate": 3.267028455011515e-06,
"loss": 0.09542588889598846,
"step": 1281
},
{
"epoch": 1.4786620530565167,
"grad_norm": 0.4870819162674872,
"learning_rate": 3.253487814305617e-06,
"loss": 0.07993606477975845,
"step": 1282
},
{
"epoch": 1.4798154555940024,
"grad_norm": 0.4775620057295734,
"learning_rate": 3.2399698386818814e-06,
"loss": 0.09210764616727829,
"step": 1283
},
{
"epoch": 1.480968858131488,
"grad_norm": 0.49379115399209844,
"learning_rate": 3.226474573554298e-06,
"loss": 0.0995272547006607,
"step": 1284
},
{
"epoch": 1.4821222606689735,
"grad_norm": 0.5054648435807929,
"learning_rate": 3.2130020642605573e-06,
"loss": 0.08597026765346527,
"step": 1285
},
{
"epoch": 1.483275663206459,
"grad_norm": 0.46426308667970434,
"learning_rate": 3.1995523560619056e-06,
"loss": 0.08014822751283646,
"step": 1286
},
{
"epoch": 1.4844290657439445,
"grad_norm": 0.4870119788915944,
"learning_rate": 3.186125494142984e-06,
"loss": 0.093277208507061,
"step": 1287
},
{
"epoch": 1.4855824682814303,
"grad_norm": 0.5057666189329207,
"learning_rate": 3.1727215236116814e-06,
"loss": 0.09922486543655396,
"step": 1288
},
{
"epoch": 1.4867358708189158,
"grad_norm": 0.5190649806826937,
"learning_rate": 3.1593404894989856e-06,
"loss": 0.11097384989261627,
"step": 1289
},
{
"epoch": 1.4878892733564013,
"grad_norm": 0.46614595184097274,
"learning_rate": 3.1459824367588245e-06,
"loss": 0.09956075251102448,
"step": 1290
},
{
"epoch": 1.489042675893887,
"grad_norm": 0.46738228045425906,
"learning_rate": 3.1326474102679226e-06,
"loss": 0.0766315907239914,
"step": 1291
},
{
"epoch": 1.4901960784313726,
"grad_norm": 0.5010034301679498,
"learning_rate": 3.1193354548256462e-06,
"loss": 0.0758824273943901,
"step": 1292
},
{
"epoch": 1.491349480968858,
"grad_norm": 0.44913856528546603,
"learning_rate": 3.106046615153855e-06,
"loss": 0.09615721553564072,
"step": 1293
},
{
"epoch": 1.4925028835063436,
"grad_norm": 0.4779075514152234,
"learning_rate": 3.0927809358967477e-06,
"loss": 0.06804818660020828,
"step": 1294
},
{
"epoch": 1.4936562860438292,
"grad_norm": 0.49174963986730746,
"learning_rate": 3.079538461620718e-06,
"loss": 0.09505069255828857,
"step": 1295
},
{
"epoch": 1.494809688581315,
"grad_norm": 0.47676195743606614,
"learning_rate": 3.0663192368142003e-06,
"loss": 0.08175475895404816,
"step": 1296
},
{
"epoch": 1.4959630911188004,
"grad_norm": 0.4747902256238471,
"learning_rate": 3.053123305887522e-06,
"loss": 0.09949962794780731,
"step": 1297
},
{
"epoch": 1.497116493656286,
"grad_norm": 0.44978884369736244,
"learning_rate": 3.0399507131727557e-06,
"loss": 0.07849137485027313,
"step": 1298
},
{
"epoch": 1.4982698961937717,
"grad_norm": 0.47983130797779766,
"learning_rate": 3.0268015029235655e-06,
"loss": 0.09223736822605133,
"step": 1299
},
{
"epoch": 1.4994232987312572,
"grad_norm": 0.4582557971176923,
"learning_rate": 3.0136757193150634e-06,
"loss": 0.08704730123281479,
"step": 1300
},
{
"epoch": 1.5005767012687428,
"grad_norm": 0.4607537945196001,
"learning_rate": 3.000573406443661e-06,
"loss": 0.08720873296260834,
"step": 1301
},
{
"epoch": 1.5017301038062283,
"grad_norm": 0.509404914003974,
"learning_rate": 2.987494608326915e-06,
"loss": 0.09135353565216064,
"step": 1302
},
{
"epoch": 1.5028835063437138,
"grad_norm": 0.48040785245454265,
"learning_rate": 2.974439368903387e-06,
"loss": 0.08743444085121155,
"step": 1303
},
{
"epoch": 1.5040369088811996,
"grad_norm": 0.4468863857310601,
"learning_rate": 2.9614077320324895e-06,
"loss": 0.07982537150382996,
"step": 1304
},
{
"epoch": 1.505190311418685,
"grad_norm": 0.4372443404801211,
"learning_rate": 2.9483997414943465e-06,
"loss": 0.07864395529031754,
"step": 1305
},
{
"epoch": 1.5063437139561708,
"grad_norm": 0.4249041897848328,
"learning_rate": 2.9354154409896386e-06,
"loss": 0.07785080373287201,
"step": 1306
},
{
"epoch": 1.5074971164936564,
"grad_norm": 0.4568896141987471,
"learning_rate": 2.9224548741394563e-06,
"loss": 0.08192753791809082,
"step": 1307
},
{
"epoch": 1.508650519031142,
"grad_norm": 0.4480543278621993,
"learning_rate": 2.909518084485161e-06,
"loss": 0.0739806592464447,
"step": 1308
},
{
"epoch": 1.5098039215686274,
"grad_norm": 0.4552400275038784,
"learning_rate": 2.8966051154882314e-06,
"loss": 0.0824100524187088,
"step": 1309
},
{
"epoch": 1.510957324106113,
"grad_norm": 0.4614725998369817,
"learning_rate": 2.883716010530122e-06,
"loss": 0.08650326728820801,
"step": 1310
},
{
"epoch": 1.5121107266435985,
"grad_norm": 0.4445475080856864,
"learning_rate": 2.870850812912113e-06,
"loss": 0.09401756525039673,
"step": 1311
},
{
"epoch": 1.5132641291810842,
"grad_norm": 0.5239686900471687,
"learning_rate": 2.8580095658551686e-06,
"loss": 0.08210277557373047,
"step": 1312
},
{
"epoch": 1.5144175317185697,
"grad_norm": 0.5256139316819979,
"learning_rate": 2.8451923124997917e-06,
"loss": 0.10271896421909332,
"step": 1313
},
{
"epoch": 1.5155709342560555,
"grad_norm": 0.47760272354744876,
"learning_rate": 2.8323990959058746e-06,
"loss": 0.09568972140550613,
"step": 1314
},
{
"epoch": 1.516724336793541,
"grad_norm": 0.4678384731802768,
"learning_rate": 2.81962995905256e-06,
"loss": 0.08159816265106201,
"step": 1315
},
{
"epoch": 1.5178777393310265,
"grad_norm": 0.46947841871109197,
"learning_rate": 2.806884944838094e-06,
"loss": 0.08716906607151031,
"step": 1316
},
{
"epoch": 1.519031141868512,
"grad_norm": 0.4537873084479394,
"learning_rate": 2.7941640960796813e-06,
"loss": 0.08128490298986435,
"step": 1317
},
{
"epoch": 1.5201845444059976,
"grad_norm": 0.49211237790082846,
"learning_rate": 2.7814674555133415e-06,
"loss": 0.10387013107538223,
"step": 1318
},
{
"epoch": 1.5213379469434831,
"grad_norm": 0.5273586426540612,
"learning_rate": 2.768795065793768e-06,
"loss": 0.07975222170352936,
"step": 1319
},
{
"epoch": 1.5224913494809689,
"grad_norm": 0.46766312311947245,
"learning_rate": 2.75614696949418e-06,
"loss": 0.09135770797729492,
"step": 1320
},
{
"epoch": 1.5236447520184544,
"grad_norm": 0.484836275898211,
"learning_rate": 2.743523209106187e-06,
"loss": 0.08969856798648834,
"step": 1321
},
{
"epoch": 1.5247981545559401,
"grad_norm": 0.4716929562226794,
"learning_rate": 2.7309238270396353e-06,
"loss": 0.08703026175498962,
"step": 1322
},
{
"epoch": 1.5259515570934257,
"grad_norm": 0.5352096769449328,
"learning_rate": 2.7183488656224765e-06,
"loss": 0.08536536991596222,
"step": 1323
},
{
"epoch": 1.5271049596309112,
"grad_norm": 0.46019957852094445,
"learning_rate": 2.7057983671006196e-06,
"loss": 0.07501795887947083,
"step": 1324
},
{
"epoch": 1.5282583621683967,
"grad_norm": 0.478791279353336,
"learning_rate": 2.6932723736377876e-06,
"loss": 0.08944360911846161,
"step": 1325
},
{
"epoch": 1.5294117647058822,
"grad_norm": 0.4896290702487815,
"learning_rate": 2.680770927315379e-06,
"loss": 0.09359028935432434,
"step": 1326
},
{
"epoch": 1.5305651672433678,
"grad_norm": 0.48764918490149256,
"learning_rate": 2.6682940701323267e-06,
"loss": 0.09980146586894989,
"step": 1327
},
{
"epoch": 1.5317185697808535,
"grad_norm": 0.4966959766812287,
"learning_rate": 2.655841844004956e-06,
"loss": 0.10196216404438019,
"step": 1328
},
{
"epoch": 1.532871972318339,
"grad_norm": 0.5138447454130256,
"learning_rate": 2.6434142907668413e-06,
"loss": 0.09396596997976303,
"step": 1329
},
{
"epoch": 1.5340253748558248,
"grad_norm": 0.5200000537792457,
"learning_rate": 2.631011452168669e-06,
"loss": 0.0923609808087349,
"step": 1330
},
{
"epoch": 1.5351787773933103,
"grad_norm": 0.4708018686405983,
"learning_rate": 2.6186333698780975e-06,
"loss": 0.07654425501823425,
"step": 1331
},
{
"epoch": 1.5363321799307958,
"grad_norm": 0.4482541717901647,
"learning_rate": 2.6062800854796134e-06,
"loss": 0.07577718049287796,
"step": 1332
},
{
"epoch": 1.5374855824682814,
"grad_norm": 0.5159675112351456,
"learning_rate": 2.5939516404743948e-06,
"loss": 0.09382124245166779,
"step": 1333
},
{
"epoch": 1.538638985005767,
"grad_norm": 0.5298508845286507,
"learning_rate": 2.5816480762801734e-06,
"loss": 0.09236343204975128,
"step": 1334
},
{
"epoch": 1.5397923875432526,
"grad_norm": 0.502235894097096,
"learning_rate": 2.569369434231088e-06,
"loss": 0.07773742079734802,
"step": 1335
},
{
"epoch": 1.5409457900807382,
"grad_norm": 0.4803190872895656,
"learning_rate": 2.557115755577557e-06,
"loss": 0.0805857926607132,
"step": 1336
},
{
"epoch": 1.542099192618224,
"grad_norm": 0.49598457303746196,
"learning_rate": 2.5448870814861304e-06,
"loss": 0.09833575785160065,
"step": 1337
},
{
"epoch": 1.5432525951557095,
"grad_norm": 0.4557903783412528,
"learning_rate": 2.532683453039353e-06,
"loss": 0.07950516045093536,
"step": 1338
},
{
"epoch": 1.544405997693195,
"grad_norm": 0.4802699009407043,
"learning_rate": 2.5205049112356327e-06,
"loss": 0.0957021415233612,
"step": 1339
},
{
"epoch": 1.5455594002306805,
"grad_norm": 0.46865993750266616,
"learning_rate": 2.5083514969890933e-06,
"loss": 0.07546206563711166,
"step": 1340
},
{
"epoch": 1.546712802768166,
"grad_norm": 0.5280452319304095,
"learning_rate": 2.496223251129447e-06,
"loss": 0.08399003744125366,
"step": 1341
},
{
"epoch": 1.5478662053056516,
"grad_norm": 0.5112960833721749,
"learning_rate": 2.4841202144018463e-06,
"loss": 0.10424164682626724,
"step": 1342
},
{
"epoch": 1.5490196078431373,
"grad_norm": 0.471128799364733,
"learning_rate": 2.472042427466759e-06,
"loss": 0.07990825176239014,
"step": 1343
},
{
"epoch": 1.5501730103806228,
"grad_norm": 0.5881998817050176,
"learning_rate": 2.459989930899821e-06,
"loss": 0.0915984958410263,
"step": 1344
},
{
"epoch": 1.5513264129181086,
"grad_norm": 0.4887858596396505,
"learning_rate": 2.447962765191706e-06,
"loss": 0.08251962065696716,
"step": 1345
},
{
"epoch": 1.552479815455594,
"grad_norm": 0.47697520954249195,
"learning_rate": 2.435960970747989e-06,
"loss": 0.09595407545566559,
"step": 1346
},
{
"epoch": 1.5536332179930796,
"grad_norm": 0.4541861757438918,
"learning_rate": 2.423984587889009e-06,
"loss": 0.07221311330795288,
"step": 1347
},
{
"epoch": 1.5547866205305652,
"grad_norm": 0.4527483141535834,
"learning_rate": 2.4120336568497338e-06,
"loss": 0.07840929925441742,
"step": 1348
},
{
"epoch": 1.5559400230680507,
"grad_norm": 0.4546804496733842,
"learning_rate": 2.400108217779624e-06,
"loss": 0.089170441031456,
"step": 1349
},
{
"epoch": 1.5570934256055362,
"grad_norm": 0.44168264842841115,
"learning_rate": 2.3882083107425037e-06,
"loss": 0.07781291007995605,
"step": 1350
},
{
"epoch": 1.558246828143022,
"grad_norm": 0.4724885545044033,
"learning_rate": 2.376333975716414e-06,
"loss": 0.0964050441980362,
"step": 1351
},
{
"epoch": 1.5594002306805075,
"grad_norm": 0.46155436502740904,
"learning_rate": 2.364485252593494e-06,
"loss": 0.07417406141757965,
"step": 1352
},
{
"epoch": 1.5605536332179932,
"grad_norm": 0.4919482731525964,
"learning_rate": 2.3526621811798343e-06,
"loss": 0.10151129961013794,
"step": 1353
},
{
"epoch": 1.5617070357554788,
"grad_norm": 0.4707856830627523,
"learning_rate": 2.3408648011953483e-06,
"loss": 0.08807054162025452,
"step": 1354
},
{
"epoch": 1.5628604382929643,
"grad_norm": 0.4685395730336516,
"learning_rate": 2.3290931522736404e-06,
"loss": 0.08405694365501404,
"step": 1355
},
{
"epoch": 1.5640138408304498,
"grad_norm": 0.430636221301278,
"learning_rate": 2.3173472739618687e-06,
"loss": 0.07221803069114685,
"step": 1356
},
{
"epoch": 1.5651672433679353,
"grad_norm": 0.4841830758366386,
"learning_rate": 2.3056272057206153e-06,
"loss": 0.08036793768405914,
"step": 1357
},
{
"epoch": 1.5663206459054209,
"grad_norm": 0.44485791677547865,
"learning_rate": 2.2939329869237536e-06,
"loss": 0.08544600754976273,
"step": 1358
},
{
"epoch": 1.5674740484429066,
"grad_norm": 0.4931248172786763,
"learning_rate": 2.2822646568583127e-06,
"loss": 0.06656067818403244,
"step": 1359
},
{
"epoch": 1.5686274509803921,
"grad_norm": 0.4798025904049733,
"learning_rate": 2.2706222547243508e-06,
"loss": 0.08108806610107422,
"step": 1360
},
{
"epoch": 1.5697808535178779,
"grad_norm": 0.4980444548207408,
"learning_rate": 2.2590058196348185e-06,
"loss": 0.07189477980136871,
"step": 1361
},
{
"epoch": 1.5709342560553634,
"grad_norm": 0.49943324270836525,
"learning_rate": 2.247415390615426e-06,
"loss": 0.09402552992105484,
"step": 1362
},
{
"epoch": 1.572087658592849,
"grad_norm": 0.4833006261860564,
"learning_rate": 2.235851006604525e-06,
"loss": 0.09311932325363159,
"step": 1363
},
{
"epoch": 1.5732410611303345,
"grad_norm": 0.5102492544972655,
"learning_rate": 2.224312706452959e-06,
"loss": 0.1014283299446106,
"step": 1364
},
{
"epoch": 1.57439446366782,
"grad_norm": 0.4773779073248023,
"learning_rate": 2.2128005289239474e-06,
"loss": 0.09416399896144867,
"step": 1365
},
{
"epoch": 1.5755478662053055,
"grad_norm": 0.43706723000455283,
"learning_rate": 2.2013145126929493e-06,
"loss": 0.07273417711257935,
"step": 1366
},
{
"epoch": 1.5767012687427913,
"grad_norm": 0.48380029090422805,
"learning_rate": 2.189854696347531e-06,
"loss": 0.08814026415348053,
"step": 1367
},
{
"epoch": 1.5778546712802768,
"grad_norm": 0.4717116528785098,
"learning_rate": 2.1784211183872426e-06,
"loss": 0.0814656913280487,
"step": 1368
},
{
"epoch": 1.5790080738177625,
"grad_norm": 0.4652691224534869,
"learning_rate": 2.167013817223487e-06,
"loss": 0.09296400099992752,
"step": 1369
},
{
"epoch": 1.580161476355248,
"grad_norm": 0.4384942663387739,
"learning_rate": 2.1556328311793837e-06,
"loss": 0.0817423015832901,
"step": 1370
},
{
"epoch": 1.5813148788927336,
"grad_norm": 0.4540520454775624,
"learning_rate": 2.1442781984896522e-06,
"loss": 0.0771784633398056,
"step": 1371
},
{
"epoch": 1.5824682814302191,
"grad_norm": 0.4647828220630081,
"learning_rate": 2.132949957300474e-06,
"loss": 0.07959800958633423,
"step": 1372
},
{
"epoch": 1.5836216839677046,
"grad_norm": 0.48009671369171025,
"learning_rate": 2.1216481456693684e-06,
"loss": 0.10326890647411346,
"step": 1373
},
{
"epoch": 1.5847750865051902,
"grad_norm": 0.44420862944948075,
"learning_rate": 2.110372801565065e-06,
"loss": 0.08022865653038025,
"step": 1374
},
{
"epoch": 1.585928489042676,
"grad_norm": 0.4356700796342711,
"learning_rate": 2.099123962867372e-06,
"loss": 0.07612542808055878,
"step": 1375
},
{
"epoch": 1.5870818915801614,
"grad_norm": 0.46595926364414136,
"learning_rate": 2.087901667367054e-06,
"loss": 0.07858747243881226,
"step": 1376
},
{
"epoch": 1.5882352941176472,
"grad_norm": 0.5248042723920031,
"learning_rate": 2.076705952765705e-06,
"loss": 0.11731183528900146,
"step": 1377
},
{
"epoch": 1.5893886966551327,
"grad_norm": 0.45926025826019834,
"learning_rate": 2.065536856675615e-06,
"loss": 0.07401393353939056,
"step": 1378
},
{
"epoch": 1.5905420991926182,
"grad_norm": 0.4810453313566103,
"learning_rate": 2.054394416619654e-06,
"loss": 0.09008114039897919,
"step": 1379
},
{
"epoch": 1.5916955017301038,
"grad_norm": 0.5455605667090078,
"learning_rate": 2.043278670031136e-06,
"loss": 0.11334876716136932,
"step": 1380
},
{
"epoch": 1.5928489042675893,
"grad_norm": 0.45179588118214276,
"learning_rate": 2.032189654253698e-06,
"loss": 0.0785571038722992,
"step": 1381
},
{
"epoch": 1.5940023068050748,
"grad_norm": 0.4649649139460803,
"learning_rate": 2.021127406541178e-06,
"loss": 0.09663601219654083,
"step": 1382
},
{
"epoch": 1.5951557093425606,
"grad_norm": 0.46222742751567936,
"learning_rate": 2.010091964057481e-06,
"loss": 0.07708097994327545,
"step": 1383
},
{
"epoch": 1.596309111880046,
"grad_norm": 0.46281674505717774,
"learning_rate": 1.9990833638764618e-06,
"loss": 0.09284300357103348,
"step": 1384
},
{
"epoch": 1.5974625144175318,
"grad_norm": 0.45959661576071786,
"learning_rate": 1.988101642981799e-06,
"loss": 0.07523857057094574,
"step": 1385
},
{
"epoch": 1.5986159169550174,
"grad_norm": 0.49486427317709786,
"learning_rate": 1.977146838266869e-06,
"loss": 0.08660058677196503,
"step": 1386
},
{
"epoch": 1.599769319492503,
"grad_norm": 0.45390093743051224,
"learning_rate": 1.9662189865346137e-06,
"loss": 0.08285167813301086,
"step": 1387
},
{
"epoch": 1.6009227220299884,
"grad_norm": 0.4669156667384996,
"learning_rate": 1.955318124497443e-06,
"loss": 0.08547037094831467,
"step": 1388
},
{
"epoch": 1.602076124567474,
"grad_norm": 0.47737450075841664,
"learning_rate": 1.94444428877708e-06,
"loss": 0.10104725509881973,
"step": 1389
},
{
"epoch": 1.6032295271049595,
"grad_norm": 0.4743361661786307,
"learning_rate": 1.9335975159044584e-06,
"loss": 0.07942627370357513,
"step": 1390
},
{
"epoch": 1.6043829296424452,
"grad_norm": 0.501098365260228,
"learning_rate": 1.922777842319591e-06,
"loss": 0.0976376160979271,
"step": 1391
},
{
"epoch": 1.6055363321799307,
"grad_norm": 0.49895338255033495,
"learning_rate": 1.9119853043714533e-06,
"loss": 0.08906835317611694,
"step": 1392
},
{
"epoch": 1.6066897347174165,
"grad_norm": 0.4821994052776269,
"learning_rate": 1.9012199383178532e-06,
"loss": 0.10265209525823593,
"step": 1393
},
{
"epoch": 1.607843137254902,
"grad_norm": 0.5228555305064182,
"learning_rate": 1.8904817803253184e-06,
"loss": 0.09977765381336212,
"step": 1394
},
{
"epoch": 1.6089965397923875,
"grad_norm": 0.4648204480948555,
"learning_rate": 1.8797708664689695e-06,
"loss": 0.09876848757266998,
"step": 1395
},
{
"epoch": 1.610149942329873,
"grad_norm": 0.4167415412035481,
"learning_rate": 1.8690872327323905e-06,
"loss": 0.07206963002681732,
"step": 1396
},
{
"epoch": 1.6113033448673586,
"grad_norm": 0.4340651847386978,
"learning_rate": 1.8584309150075353e-06,
"loss": 0.07514382153749466,
"step": 1397
},
{
"epoch": 1.6124567474048441,
"grad_norm": 0.42315454718296824,
"learning_rate": 1.8478019490945743e-06,
"loss": 0.0760129913687706,
"step": 1398
},
{
"epoch": 1.6136101499423299,
"grad_norm": 0.4672473366267868,
"learning_rate": 1.8372003707017937e-06,
"loss": 0.08932282030582428,
"step": 1399
},
{
"epoch": 1.6147635524798154,
"grad_norm": 0.500400790655529,
"learning_rate": 1.8266262154454706e-06,
"loss": 0.1011243611574173,
"step": 1400
},
{
"epoch": 1.6159169550173011,
"grad_norm": 0.4772250066364624,
"learning_rate": 1.8160795188497527e-06,
"loss": 0.08439259976148605,
"step": 1401
},
{
"epoch": 1.6170703575547867,
"grad_norm": 0.44025560697241145,
"learning_rate": 1.8055603163465396e-06,
"loss": 0.07937301695346832,
"step": 1402
},
{
"epoch": 1.6182237600922722,
"grad_norm": 0.4500795400096256,
"learning_rate": 1.7950686432753662e-06,
"loss": 0.0849744975566864,
"step": 1403
},
{
"epoch": 1.6193771626297577,
"grad_norm": 0.4731363615462152,
"learning_rate": 1.7846045348832796e-06,
"loss": 0.09475409239530563,
"step": 1404
},
{
"epoch": 1.6205305651672433,
"grad_norm": 0.4766842314773294,
"learning_rate": 1.7741680263247164e-06,
"loss": 0.08802954852581024,
"step": 1405
},
{
"epoch": 1.621683967704729,
"grad_norm": 0.46588627868824817,
"learning_rate": 1.763759152661403e-06,
"loss": 0.09397666901350021,
"step": 1406
},
{
"epoch": 1.6228373702422145,
"grad_norm": 0.524406245154532,
"learning_rate": 1.7533779488622193e-06,
"loss": 0.09133845567703247,
"step": 1407
},
{
"epoch": 1.6239907727797003,
"grad_norm": 0.47287785474609045,
"learning_rate": 1.743024449803087e-06,
"loss": 0.09165795147418976,
"step": 1408
},
{
"epoch": 1.6251441753171858,
"grad_norm": 0.4510902353418461,
"learning_rate": 1.732698690266853e-06,
"loss": 0.08350284397602081,
"step": 1409
},
{
"epoch": 1.6262975778546713,
"grad_norm": 0.49654791434818857,
"learning_rate": 1.7224007049431757e-06,
"loss": 0.0892408937215805,
"step": 1410
},
{
"epoch": 1.6274509803921569,
"grad_norm": 0.5059522740196359,
"learning_rate": 1.7121305284284029e-06,
"loss": 0.07867948710918427,
"step": 1411
},
{
"epoch": 1.6286043829296424,
"grad_norm": 0.5018062350141185,
"learning_rate": 1.7018881952254529e-06,
"loss": 0.07980003952980042,
"step": 1412
},
{
"epoch": 1.629757785467128,
"grad_norm": 0.4898716748941718,
"learning_rate": 1.6916737397437178e-06,
"loss": 0.09205864369869232,
"step": 1413
},
{
"epoch": 1.6309111880046137,
"grad_norm": 0.47424565404219543,
"learning_rate": 1.6814871962989154e-06,
"loss": 0.08973643183708191,
"step": 1414
},
{
"epoch": 1.6320645905420992,
"grad_norm": 0.48476944488603824,
"learning_rate": 1.6713285991130102e-06,
"loss": 0.0902041420340538,
"step": 1415
},
{
"epoch": 1.633217993079585,
"grad_norm": 0.45797955543797086,
"learning_rate": 1.661197982314068e-06,
"loss": 0.0835295170545578,
"step": 1416
},
{
"epoch": 1.6343713956170705,
"grad_norm": 0.5279720058747204,
"learning_rate": 1.6510953799361584e-06,
"loss": 0.06491390615701675,
"step": 1417
},
{
"epoch": 1.635524798154556,
"grad_norm": 0.5216500014480774,
"learning_rate": 1.641020825919235e-06,
"loss": 0.0802568569779396,
"step": 1418
},
{
"epoch": 1.6366782006920415,
"grad_norm": 0.48556365624716946,
"learning_rate": 1.630974354109025e-06,
"loss": 0.09327148646116257,
"step": 1419
},
{
"epoch": 1.637831603229527,
"grad_norm": 0.5045466142896784,
"learning_rate": 1.620955998256909e-06,
"loss": 0.07720756530761719,
"step": 1420
},
{
"epoch": 1.6389850057670126,
"grad_norm": 0.421988607868862,
"learning_rate": 1.6109657920198086e-06,
"loss": 0.06940457969903946,
"step": 1421
},
{
"epoch": 1.6401384083044983,
"grad_norm": 0.4441112944599818,
"learning_rate": 1.6010037689600876e-06,
"loss": 0.07340461015701294,
"step": 1422
},
{
"epoch": 1.6412918108419838,
"grad_norm": 0.45941057175724576,
"learning_rate": 1.5910699625454106e-06,
"loss": 0.09039893746376038,
"step": 1423
},
{
"epoch": 1.6424452133794696,
"grad_norm": 0.5172745622631807,
"learning_rate": 1.5811644061486653e-06,
"loss": 0.10324178636074066,
"step": 1424
},
{
"epoch": 1.643598615916955,
"grad_norm": 0.48675612764399223,
"learning_rate": 1.571287133047822e-06,
"loss": 0.07491640001535416,
"step": 1425
},
{
"epoch": 1.6447520184544406,
"grad_norm": 0.4360824622545884,
"learning_rate": 1.5614381764258346e-06,
"loss": 0.07184745371341705,
"step": 1426
},
{
"epoch": 1.6459054209919262,
"grad_norm": 0.4590311801499317,
"learning_rate": 1.5516175693705282e-06,
"loss": 0.08736727386713028,
"step": 1427
},
{
"epoch": 1.6470588235294117,
"grad_norm": 0.48523516335135053,
"learning_rate": 1.5418253448744857e-06,
"loss": 0.09634292125701904,
"step": 1428
},
{
"epoch": 1.6482122260668972,
"grad_norm": 0.4833539115968896,
"learning_rate": 1.5320615358349412e-06,
"loss": 0.08043842017650604,
"step": 1429
},
{
"epoch": 1.649365628604383,
"grad_norm": 0.46400534400476884,
"learning_rate": 1.5223261750536566e-06,
"loss": 0.08800069987773895,
"step": 1430
},
{
"epoch": 1.6505190311418685,
"grad_norm": 0.4638857746557704,
"learning_rate": 1.512619295236837e-06,
"loss": 0.07692458480596542,
"step": 1431
},
{
"epoch": 1.6516724336793542,
"grad_norm": 0.4556275876487893,
"learning_rate": 1.502940928994988e-06,
"loss": 0.08253438770771027,
"step": 1432
},
{
"epoch": 1.6528258362168398,
"grad_norm": 0.44518591627666665,
"learning_rate": 1.493291108842837e-06,
"loss": 0.08092884719371796,
"step": 1433
},
{
"epoch": 1.6539792387543253,
"grad_norm": 0.4569443268718005,
"learning_rate": 1.4836698671992022e-06,
"loss": 0.08261385560035706,
"step": 1434
},
{
"epoch": 1.6551326412918108,
"grad_norm": 0.46375974299843675,
"learning_rate": 1.474077236386895e-06,
"loss": 0.09235814213752747,
"step": 1435
},
{
"epoch": 1.6562860438292963,
"grad_norm": 0.4632311408525821,
"learning_rate": 1.4645132486326075e-06,
"loss": 0.07973376661539078,
"step": 1436
},
{
"epoch": 1.6574394463667819,
"grad_norm": 0.46277013957012975,
"learning_rate": 1.4549779360667981e-06,
"loss": 0.08445103466510773,
"step": 1437
},
{
"epoch": 1.6585928489042676,
"grad_norm": 0.4684508015210134,
"learning_rate": 1.4454713307236045e-06,
"loss": 0.09547249972820282,
"step": 1438
},
{
"epoch": 1.6597462514417531,
"grad_norm": 0.4821962986134515,
"learning_rate": 1.4359934645407036e-06,
"loss": 0.10701487958431244,
"step": 1439
},
{
"epoch": 1.6608996539792389,
"grad_norm": 0.4297310725744812,
"learning_rate": 1.4265443693592406e-06,
"loss": 0.07835353165864944,
"step": 1440
},
{
"epoch": 1.6620530565167244,
"grad_norm": 0.44728050152604865,
"learning_rate": 1.4171240769236905e-06,
"loss": 0.08186867833137512,
"step": 1441
},
{
"epoch": 1.66320645905421,
"grad_norm": 0.43279210942801805,
"learning_rate": 1.407732618881772e-06,
"loss": 0.08379806578159332,
"step": 1442
},
{
"epoch": 1.6643598615916955,
"grad_norm": 0.4730773951230568,
"learning_rate": 1.398370026784328e-06,
"loss": 0.07930352538824081,
"step": 1443
},
{
"epoch": 1.665513264129181,
"grad_norm": 0.46218937504740176,
"learning_rate": 1.3890363320852319e-06,
"loss": 0.08426142483949661,
"step": 1444
},
{
"epoch": 1.6666666666666665,
"grad_norm": 0.4473524756155622,
"learning_rate": 1.3797315661412735e-06,
"loss": 0.08007732033729553,
"step": 1445
},
{
"epoch": 1.6678200692041523,
"grad_norm": 0.44834782711840965,
"learning_rate": 1.3704557602120472e-06,
"loss": 0.07670222222805023,
"step": 1446
},
{
"epoch": 1.6689734717416378,
"grad_norm": 0.4810971170763557,
"learning_rate": 1.3612089454598721e-06,
"loss": 0.08076446503400803,
"step": 1447
},
{
"epoch": 1.6701268742791235,
"grad_norm": 0.5019426364307826,
"learning_rate": 1.3519911529496531e-06,
"loss": 0.10300135612487793,
"step": 1448
},
{
"epoch": 1.671280276816609,
"grad_norm": 0.44798478294164157,
"learning_rate": 1.3428024136488093e-06,
"loss": 0.0750141441822052,
"step": 1449
},
{
"epoch": 1.6724336793540946,
"grad_norm": 0.49961313115823347,
"learning_rate": 1.3336427584271449e-06,
"loss": 0.09704729914665222,
"step": 1450
},
{
"epoch": 1.6735870818915801,
"grad_norm": 0.4496668805700195,
"learning_rate": 1.3245122180567604e-06,
"loss": 0.06924331933259964,
"step": 1451
},
{
"epoch": 1.6747404844290656,
"grad_norm": 0.4364460205980534,
"learning_rate": 1.31541082321194e-06,
"loss": 0.0738469660282135,
"step": 1452
},
{
"epoch": 1.6758938869665512,
"grad_norm": 0.46359328858484417,
"learning_rate": 1.3063386044690563e-06,
"loss": 0.07608649134635925,
"step": 1453
},
{
"epoch": 1.677047289504037,
"grad_norm": 0.42293470199874916,
"learning_rate": 1.2972955923064645e-06,
"loss": 0.07523155212402344,
"step": 1454
},
{
"epoch": 1.6782006920415224,
"grad_norm": 0.4677842096033532,
"learning_rate": 1.288281817104391e-06,
"loss": 0.08424337208271027,
"step": 1455
},
{
"epoch": 1.6793540945790082,
"grad_norm": 0.5186131110059888,
"learning_rate": 1.2792973091448569e-06,
"loss": 0.11744806170463562,
"step": 1456
},
{
"epoch": 1.6805074971164937,
"grad_norm": 0.46790854557234035,
"learning_rate": 1.270342098611539e-06,
"loss": 0.08502757549285889,
"step": 1457
},
{
"epoch": 1.6816608996539792,
"grad_norm": 0.5911981912334355,
"learning_rate": 1.2614162155897058e-06,
"loss": 0.0732482522726059,
"step": 1458
},
{
"epoch": 1.6828143021914648,
"grad_norm": 0.460450691050943,
"learning_rate": 1.2525196900660897e-06,
"loss": 0.08331382274627686,
"step": 1459
},
{
"epoch": 1.6839677047289503,
"grad_norm": 0.5095501937863663,
"learning_rate": 1.2436525519287979e-06,
"loss": 0.07180708646774292,
"step": 1460
},
{
"epoch": 1.6851211072664358,
"grad_norm": 0.5228051281470011,
"learning_rate": 1.2348148309672126e-06,
"loss": 0.06954856216907501,
"step": 1461
},
{
"epoch": 1.6862745098039216,
"grad_norm": 0.48898509292276204,
"learning_rate": 1.2260065568718805e-06,
"loss": 0.09196717292070389,
"step": 1462
},
{
"epoch": 1.687427912341407,
"grad_norm": 0.47516351145019925,
"learning_rate": 1.217227759234434e-06,
"loss": 0.09182867407798767,
"step": 1463
},
{
"epoch": 1.6885813148788928,
"grad_norm": 0.48748124287291805,
"learning_rate": 1.2084784675474614e-06,
"loss": 0.08330394327640533,
"step": 1464
},
{
"epoch": 1.6897347174163784,
"grad_norm": 0.4921726401264911,
"learning_rate": 1.1997587112044451e-06,
"loss": 0.10150840133428574,
"step": 1465
},
{
"epoch": 1.690888119953864,
"grad_norm": 0.47673334355161395,
"learning_rate": 1.1910685194996209e-06,
"loss": 0.08911114186048508,
"step": 1466
},
{
"epoch": 1.6920415224913494,
"grad_norm": 0.4845189520555881,
"learning_rate": 1.1824079216279216e-06,
"loss": 0.08437485992908478,
"step": 1467
},
{
"epoch": 1.693194925028835,
"grad_norm": 0.4974388663695158,
"learning_rate": 1.173776946684846e-06,
"loss": 0.06775881350040436,
"step": 1468
},
{
"epoch": 1.6943483275663207,
"grad_norm": 0.45145139963465314,
"learning_rate": 1.165175623666377e-06,
"loss": 0.08223802596330643,
"step": 1469
},
{
"epoch": 1.6955017301038062,
"grad_norm": 0.48320515041438333,
"learning_rate": 1.1566039814688857e-06,
"loss": 0.09479266405105591,
"step": 1470
},
{
"epoch": 1.696655132641292,
"grad_norm": 0.4267698791027021,
"learning_rate": 1.148062048889017e-06,
"loss": 0.07235585153102875,
"step": 1471
},
{
"epoch": 1.6978085351787775,
"grad_norm": 0.485889186860865,
"learning_rate": 1.1395498546236262e-06,
"loss": 0.09909984469413757,
"step": 1472
},
{
"epoch": 1.698961937716263,
"grad_norm": 0.4626572028071968,
"learning_rate": 1.1310674272696387e-06,
"loss": 0.08447360247373581,
"step": 1473
},
{
"epoch": 1.7001153402537486,
"grad_norm": 0.44429215295312086,
"learning_rate": 1.1226147953239974e-06,
"loss": 0.08187257498502731,
"step": 1474
},
{
"epoch": 1.701268742791234,
"grad_norm": 0.43557485325415785,
"learning_rate": 1.1141919871835306e-06,
"loss": 0.07555612176656723,
"step": 1475
},
{
"epoch": 1.7024221453287196,
"grad_norm": 0.4268713660472037,
"learning_rate": 1.1057990311448863e-06,
"loss": 0.076148621737957,
"step": 1476
},
{
"epoch": 1.7035755478662054,
"grad_norm": 0.4259576917348061,
"learning_rate": 1.0974359554044133e-06,
"loss": 0.07155287265777588,
"step": 1477
},
{
"epoch": 1.7047289504036909,
"grad_norm": 0.4955969857470323,
"learning_rate": 1.089102788058084e-06,
"loss": 0.098116934299469,
"step": 1478
},
{
"epoch": 1.7058823529411766,
"grad_norm": 0.4454206673422632,
"learning_rate": 1.0807995571013897e-06,
"loss": 0.08344689756631851,
"step": 1479
},
{
"epoch": 1.7070357554786622,
"grad_norm": 0.43717686755433055,
"learning_rate": 1.0725262904292432e-06,
"loss": 0.07548746466636658,
"step": 1480
},
{
"epoch": 1.7081891580161477,
"grad_norm": 0.44082015310347494,
"learning_rate": 1.0642830158359064e-06,
"loss": 0.07321727275848389,
"step": 1481
},
{
"epoch": 1.7093425605536332,
"grad_norm": 0.405103073054465,
"learning_rate": 1.056069761014865e-06,
"loss": 0.07239045947790146,
"step": 1482
},
{
"epoch": 1.7104959630911187,
"grad_norm": 0.44903271650882476,
"learning_rate": 1.04788655355877e-06,
"loss": 0.08990779519081116,
"step": 1483
},
{
"epoch": 1.7116493656286043,
"grad_norm": 0.4258973549837027,
"learning_rate": 1.0397334209593102e-06,
"loss": 0.06993109732866287,
"step": 1484
},
{
"epoch": 1.71280276816609,
"grad_norm": 0.45913599764343144,
"learning_rate": 1.0316103906071507e-06,
"loss": 0.08928606659173965,
"step": 1485
},
{
"epoch": 1.7139561707035755,
"grad_norm": 0.45983098159948105,
"learning_rate": 1.0235174897918232e-06,
"loss": 0.08836518973112106,
"step": 1486
},
{
"epoch": 1.7151095732410613,
"grad_norm": 0.48118095180030135,
"learning_rate": 1.0154547457016305e-06,
"loss": 0.09802086651325226,
"step": 1487
},
{
"epoch": 1.7162629757785468,
"grad_norm": 0.4706970466308402,
"learning_rate": 1.0074221854235777e-06,
"loss": 0.0726025179028511,
"step": 1488
},
{
"epoch": 1.7174163783160323,
"grad_norm": 0.4495504793887915,
"learning_rate": 9.994198359432506e-07,
"loss": 0.0954207181930542,
"step": 1489
},
{
"epoch": 1.7185697808535179,
"grad_norm": 0.46671649937440113,
"learning_rate": 9.914477241447584e-07,
"loss": 0.07251942157745361,
"step": 1490
},
{
"epoch": 1.7197231833910034,
"grad_norm": 0.4422390210675713,
"learning_rate": 9.835058768106077e-07,
"loss": 0.07946689426898956,
"step": 1491
},
{
"epoch": 1.720876585928489,
"grad_norm": 0.4506757990365786,
"learning_rate": 9.755943206216489e-07,
"loss": 0.07631538808345795,
"step": 1492
},
{
"epoch": 1.7220299884659747,
"grad_norm": 0.4392443548716252,
"learning_rate": 9.677130821569502e-07,
"loss": 0.07742469012737274,
"step": 1493
},
{
"epoch": 1.7231833910034602,
"grad_norm": 0.43993453413797134,
"learning_rate": 9.598621878937441e-07,
"loss": 0.07891791313886642,
"step": 1494
},
{
"epoch": 1.724336793540946,
"grad_norm": 0.45599642280930497,
"learning_rate": 9.520416642073116e-07,
"loss": 0.09408499300479889,
"step": 1495
},
{
"epoch": 1.7254901960784315,
"grad_norm": 0.4803788795769065,
"learning_rate": 9.442515373708982e-07,
"loss": 0.09352752566337585,
"step": 1496
},
{
"epoch": 1.726643598615917,
"grad_norm": 0.4755463488975102,
"learning_rate": 9.36491833555645e-07,
"loss": 0.08097809553146362,
"step": 1497
},
{
"epoch": 1.7277970011534025,
"grad_norm": 0.48944055590821245,
"learning_rate": 9.2876257883047e-07,
"loss": 0.09281191229820251,
"step": 1498
},
{
"epoch": 1.728950403690888,
"grad_norm": 0.4499027096405317,
"learning_rate": 9.210637991620109e-07,
"loss": 0.08482381701469421,
"step": 1499
},
{
"epoch": 1.7301038062283736,
"grad_norm": 0.44368710391297017,
"learning_rate": 9.133955204145107e-07,
"loss": 0.08075851202011108,
"step": 1500
},
{
"epoch": 1.7312572087658593,
"grad_norm": 0.45659818733920193,
"learning_rate": 9.057577683497565e-07,
"loss": 0.08566870540380478,
"step": 1501
},
{
"epoch": 1.7324106113033448,
"grad_norm": 0.43933397427520027,
"learning_rate": 8.981505686269676e-07,
"loss": 0.07309457659721375,
"step": 1502
},
{
"epoch": 1.7335640138408306,
"grad_norm": 0.4771997285323543,
"learning_rate": 8.90573946802733e-07,
"loss": 0.10212794691324234,
"step": 1503
},
{
"epoch": 1.7347174163783161,
"grad_norm": 0.48726900735527373,
"learning_rate": 8.830279283309106e-07,
"loss": 0.09845606982707977,
"step": 1504
},
{
"epoch": 1.7358708189158016,
"grad_norm": 0.4349327037503685,
"learning_rate": 8.755125385625373e-07,
"loss": 0.0815720185637474,
"step": 1505
},
{
"epoch": 1.7370242214532872,
"grad_norm": 0.4554515398515434,
"learning_rate": 8.680278027457689e-07,
"loss": 0.08219771087169647,
"step": 1506
},
{
"epoch": 1.7381776239907727,
"grad_norm": 0.4814922358118263,
"learning_rate": 8.605737460257602e-07,
"loss": 0.09699362516403198,
"step": 1507
},
{
"epoch": 1.7393310265282582,
"grad_norm": 0.4643274420884982,
"learning_rate": 8.53150393444615e-07,
"loss": 0.09894633293151855,
"step": 1508
},
{
"epoch": 1.740484429065744,
"grad_norm": 0.43847382847147887,
"learning_rate": 8.457577699412711e-07,
"loss": 0.08044122159481049,
"step": 1509
},
{
"epoch": 1.7416378316032295,
"grad_norm": 0.44599904023252984,
"learning_rate": 8.383959003514442e-07,
"loss": 0.08290042728185654,
"step": 1510
},
{
"epoch": 1.7427912341407152,
"grad_norm": 0.5055750560668447,
"learning_rate": 8.310648094075258e-07,
"loss": 0.10992328822612762,
"step": 1511
},
{
"epoch": 1.7439446366782008,
"grad_norm": 0.4407404075358887,
"learning_rate": 8.237645217384982e-07,
"loss": 0.069158174097538,
"step": 1512
},
{
"epoch": 1.7450980392156863,
"grad_norm": 0.4213222355503362,
"learning_rate": 8.164950618698764e-07,
"loss": 0.07143129408359528,
"step": 1513
},
{
"epoch": 1.7462514417531718,
"grad_norm": 0.46258036246730944,
"learning_rate": 8.092564542235904e-07,
"loss": 0.09127770364284515,
"step": 1514
},
{
"epoch": 1.7474048442906573,
"grad_norm": 0.5082489152845423,
"learning_rate": 8.020487231179364e-07,
"loss": 0.0666658878326416,
"step": 1515
},
{
"epoch": 1.7485582468281429,
"grad_norm": 0.4384888603887615,
"learning_rate": 7.94871892767467e-07,
"loss": 0.07384112477302551,
"step": 1516
},
{
"epoch": 1.7497116493656286,
"grad_norm": 0.44897067540871016,
"learning_rate": 7.877259872829357e-07,
"loss": 0.07666268199682236,
"step": 1517
},
{
"epoch": 1.7508650519031141,
"grad_norm": 0.4524270898342402,
"learning_rate": 7.806110306711911e-07,
"loss": 0.07810153812170029,
"step": 1518
},
{
"epoch": 1.7520184544406,
"grad_norm": 0.4314979791805348,
"learning_rate": 7.735270468351186e-07,
"loss": 0.07891873270273209,
"step": 1519
},
{
"epoch": 1.7531718569780854,
"grad_norm": 0.43919913137104305,
"learning_rate": 7.664740595735454e-07,
"loss": 0.07492205500602722,
"step": 1520
},
{
"epoch": 1.754325259515571,
"grad_norm": 0.4307252405114126,
"learning_rate": 7.594520925811588e-07,
"loss": 0.07286188006401062,
"step": 1521
},
{
"epoch": 1.7554786620530565,
"grad_norm": 0.45633209192986746,
"learning_rate": 7.524611694484451e-07,
"loss": 0.08578339219093323,
"step": 1522
},
{
"epoch": 1.756632064590542,
"grad_norm": 0.5035949141532603,
"learning_rate": 7.455013136615851e-07,
"loss": 0.10408972948789597,
"step": 1523
},
{
"epoch": 1.7577854671280275,
"grad_norm": 0.4337749211573058,
"learning_rate": 7.385725486024009e-07,
"loss": 0.08168213814496994,
"step": 1524
},
{
"epoch": 1.7589388696655133,
"grad_norm": 0.45400551627474794,
"learning_rate": 7.316748975482513e-07,
"loss": 0.09223606437444687,
"step": 1525
},
{
"epoch": 1.7600922722029988,
"grad_norm": 0.47174655886042643,
"learning_rate": 7.24808383671981e-07,
"loss": 0.09019721299409866,
"step": 1526
},
{
"epoch": 1.7612456747404845,
"grad_norm": 0.45718276088468307,
"learning_rate": 7.179730300418141e-07,
"loss": 0.0904451459646225,
"step": 1527
},
{
"epoch": 1.76239907727797,
"grad_norm": 0.4445492045967517,
"learning_rate": 7.111688596212984e-07,
"loss": 0.07425994426012039,
"step": 1528
},
{
"epoch": 1.7635524798154556,
"grad_norm": 0.42865748608475324,
"learning_rate": 7.043958952692265e-07,
"loss": 0.08636932075023651,
"step": 1529
},
{
"epoch": 1.7647058823529411,
"grad_norm": 0.44599012769381186,
"learning_rate": 6.976541597395403e-07,
"loss": 0.08930586278438568,
"step": 1530
},
{
"epoch": 1.7658592848904267,
"grad_norm": 0.47562326982944825,
"learning_rate": 6.909436756812804e-07,
"loss": 0.07114845514297485,
"step": 1531
},
{
"epoch": 1.7670126874279122,
"grad_norm": 0.4663857736503539,
"learning_rate": 6.842644656384856e-07,
"loss": 0.09801007807254791,
"step": 1532
},
{
"epoch": 1.768166089965398,
"grad_norm": 0.428343681695422,
"learning_rate": 6.77616552050141e-07,
"loss": 0.08126062154769897,
"step": 1533
},
{
"epoch": 1.7693194925028837,
"grad_norm": 0.4255177150160356,
"learning_rate": 6.709999572500791e-07,
"loss": 0.0871000736951828,
"step": 1534
},
{
"epoch": 1.7704728950403692,
"grad_norm": 0.4745019623191183,
"learning_rate": 6.644147034669246e-07,
"loss": 0.09258916229009628,
"step": 1535
},
{
"epoch": 1.7716262975778547,
"grad_norm": 0.428973109970304,
"learning_rate": 6.578608128240061e-07,
"loss": 0.07381710410118103,
"step": 1536
},
{
"epoch": 1.7727797001153403,
"grad_norm": 0.47459779413168707,
"learning_rate": 6.513383073392876e-07,
"loss": 0.10765878856182098,
"step": 1537
},
{
"epoch": 1.7739331026528258,
"grad_norm": 0.4358260874345844,
"learning_rate": 6.448472089252999e-07,
"loss": 0.0855957493185997,
"step": 1538
},
{
"epoch": 1.7750865051903113,
"grad_norm": 0.46605478899233116,
"learning_rate": 6.383875393890493e-07,
"loss": 0.09523581713438034,
"step": 1539
},
{
"epoch": 1.776239907727797,
"grad_norm": 0.4255889716490659,
"learning_rate": 6.319593204319696e-07,
"loss": 0.07856802642345428,
"step": 1540
},
{
"epoch": 1.7773933102652826,
"grad_norm": 0.46800554995568655,
"learning_rate": 6.255625736498216e-07,
"loss": 0.09822478890419006,
"step": 1541
},
{
"epoch": 1.7785467128027683,
"grad_norm": 0.42810466722814056,
"learning_rate": 6.191973205326474e-07,
"loss": 0.08619797229766846,
"step": 1542
},
{
"epoch": 1.7797001153402539,
"grad_norm": 0.4181556128280205,
"learning_rate": 6.128635824646745e-07,
"loss": 0.07541541755199432,
"step": 1543
},
{
"epoch": 1.7808535178777394,
"grad_norm": 0.4355239285543818,
"learning_rate": 6.065613807242654e-07,
"loss": 0.08572019636631012,
"step": 1544
},
{
"epoch": 1.782006920415225,
"grad_norm": 0.5036666959639232,
"learning_rate": 6.002907364838262e-07,
"loss": 0.12350749969482422,
"step": 1545
},
{
"epoch": 1.7831603229527104,
"grad_norm": 0.4293851185965131,
"learning_rate": 5.940516708097455e-07,
"loss": 0.08298186957836151,
"step": 1546
},
{
"epoch": 1.784313725490196,
"grad_norm": 0.44593161569664913,
"learning_rate": 5.878442046623333e-07,
"loss": 0.09005622565746307,
"step": 1547
},
{
"epoch": 1.7854671280276817,
"grad_norm": 0.4277136263024806,
"learning_rate": 5.816683588957262e-07,
"loss": 0.07945345342159271,
"step": 1548
},
{
"epoch": 1.7866205305651672,
"grad_norm": 0.4388313780262988,
"learning_rate": 5.755241542578426e-07,
"loss": 0.08018314838409424,
"step": 1549
},
{
"epoch": 1.787773933102653,
"grad_norm": 0.4160760184428865,
"learning_rate": 5.694116113902903e-07,
"loss": 0.08205653727054596,
"step": 1550
},
{
"epoch": 1.7889273356401385,
"grad_norm": 0.4188550865112811,
"learning_rate": 5.633307508283237e-07,
"loss": 0.0722522884607315,
"step": 1551
},
{
"epoch": 1.790080738177624,
"grad_norm": 0.4193884319412575,
"learning_rate": 5.572815930007447e-07,
"loss": 0.0737803727388382,
"step": 1552
},
{
"epoch": 1.7912341407151096,
"grad_norm": 0.46123409881175936,
"learning_rate": 5.512641582298572e-07,
"loss": 0.08167803287506104,
"step": 1553
},
{
"epoch": 1.792387543252595,
"grad_norm": 0.4970555710861486,
"learning_rate": 5.452784667313904e-07,
"loss": 0.11562861502170563,
"step": 1554
},
{
"epoch": 1.7935409457900806,
"grad_norm": 0.4407060296690412,
"learning_rate": 5.393245386144252e-07,
"loss": 0.08366487920284271,
"step": 1555
},
{
"epoch": 1.7946943483275664,
"grad_norm": 0.39001439312942243,
"learning_rate": 5.334023938813448e-07,
"loss": 0.06376579403877258,
"step": 1556
},
{
"epoch": 1.7958477508650519,
"grad_norm": 0.4564730108563335,
"learning_rate": 5.275120524277388e-07,
"loss": 0.08829472959041595,
"step": 1557
},
{
"epoch": 1.7970011534025376,
"grad_norm": 0.49730326579485806,
"learning_rate": 5.216535340423701e-07,
"loss": 0.0621301606297493,
"step": 1558
},
{
"epoch": 1.7981545559400232,
"grad_norm": 0.5300213972214189,
"learning_rate": 5.158268584070758e-07,
"loss": 0.09604474157094955,
"step": 1559
},
{
"epoch": 1.7993079584775087,
"grad_norm": 0.42669002899792163,
"learning_rate": 5.100320450967277e-07,
"loss": 0.07681753486394882,
"step": 1560
},
{
"epoch": 1.8004613610149942,
"grad_norm": 0.42812710801454373,
"learning_rate": 5.042691135791489e-07,
"loss": 0.08376656472682953,
"step": 1561
},
{
"epoch": 1.8016147635524797,
"grad_norm": 0.46331831884899805,
"learning_rate": 4.985380832150577e-07,
"loss": 0.0920184850692749,
"step": 1562
},
{
"epoch": 1.8027681660899653,
"grad_norm": 0.4451679840859192,
"learning_rate": 4.928389732579986e-07,
"loss": 0.06738371402025223,
"step": 1563
},
{
"epoch": 1.803921568627451,
"grad_norm": 0.44458355093337815,
"learning_rate": 4.871718028542772e-07,
"loss": 0.08388322591781616,
"step": 1564
},
{
"epoch": 1.8050749711649365,
"grad_norm": 0.473514177874576,
"learning_rate": 4.815365910429037e-07,
"loss": 0.09538695216178894,
"step": 1565
},
{
"epoch": 1.8062283737024223,
"grad_norm": 0.4446521961321896,
"learning_rate": 4.7593335675551135e-07,
"loss": 0.08398496359586716,
"step": 1566
},
{
"epoch": 1.8073817762399078,
"grad_norm": 0.41352394090821665,
"learning_rate": 4.7036211881631767e-07,
"loss": 0.07504066824913025,
"step": 1567
},
{
"epoch": 1.8085351787773933,
"grad_norm": 0.4441737530372475,
"learning_rate": 4.648228959420342e-07,
"loss": 0.07978780567646027,
"step": 1568
},
{
"epoch": 1.8096885813148789,
"grad_norm": 0.4553339032937663,
"learning_rate": 4.5931570674182966e-07,
"loss": 0.08527640998363495,
"step": 1569
},
{
"epoch": 1.8108419838523644,
"grad_norm": 0.4388152400730318,
"learning_rate": 4.538405697172432e-07,
"loss": 0.07549992948770523,
"step": 1570
},
{
"epoch": 1.81199538638985,
"grad_norm": 0.46471676459373545,
"learning_rate": 4.4839750326214125e-07,
"loss": 0.08667217195034027,
"step": 1571
},
{
"epoch": 1.8131487889273357,
"grad_norm": 0.4694368606653923,
"learning_rate": 4.4298652566264755e-07,
"loss": 0.09091932326555252,
"step": 1572
},
{
"epoch": 1.8143021914648212,
"grad_norm": 0.5435468716615187,
"learning_rate": 4.376076550970776e-07,
"loss": 0.11879110336303711,
"step": 1573
},
{
"epoch": 1.815455594002307,
"grad_norm": 0.4274440120328082,
"learning_rate": 4.322609096358921e-07,
"loss": 0.07668593525886536,
"step": 1574
},
{
"epoch": 1.8166089965397925,
"grad_norm": 0.43761748057359035,
"learning_rate": 4.2694630724161466e-07,
"loss": 0.08618499338626862,
"step": 1575
},
{
"epoch": 1.817762399077278,
"grad_norm": 0.43108323793163694,
"learning_rate": 4.216638657687944e-07,
"loss": 0.0833648145198822,
"step": 1576
},
{
"epoch": 1.8189158016147635,
"grad_norm": 0.4442330488442909,
"learning_rate": 4.164136029639265e-07,
"loss": 0.07683166861534119,
"step": 1577
},
{
"epoch": 1.820069204152249,
"grad_norm": 0.466021493757971,
"learning_rate": 4.111955364654063e-07,
"loss": 0.09610314667224884,
"step": 1578
},
{
"epoch": 1.8212226066897346,
"grad_norm": 0.47669388242755495,
"learning_rate": 4.060096838034633e-07,
"loss": 0.0776035264134407,
"step": 1579
},
{
"epoch": 1.8223760092272203,
"grad_norm": 0.4253858800834901,
"learning_rate": 4.008560624001012e-07,
"loss": 0.0778421014547348,
"step": 1580
},
{
"epoch": 1.8235294117647058,
"grad_norm": 0.4397451699349066,
"learning_rate": 3.9573468956904835e-07,
"loss": 0.08073864877223969,
"step": 1581
},
{
"epoch": 1.8246828143021916,
"grad_norm": 0.4330318881987894,
"learning_rate": 3.906455825156852e-07,
"loss": 0.07604895532131195,
"step": 1582
},
{
"epoch": 1.8258362168396771,
"grad_norm": 0.4317540083369736,
"learning_rate": 3.855887583370033e-07,
"loss": 0.07334481179714203,
"step": 1583
},
{
"epoch": 1.8269896193771626,
"grad_norm": 0.4090697533273083,
"learning_rate": 3.8056423402152874e-07,
"loss": 0.07233726233243942,
"step": 1584
},
{
"epoch": 1.8281430219146482,
"grad_norm": 0.4204702141628341,
"learning_rate": 3.755720264492857e-07,
"loss": 0.07384037226438522,
"step": 1585
},
{
"epoch": 1.8292964244521337,
"grad_norm": 0.4646364964139814,
"learning_rate": 3.7061215239172164e-07,
"loss": 0.06530794501304626,
"step": 1586
},
{
"epoch": 1.8304498269896192,
"grad_norm": 0.4161062774550697,
"learning_rate": 3.656846285116622e-07,
"loss": 0.07318238914012909,
"step": 1587
},
{
"epoch": 1.831603229527105,
"grad_norm": 0.41613850258619023,
"learning_rate": 3.6078947136325207e-07,
"loss": 0.07586484402418137,
"step": 1588
},
{
"epoch": 1.8327566320645905,
"grad_norm": 0.451071309901348,
"learning_rate": 3.5592669739189734e-07,
"loss": 0.08965134620666504,
"step": 1589
},
{
"epoch": 1.8339100346020762,
"grad_norm": 0.4574965179866986,
"learning_rate": 3.5109632293421457e-07,
"loss": 0.08812291920185089,
"step": 1590
},
{
"epoch": 1.8350634371395618,
"grad_norm": 0.4764553035302359,
"learning_rate": 3.462983642179696e-07,
"loss": 0.09897389262914658,
"step": 1591
},
{
"epoch": 1.8362168396770473,
"grad_norm": 0.493653034502569,
"learning_rate": 3.415328373620308e-07,
"loss": 0.08953029662370682,
"step": 1592
},
{
"epoch": 1.8373702422145328,
"grad_norm": 0.4660211257202185,
"learning_rate": 3.3679975837630605e-07,
"loss": 0.09565223008394241,
"step": 1593
},
{
"epoch": 1.8385236447520183,
"grad_norm": 0.4181869548249153,
"learning_rate": 3.3209914316170153e-07,
"loss": 0.07131161540746689,
"step": 1594
},
{
"epoch": 1.8396770472895039,
"grad_norm": 0.46421483658681234,
"learning_rate": 3.2743100751005064e-07,
"loss": 0.0869339108467102,
"step": 1595
},
{
"epoch": 1.8408304498269896,
"grad_norm": 0.4363336874687016,
"learning_rate": 3.227953671040784e-07,
"loss": 0.07717902958393097,
"step": 1596
},
{
"epoch": 1.8419838523644751,
"grad_norm": 0.42114605395933447,
"learning_rate": 3.181922375173374e-07,
"loss": 0.07472886145114899,
"step": 1597
},
{
"epoch": 1.843137254901961,
"grad_norm": 0.4280704036816773,
"learning_rate": 3.136216342141607e-07,
"loss": 0.0832543820142746,
"step": 1598
},
{
"epoch": 1.8442906574394464,
"grad_norm": 0.4424240118836316,
"learning_rate": 3.0908357254961e-07,
"loss": 0.08354353904724121,
"step": 1599
},
{
"epoch": 1.845444059976932,
"grad_norm": 0.43538018091024877,
"learning_rate": 3.045780677694177e-07,
"loss": 0.07791280746459961,
"step": 1600
},
{
"epoch": 1.8465974625144175,
"grad_norm": 0.5097727957276319,
"learning_rate": 3.001051350099482e-07,
"loss": 0.11559078097343445,
"step": 1601
},
{
"epoch": 1.847750865051903,
"grad_norm": 0.49107387349422355,
"learning_rate": 2.9566478929813013e-07,
"loss": 0.10190825164318085,
"step": 1602
},
{
"epoch": 1.8489042675893888,
"grad_norm": 0.4849900619675024,
"learning_rate": 2.912570455514219e-07,
"loss": 0.09192824363708496,
"step": 1603
},
{
"epoch": 1.8500576701268743,
"grad_norm": 0.4462088204718826,
"learning_rate": 2.868819185777527e-07,
"loss": 0.07325074821710587,
"step": 1604
},
{
"epoch": 1.85121107266436,
"grad_norm": 0.43392142141994267,
"learning_rate": 2.825394230754741e-07,
"loss": 0.07683129608631134,
"step": 1605
},
{
"epoch": 1.8523644752018456,
"grad_norm": 0.4414815633382736,
"learning_rate": 2.7822957363331294e-07,
"loss": 0.07680347561836243,
"step": 1606
},
{
"epoch": 1.853517877739331,
"grad_norm": 0.4473949212682347,
"learning_rate": 2.7395238473031827e-07,
"loss": 0.07681342959403992,
"step": 1607
},
{
"epoch": 1.8546712802768166,
"grad_norm": 0.4758428333495726,
"learning_rate": 2.6970787073581916e-07,
"loss": 0.09116753935813904,
"step": 1608
},
{
"epoch": 1.8558246828143021,
"grad_norm": 0.4411811843542647,
"learning_rate": 2.6549604590936694e-07,
"loss": 0.08624029159545898,
"step": 1609
},
{
"epoch": 1.8569780853517877,
"grad_norm": 0.49540805599997945,
"learning_rate": 2.6131692440070077e-07,
"loss": 0.10415154695510864,
"step": 1610
},
{
"epoch": 1.8581314878892734,
"grad_norm": 0.4767407964305641,
"learning_rate": 2.5717052024968324e-07,
"loss": 0.06551147252321243,
"step": 1611
},
{
"epoch": 1.859284890426759,
"grad_norm": 0.47139939274334236,
"learning_rate": 2.530568473862671e-07,
"loss": 0.09569668769836426,
"step": 1612
},
{
"epoch": 1.8604382929642447,
"grad_norm": 0.46738087412635604,
"learning_rate": 2.489759196304442e-07,
"loss": 0.09675334393978119,
"step": 1613
},
{
"epoch": 1.8615916955017302,
"grad_norm": 0.4734069066591148,
"learning_rate": 2.4492775069219543e-07,
"loss": 0.09361612051725388,
"step": 1614
},
{
"epoch": 1.8627450980392157,
"grad_norm": 0.45836974965808713,
"learning_rate": 2.4091235417144976e-07,
"loss": 0.09062512218952179,
"step": 1615
},
{
"epoch": 1.8638985005767013,
"grad_norm": 0.43371138727719816,
"learning_rate": 2.369297435580342e-07,
"loss": 0.08059115707874298,
"step": 1616
},
{
"epoch": 1.8650519031141868,
"grad_norm": 0.457125986392491,
"learning_rate": 2.3297993223163617e-07,
"loss": 0.07893478125333786,
"step": 1617
},
{
"epoch": 1.8662053056516723,
"grad_norm": 0.47345214518054546,
"learning_rate": 2.2906293346174335e-07,
"loss": 0.11157450824975967,
"step": 1618
},
{
"epoch": 1.867358708189158,
"grad_norm": 0.4244912214351953,
"learning_rate": 2.2517876040761942e-07,
"loss": 0.07758339494466782,
"step": 1619
},
{
"epoch": 1.8685121107266436,
"grad_norm": 0.42878092364292414,
"learning_rate": 2.2132742611824076e-07,
"loss": 0.07187170535326004,
"step": 1620
},
{
"epoch": 1.8696655132641293,
"grad_norm": 0.41794062900626666,
"learning_rate": 2.1750894353226637e-07,
"loss": 0.07352786511182785,
"step": 1621
},
{
"epoch": 1.8708189158016149,
"grad_norm": 0.4668079973927394,
"learning_rate": 2.1372332547798691e-07,
"loss": 0.089034304022789,
"step": 1622
},
{
"epoch": 1.8719723183391004,
"grad_norm": 0.4303700685426409,
"learning_rate": 2.0997058467328356e-07,
"loss": 0.08350430428981781,
"step": 1623
},
{
"epoch": 1.873125720876586,
"grad_norm": 0.44094057208223425,
"learning_rate": 2.0625073372558592e-07,
"loss": 0.08190502226352692,
"step": 1624
},
{
"epoch": 1.8742791234140714,
"grad_norm": 0.5054910071583596,
"learning_rate": 2.0256378513182961e-07,
"loss": 0.12169057875871658,
"step": 1625
},
{
"epoch": 1.875432525951557,
"grad_norm": 0.4761766098723886,
"learning_rate": 1.989097512784166e-07,
"loss": 0.09786316007375717,
"step": 1626
},
{
"epoch": 1.8765859284890427,
"grad_norm": 0.4315949684236786,
"learning_rate": 1.952886444411639e-07,
"loss": 0.08479120582342148,
"step": 1627
},
{
"epoch": 1.8777393310265282,
"grad_norm": 0.48894515658086257,
"learning_rate": 1.917004767852748e-07,
"loss": 0.10730808973312378,
"step": 1628
},
{
"epoch": 1.878892733564014,
"grad_norm": 0.46937629570278516,
"learning_rate": 1.8814526036529335e-07,
"loss": 0.08613970875740051,
"step": 1629
},
{
"epoch": 1.8800461361014995,
"grad_norm": 0.4512047469634718,
"learning_rate": 1.8462300712505877e-07,
"loss": 0.09288708865642548,
"step": 1630
},
{
"epoch": 1.881199538638985,
"grad_norm": 0.42935808738236647,
"learning_rate": 1.8113372889767223e-07,
"loss": 0.07483093440532684,
"step": 1631
},
{
"epoch": 1.8823529411764706,
"grad_norm": 0.40865508921823646,
"learning_rate": 1.7767743740545573e-07,
"loss": 0.07794409990310669,
"step": 1632
},
{
"epoch": 1.883506343713956,
"grad_norm": 0.4064589762210611,
"learning_rate": 1.7425414425990773e-07,
"loss": 0.07473921775817871,
"step": 1633
},
{
"epoch": 1.8846597462514416,
"grad_norm": 0.44669866561855986,
"learning_rate": 1.7086386096167085e-07,
"loss": 0.07308952510356903,
"step": 1634
},
{
"epoch": 1.8858131487889274,
"grad_norm": 0.44730890109404575,
"learning_rate": 1.6750659890049205e-07,
"loss": 0.09215351939201355,
"step": 1635
},
{
"epoch": 1.8869665513264129,
"grad_norm": 0.43529158444589533,
"learning_rate": 1.6418236935517695e-07,
"loss": 0.08526428043842316,
"step": 1636
},
{
"epoch": 1.8881199538638986,
"grad_norm": 0.4590106225705836,
"learning_rate": 1.6089118349356113e-07,
"loss": 0.10203632712364197,
"step": 1637
},
{
"epoch": 1.8892733564013842,
"grad_norm": 0.4227366499868551,
"learning_rate": 1.5763305237246894e-07,
"loss": 0.07506399601697922,
"step": 1638
},
{
"epoch": 1.8904267589388697,
"grad_norm": 0.4697073041474404,
"learning_rate": 1.544079869376769e-07,
"loss": 0.09345696866512299,
"step": 1639
},
{
"epoch": 1.8915801614763552,
"grad_norm": 0.4007007227217427,
"learning_rate": 1.5121599802387478e-07,
"loss": 0.07455568015575409,
"step": 1640
},
{
"epoch": 1.8927335640138407,
"grad_norm": 0.4049578621545422,
"learning_rate": 1.4805709635463127e-07,
"loss": 0.07587827742099762,
"step": 1641
},
{
"epoch": 1.8938869665513263,
"grad_norm": 0.4224264243200236,
"learning_rate": 1.4493129254236072e-07,
"loss": 0.08352547883987427,
"step": 1642
},
{
"epoch": 1.895040369088812,
"grad_norm": 0.4496137800009015,
"learning_rate": 1.418385970882785e-07,
"loss": 0.0929284393787384,
"step": 1643
},
{
"epoch": 1.8961937716262975,
"grad_norm": 0.41682572537414614,
"learning_rate": 1.3877902038237912e-07,
"loss": 0.08300832659006119,
"step": 1644
},
{
"epoch": 1.8973471741637833,
"grad_norm": 0.46402453794941906,
"learning_rate": 1.3575257270338592e-07,
"loss": 0.09526363760232925,
"step": 1645
},
{
"epoch": 1.8985005767012688,
"grad_norm": 0.4506811921401444,
"learning_rate": 1.327592642187303e-07,
"loss": 0.0772494524717331,
"step": 1646
},
{
"epoch": 1.8996539792387543,
"grad_norm": 0.4428827755567893,
"learning_rate": 1.297991049845082e-07,
"loss": 0.08490500599145889,
"step": 1647
},
{
"epoch": 1.9008073817762399,
"grad_norm": 0.44920677192117986,
"learning_rate": 1.2687210494545244e-07,
"loss": 0.09368325769901276,
"step": 1648
},
{
"epoch": 1.9019607843137254,
"grad_norm": 0.4524378844291378,
"learning_rate": 1.239782739348938e-07,
"loss": 0.08469931036233902,
"step": 1649
},
{
"epoch": 1.903114186851211,
"grad_norm": 0.47619748971697734,
"learning_rate": 1.211176216747323e-07,
"loss": 0.10856804251670837,
"step": 1650
},
{
"epoch": 1.9042675893886967,
"grad_norm": 0.4357712412075289,
"learning_rate": 1.1829015777540587e-07,
"loss": 0.07768280059099197,
"step": 1651
},
{
"epoch": 1.9054209919261822,
"grad_norm": 0.47462221849996417,
"learning_rate": 1.1549589173585063e-07,
"loss": 0.08845613896846771,
"step": 1652
},
{
"epoch": 1.906574394463668,
"grad_norm": 0.4293193289700487,
"learning_rate": 1.127348329434752e-07,
"loss": 0.07947489619255066,
"step": 1653
},
{
"epoch": 1.9077277970011535,
"grad_norm": 0.4498848773960795,
"learning_rate": 1.1000699067412857e-07,
"loss": 0.09646248817443848,
"step": 1654
},
{
"epoch": 1.908881199538639,
"grad_norm": 0.4162684175414704,
"learning_rate": 1.0731237409206784e-07,
"loss": 0.06901779770851135,
"step": 1655
},
{
"epoch": 1.9100346020761245,
"grad_norm": 0.4294002051866348,
"learning_rate": 1.0465099224992614e-07,
"loss": 0.07758603990077972,
"step": 1656
},
{
"epoch": 1.91118800461361,
"grad_norm": 0.4329597944770198,
"learning_rate": 1.0202285408868584e-07,
"loss": 0.08867383748292923,
"step": 1657
},
{
"epoch": 1.9123414071510956,
"grad_norm": 0.43464193563952697,
"learning_rate": 9.942796843764424e-08,
"loss": 0.08981835842132568,
"step": 1658
},
{
"epoch": 1.9134948096885813,
"grad_norm": 0.4450584025269132,
"learning_rate": 9.686634401438578e-08,
"loss": 0.08686129748821259,
"step": 1659
},
{
"epoch": 1.9146482122260668,
"grad_norm": 0.4221967208407953,
"learning_rate": 9.433798942475647e-08,
"loss": 0.08194129914045334,
"step": 1660
},
{
"epoch": 1.9158016147635526,
"grad_norm": 0.40284285035729195,
"learning_rate": 9.184291316282734e-08,
"loss": 0.07204432785511017,
"step": 1661
},
{
"epoch": 1.9169550173010381,
"grad_norm": 0.4413042973564617,
"learning_rate": 8.938112361087215e-08,
"loss": 0.09056876599788666,
"step": 1662
},
{
"epoch": 1.9181084198385236,
"grad_norm": 0.45152870994842365,
"learning_rate": 8.69526290393352e-08,
"loss": 0.09037656337022781,
"step": 1663
},
{
"epoch": 1.9192618223760092,
"grad_norm": 0.4395421653725749,
"learning_rate": 8.455743760680812e-08,
"loss": 0.08263800293207169,
"step": 1664
},
{
"epoch": 1.9204152249134947,
"grad_norm": 0.48690656543100486,
"learning_rate": 8.219555735999862e-08,
"loss": 0.09745900332927704,
"step": 1665
},
{
"epoch": 1.9215686274509802,
"grad_norm": 0.47727669351065555,
"learning_rate": 7.986699623370397e-08,
"loss": 0.09276721626520157,
"step": 1666
},
{
"epoch": 1.922722029988466,
"grad_norm": 0.4334378193313721,
"learning_rate": 7.757176205078542e-08,
"loss": 0.08352711796760559,
"step": 1667
},
{
"epoch": 1.9238754325259517,
"grad_norm": 0.4392002003646621,
"learning_rate": 7.530986252214157e-08,
"loss": 0.08440464735031128,
"step": 1668
},
{
"epoch": 1.9250288350634372,
"grad_norm": 0.4898265765050765,
"learning_rate": 7.308130524668389e-08,
"loss": 0.11039169132709503,
"step": 1669
},
{
"epoch": 1.9261822376009228,
"grad_norm": 0.4689937470815329,
"learning_rate": 7.088609771130572e-08,
"loss": 0.09221184253692627,
"step": 1670
},
{
"epoch": 1.9273356401384083,
"grad_norm": 0.44618964761752694,
"learning_rate": 6.872424729086447e-08,
"loss": 0.10166947543621063,
"step": 1671
},
{
"epoch": 1.9284890426758938,
"grad_norm": 0.44363808214869627,
"learning_rate": 6.65957612481527e-08,
"loss": 0.07165621966123581,
"step": 1672
},
{
"epoch": 1.9296424452133794,
"grad_norm": 0.4297449915434851,
"learning_rate": 6.450064673387491e-08,
"loss": 0.08633308112621307,
"step": 1673
},
{
"epoch": 1.930795847750865,
"grad_norm": 0.4507618256018681,
"learning_rate": 6.24389107866219e-08,
"loss": 0.08759327232837677,
"step": 1674
},
{
"epoch": 1.9319492502883506,
"grad_norm": 0.45543239318792605,
"learning_rate": 6.041056033284864e-08,
"loss": 0.10010764002799988,
"step": 1675
},
{
"epoch": 1.9331026528258364,
"grad_norm": 0.40644024251284744,
"learning_rate": 5.841560218685205e-08,
"loss": 0.07530717551708221,
"step": 1676
},
{
"epoch": 1.934256055363322,
"grad_norm": 0.4618201653360246,
"learning_rate": 5.6454043050743204e-08,
"loss": 0.09785068035125732,
"step": 1677
},
{
"epoch": 1.9354094579008074,
"grad_norm": 0.4745989896818971,
"learning_rate": 5.452588951443405e-08,
"loss": 0.09745068848133087,
"step": 1678
},
{
"epoch": 1.936562860438293,
"grad_norm": 0.41159117643376797,
"learning_rate": 5.2631148055602986e-08,
"loss": 0.07882726192474365,
"step": 1679
},
{
"epoch": 1.9377162629757785,
"grad_norm": 0.4598826256281131,
"learning_rate": 5.076982503968597e-08,
"loss": 0.08580614626407623,
"step": 1680
},
{
"epoch": 1.938869665513264,
"grad_norm": 0.4370098798290825,
"learning_rate": 4.894192671984321e-08,
"loss": 0.08357317745685577,
"step": 1681
},
{
"epoch": 1.9400230680507498,
"grad_norm": 0.4398965700292615,
"learning_rate": 4.714745923694808e-08,
"loss": 0.08020699769258499,
"step": 1682
},
{
"epoch": 1.9411764705882353,
"grad_norm": 0.44009680393742334,
"learning_rate": 4.538642861956044e-08,
"loss": 0.08874964714050293,
"step": 1683
},
{
"epoch": 1.942329873125721,
"grad_norm": 0.4533413355355902,
"learning_rate": 4.3658840783906695e-08,
"loss": 0.09573370963335037,
"step": 1684
},
{
"epoch": 1.9434832756632066,
"grad_norm": 0.4333525216486871,
"learning_rate": 4.1964701533864226e-08,
"loss": 0.08738063275814056,
"step": 1685
},
{
"epoch": 1.944636678200692,
"grad_norm": 0.8805372679339996,
"learning_rate": 4.030401656093697e-08,
"loss": 0.07910633832216263,
"step": 1686
},
{
"epoch": 1.9457900807381776,
"grad_norm": 0.4418864869239784,
"learning_rate": 3.867679144423986e-08,
"loss": 0.09160259366035461,
"step": 1687
},
{
"epoch": 1.9469434832756631,
"grad_norm": 0.4236487027391057,
"learning_rate": 3.7083031650477776e-08,
"loss": 0.08856689929962158,
"step": 1688
},
{
"epoch": 1.9480968858131487,
"grad_norm": 0.4051298041617625,
"learning_rate": 3.5522742533926626e-08,
"loss": 0.07460395991802216,
"step": 1689
},
{
"epoch": 1.9492502883506344,
"grad_norm": 0.4348792244264226,
"learning_rate": 3.399592933641893e-08,
"loss": 0.08177849650382996,
"step": 1690
},
{
"epoch": 1.95040369088812,
"grad_norm": 0.44100629268998814,
"learning_rate": 3.2502597187322736e-08,
"loss": 0.08310138434171677,
"step": 1691
},
{
"epoch": 1.9515570934256057,
"grad_norm": 0.4161821447451508,
"learning_rate": 3.104275110352717e-08,
"loss": 0.07306216657161713,
"step": 1692
},
{
"epoch": 1.9527104959630912,
"grad_norm": 0.4205460743868366,
"learning_rate": 2.9616395989421342e-08,
"loss": 0.0771905854344368,
"step": 1693
},
{
"epoch": 1.9538638985005767,
"grad_norm": 0.414517624269698,
"learning_rate": 2.8223536636883265e-08,
"loss": 0.07649216055870056,
"step": 1694
},
{
"epoch": 1.9550173010380623,
"grad_norm": 0.41892503570215905,
"learning_rate": 2.6864177725259843e-08,
"loss": 0.07345999032258987,
"step": 1695
},
{
"epoch": 1.9561707035755478,
"grad_norm": 0.44896321506968956,
"learning_rate": 2.5538323821351352e-08,
"loss": 0.0886496901512146,
"step": 1696
},
{
"epoch": 1.9573241061130333,
"grad_norm": 0.40584053731616787,
"learning_rate": 2.424597937940032e-08,
"loss": 0.07106611877679825,
"step": 1697
},
{
"epoch": 1.958477508650519,
"grad_norm": 0.43058316716909206,
"learning_rate": 2.298714874106933e-08,
"loss": 0.08795999735593796,
"step": 1698
},
{
"epoch": 1.9596309111880046,
"grad_norm": 0.40141925468706763,
"learning_rate": 2.1761836135433257e-08,
"loss": 0.07311815023422241,
"step": 1699
},
{
"epoch": 1.9607843137254903,
"grad_norm": 0.4714205008344836,
"learning_rate": 2.057004567896148e-08,
"loss": 0.09680025279521942,
"step": 1700
},
{
"epoch": 1.9619377162629759,
"grad_norm": 0.42483282768669783,
"learning_rate": 1.941178137550459e-08,
"loss": 0.0846504345536232,
"step": 1701
},
{
"epoch": 1.9630911188004614,
"grad_norm": 0.4082301263527964,
"learning_rate": 1.8287047116282154e-08,
"loss": 0.06939003616571426,
"step": 1702
},
{
"epoch": 1.964244521337947,
"grad_norm": 0.4275511555463328,
"learning_rate": 1.719584667986829e-08,
"loss": 0.07961012423038483,
"step": 1703
},
{
"epoch": 1.9653979238754324,
"grad_norm": 0.4169519116920383,
"learning_rate": 1.6138183732177236e-08,
"loss": 0.08381155133247375,
"step": 1704
},
{
"epoch": 1.966551326412918,
"grad_norm": 0.4164496063867734,
"learning_rate": 1.5114061826456696e-08,
"loss": 0.07898962497711182,
"step": 1705
},
{
"epoch": 1.9677047289504037,
"grad_norm": 0.4179585266322974,
"learning_rate": 1.4123484403270049e-08,
"loss": 0.07682450115680695,
"step": 1706
},
{
"epoch": 1.9688581314878892,
"grad_norm": 0.43216775289592646,
"learning_rate": 1.3166454790488614e-08,
"loss": 0.08242955803871155,
"step": 1707
},
{
"epoch": 1.970011534025375,
"grad_norm": 0.49621437373318783,
"learning_rate": 1.224297620327608e-08,
"loss": 0.1118517518043518,
"step": 1708
},
{
"epoch": 1.9711649365628605,
"grad_norm": 0.4506955187985243,
"learning_rate": 1.1353051744085185e-08,
"loss": 0.07482790946960449,
"step": 1709
},
{
"epoch": 1.972318339100346,
"grad_norm": 0.4064634839279055,
"learning_rate": 1.0496684402638847e-08,
"loss": 0.07155830413103104,
"step": 1710
},
{
"epoch": 1.9734717416378316,
"grad_norm": 0.4080577473861031,
"learning_rate": 9.673877055924596e-09,
"loss": 0.07682958245277405,
"step": 1711
},
{
"epoch": 1.974625144175317,
"grad_norm": 0.43842125076477284,
"learning_rate": 8.884632468186827e-09,
"loss": 0.08658139407634735,
"step": 1712
},
{
"epoch": 1.9757785467128026,
"grad_norm": 0.45587186644927224,
"learning_rate": 8.128953290911235e-09,
"loss": 0.08554649353027344,
"step": 1713
},
{
"epoch": 1.9769319492502884,
"grad_norm": 0.438344533662239,
"learning_rate": 7.4068420628226055e-09,
"loss": 0.08503372222185135,
"step": 1714
},
{
"epoch": 1.978085351787774,
"grad_norm": 0.48651313336330354,
"learning_rate": 6.718301209872602e-09,
"loss": 0.06198807805776596,
"step": 1715
},
{
"epoch": 1.9792387543252596,
"grad_norm": 0.4272538309296738,
"learning_rate": 6.063333045230879e-09,
"loss": 0.08505403250455856,
"step": 1716
},
{
"epoch": 1.9803921568627452,
"grad_norm": 0.435323781601703,
"learning_rate": 5.441939769280646e-09,
"loss": 0.08017569035291672,
"step": 1717
},
{
"epoch": 1.9815455594002307,
"grad_norm": 0.3956387481382087,
"learning_rate": 4.854123469608673e-09,
"loss": 0.07134771347045898,
"step": 1718
},
{
"epoch": 1.9826989619377162,
"grad_norm": 0.45995413961064047,
"learning_rate": 4.2998861209997415e-09,
"loss": 0.0966772586107254,
"step": 1719
},
{
"epoch": 1.9838523644752017,
"grad_norm": 0.39782283495144244,
"learning_rate": 3.7792295854277575e-09,
"loss": 0.0716339647769928,
"step": 1720
},
{
"epoch": 1.9850057670126873,
"grad_norm": 0.4369240107589842,
"learning_rate": 3.2921556120513178e-09,
"loss": 0.08201077580451965,
"step": 1721
},
{
"epoch": 1.986159169550173,
"grad_norm": 0.4442749928596274,
"learning_rate": 2.8386658372081543e-09,
"loss": 0.07989010959863663,
"step": 1722
},
{
"epoch": 1.9873125720876585,
"grad_norm": 0.42967459816649856,
"learning_rate": 2.4187617844095846e-09,
"loss": 0.08655820786952972,
"step": 1723
},
{
"epoch": 1.9884659746251443,
"grad_norm": 0.43237966786939785,
"learning_rate": 2.032444864333849e-09,
"loss": 0.08806741237640381,
"step": 1724
},
{
"epoch": 1.9896193771626298,
"grad_norm": 0.41616303114284064,
"learning_rate": 1.6797163748227819e-09,
"loss": 0.07314859330654144,
"step": 1725
},
{
"epoch": 1.9907727797001153,
"grad_norm": 0.4063570334407316,
"learning_rate": 1.360577500877369e-09,
"loss": 0.07585233449935913,
"step": 1726
},
{
"epoch": 1.9919261822376009,
"grad_norm": 0.45500003676904655,
"learning_rate": 1.075029314651088e-09,
"loss": 0.09444887936115265,
"step": 1727
},
{
"epoch": 1.9930795847750864,
"grad_norm": 0.4343399697541974,
"learning_rate": 8.230727754521273e-10,
"loss": 0.08475548774003983,
"step": 1728
},
{
"epoch": 1.994232987312572,
"grad_norm": 0.4373280556000089,
"learning_rate": 6.047087297345045e-10,
"loss": 0.09195937216281891,
"step": 1729
},
{
"epoch": 1.9953863898500577,
"grad_norm": 0.44219397794964116,
"learning_rate": 4.1993791109806727e-10,
"loss": 0.08413922041654587,
"step": 1730
},
{
"epoch": 1.9965397923875432,
"grad_norm": 0.40031592780024006,
"learning_rate": 2.687609402851621e-10,
"loss": 0.06829897314310074,
"step": 1731
},
{
"epoch": 1.997693194925029,
"grad_norm": 0.48330964609960303,
"learning_rate": 1.511783251784138e-10,
"loss": 0.10645907372236252,
"step": 1732
},
{
"epoch": 1.9988465974625145,
"grad_norm": 0.444685741890041,
"learning_rate": 6.719046079961545e-11,
"loss": 0.09139378368854523,
"step": 1733
},
{
"epoch": 2.0,
"grad_norm": 0.37410386655851224,
"learning_rate": 1.6797629307507835e-11,
"loss": 0.06574840098619461,
"step": 1734
},
{
"epoch": 2.0,
"step": 1734,
"total_flos": 3975542972350464.0,
"train_loss": 0.12377502597791101,
"train_runtime": 47384.8787,
"train_samples_per_second": 0.146,
"train_steps_per_second": 0.037
}
],
"logging_steps": 1,
"max_steps": 1734,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3975542972350464.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}