Text Ranking
sentence-transformers
Safetensors
xlm-roberta
cross-encoder
reranker
Generated from Trainer
dataset_size:3299295
loss:BinaryCrossEntropyLoss
Eval Results (legacy)
text-embeddings-inference
Instructions to use Starkdoorstep12/entity-resolution-crossencoder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use Starkdoorstep12/entity-resolution-crossencoder with sentence-transformers:
from sentence_transformers import CrossEncoder model = CrossEncoder("Starkdoorstep12/entity-resolution-crossencoder") query = "Which planet is known as the Red Planet?" passages = [ "Venus is often called Earth's twin because of its similar size and proximity.", "Mars, known for its reddish appearance, is often referred to as the Red Planet.", "Jupiter, the largest planet in our solar system, has a prominent red spot.", "Saturn, famous for its rings, is sometimes mistaken for the Red Planet." ] scores = model.predict([(query, passage) for passage in passages]) print(scores) - Notebooks
- Google Colab
- Kaggle
Download checkpoint-103104/trainer_state.json from Starkdoorstep12/entity-resolution-crossencoder: direct link, hf CLI and curl.
- Browser
- Download file 394 kB
-
https://huggingface.co/Starkdoorstep12/entity-resolution-crossencoder/resolve/main/checkpoint-103104/trainer_state.json
- Command line
-
hf download hf://Starkdoorstep12/entity-resolution-crossencoder/checkpoint-103104/trainer_state.json
-
curl -L -o trainer_state.json https://huggingface.co/Starkdoorstep12/entity-resolution-crossencoder/resolve/main/checkpoint-103104/trainer_state.json
394 kB
{
"best_global_step": 103104,
"best_metric": 0.9996460692260924,
"best_model_checkpoint": "../ce_model_100k/checkpoint-103104",
"epoch": 2.0,
"eval_steps": 500,
"global_step": 103104,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0009698944754810676,
"grad_norm": 27.08783531188965,
"learning_rate": 9.504412763068568e-08,
"loss": 1.3570481872558593,
"step": 50
},
{
"epoch": 0.0019397889509621352,
"grad_norm": 20.9023494720459,
"learning_rate": 1.92027931335467e-07,
"loss": 1.1735592651367188,
"step": 100
},
{
"epoch": 0.002909683426443203,
"grad_norm": 8.505711555480957,
"learning_rate": 2.890117350402483e-07,
"loss": 0.8990924072265625,
"step": 150
},
{
"epoch": 0.0038795779019242705,
"grad_norm": 4.790644645690918,
"learning_rate": 3.859955387450296e-07,
"loss": 0.6421822357177734,
"step": 200
},
{
"epoch": 0.004849472377405338,
"grad_norm": 5.505950927734375,
"learning_rate": 4.829793424498109e-07,
"loss": 0.45252552032470705,
"step": 250
},
{
"epoch": 0.005819366852886406,
"grad_norm": 6.148208141326904,
"learning_rate": 5.799631461545922e-07,
"loss": 0.3296080780029297,
"step": 300
},
{
"epoch": 0.006789261328367473,
"grad_norm": 2.746929168701172,
"learning_rate": 6.769469498593735e-07,
"loss": 0.21170696258544922,
"step": 350
},
{
"epoch": 0.007759155803848541,
"grad_norm": 3.268200397491455,
"learning_rate": 7.739307535641548e-07,
"loss": 0.2188690185546875,
"step": 400
},
{
"epoch": 0.008729050279329608,
"grad_norm": 4.391243934631348,
"learning_rate": 8.709145572689362e-07,
"loss": 0.21369564056396484,
"step": 450
},
{
"epoch": 0.009698944754810677,
"grad_norm": 3.1883127689361572,
"learning_rate": 9.678983609737174e-07,
"loss": 0.17295127868652344,
"step": 500
},
{
"epoch": 0.010668839230291743,
"grad_norm": 3.346428394317627,
"learning_rate": 1.0648821646784986e-06,
"loss": 0.16976539611816407,
"step": 550
},
{
"epoch": 0.011638733705772812,
"grad_norm": 2.9807257652282715,
"learning_rate": 1.1618659683832801e-06,
"loss": 0.16226112365722656,
"step": 600
},
{
"epoch": 0.01260862818125388,
"grad_norm": 2.836063861846924,
"learning_rate": 1.2588497720880614e-06,
"loss": 0.1293662452697754,
"step": 650
},
{
"epoch": 0.013578522656734947,
"grad_norm": 4.722393035888672,
"learning_rate": 1.3558335757928427e-06,
"loss": 0.12232763290405274,
"step": 700
},
{
"epoch": 0.014548417132216015,
"grad_norm": 1.9236106872558594,
"learning_rate": 1.4528173794976242e-06,
"loss": 0.12566470146179198,
"step": 750
},
{
"epoch": 0.015518311607697082,
"grad_norm": 1.82411789894104,
"learning_rate": 1.5498011832024054e-06,
"loss": 0.11197481155395508,
"step": 800
},
{
"epoch": 0.01648820608317815,
"grad_norm": 1.993725299835205,
"learning_rate": 1.6467849869071867e-06,
"loss": 0.11361078262329101,
"step": 850
},
{
"epoch": 0.017458100558659217,
"grad_norm": 2.9138834476470947,
"learning_rate": 1.743768790611968e-06,
"loss": 0.10107772827148437,
"step": 900
},
{
"epoch": 0.018427995034140287,
"grad_norm": 4.403861999511719,
"learning_rate": 1.8407525943167492e-06,
"loss": 0.09816398620605468,
"step": 950
},
{
"epoch": 0.019397889509621354,
"grad_norm": 2.7256076335906982,
"learning_rate": 1.9377363980215305e-06,
"loss": 0.10918959617614746,
"step": 1000
},
{
"epoch": 0.02036778398510242,
"grad_norm": 3.7052271366119385,
"learning_rate": 2.0347202017263118e-06,
"loss": 0.09358957290649414,
"step": 1050
},
{
"epoch": 0.021337678460583487,
"grad_norm": 2.7027766704559326,
"learning_rate": 2.131704005431093e-06,
"loss": 0.0991772747039795,
"step": 1100
},
{
"epoch": 0.022307572936064557,
"grad_norm": 2.5295498371124268,
"learning_rate": 2.2286878091358748e-06,
"loss": 0.09076035499572754,
"step": 1150
},
{
"epoch": 0.023277467411545624,
"grad_norm": 6.041042804718018,
"learning_rate": 2.325671612840656e-06,
"loss": 0.07770174503326416,
"step": 1200
},
{
"epoch": 0.02424736188702669,
"grad_norm": 2.3193199634552,
"learning_rate": 2.4226554165454373e-06,
"loss": 0.08126268386840821,
"step": 1250
},
{
"epoch": 0.02521725636250776,
"grad_norm": 2.82558274269104,
"learning_rate": 2.5196392202502186e-06,
"loss": 0.07243297100067139,
"step": 1300
},
{
"epoch": 0.026187150837988827,
"grad_norm": 3.5613627433776855,
"learning_rate": 2.6166230239549994e-06,
"loss": 0.07283824920654297,
"step": 1350
},
{
"epoch": 0.027157045313469894,
"grad_norm": 2.2605931758880615,
"learning_rate": 2.713606827659781e-06,
"loss": 0.07112550735473633,
"step": 1400
},
{
"epoch": 0.028126939788950964,
"grad_norm": 2.785177230834961,
"learning_rate": 2.8105906313645624e-06,
"loss": 0.07532407760620118,
"step": 1450
},
{
"epoch": 0.02909683426443203,
"grad_norm": 4.442914009094238,
"learning_rate": 2.9075744350693437e-06,
"loss": 0.06706559181213378,
"step": 1500
},
{
"epoch": 0.030066728739913097,
"grad_norm": 3.779153347015381,
"learning_rate": 3.004558238774125e-06,
"loss": 0.0789792537689209,
"step": 1550
},
{
"epoch": 0.031036623215394164,
"grad_norm": 1.5070732831954956,
"learning_rate": 3.101542042478906e-06,
"loss": 0.06849954128265381,
"step": 1600
},
{
"epoch": 0.03200651769087523,
"grad_norm": 3.2382469177246094,
"learning_rate": 3.1985258461836875e-06,
"loss": 0.06706800460815429,
"step": 1650
},
{
"epoch": 0.0329764121663563,
"grad_norm": 2.562574625015259,
"learning_rate": 3.2955096498884687e-06,
"loss": 0.06839064598083495,
"step": 1700
},
{
"epoch": 0.03394630664183737,
"grad_norm": 2.8584535121917725,
"learning_rate": 3.39249345359325e-06,
"loss": 0.05964101314544678,
"step": 1750
},
{
"epoch": 0.034916201117318434,
"grad_norm": 2.199098825454712,
"learning_rate": 3.4894772572980317e-06,
"loss": 0.052366929054260256,
"step": 1800
},
{
"epoch": 0.035886095592799504,
"grad_norm": 1.8702822923660278,
"learning_rate": 3.586461061002813e-06,
"loss": 0.06610711574554444,
"step": 1850
},
{
"epoch": 0.036855990068280574,
"grad_norm": 1.4206616878509521,
"learning_rate": 3.6834448647075942e-06,
"loss": 0.06082666397094726,
"step": 1900
},
{
"epoch": 0.03782588454376164,
"grad_norm": 2.163145065307617,
"learning_rate": 3.7804286684123755e-06,
"loss": 0.0696146821975708,
"step": 1950
},
{
"epoch": 0.03879577901924271,
"grad_norm": 5.193337917327881,
"learning_rate": 3.877412472117157e-06,
"loss": 0.06211214542388916,
"step": 2000
},
{
"epoch": 0.03976567349472378,
"grad_norm": 8.509782791137695,
"learning_rate": 3.974396275821938e-06,
"loss": 0.04595749378204346,
"step": 2050
},
{
"epoch": 0.04073556797020484,
"grad_norm": 3.032975435256958,
"learning_rate": 4.07138007952672e-06,
"loss": 0.05319104671478272,
"step": 2100
},
{
"epoch": 0.04170546244568591,
"grad_norm": 2.520284414291382,
"learning_rate": 4.1683638832315e-06,
"loss": 0.051291098594665525,
"step": 2150
},
{
"epoch": 0.042675356921166974,
"grad_norm": 2.1702213287353516,
"learning_rate": 4.265347686936282e-06,
"loss": 0.05073753833770752,
"step": 2200
},
{
"epoch": 0.043645251396648044,
"grad_norm": 2.203334093093872,
"learning_rate": 4.3623314906410636e-06,
"loss": 0.06516064643859863,
"step": 2250
},
{
"epoch": 0.044615145872129114,
"grad_norm": 1.5464969873428345,
"learning_rate": 4.459315294345845e-06,
"loss": 0.05410961151123047,
"step": 2300
},
{
"epoch": 0.04558504034761018,
"grad_norm": 3.1183531284332275,
"learning_rate": 4.556299098050626e-06,
"loss": 0.058303179740905764,
"step": 2350
},
{
"epoch": 0.04655493482309125,
"grad_norm": 5.776540279388428,
"learning_rate": 4.653282901755407e-06,
"loss": 0.04615755558013916,
"step": 2400
},
{
"epoch": 0.04752482929857232,
"grad_norm": 1.5153908729553223,
"learning_rate": 4.750266705460189e-06,
"loss": 0.049284563064575196,
"step": 2450
},
{
"epoch": 0.04849472377405338,
"grad_norm": 1.6825062036514282,
"learning_rate": 4.84725050916497e-06,
"loss": 0.053444743156433105,
"step": 2500
},
{
"epoch": 0.04946461824953445,
"grad_norm": 2.0927391052246094,
"learning_rate": 4.944234312869751e-06,
"loss": 0.055415797233581546,
"step": 2550
},
{
"epoch": 0.05043451272501552,
"grad_norm": 0.8396100997924805,
"learning_rate": 5.0412181165745325e-06,
"loss": 0.043874931335449216,
"step": 2600
},
{
"epoch": 0.051404407200496584,
"grad_norm": 3.5946476459503174,
"learning_rate": 5.138201920279315e-06,
"loss": 0.046492481231689455,
"step": 2650
},
{
"epoch": 0.052374301675977654,
"grad_norm": 6.901855945587158,
"learning_rate": 5.235185723984095e-06,
"loss": 0.047220892906188965,
"step": 2700
},
{
"epoch": 0.053344196151458724,
"grad_norm": 3.529270648956299,
"learning_rate": 5.332169527688876e-06,
"loss": 0.05250232219696045,
"step": 2750
},
{
"epoch": 0.05431409062693979,
"grad_norm": 2.119624376296997,
"learning_rate": 5.4291533313936576e-06,
"loss": 0.04291744709014893,
"step": 2800
},
{
"epoch": 0.05528398510242086,
"grad_norm": 1.4964957237243652,
"learning_rate": 5.52613713509844e-06,
"loss": 0.05332083702087402,
"step": 2850
},
{
"epoch": 0.05625387957790193,
"grad_norm": 1.1058987379074097,
"learning_rate": 5.62312093880322e-06,
"loss": 0.0486084508895874,
"step": 2900
},
{
"epoch": 0.05722377405338299,
"grad_norm": 3.1241092681884766,
"learning_rate": 5.720104742508001e-06,
"loss": 0.04736727237701416,
"step": 2950
},
{
"epoch": 0.05819366852886406,
"grad_norm": 2.7152745723724365,
"learning_rate": 5.817088546212783e-06,
"loss": 0.03962032318115234,
"step": 3000
},
{
"epoch": 0.059163563004345124,
"grad_norm": 3.2394368648529053,
"learning_rate": 5.914072349917565e-06,
"loss": 0.04241170883178711,
"step": 3050
},
{
"epoch": 0.060133457479826194,
"grad_norm": 2.0426104068756104,
"learning_rate": 6.011056153622345e-06,
"loss": 0.04628231525421143,
"step": 3100
},
{
"epoch": 0.061103351955307264,
"grad_norm": 2.0696940422058105,
"learning_rate": 6.1080399573271265e-06,
"loss": 0.042524590492248535,
"step": 3150
},
{
"epoch": 0.06207324643078833,
"grad_norm": 3.846797466278076,
"learning_rate": 6.2050237610319086e-06,
"loss": 0.048424487113952634,
"step": 3200
},
{
"epoch": 0.0630431409062694,
"grad_norm": 3.7898294925689697,
"learning_rate": 6.30200756473669e-06,
"loss": 0.04111265182495117,
"step": 3250
},
{
"epoch": 0.06401303538175046,
"grad_norm": 4.234498023986816,
"learning_rate": 6.39899136844147e-06,
"loss": 0.04294558048248291,
"step": 3300
},
{
"epoch": 0.06498292985723153,
"grad_norm": 3.7738075256347656,
"learning_rate": 6.4959751721462515e-06,
"loss": 0.029208140373229982,
"step": 3350
},
{
"epoch": 0.0659528243327126,
"grad_norm": 0.26079168915748596,
"learning_rate": 6.592958975851034e-06,
"loss": 0.038891749382019045,
"step": 3400
},
{
"epoch": 0.06692271880819367,
"grad_norm": 3.0058045387268066,
"learning_rate": 6.689942779555815e-06,
"loss": 0.03728559732437134,
"step": 3450
},
{
"epoch": 0.06789261328367474,
"grad_norm": 3.358690023422241,
"learning_rate": 6.786926583260595e-06,
"loss": 0.046622242927551266,
"step": 3500
},
{
"epoch": 0.0688625077591558,
"grad_norm": 1.2739336490631104,
"learning_rate": 6.8839103869653775e-06,
"loss": 0.03435472249984741,
"step": 3550
},
{
"epoch": 0.06983240223463687,
"grad_norm": 4.971105098724365,
"learning_rate": 6.980894190670159e-06,
"loss": 0.03709745168685913,
"step": 3600
},
{
"epoch": 0.07080229671011794,
"grad_norm": 2.439927577972412,
"learning_rate": 7.07787799437494e-06,
"loss": 0.037032637596130374,
"step": 3650
},
{
"epoch": 0.07177219118559901,
"grad_norm": 1.6749036312103271,
"learning_rate": 7.1748617980797204e-06,
"loss": 0.04503728866577148,
"step": 3700
},
{
"epoch": 0.07274208566108008,
"grad_norm": 0.7645151019096375,
"learning_rate": 7.2718456017845026e-06,
"loss": 0.03172181129455567,
"step": 3750
},
{
"epoch": 0.07371198013656115,
"grad_norm": 1.902260184288025,
"learning_rate": 7.368829405489284e-06,
"loss": 0.04537493228912354,
"step": 3800
},
{
"epoch": 0.0746818746120422,
"grad_norm": 1.6529265642166138,
"learning_rate": 7.465813209194065e-06,
"loss": 0.04448251724243164,
"step": 3850
},
{
"epoch": 0.07565176908752327,
"grad_norm": 1.9685674905776978,
"learning_rate": 7.562797012898847e-06,
"loss": 0.04649857997894287,
"step": 3900
},
{
"epoch": 0.07662166356300434,
"grad_norm": 1.3922754526138306,
"learning_rate": 7.659780816603628e-06,
"loss": 0.03474323511123657,
"step": 3950
},
{
"epoch": 0.07759155803848541,
"grad_norm": 3.2225141525268555,
"learning_rate": 7.756764620308409e-06,
"loss": 0.04120240688323975,
"step": 4000
},
{
"epoch": 0.07856145251396648,
"grad_norm": 0.19655612111091614,
"learning_rate": 7.85374842401319e-06,
"loss": 0.03230772733688354,
"step": 4050
},
{
"epoch": 0.07953134698944755,
"grad_norm": 1.9164402484893799,
"learning_rate": 7.950732227717971e-06,
"loss": 0.036763713359832764,
"step": 4100
},
{
"epoch": 0.08050124146492861,
"grad_norm": 0.3221072256565094,
"learning_rate": 8.047716031422754e-06,
"loss": 0.04401978492736816,
"step": 4150
},
{
"epoch": 0.08147113594040968,
"grad_norm": 2.6497626304626465,
"learning_rate": 8.144699835127534e-06,
"loss": 0.03281592607498169,
"step": 4200
},
{
"epoch": 0.08244103041589075,
"grad_norm": 0.2473970353603363,
"learning_rate": 8.241683638832314e-06,
"loss": 0.03679916143417358,
"step": 4250
},
{
"epoch": 0.08341092489137182,
"grad_norm": 5.424220085144043,
"learning_rate": 8.338667442537097e-06,
"loss": 0.03891810178756714,
"step": 4300
},
{
"epoch": 0.08438081936685289,
"grad_norm": 3.310107707977295,
"learning_rate": 8.435651246241879e-06,
"loss": 0.04210967063903809,
"step": 4350
},
{
"epoch": 0.08535071384233395,
"grad_norm": 1.6154475212097168,
"learning_rate": 8.532635049946659e-06,
"loss": 0.03312542200088501,
"step": 4400
},
{
"epoch": 0.08632060831781502,
"grad_norm": 1.5331177711486816,
"learning_rate": 8.629618853651441e-06,
"loss": 0.033189141750335695,
"step": 4450
},
{
"epoch": 0.08729050279329609,
"grad_norm": 1.4805681705474854,
"learning_rate": 8.726602657356222e-06,
"loss": 0.040745387077331545,
"step": 4500
},
{
"epoch": 0.08826039726877716,
"grad_norm": 3.989293336868286,
"learning_rate": 8.823586461061004e-06,
"loss": 0.027628159523010253,
"step": 4550
},
{
"epoch": 0.08923029174425823,
"grad_norm": 0.6511023044586182,
"learning_rate": 8.920570264765784e-06,
"loss": 0.036110734939575194,
"step": 4600
},
{
"epoch": 0.0902001862197393,
"grad_norm": 3.284632444381714,
"learning_rate": 9.017554068470566e-06,
"loss": 0.02985992670059204,
"step": 4650
},
{
"epoch": 0.09117008069522035,
"grad_norm": 2.6430792808532715,
"learning_rate": 9.114537872175347e-06,
"loss": 0.02530299186706543,
"step": 4700
},
{
"epoch": 0.09213997517070142,
"grad_norm": 1.503933310508728,
"learning_rate": 9.211521675880129e-06,
"loss": 0.03698864698410034,
"step": 4750
},
{
"epoch": 0.0931098696461825,
"grad_norm": 0.16252748668193817,
"learning_rate": 9.308505479584911e-06,
"loss": 0.030392026901245116,
"step": 4800
},
{
"epoch": 0.09407976412166356,
"grad_norm": 0.021069668233394623,
"learning_rate": 9.405489283289691e-06,
"loss": 0.03479726791381836,
"step": 4850
},
{
"epoch": 0.09504965859714463,
"grad_norm": 1.2566858530044556,
"learning_rate": 9.502473086994472e-06,
"loss": 0.030181329250335693,
"step": 4900
},
{
"epoch": 0.0960195530726257,
"grad_norm": 0.4371766448020935,
"learning_rate": 9.599456890699254e-06,
"loss": 0.032189674377441406,
"step": 4950
},
{
"epoch": 0.09698944754810676,
"grad_norm": 0.37218987941741943,
"learning_rate": 9.696440694404036e-06,
"loss": 0.031464502811431885,
"step": 5000
},
{
"epoch": 0.09795934202358783,
"grad_norm": 2.5261168479919434,
"learning_rate": 9.793424498108816e-06,
"loss": 0.04022056102752686,
"step": 5050
},
{
"epoch": 0.0989292364990689,
"grad_norm": 1.2079966068267822,
"learning_rate": 9.890408301813597e-06,
"loss": 0.03203359603881836,
"step": 5100
},
{
"epoch": 0.09989913097454997,
"grad_norm": 2.1484744548797607,
"learning_rate": 9.987392105518379e-06,
"loss": 0.03688978433609009,
"step": 5150
},
{
"epoch": 0.10086902545003104,
"grad_norm": 1.4678980112075806,
"learning_rate": 1.0084375909223161e-05,
"loss": 0.027073802947998046,
"step": 5200
},
{
"epoch": 0.1018389199255121,
"grad_norm": 1.1997852325439453,
"learning_rate": 1.0181359712927943e-05,
"loss": 0.03720566034317017,
"step": 5250
},
{
"epoch": 0.10280881440099317,
"grad_norm": 1.260424017906189,
"learning_rate": 1.0278343516632722e-05,
"loss": 0.034733655452728274,
"step": 5300
},
{
"epoch": 0.10377870887647424,
"grad_norm": 2.1502304077148438,
"learning_rate": 1.0375327320337504e-05,
"loss": 0.03305077314376831,
"step": 5350
},
{
"epoch": 0.10474860335195531,
"grad_norm": 1.391677737236023,
"learning_rate": 1.0472311124042285e-05,
"loss": 0.03419046640396118,
"step": 5400
},
{
"epoch": 0.10571849782743638,
"grad_norm": 4.9873175621032715,
"learning_rate": 1.0569294927747067e-05,
"loss": 0.0364527153968811,
"step": 5450
},
{
"epoch": 0.10668839230291745,
"grad_norm": 0.039271898567676544,
"learning_rate": 1.0666278731451849e-05,
"loss": 0.03177512168884277,
"step": 5500
},
{
"epoch": 0.1076582867783985,
"grad_norm": 3.9023807048797607,
"learning_rate": 1.076326253515663e-05,
"loss": 0.017125914096832274,
"step": 5550
},
{
"epoch": 0.10862818125387957,
"grad_norm": 1.3851922750473022,
"learning_rate": 1.0860246338861411e-05,
"loss": 0.032180070877075195,
"step": 5600
},
{
"epoch": 0.10959807572936064,
"grad_norm": 1.3358694314956665,
"learning_rate": 1.0957230142566193e-05,
"loss": 0.036840176582336424,
"step": 5650
},
{
"epoch": 0.11056797020484171,
"grad_norm": 0.08547809720039368,
"learning_rate": 1.1054213946270974e-05,
"loss": 0.030385522842407225,
"step": 5700
},
{
"epoch": 0.11153786468032278,
"grad_norm": 4.534387588500977,
"learning_rate": 1.1151197749975754e-05,
"loss": 0.029108104705810548,
"step": 5750
},
{
"epoch": 0.11250775915580385,
"grad_norm": 3.5855824947357178,
"learning_rate": 1.1248181553680536e-05,
"loss": 0.025364418029785157,
"step": 5800
},
{
"epoch": 0.11347765363128491,
"grad_norm": 0.6006166934967041,
"learning_rate": 1.1345165357385317e-05,
"loss": 0.032281730175018314,
"step": 5850
},
{
"epoch": 0.11444754810676598,
"grad_norm": 0.09923648834228516,
"learning_rate": 1.1442149161090099e-05,
"loss": 0.033263790607452395,
"step": 5900
},
{
"epoch": 0.11541744258224705,
"grad_norm": 1.4193023443222046,
"learning_rate": 1.153913296479488e-05,
"loss": 0.034170238971710204,
"step": 5950
},
{
"epoch": 0.11638733705772812,
"grad_norm": 1.0424859523773193,
"learning_rate": 1.1636116768499661e-05,
"loss": 0.032919149398803714,
"step": 6000
},
{
"epoch": 0.11735723153320919,
"grad_norm": 0.7413951754570007,
"learning_rate": 1.1733100572204444e-05,
"loss": 0.020133750438690187,
"step": 6050
},
{
"epoch": 0.11832712600869025,
"grad_norm": 0.46916237473487854,
"learning_rate": 1.1830084375909224e-05,
"loss": 0.03417280435562134,
"step": 6100
},
{
"epoch": 0.11929702048417132,
"grad_norm": 0.17570248246192932,
"learning_rate": 1.1927068179614006e-05,
"loss": 0.032292633056640624,
"step": 6150
},
{
"epoch": 0.12026691495965239,
"grad_norm": 2.131899118423462,
"learning_rate": 1.2024051983318788e-05,
"loss": 0.03043987512588501,
"step": 6200
},
{
"epoch": 0.12123680943513346,
"grad_norm": 0.1402285248041153,
"learning_rate": 1.2121035787023567e-05,
"loss": 0.03259751558303833,
"step": 6250
},
{
"epoch": 0.12220670391061453,
"grad_norm": 0.07876459509134293,
"learning_rate": 1.2218019590728349e-05,
"loss": 0.025948736667633057,
"step": 6300
},
{
"epoch": 0.1231765983860956,
"grad_norm": 3.701195240020752,
"learning_rate": 1.2315003394433131e-05,
"loss": 0.04177140235900879,
"step": 6350
},
{
"epoch": 0.12414649286157665,
"grad_norm": 3.1167237758636475,
"learning_rate": 1.2411987198137912e-05,
"loss": 0.02722917318344116,
"step": 6400
},
{
"epoch": 0.12511638733705774,
"grad_norm": 0.014126094989478588,
"learning_rate": 1.2508971001842694e-05,
"loss": 0.022621097564697264,
"step": 6450
},
{
"epoch": 0.1260862818125388,
"grad_norm": 0.05082406848669052,
"learning_rate": 1.2605954805547476e-05,
"loss": 0.023705952167510987,
"step": 6500
},
{
"epoch": 0.12705617628801985,
"grad_norm": 0.4131111204624176,
"learning_rate": 1.2702938609252256e-05,
"loss": 0.027076613903045655,
"step": 6550
},
{
"epoch": 0.12802607076350092,
"grad_norm": 0.4441278874874115,
"learning_rate": 1.2799922412957038e-05,
"loss": 0.021566200256347656,
"step": 6600
},
{
"epoch": 0.128995965238982,
"grad_norm": 0.2770746648311615,
"learning_rate": 1.2896906216661817e-05,
"loss": 0.024446680545806884,
"step": 6650
},
{
"epoch": 0.12996585971446306,
"grad_norm": 3.7211852073669434,
"learning_rate": 1.29938900203666e-05,
"loss": 0.03098001956939697,
"step": 6700
},
{
"epoch": 0.13093575418994413,
"grad_norm": 3.914680242538452,
"learning_rate": 1.3090873824071381e-05,
"loss": 0.02592698097229004,
"step": 6750
},
{
"epoch": 0.1319056486654252,
"grad_norm": 1.94025456905365,
"learning_rate": 1.3187857627776162e-05,
"loss": 0.025665268898010254,
"step": 6800
},
{
"epoch": 0.13287554314090627,
"grad_norm": 2.094653606414795,
"learning_rate": 1.3284841431480944e-05,
"loss": 0.02358160972595215,
"step": 6850
},
{
"epoch": 0.13384543761638734,
"grad_norm": 1.1335976123809814,
"learning_rate": 1.3381825235185726e-05,
"loss": 0.02783395051956177,
"step": 6900
},
{
"epoch": 0.1348153320918684,
"grad_norm": 0.3770068287849426,
"learning_rate": 1.3478809038890506e-05,
"loss": 0.0290922474861145,
"step": 6950
},
{
"epoch": 0.13578522656734948,
"grad_norm": 1.6042747497558594,
"learning_rate": 1.3575792842595289e-05,
"loss": 0.02653456449508667,
"step": 7000
},
{
"epoch": 0.13675512104283055,
"grad_norm": 3.5152595043182373,
"learning_rate": 1.367277664630007e-05,
"loss": 0.022409119606018067,
"step": 7050
},
{
"epoch": 0.1377250155183116,
"grad_norm": 0.5457113981246948,
"learning_rate": 1.376976045000485e-05,
"loss": 0.016104856729507445,
"step": 7100
},
{
"epoch": 0.13869490999379266,
"grad_norm": 3.495863199234009,
"learning_rate": 1.3866744253709632e-05,
"loss": 0.03198759555816651,
"step": 7150
},
{
"epoch": 0.13966480446927373,
"grad_norm": 2.9386863708496094,
"learning_rate": 1.3963728057414412e-05,
"loss": 0.03470186471939087,
"step": 7200
},
{
"epoch": 0.1406346989447548,
"grad_norm": 1.2066242694854736,
"learning_rate": 1.4060711861119194e-05,
"loss": 0.027127294540405272,
"step": 7250
},
{
"epoch": 0.14160459342023587,
"grad_norm": 5.149142742156982,
"learning_rate": 1.4157695664823976e-05,
"loss": 0.02857618570327759,
"step": 7300
},
{
"epoch": 0.14257448789571694,
"grad_norm": 0.37447404861450195,
"learning_rate": 1.4254679468528757e-05,
"loss": 0.030744616985321046,
"step": 7350
},
{
"epoch": 0.14354438237119801,
"grad_norm": 0.4187411069869995,
"learning_rate": 1.4351663272233539e-05,
"loss": 0.03270892858505249,
"step": 7400
},
{
"epoch": 0.14451427684667909,
"grad_norm": 0.10034716874361038,
"learning_rate": 1.4448647075938321e-05,
"loss": 0.027905380725860594,
"step": 7450
},
{
"epoch": 0.14548417132216016,
"grad_norm": 0.11009234189987183,
"learning_rate": 1.45456308796431e-05,
"loss": 0.02118866443634033,
"step": 7500
},
{
"epoch": 0.14645406579764123,
"grad_norm": 2.6591553688049316,
"learning_rate": 1.4642614683347882e-05,
"loss": 0.03232965230941773,
"step": 7550
},
{
"epoch": 0.1474239602731223,
"grad_norm": 0.44522708654403687,
"learning_rate": 1.4739598487052664e-05,
"loss": 0.027144644260406494,
"step": 7600
},
{
"epoch": 0.14839385474860337,
"grad_norm": 2.197603464126587,
"learning_rate": 1.4836582290757444e-05,
"loss": 0.029556453227996826,
"step": 7650
},
{
"epoch": 0.1493637492240844,
"grad_norm": 2.7628333568573,
"learning_rate": 1.4933566094462226e-05,
"loss": 0.03335941553115845,
"step": 7700
},
{
"epoch": 0.15033364369956548,
"grad_norm": 0.8265947103500366,
"learning_rate": 1.5030549898167009e-05,
"loss": 0.02280752420425415,
"step": 7750
},
{
"epoch": 0.15130353817504655,
"grad_norm": 0.020113505423069,
"learning_rate": 1.5127533701871789e-05,
"loss": 0.0231875205039978,
"step": 7800
},
{
"epoch": 0.15227343265052762,
"grad_norm": 0.6898635029792786,
"learning_rate": 1.5224517505576571e-05,
"loss": 0.024838275909423828,
"step": 7850
},
{
"epoch": 0.1532433271260087,
"grad_norm": 1.7432154417037964,
"learning_rate": 1.532150130928135e-05,
"loss": 0.030983116626739502,
"step": 7900
},
{
"epoch": 0.15421322160148976,
"grad_norm": 5.397827625274658,
"learning_rate": 1.5418485112986132e-05,
"loss": 0.033342487812042236,
"step": 7950
},
{
"epoch": 0.15518311607697083,
"grad_norm": 0.5413951873779297,
"learning_rate": 1.5515468916690914e-05,
"loss": 0.029320619106292724,
"step": 8000
},
{
"epoch": 0.1561530105524519,
"grad_norm": 0.32360729575157166,
"learning_rate": 1.5612452720395693e-05,
"loss": 0.027219028472900392,
"step": 8050
},
{
"epoch": 0.15712290502793297,
"grad_norm": 0.7686426043510437,
"learning_rate": 1.5709436524100475e-05,
"loss": 0.0243973708152771,
"step": 8100
},
{
"epoch": 0.15809279950341404,
"grad_norm": 2.9169368743896484,
"learning_rate": 1.5806420327805257e-05,
"loss": 0.029242515563964844,
"step": 8150
},
{
"epoch": 0.1590626939788951,
"grad_norm": 5.685990333557129,
"learning_rate": 1.590340413151004e-05,
"loss": 0.025703225135803223,
"step": 8200
},
{
"epoch": 0.16003258845437615,
"grad_norm": 0.22357097268104553,
"learning_rate": 1.600038793521482e-05,
"loss": 0.015003883838653564,
"step": 8250
},
{
"epoch": 0.16100248292985722,
"grad_norm": 1.8692313432693481,
"learning_rate": 1.6097371738919603e-05,
"loss": 0.016555553674697875,
"step": 8300
},
{
"epoch": 0.1619723774053383,
"grad_norm": 0.47332775592803955,
"learning_rate": 1.6194355542624382e-05,
"loss": 0.022484793663024902,
"step": 8350
},
{
"epoch": 0.16294227188081936,
"grad_norm": 2.3480517864227295,
"learning_rate": 1.6291339346329164e-05,
"loss": 0.022541608810424805,
"step": 8400
},
{
"epoch": 0.16391216635630043,
"grad_norm": 0.652316689491272,
"learning_rate": 1.6388323150033943e-05,
"loss": 0.026096351146697998,
"step": 8450
},
{
"epoch": 0.1648820608317815,
"grad_norm": 0.15471266210079193,
"learning_rate": 1.6485306953738725e-05,
"loss": 0.02419330596923828,
"step": 8500
},
{
"epoch": 0.16585195530726257,
"grad_norm": 0.15628892183303833,
"learning_rate": 1.6582290757443507e-05,
"loss": 0.02312201976776123,
"step": 8550
},
{
"epoch": 0.16682184978274364,
"grad_norm": 1.5839548110961914,
"learning_rate": 1.667927456114829e-05,
"loss": 0.02558695077896118,
"step": 8600
},
{
"epoch": 0.1677917442582247,
"grad_norm": 2.1578922271728516,
"learning_rate": 1.677625836485307e-05,
"loss": 0.024801111221313475,
"step": 8650
},
{
"epoch": 0.16876163873370578,
"grad_norm": 2.2771260738372803,
"learning_rate": 1.6873242168557854e-05,
"loss": 0.034369521141052246,
"step": 8700
},
{
"epoch": 0.16973153320918685,
"grad_norm": 3.0908563137054443,
"learning_rate": 1.6970225972262632e-05,
"loss": 0.027252516746520995,
"step": 8750
},
{
"epoch": 0.1707014276846679,
"grad_norm": 2.2698142528533936,
"learning_rate": 1.7067209775967414e-05,
"loss": 0.02624335289001465,
"step": 8800
},
{
"epoch": 0.17167132216014896,
"grad_norm": 0.08631814271211624,
"learning_rate": 1.7164193579672196e-05,
"loss": 0.025639703273773195,
"step": 8850
},
{
"epoch": 0.17264121663563003,
"grad_norm": 0.5212802886962891,
"learning_rate": 1.7261177383376975e-05,
"loss": 0.02779716730117798,
"step": 8900
},
{
"epoch": 0.1736111111111111,
"grad_norm": 0.025622159242630005,
"learning_rate": 1.7358161187081757e-05,
"loss": 0.018600006103515625,
"step": 8950
},
{
"epoch": 0.17458100558659218,
"grad_norm": 2.3365044593811035,
"learning_rate": 1.745514499078654e-05,
"loss": 0.031590254306793214,
"step": 9000
},
{
"epoch": 0.17555090006207325,
"grad_norm": 0.44256386160850525,
"learning_rate": 1.755212879449132e-05,
"loss": 0.024406595230102537,
"step": 9050
},
{
"epoch": 0.17652079453755432,
"grad_norm": 3.0731847286224365,
"learning_rate": 1.7649112598196104e-05,
"loss": 0.01981626272201538,
"step": 9100
},
{
"epoch": 0.17749068901303539,
"grad_norm": 2.8878190517425537,
"learning_rate": 1.7746096401900882e-05,
"loss": 0.02611752986907959,
"step": 9150
},
{
"epoch": 0.17846058348851646,
"grad_norm": 0.19337379932403564,
"learning_rate": 1.7843080205605665e-05,
"loss": 0.025720217227935792,
"step": 9200
},
{
"epoch": 0.17943047796399753,
"grad_norm": 0.3339148461818695,
"learning_rate": 1.7940064009310447e-05,
"loss": 0.025402896404266358,
"step": 9250
},
{
"epoch": 0.1804003724394786,
"grad_norm": 2.3808434009552,
"learning_rate": 1.803704781301523e-05,
"loss": 0.017128294706344603,
"step": 9300
},
{
"epoch": 0.18137026691495967,
"grad_norm": 0.07367683202028275,
"learning_rate": 1.813403161672001e-05,
"loss": 0.02833383560180664,
"step": 9350
},
{
"epoch": 0.1823401613904407,
"grad_norm": 1.283496618270874,
"learning_rate": 1.8231015420424793e-05,
"loss": 0.02420175790786743,
"step": 9400
},
{
"epoch": 0.18331005586592178,
"grad_norm": 0.14021866023540497,
"learning_rate": 1.8327999224129572e-05,
"loss": 0.023649415969848632,
"step": 9450
},
{
"epoch": 0.18427995034140285,
"grad_norm": 2.434354782104492,
"learning_rate": 1.8424983027834354e-05,
"loss": 0.026294608116149903,
"step": 9500
},
{
"epoch": 0.18524984481688392,
"grad_norm": 0.020875701680779457,
"learning_rate": 1.8521966831539136e-05,
"loss": 0.026679866313934326,
"step": 9550
},
{
"epoch": 0.186219739292365,
"grad_norm": 0.13525816798210144,
"learning_rate": 1.8618950635243915e-05,
"loss": 0.020075747966766356,
"step": 9600
},
{
"epoch": 0.18718963376784606,
"grad_norm": 0.14159230887889862,
"learning_rate": 1.8715934438948697e-05,
"loss": 0.019264304637908937,
"step": 9650
},
{
"epoch": 0.18815952824332713,
"grad_norm": 2.174453020095825,
"learning_rate": 1.881291824265348e-05,
"loss": 0.026439454555511475,
"step": 9700
},
{
"epoch": 0.1891294227188082,
"grad_norm": 0.7893092036247253,
"learning_rate": 1.890990204635826e-05,
"loss": 0.024656739234924317,
"step": 9750
},
{
"epoch": 0.19009931719428927,
"grad_norm": 1.0362181663513184,
"learning_rate": 1.9006885850063043e-05,
"loss": 0.02303797960281372,
"step": 9800
},
{
"epoch": 0.19106921166977034,
"grad_norm": 4.287092208862305,
"learning_rate": 1.9103869653767822e-05,
"loss": 0.017587192058563232,
"step": 9850
},
{
"epoch": 0.1920391061452514,
"grad_norm": 0.8740363717079163,
"learning_rate": 1.9200853457472604e-05,
"loss": 0.023234093189239503,
"step": 9900
},
{
"epoch": 0.19300900062073245,
"grad_norm": 0.008943901397287846,
"learning_rate": 1.9297837261177386e-05,
"loss": 0.017660714387893676,
"step": 9950
},
{
"epoch": 0.19397889509621352,
"grad_norm": 0.6272763013839722,
"learning_rate": 1.9394821064882165e-05,
"loss": 0.024134364128112793,
"step": 10000
},
{
"epoch": 0.1949487895716946,
"grad_norm": 0.04520373046398163,
"learning_rate": 1.9491804868586947e-05,
"loss": 0.01658909797668457,
"step": 10050
},
{
"epoch": 0.19591868404717566,
"grad_norm": 0.01031877938657999,
"learning_rate": 1.958878867229173e-05,
"loss": 0.021297156810760498,
"step": 10100
},
{
"epoch": 0.19688857852265673,
"grad_norm": 0.3766087293624878,
"learning_rate": 1.968577247599651e-05,
"loss": 0.023214969635009765,
"step": 10150
},
{
"epoch": 0.1978584729981378,
"grad_norm": 0.10460980236530304,
"learning_rate": 1.9782756279701293e-05,
"loss": 0.029245076179504396,
"step": 10200
},
{
"epoch": 0.19882836747361887,
"grad_norm": 0.547511875629425,
"learning_rate": 1.9879740083406075e-05,
"loss": 0.021636698246002197,
"step": 10250
},
{
"epoch": 0.19979826194909994,
"grad_norm": 4.814013957977295,
"learning_rate": 1.9976723887110854e-05,
"loss": 0.020318408012390137,
"step": 10300
},
{
"epoch": 0.200768156424581,
"grad_norm": 0.44551169872283936,
"learning_rate": 1.9991809727026826e-05,
"loss": 0.018024702072143555,
"step": 10350
},
{
"epoch": 0.20173805090006208,
"grad_norm": 0.2480621486902237,
"learning_rate": 1.998103305206212e-05,
"loss": 0.022806077003479003,
"step": 10400
},
{
"epoch": 0.20270794537554315,
"grad_norm": 1.2251558303833008,
"learning_rate": 1.9970256377097412e-05,
"loss": 0.02145651340484619,
"step": 10450
},
{
"epoch": 0.2036778398510242,
"grad_norm": 2.56076717376709,
"learning_rate": 1.9959479702132705e-05,
"loss": 0.02443584203720093,
"step": 10500
},
{
"epoch": 0.20464773432650527,
"grad_norm": 0.8434218764305115,
"learning_rate": 1.9948703027167998e-05,
"loss": 0.026762964725494384,
"step": 10550
},
{
"epoch": 0.20561762880198634,
"grad_norm": 0.905733048915863,
"learning_rate": 1.993792635220329e-05,
"loss": 0.027904486656188963,
"step": 10600
},
{
"epoch": 0.2065875232774674,
"grad_norm": 1.2831109762191772,
"learning_rate": 1.9927149677238588e-05,
"loss": 0.02265932321548462,
"step": 10650
},
{
"epoch": 0.20755741775294848,
"grad_norm": 0.9724456667900085,
"learning_rate": 1.991637300227388e-05,
"loss": 0.027907648086547852,
"step": 10700
},
{
"epoch": 0.20852731222842955,
"grad_norm": 1.096745491027832,
"learning_rate": 1.9905596327309174e-05,
"loss": 0.01947831153869629,
"step": 10750
},
{
"epoch": 0.20949720670391062,
"grad_norm": 0.32208794355392456,
"learning_rate": 1.9894819652344467e-05,
"loss": 0.02368006229400635,
"step": 10800
},
{
"epoch": 0.21046710117939169,
"grad_norm": 3.681391954421997,
"learning_rate": 1.988404297737976e-05,
"loss": 0.02219386100769043,
"step": 10850
},
{
"epoch": 0.21143699565487276,
"grad_norm": 0.14404571056365967,
"learning_rate": 1.9873266302415054e-05,
"loss": 0.017421839237213136,
"step": 10900
},
{
"epoch": 0.21240689013035383,
"grad_norm": 0.6221453547477722,
"learning_rate": 1.9862489627450347e-05,
"loss": 0.013101431131362916,
"step": 10950
},
{
"epoch": 0.2133767846058349,
"grad_norm": 0.09375086426734924,
"learning_rate": 1.9851712952485643e-05,
"loss": 0.041983208656311034,
"step": 11000
},
{
"epoch": 0.21434667908131597,
"grad_norm": 4.174381256103516,
"learning_rate": 1.9840936277520933e-05,
"loss": 0.02042308568954468,
"step": 11050
},
{
"epoch": 0.215316573556797,
"grad_norm": 0.5037624835968018,
"learning_rate": 1.983015960255623e-05,
"loss": 0.018624961376190186,
"step": 11100
},
{
"epoch": 0.21628646803227808,
"grad_norm": 3.348021984100342,
"learning_rate": 1.9819382927591523e-05,
"loss": 0.03587321519851685,
"step": 11150
},
{
"epoch": 0.21725636250775915,
"grad_norm": 1.0664407014846802,
"learning_rate": 1.9808606252626816e-05,
"loss": 0.027515835762023926,
"step": 11200
},
{
"epoch": 0.21822625698324022,
"grad_norm": 0.25070491433143616,
"learning_rate": 1.979782957766211e-05,
"loss": 0.027012321949005127,
"step": 11250
},
{
"epoch": 0.2191961514587213,
"grad_norm": 0.36331242322921753,
"learning_rate": 1.9787052902697402e-05,
"loss": 0.03157989263534546,
"step": 11300
},
{
"epoch": 0.22016604593420236,
"grad_norm": 0.09230268746614456,
"learning_rate": 1.97762762277327e-05,
"loss": 0.02538405418395996,
"step": 11350
},
{
"epoch": 0.22113594040968343,
"grad_norm": 0.10893985629081726,
"learning_rate": 1.9765499552767992e-05,
"loss": 0.02563972234725952,
"step": 11400
},
{
"epoch": 0.2221058348851645,
"grad_norm": 0.023250911384820938,
"learning_rate": 1.9754722877803285e-05,
"loss": 0.02095996618270874,
"step": 11450
},
{
"epoch": 0.22307572936064557,
"grad_norm": 0.5929452180862427,
"learning_rate": 1.974394620283858e-05,
"loss": 0.01749273419380188,
"step": 11500
},
{
"epoch": 0.22404562383612664,
"grad_norm": 0.32036545872688293,
"learning_rate": 1.973316952787387e-05,
"loss": 0.025547654628753663,
"step": 11550
},
{
"epoch": 0.2250155183116077,
"grad_norm": 0.027903523296117783,
"learning_rate": 1.9722392852909165e-05,
"loss": 0.01829303503036499,
"step": 11600
},
{
"epoch": 0.22598541278708875,
"grad_norm": 0.33727318048477173,
"learning_rate": 1.9711616177944458e-05,
"loss": 0.018667610883712767,
"step": 11650
},
{
"epoch": 0.22695530726256982,
"grad_norm": 0.23035697638988495,
"learning_rate": 1.970083950297975e-05,
"loss": 0.024986166954040528,
"step": 11700
},
{
"epoch": 0.2279252017380509,
"grad_norm": 3.0287418365478516,
"learning_rate": 1.9690062828015048e-05,
"loss": 0.01976940631866455,
"step": 11750
},
{
"epoch": 0.22889509621353196,
"grad_norm": 0.32429778575897217,
"learning_rate": 1.9679286153050337e-05,
"loss": 0.022787065505981446,
"step": 11800
},
{
"epoch": 0.22986499068901303,
"grad_norm": 0.9836875796318054,
"learning_rate": 1.9668509478085634e-05,
"loss": 0.022979652881622313,
"step": 11850
},
{
"epoch": 0.2308348851644941,
"grad_norm": 3.3939058780670166,
"learning_rate": 1.9657732803120927e-05,
"loss": 0.027675604820251463,
"step": 11900
},
{
"epoch": 0.23180477963997517,
"grad_norm": 0.24185506999492645,
"learning_rate": 1.964695612815622e-05,
"loss": 0.022160177230834962,
"step": 11950
},
{
"epoch": 0.23277467411545624,
"grad_norm": 1.4770416021347046,
"learning_rate": 1.9636179453191513e-05,
"loss": 0.02430246114730835,
"step": 12000
},
{
"epoch": 0.2337445685909373,
"grad_norm": 0.16257409751415253,
"learning_rate": 1.9625402778226806e-05,
"loss": 0.02405242681503296,
"step": 12050
},
{
"epoch": 0.23471446306641838,
"grad_norm": 0.26366299390792847,
"learning_rate": 1.9614626103262103e-05,
"loss": 0.021252565383911133,
"step": 12100
},
{
"epoch": 0.23568435754189945,
"grad_norm": 0.3956337869167328,
"learning_rate": 1.9603849428297396e-05,
"loss": 0.012004565000534057,
"step": 12150
},
{
"epoch": 0.2366542520173805,
"grad_norm": 0.26303809881210327,
"learning_rate": 1.959307275333269e-05,
"loss": 0.02913022756576538,
"step": 12200
},
{
"epoch": 0.23762414649286157,
"grad_norm": 0.4916934669017792,
"learning_rate": 1.9582296078367983e-05,
"loss": 0.020663721561431883,
"step": 12250
},
{
"epoch": 0.23859404096834264,
"grad_norm": 0.2867946922779083,
"learning_rate": 1.9571519403403276e-05,
"loss": 0.026198358535766603,
"step": 12300
},
{
"epoch": 0.2395639354438237,
"grad_norm": 0.3280359208583832,
"learning_rate": 1.956074272843857e-05,
"loss": 0.023178770542144775,
"step": 12350
},
{
"epoch": 0.24053382991930478,
"grad_norm": 0.15737749636173248,
"learning_rate": 1.9549966053473862e-05,
"loss": 0.01725036382675171,
"step": 12400
},
{
"epoch": 0.24150372439478585,
"grad_norm": 0.1503092646598816,
"learning_rate": 1.9539189378509155e-05,
"loss": 0.03060781478881836,
"step": 12450
},
{
"epoch": 0.24247361887026692,
"grad_norm": 0.5908766984939575,
"learning_rate": 1.9528412703544452e-05,
"loss": 0.012551853656768799,
"step": 12500
},
{
"epoch": 0.24344351334574799,
"grad_norm": 1.315533995628357,
"learning_rate": 1.951763602857974e-05,
"loss": 0.032272634506225584,
"step": 12550
},
{
"epoch": 0.24441340782122906,
"grad_norm": 1.0292696952819824,
"learning_rate": 1.9506859353615038e-05,
"loss": 0.019478362798690797,
"step": 12600
},
{
"epoch": 0.24538330229671013,
"grad_norm": 1.0259524583816528,
"learning_rate": 1.949608267865033e-05,
"loss": 0.015265188217163085,
"step": 12650
},
{
"epoch": 0.2463531967721912,
"grad_norm": 0.45366767048835754,
"learning_rate": 1.9485306003685624e-05,
"loss": 0.0281591534614563,
"step": 12700
},
{
"epoch": 0.24732309124767227,
"grad_norm": 0.01454345602542162,
"learning_rate": 1.9474529328720918e-05,
"loss": 0.025230038166046142,
"step": 12750
},
{
"epoch": 0.2482929857231533,
"grad_norm": 0.04741954058408737,
"learning_rate": 1.946375265375621e-05,
"loss": 0.02140932559967041,
"step": 12800
},
{
"epoch": 0.24926288019863438,
"grad_norm": 1.1716814041137695,
"learning_rate": 1.9452975978791507e-05,
"loss": 0.023408467769622802,
"step": 12850
},
{
"epoch": 0.2502327746741155,
"grad_norm": 0.11229647696018219,
"learning_rate": 1.9442199303826797e-05,
"loss": 0.021542866230010987,
"step": 12900
},
{
"epoch": 0.25120266914959655,
"grad_norm": 2.3136942386627197,
"learning_rate": 1.9431422628862094e-05,
"loss": 0.01724626064300537,
"step": 12950
},
{
"epoch": 0.2521725636250776,
"grad_norm": 0.7710976004600525,
"learning_rate": 1.9420645953897387e-05,
"loss": 0.019175268411636352,
"step": 13000
},
{
"epoch": 0.25314245810055863,
"grad_norm": 0.05378899350762367,
"learning_rate": 1.940986927893268e-05,
"loss": 0.01597250461578369,
"step": 13050
},
{
"epoch": 0.2541123525760397,
"grad_norm": 0.36054426431655884,
"learning_rate": 1.9399092603967973e-05,
"loss": 0.0254284930229187,
"step": 13100
},
{
"epoch": 0.25508224705152077,
"grad_norm": 0.0840875431895256,
"learning_rate": 1.9388315929003266e-05,
"loss": 0.016272796392440794,
"step": 13150
},
{
"epoch": 0.25605214152700184,
"grad_norm": 0.7416884303092957,
"learning_rate": 1.937753925403856e-05,
"loss": 0.015000748634338378,
"step": 13200
},
{
"epoch": 0.2570220360024829,
"grad_norm": 0.5830755829811096,
"learning_rate": 1.9366762579073856e-05,
"loss": 0.023126914501190185,
"step": 13250
},
{
"epoch": 0.257991930477964,
"grad_norm": 1.6759475469589233,
"learning_rate": 1.9355985904109146e-05,
"loss": 0.02584106683731079,
"step": 13300
},
{
"epoch": 0.25896182495344505,
"grad_norm": 1.9239856004714966,
"learning_rate": 1.9345209229144442e-05,
"loss": 0.023104896545410158,
"step": 13350
},
{
"epoch": 0.2599317194289261,
"grad_norm": 0.20403368771076202,
"learning_rate": 1.9334432554179735e-05,
"loss": 0.01455102801322937,
"step": 13400
},
{
"epoch": 0.2609016139044072,
"grad_norm": 1.2329163551330566,
"learning_rate": 1.932365587921503e-05,
"loss": 0.020922226905822752,
"step": 13450
},
{
"epoch": 0.26187150837988826,
"grad_norm": 0.1310439109802246,
"learning_rate": 1.931287920425032e-05,
"loss": 0.016660687923431398,
"step": 13500
},
{
"epoch": 0.26284140285536933,
"grad_norm": 0.07179796695709229,
"learning_rate": 1.9302102529285615e-05,
"loss": 0.021798677444458008,
"step": 13550
},
{
"epoch": 0.2638112973308504,
"grad_norm": 0.23347480595111847,
"learning_rate": 1.929132585432091e-05,
"loss": 0.016233474016189575,
"step": 13600
},
{
"epoch": 0.2647811918063315,
"grad_norm": 0.4317619502544403,
"learning_rate": 1.92805491793562e-05,
"loss": 0.02409332036972046,
"step": 13650
},
{
"epoch": 0.26575108628181254,
"grad_norm": 0.1272326558828354,
"learning_rate": 1.9269772504391498e-05,
"loss": 0.020026631355285644,
"step": 13700
},
{
"epoch": 0.2667209807572936,
"grad_norm": 1.5374191999435425,
"learning_rate": 1.925899582942679e-05,
"loss": 0.025140745639801024,
"step": 13750
},
{
"epoch": 0.2676908752327747,
"grad_norm": 0.95331209897995,
"learning_rate": 1.9248219154462084e-05,
"loss": 0.01790245294570923,
"step": 13800
},
{
"epoch": 0.26866076970825575,
"grad_norm": 1.4951375722885132,
"learning_rate": 1.9237442479497377e-05,
"loss": 0.018575254678726196,
"step": 13850
},
{
"epoch": 0.2696306641837368,
"grad_norm": 0.6143473982810974,
"learning_rate": 1.922666580453267e-05,
"loss": 0.021291687488555908,
"step": 13900
},
{
"epoch": 0.2706005586592179,
"grad_norm": 0.02985381707549095,
"learning_rate": 1.9215889129567964e-05,
"loss": 0.023942198753356934,
"step": 13950
},
{
"epoch": 0.27157045313469896,
"grad_norm": 0.027741987258195877,
"learning_rate": 1.920511245460326e-05,
"loss": 0.015487182140350341,
"step": 14000
},
{
"epoch": 0.27254034761018003,
"grad_norm": 0.10407145321369171,
"learning_rate": 1.919433577963855e-05,
"loss": 0.014503239393234254,
"step": 14050
},
{
"epoch": 0.2735102420856611,
"grad_norm": 0.023095162585377693,
"learning_rate": 1.9183559104673846e-05,
"loss": 0.014814221858978271,
"step": 14100
},
{
"epoch": 0.2744801365611422,
"grad_norm": 0.4851715862751007,
"learning_rate": 1.917278242970914e-05,
"loss": 0.019852250814437866,
"step": 14150
},
{
"epoch": 0.2754500310366232,
"grad_norm": 0.27584078907966614,
"learning_rate": 1.9162005754744433e-05,
"loss": 0.020890560150146485,
"step": 14200
},
{
"epoch": 0.27641992551210426,
"grad_norm": 0.7808701992034912,
"learning_rate": 1.9151229079779726e-05,
"loss": 0.018995105028152465,
"step": 14250
},
{
"epoch": 0.27738981998758533,
"grad_norm": 3.7738544940948486,
"learning_rate": 1.914045240481502e-05,
"loss": 0.02296557426452637,
"step": 14300
},
{
"epoch": 0.2783597144630664,
"grad_norm": 0.4710941016674042,
"learning_rate": 1.9129675729850316e-05,
"loss": 0.019027496576309203,
"step": 14350
},
{
"epoch": 0.27932960893854747,
"grad_norm": 1.0808541774749756,
"learning_rate": 1.9118899054885605e-05,
"loss": 0.02016347408294678,
"step": 14400
},
{
"epoch": 0.28029950341402854,
"grad_norm": 1.8925045728683472,
"learning_rate": 1.9108122379920902e-05,
"loss": 0.022477951049804688,
"step": 14450
},
{
"epoch": 0.2812693978895096,
"grad_norm": 5.328800678253174,
"learning_rate": 1.9097345704956195e-05,
"loss": 0.020785722732543945,
"step": 14500
},
{
"epoch": 0.2822392923649907,
"grad_norm": 0.2927640974521637,
"learning_rate": 1.9086569029991488e-05,
"loss": 0.017422797679901122,
"step": 14550
},
{
"epoch": 0.28320918684047175,
"grad_norm": 0.3091449737548828,
"learning_rate": 1.907579235502678e-05,
"loss": 0.01894198179244995,
"step": 14600
},
{
"epoch": 0.2841790813159528,
"grad_norm": 0.052439186722040176,
"learning_rate": 1.9065015680062075e-05,
"loss": 0.013963347673416138,
"step": 14650
},
{
"epoch": 0.2851489757914339,
"grad_norm": 0.23640276491641998,
"learning_rate": 1.9054239005097368e-05,
"loss": 0.01742790937423706,
"step": 14700
},
{
"epoch": 0.28611887026691496,
"grad_norm": 0.2411554753780365,
"learning_rate": 1.9043462330132664e-05,
"loss": 0.02314033269882202,
"step": 14750
},
{
"epoch": 0.28708876474239603,
"grad_norm": 0.8846039772033691,
"learning_rate": 1.9032685655167954e-05,
"loss": 0.019311647415161132,
"step": 14800
},
{
"epoch": 0.2880586592178771,
"grad_norm": 0.7877144813537598,
"learning_rate": 1.902190898020325e-05,
"loss": 0.01981956124305725,
"step": 14850
},
{
"epoch": 0.28902855369335817,
"grad_norm": 0.0969516783952713,
"learning_rate": 1.9011132305238544e-05,
"loss": 0.027691679000854494,
"step": 14900
},
{
"epoch": 0.28999844816883924,
"grad_norm": 0.04294842481613159,
"learning_rate": 1.9000355630273837e-05,
"loss": 0.019661370515823364,
"step": 14950
},
{
"epoch": 0.2909683426443203,
"grad_norm": 0.19237542152404785,
"learning_rate": 1.898957895530913e-05,
"loss": 0.013877005577087402,
"step": 15000
},
{
"epoch": 0.2919382371198014,
"grad_norm": 3.7419471740722656,
"learning_rate": 1.8978802280344423e-05,
"loss": 0.017435097694396974,
"step": 15050
},
{
"epoch": 0.29290813159528245,
"grad_norm": 0.4388321340084076,
"learning_rate": 1.896802560537972e-05,
"loss": 0.02793212413787842,
"step": 15100
},
{
"epoch": 0.2938780260707635,
"grad_norm": 1.2302663326263428,
"learning_rate": 1.895724893041501e-05,
"loss": 0.016931357383728026,
"step": 15150
},
{
"epoch": 0.2948479205462446,
"grad_norm": 0.5854755640029907,
"learning_rate": 1.8946472255450306e-05,
"loss": 0.013195137977600097,
"step": 15200
},
{
"epoch": 0.29581781502172566,
"grad_norm": 1.1579340696334839,
"learning_rate": 1.89356955804856e-05,
"loss": 0.015953311920166014,
"step": 15250
},
{
"epoch": 0.29678770949720673,
"grad_norm": 2.215506076812744,
"learning_rate": 1.8924918905520892e-05,
"loss": 0.022483954429626463,
"step": 15300
},
{
"epoch": 0.29775760397268775,
"grad_norm": 0.6977174282073975,
"learning_rate": 1.8914142230556186e-05,
"loss": 0.019263370037078856,
"step": 15350
},
{
"epoch": 0.2987274984481688,
"grad_norm": 0.839987576007843,
"learning_rate": 1.890336555559148e-05,
"loss": 0.01937116861343384,
"step": 15400
},
{
"epoch": 0.2996973929236499,
"grad_norm": 3.004866123199463,
"learning_rate": 1.8892588880626772e-05,
"loss": 0.016002275943756104,
"step": 15450
},
{
"epoch": 0.30066728739913096,
"grad_norm": 0.3348938822746277,
"learning_rate": 1.888181220566207e-05,
"loss": 0.019188443422317503,
"step": 15500
},
{
"epoch": 0.301637181874612,
"grad_norm": 0.038557019084692,
"learning_rate": 1.8871035530697358e-05,
"loss": 0.024613924026489258,
"step": 15550
},
{
"epoch": 0.3026070763500931,
"grad_norm": 0.04510299116373062,
"learning_rate": 1.8860258855732655e-05,
"loss": 0.02006476402282715,
"step": 15600
},
{
"epoch": 0.30357697082557417,
"grad_norm": 0.10527459532022476,
"learning_rate": 1.8849482180767948e-05,
"loss": 0.015244543552398682,
"step": 15650
},
{
"epoch": 0.30454686530105524,
"grad_norm": 0.10243597626686096,
"learning_rate": 1.883870550580324e-05,
"loss": 0.021969802379608154,
"step": 15700
},
{
"epoch": 0.3055167597765363,
"grad_norm": 0.03891425207257271,
"learning_rate": 1.8827928830838534e-05,
"loss": 0.024867937564849854,
"step": 15750
},
{
"epoch": 0.3064866542520174,
"grad_norm": 0.9363669753074646,
"learning_rate": 1.8817152155873827e-05,
"loss": 0.016571789979934692,
"step": 15800
},
{
"epoch": 0.30745654872749845,
"grad_norm": 2.1093783378601074,
"learning_rate": 1.8806375480909124e-05,
"loss": 0.016006827354431152,
"step": 15850
},
{
"epoch": 0.3084264432029795,
"grad_norm": 0.48025721311569214,
"learning_rate": 1.8795598805944414e-05,
"loss": 0.013361433744430542,
"step": 15900
},
{
"epoch": 0.3093963376784606,
"grad_norm": 0.13708601891994476,
"learning_rate": 1.878482213097971e-05,
"loss": 0.022574400901794432,
"step": 15950
},
{
"epoch": 0.31036623215394166,
"grad_norm": 2.7387800216674805,
"learning_rate": 1.8774045456015003e-05,
"loss": 0.01731914758682251,
"step": 16000
},
{
"epoch": 0.3113361266294227,
"grad_norm": 0.6595755815505981,
"learning_rate": 1.8763268781050297e-05,
"loss": 0.01859807252883911,
"step": 16050
},
{
"epoch": 0.3123060211049038,
"grad_norm": 0.014425593428313732,
"learning_rate": 1.875249210608559e-05,
"loss": 0.01530753493309021,
"step": 16100
},
{
"epoch": 0.31327591558038487,
"grad_norm": 1.7439346313476562,
"learning_rate": 1.8741715431120883e-05,
"loss": 0.020837481021881103,
"step": 16150
},
{
"epoch": 0.31424581005586594,
"grad_norm": 0.39964616298675537,
"learning_rate": 1.8730938756156176e-05,
"loss": 0.01951762557029724,
"step": 16200
},
{
"epoch": 0.315215704531347,
"grad_norm": 0.0767185389995575,
"learning_rate": 1.872016208119147e-05,
"loss": 0.015181030035018922,
"step": 16250
},
{
"epoch": 0.3161855990068281,
"grad_norm": 3.489952325820923,
"learning_rate": 1.8709385406226762e-05,
"loss": 0.023023810386657715,
"step": 16300
},
{
"epoch": 0.31715549348230915,
"grad_norm": 0.07285687327384949,
"learning_rate": 1.869860873126206e-05,
"loss": 0.014600449800491333,
"step": 16350
},
{
"epoch": 0.3181253879577902,
"grad_norm": 2.945131540298462,
"learning_rate": 1.8687832056297352e-05,
"loss": 0.016465651988983154,
"step": 16400
},
{
"epoch": 0.3190952824332713,
"grad_norm": 0.14832748472690582,
"learning_rate": 1.8677055381332645e-05,
"loss": 0.02278857946395874,
"step": 16450
},
{
"epoch": 0.3200651769087523,
"grad_norm": 0.034537095576524734,
"learning_rate": 1.866627870636794e-05,
"loss": 0.01592418909072876,
"step": 16500
},
{
"epoch": 0.3210350713842334,
"grad_norm": 2.709415912628174,
"learning_rate": 1.865550203140323e-05,
"loss": 0.023964059352874757,
"step": 16550
},
{
"epoch": 0.32200496585971444,
"grad_norm": 0.0847368910908699,
"learning_rate": 1.8644725356438528e-05,
"loss": 0.01850275754928589,
"step": 16600
},
{
"epoch": 0.3229748603351955,
"grad_norm": 0.02308146096765995,
"learning_rate": 1.8633948681473818e-05,
"loss": 0.01678106427192688,
"step": 16650
},
{
"epoch": 0.3239447548106766,
"grad_norm": 1.0556334257125854,
"learning_rate": 1.8623172006509114e-05,
"loss": 0.013808093070983886,
"step": 16700
},
{
"epoch": 0.32491464928615765,
"grad_norm": 0.22287754714488983,
"learning_rate": 1.8612395331544408e-05,
"loss": 0.022042360305786133,
"step": 16750
},
{
"epoch": 0.3258845437616387,
"grad_norm": 0.08053021878004074,
"learning_rate": 1.86016186565797e-05,
"loss": 0.016601924896240235,
"step": 16800
},
{
"epoch": 0.3268544382371198,
"grad_norm": 0.4292014539241791,
"learning_rate": 1.8590841981614994e-05,
"loss": 0.01644389271736145,
"step": 16850
},
{
"epoch": 0.32782433271260086,
"grad_norm": 4.123815059661865,
"learning_rate": 1.8580065306650287e-05,
"loss": 0.019462852478027343,
"step": 16900
},
{
"epoch": 0.32879422718808193,
"grad_norm": 0.13867676258087158,
"learning_rate": 1.856928863168558e-05,
"loss": 0.01476535677909851,
"step": 16950
},
{
"epoch": 0.329764121663563,
"grad_norm": 0.10332488268613815,
"learning_rate": 1.8558511956720873e-05,
"loss": 0.017257285118103028,
"step": 17000
},
{
"epoch": 0.3307340161390441,
"grad_norm": 0.1944606900215149,
"learning_rate": 1.854773528175617e-05,
"loss": 0.01651743769645691,
"step": 17050
},
{
"epoch": 0.33170391061452514,
"grad_norm": 0.9710404872894287,
"learning_rate": 1.8536958606791463e-05,
"loss": 0.018955093622207642,
"step": 17100
},
{
"epoch": 0.3326738050900062,
"grad_norm": 1.1976507902145386,
"learning_rate": 1.8526181931826756e-05,
"loss": 0.02301367998123169,
"step": 17150
},
{
"epoch": 0.3336436995654873,
"grad_norm": 2.7661259174346924,
"learning_rate": 1.851540525686205e-05,
"loss": 0.018362454175949096,
"step": 17200
},
{
"epoch": 0.33461359404096835,
"grad_norm": 0.08934645354747772,
"learning_rate": 1.8504628581897343e-05,
"loss": 0.016374958753585814,
"step": 17250
},
{
"epoch": 0.3355834885164494,
"grad_norm": 0.18789377808570862,
"learning_rate": 1.8493851906932636e-05,
"loss": 0.01960576295852661,
"step": 17300
},
{
"epoch": 0.3365533829919305,
"grad_norm": 0.0858871266245842,
"learning_rate": 1.8483075231967932e-05,
"loss": 0.017662898302078248,
"step": 17350
},
{
"epoch": 0.33752327746741156,
"grad_norm": 6.563319683074951,
"learning_rate": 1.8472298557003222e-05,
"loss": 0.014416933059692383,
"step": 17400
},
{
"epoch": 0.33849317194289263,
"grad_norm": 0.8837745189666748,
"learning_rate": 1.846152188203852e-05,
"loss": 0.02055607795715332,
"step": 17450
},
{
"epoch": 0.3394630664183737,
"grad_norm": 0.5295986533164978,
"learning_rate": 1.8450745207073812e-05,
"loss": 0.015508122444152832,
"step": 17500
},
{
"epoch": 0.3404329608938548,
"grad_norm": 0.06741024553775787,
"learning_rate": 1.8439968532109105e-05,
"loss": 0.01913496732711792,
"step": 17550
},
{
"epoch": 0.3414028553693358,
"grad_norm": 0.7210333943367004,
"learning_rate": 1.8429191857144398e-05,
"loss": 0.018211437463760374,
"step": 17600
},
{
"epoch": 0.34237274984481686,
"grad_norm": 0.16165460646152496,
"learning_rate": 1.841841518217969e-05,
"loss": 0.012699922323226929,
"step": 17650
},
{
"epoch": 0.34334264432029793,
"grad_norm": 0.06477090716362,
"learning_rate": 1.8407638507214984e-05,
"loss": 0.018860408067703248,
"step": 17700
},
{
"epoch": 0.344312538795779,
"grad_norm": 0.20910954475402832,
"learning_rate": 1.8396861832250278e-05,
"loss": 0.018482561111450194,
"step": 17750
},
{
"epoch": 0.34528243327126007,
"grad_norm": 2.751336097717285,
"learning_rate": 1.8386085157285574e-05,
"loss": 0.016455549001693725,
"step": 17800
},
{
"epoch": 0.34625232774674114,
"grad_norm": 0.2761375904083252,
"learning_rate": 1.8375308482320867e-05,
"loss": 0.01509534239768982,
"step": 17850
},
{
"epoch": 0.3472222222222222,
"grad_norm": 0.08134075254201889,
"learning_rate": 1.836453180735616e-05,
"loss": 0.018404864072799683,
"step": 17900
},
{
"epoch": 0.3481921166977033,
"grad_norm": 0.17783550918102264,
"learning_rate": 1.8353755132391454e-05,
"loss": 0.02004150867462158,
"step": 17950
},
{
"epoch": 0.34916201117318435,
"grad_norm": 3.4772727489471436,
"learning_rate": 1.8342978457426747e-05,
"loss": 0.012522979974746704,
"step": 18000
},
{
"epoch": 0.3501319056486654,
"grad_norm": 0.0877423882484436,
"learning_rate": 1.833220178246204e-05,
"loss": 0.014303599596023559,
"step": 18050
},
{
"epoch": 0.3511018001241465,
"grad_norm": 0.09105542302131653,
"learning_rate": 1.8321425107497336e-05,
"loss": 0.02131143808364868,
"step": 18100
},
{
"epoch": 0.35207169459962756,
"grad_norm": 0.9323895573616028,
"learning_rate": 1.8310648432532626e-05,
"loss": 0.014694654941558838,
"step": 18150
},
{
"epoch": 0.35304158907510863,
"grad_norm": 0.12636898458003998,
"learning_rate": 1.8299871757567923e-05,
"loss": 0.012409236431121826,
"step": 18200
},
{
"epoch": 0.3540114835505897,
"grad_norm": 0.01108152698725462,
"learning_rate": 1.8289095082603216e-05,
"loss": 0.009763199090957641,
"step": 18250
},
{
"epoch": 0.35498137802607077,
"grad_norm": 0.1951681524515152,
"learning_rate": 1.827831840763851e-05,
"loss": 0.012765527963638305,
"step": 18300
},
{
"epoch": 0.35595127250155184,
"grad_norm": 0.03488857299089432,
"learning_rate": 1.8267541732673802e-05,
"loss": 0.013601378202438355,
"step": 18350
},
{
"epoch": 0.3569211669770329,
"grad_norm": 0.27326229214668274,
"learning_rate": 1.8256765057709095e-05,
"loss": 0.021431758403778076,
"step": 18400
},
{
"epoch": 0.357891061452514,
"grad_norm": 0.16669131815433502,
"learning_rate": 1.824598838274439e-05,
"loss": 0.012068156003952026,
"step": 18450
},
{
"epoch": 0.35886095592799505,
"grad_norm": 0.017041217535734177,
"learning_rate": 1.823521170777968e-05,
"loss": 0.02024803400039673,
"step": 18500
},
{
"epoch": 0.3598308504034761,
"grad_norm": 0.06527240574359894,
"learning_rate": 1.8224435032814978e-05,
"loss": 0.02218601942062378,
"step": 18550
},
{
"epoch": 0.3608007448789572,
"grad_norm": 0.01943429559469223,
"learning_rate": 1.821365835785027e-05,
"loss": 0.018823349475860597,
"step": 18600
},
{
"epoch": 0.36177063935443826,
"grad_norm": 0.8811367750167847,
"learning_rate": 1.8202881682885565e-05,
"loss": 0.01956810235977173,
"step": 18650
},
{
"epoch": 0.36274053382991933,
"grad_norm": 0.26068174839019775,
"learning_rate": 1.8192105007920858e-05,
"loss": 0.018672446012496947,
"step": 18700
},
{
"epoch": 0.36371042830540035,
"grad_norm": 0.11578455567359924,
"learning_rate": 1.818132833295615e-05,
"loss": 0.02108144521713257,
"step": 18750
},
{
"epoch": 0.3646803227808814,
"grad_norm": 0.3520309627056122,
"learning_rate": 1.8170551657991444e-05,
"loss": 0.019701037406921387,
"step": 18800
},
{
"epoch": 0.3656502172563625,
"grad_norm": 0.35474830865859985,
"learning_rate": 1.815977498302674e-05,
"loss": 0.020299012660980224,
"step": 18850
},
{
"epoch": 0.36662011173184356,
"grad_norm": 0.02560902200639248,
"learning_rate": 1.814899830806203e-05,
"loss": 0.015711992979049683,
"step": 18900
},
{
"epoch": 0.3675900062073246,
"grad_norm": 1.5767252445220947,
"learning_rate": 1.8138221633097327e-05,
"loss": 0.022156121730804442,
"step": 18950
},
{
"epoch": 0.3685599006828057,
"grad_norm": 0.2603257894515991,
"learning_rate": 1.812744495813262e-05,
"loss": 0.014314546585083007,
"step": 19000
},
{
"epoch": 0.36952979515828677,
"grad_norm": 0.017517194151878357,
"learning_rate": 1.8116668283167913e-05,
"loss": 0.01578901171684265,
"step": 19050
},
{
"epoch": 0.37049968963376784,
"grad_norm": 0.25757622718811035,
"learning_rate": 1.8105891608203206e-05,
"loss": 0.01114324688911438,
"step": 19100
},
{
"epoch": 0.3714695841092489,
"grad_norm": 4.805666923522949,
"learning_rate": 1.80951149332385e-05,
"loss": 0.01052593469619751,
"step": 19150
},
{
"epoch": 0.37243947858473,
"grad_norm": 0.5105089545249939,
"learning_rate": 1.8084338258273796e-05,
"loss": 0.019789135456085204,
"step": 19200
},
{
"epoch": 0.37340937306021105,
"grad_norm": 2.920492649078369,
"learning_rate": 1.8073561583309086e-05,
"loss": 0.0211655068397522,
"step": 19250
},
{
"epoch": 0.3743792675356921,
"grad_norm": 0.09597882628440857,
"learning_rate": 1.8062784908344382e-05,
"loss": 0.019165738821029663,
"step": 19300
},
{
"epoch": 0.3753491620111732,
"grad_norm": 0.022123487666249275,
"learning_rate": 1.8052008233379676e-05,
"loss": 0.01578805685043335,
"step": 19350
},
{
"epoch": 0.37631905648665426,
"grad_norm": 0.14135892689228058,
"learning_rate": 1.804123155841497e-05,
"loss": 0.021332933902740478,
"step": 19400
},
{
"epoch": 0.3772889509621353,
"grad_norm": 0.3183988332748413,
"learning_rate": 1.8030454883450262e-05,
"loss": 0.020126922130584715,
"step": 19450
},
{
"epoch": 0.3782588454376164,
"grad_norm": 0.4053044319152832,
"learning_rate": 1.8019678208485555e-05,
"loss": 0.01860170841217041,
"step": 19500
},
{
"epoch": 0.37922873991309747,
"grad_norm": 1.1251856088638306,
"learning_rate": 1.8008901533520848e-05,
"loss": 0.013016501665115357,
"step": 19550
},
{
"epoch": 0.38019863438857854,
"grad_norm": 6.035063743591309,
"learning_rate": 1.799812485855614e-05,
"loss": 0.012873082160949708,
"step": 19600
},
{
"epoch": 0.3811685288640596,
"grad_norm": 0.3170987665653229,
"learning_rate": 1.7987348183591435e-05,
"loss": 0.021179401874542238,
"step": 19650
},
{
"epoch": 0.3821384233395407,
"grad_norm": 0.11812082678079605,
"learning_rate": 1.797657150862673e-05,
"loss": 0.011177585124969483,
"step": 19700
},
{
"epoch": 0.38310831781502175,
"grad_norm": 0.6298716068267822,
"learning_rate": 1.796579483366202e-05,
"loss": 0.013784595727920533,
"step": 19750
},
{
"epoch": 0.3840782122905028,
"grad_norm": 0.21435242891311646,
"learning_rate": 1.7955018158697317e-05,
"loss": 0.0177967369556427,
"step": 19800
},
{
"epoch": 0.3850481067659839,
"grad_norm": 0.023306887596845627,
"learning_rate": 1.794424148373261e-05,
"loss": 0.016050543785095215,
"step": 19850
},
{
"epoch": 0.3860180012414649,
"grad_norm": 0.636067271232605,
"learning_rate": 1.7933464808767904e-05,
"loss": 0.020504894256591796,
"step": 19900
},
{
"epoch": 0.386987895716946,
"grad_norm": 0.07343681156635284,
"learning_rate": 1.79226881338032e-05,
"loss": 0.018533397912979126,
"step": 19950
},
{
"epoch": 0.38795779019242704,
"grad_norm": 0.03663763403892517,
"learning_rate": 1.791191145883849e-05,
"loss": 0.02435717821121216,
"step": 20000
},
{
"epoch": 0.3889276846679081,
"grad_norm": 3.856032371520996,
"learning_rate": 1.7901134783873787e-05,
"loss": 0.015563478469848633,
"step": 20050
},
{
"epoch": 0.3898975791433892,
"grad_norm": 0.5419818758964539,
"learning_rate": 1.789035810890908e-05,
"loss": 0.011545709371566772,
"step": 20100
},
{
"epoch": 0.39086747361887025,
"grad_norm": 0.2501530051231384,
"learning_rate": 1.7879581433944373e-05,
"loss": 0.01580398917198181,
"step": 20150
},
{
"epoch": 0.3918373680943513,
"grad_norm": 0.32184022665023804,
"learning_rate": 1.7868804758979666e-05,
"loss": 0.019735274314880372,
"step": 20200
},
{
"epoch": 0.3928072625698324,
"grad_norm": 0.6673415303230286,
"learning_rate": 1.785802808401496e-05,
"loss": 0.0176770281791687,
"step": 20250
},
{
"epoch": 0.39377715704531346,
"grad_norm": 0.05260830745100975,
"learning_rate": 1.7847251409050252e-05,
"loss": 0.017134491205215454,
"step": 20300
},
{
"epoch": 0.39474705152079453,
"grad_norm": 0.015436123125255108,
"learning_rate": 1.7836474734085546e-05,
"loss": 0.018016055822372437,
"step": 20350
},
{
"epoch": 0.3957169459962756,
"grad_norm": 0.22676624357700348,
"learning_rate": 1.782569805912084e-05,
"loss": 0.0179782497882843,
"step": 20400
},
{
"epoch": 0.3966868404717567,
"grad_norm": 2.8003814220428467,
"learning_rate": 1.7814921384156135e-05,
"loss": 0.018881620168685914,
"step": 20450
},
{
"epoch": 0.39765673494723774,
"grad_norm": 0.08627859503030777,
"learning_rate": 1.7804144709191425e-05,
"loss": 0.019868394136428835,
"step": 20500
},
{
"epoch": 0.3986266294227188,
"grad_norm": 0.1056988537311554,
"learning_rate": 1.779336803422672e-05,
"loss": 0.013259539604187012,
"step": 20550
},
{
"epoch": 0.3995965238981999,
"grad_norm": 0.29137933254241943,
"learning_rate": 1.7782591359262015e-05,
"loss": 0.016013699769973754,
"step": 20600
},
{
"epoch": 0.40056641837368095,
"grad_norm": 2.147972822189331,
"learning_rate": 1.7771814684297308e-05,
"loss": 0.023022284507751466,
"step": 20650
},
{
"epoch": 0.401536312849162,
"grad_norm": 0.2576451301574707,
"learning_rate": 1.7761038009332604e-05,
"loss": 0.014544572830200195,
"step": 20700
},
{
"epoch": 0.4025062073246431,
"grad_norm": 0.12398266047239304,
"learning_rate": 1.7750261334367894e-05,
"loss": 0.010878020524978637,
"step": 20750
},
{
"epoch": 0.40347610180012417,
"grad_norm": 0.17603278160095215,
"learning_rate": 1.773948465940319e-05,
"loss": 0.01713404178619385,
"step": 20800
},
{
"epoch": 0.40444599627560524,
"grad_norm": 0.05527504161000252,
"learning_rate": 1.7728707984438484e-05,
"loss": 0.01166084885597229,
"step": 20850
},
{
"epoch": 0.4054158907510863,
"grad_norm": 3.0220062732696533,
"learning_rate": 1.7717931309473777e-05,
"loss": 0.018041354417800904,
"step": 20900
},
{
"epoch": 0.4063857852265674,
"grad_norm": 0.62635737657547,
"learning_rate": 1.770715463450907e-05,
"loss": 0.013640594482421876,
"step": 20950
},
{
"epoch": 0.4073556797020484,
"grad_norm": 0.03590773418545723,
"learning_rate": 1.7696377959544363e-05,
"loss": 0.01427937150001526,
"step": 21000
},
{
"epoch": 0.40832557417752946,
"grad_norm": 0.3204002380371094,
"learning_rate": 1.7685601284579657e-05,
"loss": 0.024812512397766114,
"step": 21050
},
{
"epoch": 0.40929546865301053,
"grad_norm": 0.01931946538388729,
"learning_rate": 1.767482460961495e-05,
"loss": 0.01603399157524109,
"step": 21100
},
{
"epoch": 0.4102653631284916,
"grad_norm": 0.009981716051697731,
"learning_rate": 1.7664047934650243e-05,
"loss": 0.013015174865722656,
"step": 21150
},
{
"epoch": 0.41123525760397267,
"grad_norm": 0.0961197167634964,
"learning_rate": 1.765327125968554e-05,
"loss": 0.011733157634735107,
"step": 21200
},
{
"epoch": 0.41220515207945374,
"grad_norm": 0.002572803059592843,
"learning_rate": 1.764249458472083e-05,
"loss": 0.015126293897628785,
"step": 21250
},
{
"epoch": 0.4131750465549348,
"grad_norm": 0.11168774217367172,
"learning_rate": 1.7631717909756126e-05,
"loss": 0.014364672899246216,
"step": 21300
},
{
"epoch": 0.4141449410304159,
"grad_norm": 1.0546419620513916,
"learning_rate": 1.762094123479142e-05,
"loss": 0.025698075294494627,
"step": 21350
},
{
"epoch": 0.41511483550589695,
"grad_norm": 0.1012447327375412,
"learning_rate": 1.7610164559826712e-05,
"loss": 0.018101298809051515,
"step": 21400
},
{
"epoch": 0.416084729981378,
"grad_norm": 0.03125472739338875,
"learning_rate": 1.759938788486201e-05,
"loss": 0.017162673473358155,
"step": 21450
},
{
"epoch": 0.4170546244568591,
"grad_norm": 3.074187755584717,
"learning_rate": 1.75886112098973e-05,
"loss": 0.019823522567749025,
"step": 21500
},
{
"epoch": 0.41802451893234016,
"grad_norm": 0.07163562625646591,
"learning_rate": 1.7577834534932595e-05,
"loss": 0.022766478061676025,
"step": 21550
},
{
"epoch": 0.41899441340782123,
"grad_norm": 0.45836541056632996,
"learning_rate": 1.7567057859967888e-05,
"loss": 0.017454534769058228,
"step": 21600
},
{
"epoch": 0.4199643078833023,
"grad_norm": 0.18779192864894867,
"learning_rate": 1.755628118500318e-05,
"loss": 0.013594653606414795,
"step": 21650
},
{
"epoch": 0.42093420235878337,
"grad_norm": 0.07238296419382095,
"learning_rate": 1.7545504510038474e-05,
"loss": 0.01464440107345581,
"step": 21700
},
{
"epoch": 0.42190409683426444,
"grad_norm": 1.5767980813980103,
"learning_rate": 1.7534727835073768e-05,
"loss": 0.021548926830291748,
"step": 21750
},
{
"epoch": 0.4228739913097455,
"grad_norm": 0.06450723856687546,
"learning_rate": 1.752395116010906e-05,
"loss": 0.015832515954971312,
"step": 21800
},
{
"epoch": 0.4238438857852266,
"grad_norm": 0.3058660626411438,
"learning_rate": 1.7513174485144354e-05,
"loss": 0.019747809171676636,
"step": 21850
},
{
"epoch": 0.42481378026070765,
"grad_norm": 1.969959020614624,
"learning_rate": 1.7502397810179647e-05,
"loss": 0.012038712501525878,
"step": 21900
},
{
"epoch": 0.4257836747361887,
"grad_norm": 0.026327205821871758,
"learning_rate": 1.7491621135214944e-05,
"loss": 0.013695956468582153,
"step": 21950
},
{
"epoch": 0.4267535692116698,
"grad_norm": 2.22163987159729,
"learning_rate": 1.7480844460250233e-05,
"loss": 0.017967050075531007,
"step": 22000
},
{
"epoch": 0.42772346368715086,
"grad_norm": 0.9522823095321655,
"learning_rate": 1.747006778528553e-05,
"loss": 0.016170030832290648,
"step": 22050
},
{
"epoch": 0.42869335816263193,
"grad_norm": 0.8655645847320557,
"learning_rate": 1.7459291110320823e-05,
"loss": 0.010764075517654419,
"step": 22100
},
{
"epoch": 0.42966325263811295,
"grad_norm": 0.7014385461807251,
"learning_rate": 1.7448514435356116e-05,
"loss": 0.024063951969146728,
"step": 22150
},
{
"epoch": 0.430633147113594,
"grad_norm": 0.035301998257637024,
"learning_rate": 1.7437737760391413e-05,
"loss": 0.009817119836807251,
"step": 22200
},
{
"epoch": 0.4316030415890751,
"grad_norm": 0.8742854595184326,
"learning_rate": 1.7426961085426703e-05,
"loss": 0.024521350860595703,
"step": 22250
},
{
"epoch": 0.43257293606455616,
"grad_norm": 0.18603980541229248,
"learning_rate": 1.7416184410462e-05,
"loss": 0.02033165216445923,
"step": 22300
},
{
"epoch": 0.4335428305400372,
"grad_norm": 0.16574275493621826,
"learning_rate": 1.7405407735497292e-05,
"loss": 0.014174036979675293,
"step": 22350
},
{
"epoch": 0.4345127250155183,
"grad_norm": 0.2310817986726761,
"learning_rate": 1.7394631060532585e-05,
"loss": 0.01202249050140381,
"step": 22400
},
{
"epoch": 0.43548261949099937,
"grad_norm": 0.1006666049361229,
"learning_rate": 1.738385438556788e-05,
"loss": 0.011205404996871948,
"step": 22450
},
{
"epoch": 0.43645251396648044,
"grad_norm": 0.06695405393838882,
"learning_rate": 1.7373077710603172e-05,
"loss": 0.01337580919265747,
"step": 22500
},
{
"epoch": 0.4374224084419615,
"grad_norm": 0.30177491903305054,
"learning_rate": 1.7362301035638465e-05,
"loss": 0.013815712928771973,
"step": 22550
},
{
"epoch": 0.4383923029174426,
"grad_norm": 0.08363492786884308,
"learning_rate": 1.7351524360673758e-05,
"loss": 0.014862210750579833,
"step": 22600
},
{
"epoch": 0.43936219739292365,
"grad_norm": 0.22705359756946564,
"learning_rate": 1.734074768570905e-05,
"loss": 0.010026525259017944,
"step": 22650
},
{
"epoch": 0.4403320918684047,
"grad_norm": 0.06809738278388977,
"learning_rate": 1.7329971010744348e-05,
"loss": 0.01910570740699768,
"step": 22700
},
{
"epoch": 0.4413019863438858,
"grad_norm": 0.5422145128250122,
"learning_rate": 1.731919433577964e-05,
"loss": 0.015320430994033813,
"step": 22750
},
{
"epoch": 0.44227188081936686,
"grad_norm": 0.11068514734506607,
"learning_rate": 1.7308417660814934e-05,
"loss": 0.010554667711257935,
"step": 22800
},
{
"epoch": 0.44324177529484793,
"grad_norm": 0.2909947335720062,
"learning_rate": 1.7297640985850227e-05,
"loss": 0.020208845138549803,
"step": 22850
},
{
"epoch": 0.444211669770329,
"grad_norm": 4.575692653656006,
"learning_rate": 1.728686431088552e-05,
"loss": 0.017979359626770018,
"step": 22900
},
{
"epoch": 0.44518156424581007,
"grad_norm": 0.5604790449142456,
"learning_rate": 1.7276087635920814e-05,
"loss": 0.016861741542816163,
"step": 22950
},
{
"epoch": 0.44615145872129114,
"grad_norm": 0.06400007009506226,
"learning_rate": 1.7265310960956107e-05,
"loss": 0.01869694471359253,
"step": 23000
},
{
"epoch": 0.4471213531967722,
"grad_norm": 0.25228390097618103,
"learning_rate": 1.7254534285991403e-05,
"loss": 0.011897269487380981,
"step": 23050
},
{
"epoch": 0.4480912476722533,
"grad_norm": 0.09902919083833694,
"learning_rate": 1.7243757611026693e-05,
"loss": 0.020924563407897948,
"step": 23100
},
{
"epoch": 0.44906114214773435,
"grad_norm": 2.1226212978363037,
"learning_rate": 1.723298093606199e-05,
"loss": 0.01823074698448181,
"step": 23150
},
{
"epoch": 0.4500310366232154,
"grad_norm": 0.05765779688954353,
"learning_rate": 1.7222204261097283e-05,
"loss": 0.014975235462188721,
"step": 23200
},
{
"epoch": 0.4510009310986965,
"grad_norm": 0.12496720999479294,
"learning_rate": 1.7211427586132576e-05,
"loss": 0.011880019903182984,
"step": 23250
},
{
"epoch": 0.4519708255741775,
"grad_norm": 0.04651029407978058,
"learning_rate": 1.720065091116787e-05,
"loss": 0.019456816911697386,
"step": 23300
},
{
"epoch": 0.4529407200496586,
"grad_norm": 0.2909986972808838,
"learning_rate": 1.7189874236203162e-05,
"loss": 0.016664927005767823,
"step": 23350
},
{
"epoch": 0.45391061452513964,
"grad_norm": 0.0025691871996968985,
"learning_rate": 1.7179097561238455e-05,
"loss": 0.01436832308769226,
"step": 23400
},
{
"epoch": 0.4548805090006207,
"grad_norm": 0.06631513684988022,
"learning_rate": 1.7168320886273752e-05,
"loss": 0.02095062017440796,
"step": 23450
},
{
"epoch": 0.4558504034761018,
"grad_norm": 0.056439995765686035,
"learning_rate": 1.7157544211309045e-05,
"loss": 0.014855221509933472,
"step": 23500
},
{
"epoch": 0.45682029795158285,
"grad_norm": 1.4747319221496582,
"learning_rate": 1.7146767536344338e-05,
"loss": 0.012086091041564941,
"step": 23550
},
{
"epoch": 0.4577901924270639,
"grad_norm": 0.1493714451789856,
"learning_rate": 1.713599086137963e-05,
"loss": 0.012832287549972534,
"step": 23600
},
{
"epoch": 0.458760086902545,
"grad_norm": 0.02708921581506729,
"learning_rate": 1.7125214186414925e-05,
"loss": 0.01903090238571167,
"step": 23650
},
{
"epoch": 0.45972998137802606,
"grad_norm": 3.2368743419647217,
"learning_rate": 1.7114437511450218e-05,
"loss": 0.014929980039596558,
"step": 23700
},
{
"epoch": 0.46069987585350713,
"grad_norm": 0.06653718650341034,
"learning_rate": 1.710366083648551e-05,
"loss": 0.01565203070640564,
"step": 23750
},
{
"epoch": 0.4616697703289882,
"grad_norm": 0.056955285370349884,
"learning_rate": 1.7092884161520807e-05,
"loss": 0.010287386178970338,
"step": 23800
},
{
"epoch": 0.4626396648044693,
"grad_norm": 0.28941360116004944,
"learning_rate": 1.7082107486556097e-05,
"loss": 0.020627894401550294,
"step": 23850
},
{
"epoch": 0.46360955927995035,
"grad_norm": 0.005853681359440088,
"learning_rate": 1.7071330811591394e-05,
"loss": 0.014266180992126464,
"step": 23900
},
{
"epoch": 0.4645794537554314,
"grad_norm": 0.16406269371509552,
"learning_rate": 1.7060554136626687e-05,
"loss": 0.012651551961898804,
"step": 23950
},
{
"epoch": 0.4655493482309125,
"grad_norm": 0.004302423447370529,
"learning_rate": 1.704977746166198e-05,
"loss": 0.016248624324798584,
"step": 24000
},
{
"epoch": 0.46651924270639356,
"grad_norm": 0.2806839644908905,
"learning_rate": 1.7039000786697273e-05,
"loss": 0.02316607713699341,
"step": 24050
},
{
"epoch": 0.4674891371818746,
"grad_norm": 3.055875301361084,
"learning_rate": 1.7028224111732566e-05,
"loss": 0.013369892835617065,
"step": 24100
},
{
"epoch": 0.4684590316573557,
"grad_norm": 0.1110517606139183,
"learning_rate": 1.701744743676786e-05,
"loss": 0.011998939514160156,
"step": 24150
},
{
"epoch": 0.46942892613283677,
"grad_norm": 1.5323948860168457,
"learning_rate": 1.7006670761803156e-05,
"loss": 0.017416648864746094,
"step": 24200
},
{
"epoch": 0.47039882060831784,
"grad_norm": 0.5642879605293274,
"learning_rate": 1.699589408683845e-05,
"loss": 0.01614004135131836,
"step": 24250
},
{
"epoch": 0.4713687150837989,
"grad_norm": 0.11729516088962555,
"learning_rate": 1.6985117411873742e-05,
"loss": 0.014984267950057983,
"step": 24300
},
{
"epoch": 0.47233860955928,
"grad_norm": 0.002711569657549262,
"learning_rate": 1.6974340736909036e-05,
"loss": 0.01662859559059143,
"step": 24350
},
{
"epoch": 0.473308504034761,
"grad_norm": 0.22783033549785614,
"learning_rate": 1.696356406194433e-05,
"loss": 0.015154259204864502,
"step": 24400
},
{
"epoch": 0.47427839851024206,
"grad_norm": 0.19248810410499573,
"learning_rate": 1.6952787386979622e-05,
"loss": 0.01850400447845459,
"step": 24450
},
{
"epoch": 0.47524829298572313,
"grad_norm": 0.15362098813056946,
"learning_rate": 1.6942010712014915e-05,
"loss": 0.014641754627227783,
"step": 24500
},
{
"epoch": 0.4762181874612042,
"grad_norm": 0.034069404006004333,
"learning_rate": 1.693123403705021e-05,
"loss": 0.01503227949142456,
"step": 24550
},
{
"epoch": 0.47718808193668527,
"grad_norm": 0.0034818127751350403,
"learning_rate": 1.69204573620855e-05,
"loss": 0.019913687705993652,
"step": 24600
},
{
"epoch": 0.47815797641216634,
"grad_norm": 0.0016202256083488464,
"learning_rate": 1.6909680687120798e-05,
"loss": 0.010700620412826538,
"step": 24650
},
{
"epoch": 0.4791278708876474,
"grad_norm": 0.16094791889190674,
"learning_rate": 1.689890401215609e-05,
"loss": 0.016578717231750487,
"step": 24700
},
{
"epoch": 0.4800977653631285,
"grad_norm": 0.2292594015598297,
"learning_rate": 1.6888127337191384e-05,
"loss": 0.0216058349609375,
"step": 24750
},
{
"epoch": 0.48106765983860955,
"grad_norm": 0.30840781331062317,
"learning_rate": 1.6877350662226677e-05,
"loss": 0.0105799400806427,
"step": 24800
},
{
"epoch": 0.4820375543140906,
"grad_norm": 0.3342400789260864,
"learning_rate": 1.686657398726197e-05,
"loss": 0.02784029483795166,
"step": 24850
},
{
"epoch": 0.4830074487895717,
"grad_norm": 0.23664791882038116,
"learning_rate": 1.6855797312297267e-05,
"loss": 0.019208911657333374,
"step": 24900
},
{
"epoch": 0.48397734326505276,
"grad_norm": 0.09094451367855072,
"learning_rate": 1.684502063733256e-05,
"loss": 0.012639690637588501,
"step": 24950
},
{
"epoch": 0.48494723774053383,
"grad_norm": 7.277854919433594,
"learning_rate": 1.6834243962367853e-05,
"loss": 0.021325414180755616,
"step": 25000
},
{
"epoch": 0.4859171322160149,
"grad_norm": 0.4742510914802551,
"learning_rate": 1.6823467287403147e-05,
"loss": 0.019907597303390503,
"step": 25050
},
{
"epoch": 0.48688702669149597,
"grad_norm": 1.6626486778259277,
"learning_rate": 1.681269061243844e-05,
"loss": 0.01400539517402649,
"step": 25100
},
{
"epoch": 0.48785692116697704,
"grad_norm": 0.09989999979734421,
"learning_rate": 1.6801913937473733e-05,
"loss": 0.01599094271659851,
"step": 25150
},
{
"epoch": 0.4888268156424581,
"grad_norm": 0.3337894082069397,
"learning_rate": 1.6791137262509026e-05,
"loss": 0.016563934087753297,
"step": 25200
},
{
"epoch": 0.4897967101179392,
"grad_norm": 0.014184270054101944,
"learning_rate": 1.678036058754432e-05,
"loss": 0.008327061533927918,
"step": 25250
},
{
"epoch": 0.49076660459342025,
"grad_norm": 0.6191167235374451,
"learning_rate": 1.6769583912579616e-05,
"loss": 0.01686647653579712,
"step": 25300
},
{
"epoch": 0.4917364990689013,
"grad_norm": 0.7777760028839111,
"learning_rate": 1.6758807237614906e-05,
"loss": 0.0205366587638855,
"step": 25350
},
{
"epoch": 0.4927063935443824,
"grad_norm": 0.37853196263313293,
"learning_rate": 1.6748030562650202e-05,
"loss": 0.013436235189437866,
"step": 25400
},
{
"epoch": 0.49367628801986346,
"grad_norm": 1.0371050834655762,
"learning_rate": 1.6737253887685495e-05,
"loss": 0.012231028079986573,
"step": 25450
},
{
"epoch": 0.49464618249534453,
"grad_norm": 0.02320312336087227,
"learning_rate": 1.672647721272079e-05,
"loss": 0.015402934551239013,
"step": 25500
},
{
"epoch": 0.49561607697082555,
"grad_norm": 0.045472659170627594,
"learning_rate": 1.671570053775608e-05,
"loss": 0.011728731393814086,
"step": 25550
},
{
"epoch": 0.4965859714463066,
"grad_norm": 0.33407220244407654,
"learning_rate": 1.6704923862791375e-05,
"loss": 0.01573185920715332,
"step": 25600
},
{
"epoch": 0.4975558659217877,
"grad_norm": 0.2012293040752411,
"learning_rate": 1.669414718782667e-05,
"loss": 0.015845534801483156,
"step": 25650
},
{
"epoch": 0.49852576039726876,
"grad_norm": 2.4472267627716064,
"learning_rate": 1.6683370512861964e-05,
"loss": 0.017187341451644897,
"step": 25700
},
{
"epoch": 0.49949565487274983,
"grad_norm": 0.03843652456998825,
"learning_rate": 1.6672593837897258e-05,
"loss": 0.014713480472564697,
"step": 25750
},
{
"epoch": 0.500465549348231,
"grad_norm": 0.004235955886542797,
"learning_rate": 1.666181716293255e-05,
"loss": 0.020253138542175295,
"step": 25800
},
{
"epoch": 0.501435443823712,
"grad_norm": 0.1848042607307434,
"learning_rate": 1.6651040487967844e-05,
"loss": 0.009521148204803466,
"step": 25850
},
{
"epoch": 0.5024053382991931,
"grad_norm": 0.0008410600130446255,
"learning_rate": 1.6640263813003137e-05,
"loss": 0.0132768714427948,
"step": 25900
},
{
"epoch": 0.5033752327746741,
"grad_norm": 0.6806746125221252,
"learning_rate": 1.662948713803843e-05,
"loss": 0.020576977729797365,
"step": 25950
},
{
"epoch": 0.5043451272501552,
"grad_norm": 0.13482260704040527,
"learning_rate": 1.6618710463073723e-05,
"loss": 0.018458107709884642,
"step": 26000
},
{
"epoch": 0.5053150217256362,
"grad_norm": 0.7686246633529663,
"learning_rate": 1.660793378810902e-05,
"loss": 0.015936212539672853,
"step": 26050
},
{
"epoch": 0.5062849162011173,
"grad_norm": 2.190617084503174,
"learning_rate": 1.659715711314431e-05,
"loss": 0.017016206979751588,
"step": 26100
},
{
"epoch": 0.5072548106765984,
"grad_norm": 1.0037477016448975,
"learning_rate": 1.6586380438179606e-05,
"loss": 0.014334765672683715,
"step": 26150
},
{
"epoch": 0.5082247051520794,
"grad_norm": 10.000763893127441,
"learning_rate": 1.65756037632149e-05,
"loss": 0.01537983536720276,
"step": 26200
},
{
"epoch": 0.5091945996275605,
"grad_norm": 1.0490413904190063,
"learning_rate": 1.6564827088250193e-05,
"loss": 0.015433973073959351,
"step": 26250
},
{
"epoch": 0.5101644941030415,
"grad_norm": 0.14998184144496918,
"learning_rate": 1.6554050413285486e-05,
"loss": 0.01647668480873108,
"step": 26300
},
{
"epoch": 0.5111343885785227,
"grad_norm": 0.7908254265785217,
"learning_rate": 1.654327373832078e-05,
"loss": 0.015480660200119019,
"step": 26350
},
{
"epoch": 0.5121042830540037,
"grad_norm": 0.10278303176164627,
"learning_rate": 1.6532497063356075e-05,
"loss": 0.014672179222106934,
"step": 26400
},
{
"epoch": 0.5130741775294848,
"grad_norm": 0.3102555274963379,
"learning_rate": 1.6521720388391365e-05,
"loss": 0.015705890655517578,
"step": 26450
},
{
"epoch": 0.5140440720049658,
"grad_norm": 0.32322463393211365,
"learning_rate": 1.6510943713426662e-05,
"loss": 0.00959071695804596,
"step": 26500
},
{
"epoch": 0.515013966480447,
"grad_norm": 1.4172191619873047,
"learning_rate": 1.6500167038461955e-05,
"loss": 0.020870118141174315,
"step": 26550
},
{
"epoch": 0.515983860955928,
"grad_norm": 0.027370823547244072,
"learning_rate": 1.6489390363497248e-05,
"loss": 0.015793596506118775,
"step": 26600
},
{
"epoch": 0.5169537554314091,
"grad_norm": 0.8051190376281738,
"learning_rate": 1.647861368853254e-05,
"loss": 0.019760849475860594,
"step": 26650
},
{
"epoch": 0.5179236499068901,
"grad_norm": 0.03524237126111984,
"learning_rate": 1.6467837013567834e-05,
"loss": 0.009431734085083007,
"step": 26700
},
{
"epoch": 0.5188935443823712,
"grad_norm": 0.3385278880596161,
"learning_rate": 1.6457060338603128e-05,
"loss": 0.01677426815032959,
"step": 26750
},
{
"epoch": 0.5198634388578522,
"grad_norm": 0.5713192820549011,
"learning_rate": 1.6446283663638424e-05,
"loss": 0.01522658348083496,
"step": 26800
},
{
"epoch": 0.5208333333333334,
"grad_norm": 0.10292282700538635,
"learning_rate": 1.6435506988673714e-05,
"loss": 0.014670636653900147,
"step": 26850
},
{
"epoch": 0.5218032278088144,
"grad_norm": 0.3331795334815979,
"learning_rate": 1.642473031370901e-05,
"loss": 0.010354068279266358,
"step": 26900
},
{
"epoch": 0.5227731222842955,
"grad_norm": 0.1668291538953781,
"learning_rate": 1.6413953638744304e-05,
"loss": 0.014502136707305909,
"step": 26950
},
{
"epoch": 0.5237430167597765,
"grad_norm": 1.325940728187561,
"learning_rate": 1.6403176963779597e-05,
"loss": 0.00991119146347046,
"step": 27000
},
{
"epoch": 0.5247129112352577,
"grad_norm": 0.24499039351940155,
"learning_rate": 1.639240028881489e-05,
"loss": 0.020862958431243896,
"step": 27050
},
{
"epoch": 0.5256828057107387,
"grad_norm": 0.010075472295284271,
"learning_rate": 1.6381623613850183e-05,
"loss": 0.014404611587524414,
"step": 27100
},
{
"epoch": 0.5266527001862198,
"grad_norm": 0.7762529253959656,
"learning_rate": 1.637084693888548e-05,
"loss": 0.018809754848480225,
"step": 27150
},
{
"epoch": 0.5276225946617008,
"grad_norm": 0.3500649034976959,
"learning_rate": 1.636007026392077e-05,
"loss": 0.013991461992263794,
"step": 27200
},
{
"epoch": 0.5285924891371818,
"grad_norm": 0.11990806460380554,
"learning_rate": 1.6349293588956066e-05,
"loss": 0.02525176286697388,
"step": 27250
},
{
"epoch": 0.529562383612663,
"grad_norm": 0.14011047780513763,
"learning_rate": 1.633851691399136e-05,
"loss": 0.013289759159088135,
"step": 27300
},
{
"epoch": 0.530532278088144,
"grad_norm": 0.05568653717637062,
"learning_rate": 1.6327740239026652e-05,
"loss": 0.01355094313621521,
"step": 27350
},
{
"epoch": 0.5315021725636251,
"grad_norm": 2.287010908126831,
"learning_rate": 1.6316963564061945e-05,
"loss": 0.013338940143585205,
"step": 27400
},
{
"epoch": 0.5324720670391061,
"grad_norm": 0.236273393034935,
"learning_rate": 1.630618688909724e-05,
"loss": 0.017981814146041872,
"step": 27450
},
{
"epoch": 0.5334419615145872,
"grad_norm": 0.06931453943252563,
"learning_rate": 1.6295410214132532e-05,
"loss": 0.015636544227600097,
"step": 27500
},
{
"epoch": 0.5344118559900682,
"grad_norm": 0.09620760381221771,
"learning_rate": 1.6284633539167828e-05,
"loss": 0.019642093181610108,
"step": 27550
},
{
"epoch": 0.5353817504655494,
"grad_norm": 0.7033560276031494,
"learning_rate": 1.6273856864203118e-05,
"loss": 0.021041052341461183,
"step": 27600
},
{
"epoch": 0.5363516449410304,
"grad_norm": 0.8700341582298279,
"learning_rate": 1.6263080189238415e-05,
"loss": 0.01979437828063965,
"step": 27650
},
{
"epoch": 0.5373215394165115,
"grad_norm": 0.06736158579587936,
"learning_rate": 1.6252303514273708e-05,
"loss": 0.012386791706085205,
"step": 27700
},
{
"epoch": 0.5382914338919925,
"grad_norm": 1.8305896520614624,
"learning_rate": 1.6241526839309e-05,
"loss": 0.02092858076095581,
"step": 27750
},
{
"epoch": 0.5392613283674736,
"grad_norm": 0.16352833807468414,
"learning_rate": 1.6230750164344294e-05,
"loss": 0.015582860708236694,
"step": 27800
},
{
"epoch": 0.5402312228429547,
"grad_norm": 0.7408884167671204,
"learning_rate": 1.6219973489379587e-05,
"loss": 0.015786865949630736,
"step": 27850
},
{
"epoch": 0.5412011173184358,
"grad_norm": 0.21118904650211334,
"learning_rate": 1.6209196814414884e-05,
"loss": 0.0179096782207489,
"step": 27900
},
{
"epoch": 0.5421710117939168,
"grad_norm": 0.025865672156214714,
"learning_rate": 1.6198420139450174e-05,
"loss": 0.013157802820205688,
"step": 27950
},
{
"epoch": 0.5431409062693979,
"grad_norm": 0.6807165145874023,
"learning_rate": 1.618764346448547e-05,
"loss": 0.009808352589607239,
"step": 28000
},
{
"epoch": 0.5441108007448789,
"grad_norm": 0.42911773920059204,
"learning_rate": 1.6176866789520763e-05,
"loss": 0.017121896743774415,
"step": 28050
},
{
"epoch": 0.5450806952203601,
"grad_norm": 1.887188196182251,
"learning_rate": 1.6166090114556056e-05,
"loss": 0.020593838691711427,
"step": 28100
},
{
"epoch": 0.5460505896958411,
"grad_norm": 0.3655708134174347,
"learning_rate": 1.615531343959135e-05,
"loss": 0.01862805724143982,
"step": 28150
},
{
"epoch": 0.5470204841713222,
"grad_norm": 0.6240643262863159,
"learning_rate": 1.6144536764626643e-05,
"loss": 0.012342108488082886,
"step": 28200
},
{
"epoch": 0.5479903786468032,
"grad_norm": 0.008131642825901508,
"learning_rate": 1.6133760089661936e-05,
"loss": 0.015302373170852661,
"step": 28250
},
{
"epoch": 0.5489602731222843,
"grad_norm": 0.6189226508140564,
"learning_rate": 1.6122983414697232e-05,
"loss": 0.009876628518104552,
"step": 28300
},
{
"epoch": 0.5499301675977654,
"grad_norm": 0.07442165911197662,
"learning_rate": 1.6112206739732522e-05,
"loss": 0.02081130027770996,
"step": 28350
},
{
"epoch": 0.5509000620732464,
"grad_norm": 0.6516116857528687,
"learning_rate": 1.610143006476782e-05,
"loss": 0.016367655992507935,
"step": 28400
},
{
"epoch": 0.5518699565487275,
"grad_norm": 0.01618054322898388,
"learning_rate": 1.6090653389803112e-05,
"loss": 0.02339813232421875,
"step": 28450
},
{
"epoch": 0.5528398510242085,
"grad_norm": 0.09775125980377197,
"learning_rate": 1.6079876714838405e-05,
"loss": 0.011870177984237671,
"step": 28500
},
{
"epoch": 0.5538097454996896,
"grad_norm": 0.26518115401268005,
"learning_rate": 1.6069100039873698e-05,
"loss": 0.012791706323623657,
"step": 28550
},
{
"epoch": 0.5547796399751707,
"grad_norm": 0.31162023544311523,
"learning_rate": 1.605832336490899e-05,
"loss": 0.012753716707229613,
"step": 28600
},
{
"epoch": 0.5557495344506518,
"grad_norm": 0.07928977906703949,
"learning_rate": 1.6047546689944288e-05,
"loss": 0.012891114950180053,
"step": 28650
},
{
"epoch": 0.5567194289261328,
"grad_norm": 0.19415180385112762,
"learning_rate": 1.6036770014979578e-05,
"loss": 0.011597986221313477,
"step": 28700
},
{
"epoch": 0.5576893234016139,
"grad_norm": 0.03397326171398163,
"learning_rate": 1.6025993340014874e-05,
"loss": 0.02097949981689453,
"step": 28750
},
{
"epoch": 0.5586592178770949,
"grad_norm": 1.449591040611267,
"learning_rate": 1.6015216665050167e-05,
"loss": 0.015120692253112793,
"step": 28800
},
{
"epoch": 0.5596291123525761,
"grad_norm": 0.002593772253021598,
"learning_rate": 1.600443999008546e-05,
"loss": 0.01539137601852417,
"step": 28850
},
{
"epoch": 0.5605990068280571,
"grad_norm": 1.1160919666290283,
"learning_rate": 1.5993663315120754e-05,
"loss": 0.01590176224708557,
"step": 28900
},
{
"epoch": 0.5615689013035382,
"grad_norm": 0.18525412678718567,
"learning_rate": 1.5982886640156047e-05,
"loss": 0.01746701717376709,
"step": 28950
},
{
"epoch": 0.5625387957790192,
"grad_norm": 0.10696443915367126,
"learning_rate": 1.597210996519134e-05,
"loss": 0.016462042331695556,
"step": 29000
},
{
"epoch": 0.5635086902545003,
"grad_norm": 0.22595790028572083,
"learning_rate": 1.5961333290226633e-05,
"loss": 0.015516710281372071,
"step": 29050
},
{
"epoch": 0.5644785847299814,
"grad_norm": 0.15787707269191742,
"learning_rate": 1.5950556615261926e-05,
"loss": 0.014253429174423217,
"step": 29100
},
{
"epoch": 0.5654484792054625,
"grad_norm": 0.09877102077007294,
"learning_rate": 1.5939779940297223e-05,
"loss": 0.014578464031219483,
"step": 29150
},
{
"epoch": 0.5664183736809435,
"grad_norm": 0.8764503002166748,
"learning_rate": 1.5929003265332516e-05,
"loss": 0.01088139533996582,
"step": 29200
},
{
"epoch": 0.5673882681564246,
"grad_norm": 0.3228471875190735,
"learning_rate": 1.591822659036781e-05,
"loss": 0.017314507961273193,
"step": 29250
},
{
"epoch": 0.5683581626319056,
"grad_norm": 0.22145739197731018,
"learning_rate": 1.5907449915403102e-05,
"loss": 0.017583905458450316,
"step": 29300
},
{
"epoch": 0.5693280571073868,
"grad_norm": 1.2110650539398193,
"learning_rate": 1.5896673240438396e-05,
"loss": 0.01523253321647644,
"step": 29350
},
{
"epoch": 0.5702979515828678,
"grad_norm": 6.328366279602051,
"learning_rate": 1.5885896565473692e-05,
"loss": 0.0197372305393219,
"step": 29400
},
{
"epoch": 0.5712678460583489,
"grad_norm": 2.2318108081817627,
"learning_rate": 1.5875119890508982e-05,
"loss": 0.01524843692779541,
"step": 29450
},
{
"epoch": 0.5722377405338299,
"grad_norm": 0.4928746819496155,
"learning_rate": 1.586434321554428e-05,
"loss": 0.014557511806488036,
"step": 29500
},
{
"epoch": 0.5732076350093109,
"grad_norm": 0.09176398813724518,
"learning_rate": 1.585356654057957e-05,
"loss": 0.013921273946762085,
"step": 29550
},
{
"epoch": 0.5741775294847921,
"grad_norm": 0.27174991369247437,
"learning_rate": 1.5842789865614865e-05,
"loss": 0.015718015432357787,
"step": 29600
},
{
"epoch": 0.5751474239602731,
"grad_norm": 0.9980429410934448,
"learning_rate": 1.5832013190650158e-05,
"loss": 0.01653099775314331,
"step": 29650
},
{
"epoch": 0.5761173184357542,
"grad_norm": 1.4954931735992432,
"learning_rate": 1.582123651568545e-05,
"loss": 0.01180437445640564,
"step": 29700
},
{
"epoch": 0.5770872129112352,
"grad_norm": 0.12084205448627472,
"learning_rate": 1.5810459840720744e-05,
"loss": 0.009636707305908203,
"step": 29750
},
{
"epoch": 0.5780571073867163,
"grad_norm": 0.20569948852062225,
"learning_rate": 1.5799683165756037e-05,
"loss": 0.015220273733139039,
"step": 29800
},
{
"epoch": 0.5790270018621974,
"grad_norm": 0.17094168066978455,
"learning_rate": 1.5788906490791334e-05,
"loss": 0.014539427757263183,
"step": 29850
},
{
"epoch": 0.5799968963376785,
"grad_norm": 0.16208311915397644,
"learning_rate": 1.5778129815826627e-05,
"loss": 0.012459973096847534,
"step": 29900
},
{
"epoch": 0.5809667908131595,
"grad_norm": 1.1370199918746948,
"learning_rate": 1.576735314086192e-05,
"loss": 0.016061533689498902,
"step": 29950
},
{
"epoch": 0.5819366852886406,
"grad_norm": 0.2662105858325958,
"learning_rate": 1.5756576465897213e-05,
"loss": 0.015519869327545167,
"step": 30000
},
{
"epoch": 0.5829065797641216,
"grad_norm": 0.054686788469552994,
"learning_rate": 1.5745799790932507e-05,
"loss": 0.010285030603408813,
"step": 30050
},
{
"epoch": 0.5838764742396028,
"grad_norm": 0.051669951528310776,
"learning_rate": 1.57350231159678e-05,
"loss": 0.013581469058990478,
"step": 30100
},
{
"epoch": 0.5848463687150838,
"grad_norm": 10.569610595703125,
"learning_rate": 1.5724246441003096e-05,
"loss": 0.009640730023384093,
"step": 30150
},
{
"epoch": 0.5858162631905649,
"grad_norm": 3.9102864265441895,
"learning_rate": 1.5713469766038386e-05,
"loss": 0.018037461042404175,
"step": 30200
},
{
"epoch": 0.5867861576660459,
"grad_norm": 0.061651188880205154,
"learning_rate": 1.5702693091073683e-05,
"loss": 0.008140394687652588,
"step": 30250
},
{
"epoch": 0.587756052141527,
"grad_norm": 0.048784039914608,
"learning_rate": 1.5691916416108976e-05,
"loss": 0.015503416061401367,
"step": 30300
},
{
"epoch": 0.5887259466170081,
"grad_norm": 0.1897561252117157,
"learning_rate": 1.568113974114427e-05,
"loss": 0.01574919104576111,
"step": 30350
},
{
"epoch": 0.5896958410924892,
"grad_norm": 0.011347883380949497,
"learning_rate": 1.5670363066179562e-05,
"loss": 0.015201386213302612,
"step": 30400
},
{
"epoch": 0.5906657355679702,
"grad_norm": 1.5165247917175293,
"learning_rate": 1.5659586391214855e-05,
"loss": 0.016707221269607542,
"step": 30450
},
{
"epoch": 0.5916356300434513,
"grad_norm": 0.03335001319646835,
"learning_rate": 1.564880971625015e-05,
"loss": 0.019050198793411254,
"step": 30500
},
{
"epoch": 0.5926055245189323,
"grad_norm": 0.09694509208202362,
"learning_rate": 1.563803304128544e-05,
"loss": 0.011075567007064819,
"step": 30550
},
{
"epoch": 0.5935754189944135,
"grad_norm": 1.5571444034576416,
"learning_rate": 1.5627256366320738e-05,
"loss": 0.013307219743728638,
"step": 30600
},
{
"epoch": 0.5945453134698945,
"grad_norm": 0.11216409504413605,
"learning_rate": 1.561647969135603e-05,
"loss": 0.013653246164321899,
"step": 30650
},
{
"epoch": 0.5955152079453755,
"grad_norm": 0.3622380197048187,
"learning_rate": 1.5605703016391324e-05,
"loss": 0.014558343887329102,
"step": 30700
},
{
"epoch": 0.5964851024208566,
"grad_norm": 0.043230123817920685,
"learning_rate": 1.5594926341426618e-05,
"loss": 0.011949583292007446,
"step": 30750
},
{
"epoch": 0.5974549968963376,
"grad_norm": 0.0020321407355368137,
"learning_rate": 1.558414966646191e-05,
"loss": 0.008281537890434265,
"step": 30800
},
{
"epoch": 0.5984248913718188,
"grad_norm": 0.03121757321059704,
"learning_rate": 1.5573372991497204e-05,
"loss": 0.013251700401306153,
"step": 30850
},
{
"epoch": 0.5993947858472998,
"grad_norm": 1.9448819160461426,
"learning_rate": 1.55625963165325e-05,
"loss": 0.01626870036125183,
"step": 30900
},
{
"epoch": 0.6003646803227809,
"grad_norm": 0.3821398913860321,
"learning_rate": 1.555181964156779e-05,
"loss": 0.014425595998764038,
"step": 30950
},
{
"epoch": 0.6013345747982619,
"grad_norm": 0.14565736055374146,
"learning_rate": 1.5541042966603087e-05,
"loss": 0.013312355279922486,
"step": 31000
},
{
"epoch": 0.602304469273743,
"grad_norm": 0.22762785851955414,
"learning_rate": 1.553026629163838e-05,
"loss": 0.016314936876296995,
"step": 31050
},
{
"epoch": 0.603274363749224,
"grad_norm": 0.033526696264743805,
"learning_rate": 1.5519489616673673e-05,
"loss": 0.01860446810722351,
"step": 31100
},
{
"epoch": 0.6042442582247052,
"grad_norm": 2.690331220626831,
"learning_rate": 1.5508712941708966e-05,
"loss": 0.007623982429504394,
"step": 31150
},
{
"epoch": 0.6052141527001862,
"grad_norm": 0.9849132299423218,
"learning_rate": 1.549793626674426e-05,
"loss": 0.012030644416809082,
"step": 31200
},
{
"epoch": 0.6061840471756673,
"grad_norm": 0.433536171913147,
"learning_rate": 1.5487159591779553e-05,
"loss": 0.017185583114624023,
"step": 31250
},
{
"epoch": 0.6071539416511483,
"grad_norm": 0.0022617331705987453,
"learning_rate": 1.5476382916814846e-05,
"loss": 0.013704053163528442,
"step": 31300
},
{
"epoch": 0.6081238361266295,
"grad_norm": 0.05397520959377289,
"learning_rate": 1.5465606241850142e-05,
"loss": 0.012442575693130493,
"step": 31350
},
{
"epoch": 0.6090937306021105,
"grad_norm": 0.01266697607934475,
"learning_rate": 1.5454829566885435e-05,
"loss": 0.016678683757781983,
"step": 31400
},
{
"epoch": 0.6100636250775916,
"grad_norm": 0.8497613072395325,
"learning_rate": 1.544405289192073e-05,
"loss": 0.009722362160682679,
"step": 31450
},
{
"epoch": 0.6110335195530726,
"grad_norm": 1.1832634210586548,
"learning_rate": 1.5433276216956022e-05,
"loss": 0.011429544687271118,
"step": 31500
},
{
"epoch": 0.6120034140285537,
"grad_norm": 0.36344656348228455,
"learning_rate": 1.5422499541991315e-05,
"loss": 0.01807589292526245,
"step": 31550
},
{
"epoch": 0.6129733085040348,
"grad_norm": 0.09691259264945984,
"learning_rate": 1.5411722867026608e-05,
"loss": 0.013921138048171997,
"step": 31600
},
{
"epoch": 0.6139432029795159,
"grad_norm": 1.314125895500183,
"learning_rate": 1.5400946192061905e-05,
"loss": 0.01194469690322876,
"step": 31650
},
{
"epoch": 0.6149130974549969,
"grad_norm": 0.04846692457795143,
"learning_rate": 1.5390169517097194e-05,
"loss": 0.013730260133743287,
"step": 31700
},
{
"epoch": 0.615882991930478,
"grad_norm": 0.5869594812393188,
"learning_rate": 1.537939284213249e-05,
"loss": 0.013563418388366699,
"step": 31750
},
{
"epoch": 0.616852886405959,
"grad_norm": 0.13929125666618347,
"learning_rate": 1.5368616167167784e-05,
"loss": 0.01918055772781372,
"step": 31800
},
{
"epoch": 0.61782278088144,
"grad_norm": 0.011406362988054752,
"learning_rate": 1.5357839492203077e-05,
"loss": 0.013921527862548829,
"step": 31850
},
{
"epoch": 0.6187926753569212,
"grad_norm": 0.027745971456170082,
"learning_rate": 1.534706281723837e-05,
"loss": 0.008195595145225524,
"step": 31900
},
{
"epoch": 0.6197625698324022,
"grad_norm": 0.08578634262084961,
"learning_rate": 1.5336286142273664e-05,
"loss": 0.01342859148979187,
"step": 31950
},
{
"epoch": 0.6207324643078833,
"grad_norm": 0.020374823361635208,
"learning_rate": 1.532550946730896e-05,
"loss": 0.016195051670074463,
"step": 32000
},
{
"epoch": 0.6217023587833643,
"grad_norm": 0.06262603402137756,
"learning_rate": 1.531473279234425e-05,
"loss": 0.009722581505775452,
"step": 32050
},
{
"epoch": 0.6226722532588455,
"grad_norm": 0.06660036742687225,
"learning_rate": 1.5303956117379546e-05,
"loss": 0.013888064622879028,
"step": 32100
},
{
"epoch": 0.6236421477343265,
"grad_norm": 0.2871779799461365,
"learning_rate": 1.529317944241484e-05,
"loss": 0.013345803022384644,
"step": 32150
},
{
"epoch": 0.6246120422098076,
"grad_norm": 0.008470186032354832,
"learning_rate": 1.5282402767450133e-05,
"loss": 0.012796964645385742,
"step": 32200
},
{
"epoch": 0.6255819366852886,
"grad_norm": 0.3281254768371582,
"learning_rate": 1.5271626092485426e-05,
"loss": 0.015310152769088745,
"step": 32250
},
{
"epoch": 0.6265518311607697,
"grad_norm": 0.46868330240249634,
"learning_rate": 1.526084941752072e-05,
"loss": 0.015648070573806763,
"step": 32300
},
{
"epoch": 0.6275217256362507,
"grad_norm": 0.5751120448112488,
"learning_rate": 1.5250072742556012e-05,
"loss": 0.014132370948791504,
"step": 32350
},
{
"epoch": 0.6284916201117319,
"grad_norm": 0.08697997778654099,
"learning_rate": 1.5239296067591307e-05,
"loss": 0.016367411613464354,
"step": 32400
},
{
"epoch": 0.6294615145872129,
"grad_norm": 0.19371114671230316,
"learning_rate": 1.52285193926266e-05,
"loss": 0.009997128248214722,
"step": 32450
},
{
"epoch": 0.630431409062694,
"grad_norm": 0.43518903851509094,
"learning_rate": 1.5217742717661893e-05,
"loss": 0.016916910409927367,
"step": 32500
},
{
"epoch": 0.631401303538175,
"grad_norm": 0.5762829184532166,
"learning_rate": 1.5206966042697187e-05,
"loss": 0.013439910411834717,
"step": 32550
},
{
"epoch": 0.6323711980136562,
"grad_norm": 0.022144904360175133,
"learning_rate": 1.5196189367732481e-05,
"loss": 0.009728944897651672,
"step": 32600
},
{
"epoch": 0.6333410924891372,
"grad_norm": 2.6978955268859863,
"learning_rate": 1.5185412692767773e-05,
"loss": 0.019438672065734863,
"step": 32650
},
{
"epoch": 0.6343109869646183,
"grad_norm": 0.08614096790552139,
"learning_rate": 1.5174636017803068e-05,
"loss": 0.01379317045211792,
"step": 32700
},
{
"epoch": 0.6352808814400993,
"grad_norm": 0.018508311361074448,
"learning_rate": 1.5163859342838363e-05,
"loss": 0.011779872179031372,
"step": 32750
},
{
"epoch": 0.6362507759155804,
"grad_norm": 0.044328901916742325,
"learning_rate": 1.5153082667873656e-05,
"loss": 0.011356784105300903,
"step": 32800
},
{
"epoch": 0.6372206703910615,
"grad_norm": 0.10524370521306992,
"learning_rate": 1.514230599290895e-05,
"loss": 0.014594414234161378,
"step": 32850
},
{
"epoch": 0.6381905648665426,
"grad_norm": 1.2810033559799194,
"learning_rate": 1.5131529317944242e-05,
"loss": 0.018305486440658568,
"step": 32900
},
{
"epoch": 0.6391604593420236,
"grad_norm": 0.13946221768856049,
"learning_rate": 1.5120752642979537e-05,
"loss": 0.009477558135986329,
"step": 32950
},
{
"epoch": 0.6401303538175046,
"grad_norm": 0.11215221136808395,
"learning_rate": 1.510997596801483e-05,
"loss": 0.01256560206413269,
"step": 33000
},
{
"epoch": 0.6411002482929857,
"grad_norm": 0.01814519613981247,
"learning_rate": 1.5099199293050125e-05,
"loss": 0.012335376739501953,
"step": 33050
},
{
"epoch": 0.6420701427684667,
"grad_norm": 0.2854568064212799,
"learning_rate": 1.5088422618085416e-05,
"loss": 0.021937782764434813,
"step": 33100
},
{
"epoch": 0.6430400372439479,
"grad_norm": 0.07269173115491867,
"learning_rate": 1.5077645943120711e-05,
"loss": 0.010593458414077758,
"step": 33150
},
{
"epoch": 0.6440099317194289,
"grad_norm": 0.1534934788942337,
"learning_rate": 1.5066869268156004e-05,
"loss": 0.016071219444274903,
"step": 33200
},
{
"epoch": 0.64497982619491,
"grad_norm": 0.18668745458126068,
"learning_rate": 1.5056092593191298e-05,
"loss": 0.022440412044525147,
"step": 33250
},
{
"epoch": 0.645949720670391,
"grad_norm": 0.02987876906991005,
"learning_rate": 1.504531591822659e-05,
"loss": 0.009408287405967712,
"step": 33300
},
{
"epoch": 0.6469196151458722,
"grad_norm": 0.0074963136576116085,
"learning_rate": 1.5034539243261886e-05,
"loss": 0.018162318468093873,
"step": 33350
},
{
"epoch": 0.6478895096213532,
"grad_norm": 0.8206974267959595,
"learning_rate": 1.5023762568297177e-05,
"loss": 0.01280319094657898,
"step": 33400
},
{
"epoch": 0.6488594040968343,
"grad_norm": 0.12867464125156403,
"learning_rate": 1.5012985893332472e-05,
"loss": 0.01173193097114563,
"step": 33450
},
{
"epoch": 0.6498292985723153,
"grad_norm": 0.21913231909275055,
"learning_rate": 1.5002209218367767e-05,
"loss": 0.02002969741821289,
"step": 33500
},
{
"epoch": 0.6507991930477964,
"grad_norm": 0.003716357983648777,
"learning_rate": 1.499143254340306e-05,
"loss": 0.01119181752204895,
"step": 33550
},
{
"epoch": 0.6517690875232774,
"grad_norm": 0.08030205965042114,
"learning_rate": 1.4980655868438355e-05,
"loss": 0.014008607864379883,
"step": 33600
},
{
"epoch": 0.6527389819987586,
"grad_norm": 3.1721034049987793,
"learning_rate": 1.4969879193473646e-05,
"loss": 0.013823283910751343,
"step": 33650
},
{
"epoch": 0.6537088764742396,
"grad_norm": 0.03734411299228668,
"learning_rate": 1.4959102518508941e-05,
"loss": 0.015704777240753174,
"step": 33700
},
{
"epoch": 0.6546787709497207,
"grad_norm": 0.1846647411584854,
"learning_rate": 1.4948325843544234e-05,
"loss": 0.010865488052368165,
"step": 33750
},
{
"epoch": 0.6556486654252017,
"grad_norm": 0.2678675353527069,
"learning_rate": 1.4937549168579527e-05,
"loss": 0.009887093901634217,
"step": 33800
},
{
"epoch": 0.6566185599006829,
"grad_norm": 0.010403883643448353,
"learning_rate": 1.492677249361482e-05,
"loss": 0.010661989450454712,
"step": 33850
},
{
"epoch": 0.6575884543761639,
"grad_norm": 0.0025241076946258545,
"learning_rate": 1.4915995818650115e-05,
"loss": 0.014609829187393189,
"step": 33900
},
{
"epoch": 0.658558348851645,
"grad_norm": 1.1268956661224365,
"learning_rate": 1.4905219143685407e-05,
"loss": 0.010683060884475707,
"step": 33950
},
{
"epoch": 0.659528243327126,
"grad_norm": 0.017336104065179825,
"learning_rate": 1.4894442468720702e-05,
"loss": 0.014845824241638184,
"step": 34000
},
{
"epoch": 0.660498137802607,
"grad_norm": 0.044322725385427475,
"learning_rate": 1.4883665793755995e-05,
"loss": 0.01597532391548157,
"step": 34050
},
{
"epoch": 0.6614680322780881,
"grad_norm": 0.5430675148963928,
"learning_rate": 1.487288911879129e-05,
"loss": 0.013065664768218995,
"step": 34100
},
{
"epoch": 0.6624379267535692,
"grad_norm": 0.05645228549838066,
"learning_rate": 1.4862112443826585e-05,
"loss": 0.0123112952709198,
"step": 34150
},
{
"epoch": 0.6634078212290503,
"grad_norm": 0.0009722138638608158,
"learning_rate": 1.4851335768861876e-05,
"loss": 0.011547880172729492,
"step": 34200
},
{
"epoch": 0.6643777157045313,
"grad_norm": 0.1503857523202896,
"learning_rate": 1.4840559093897171e-05,
"loss": 0.017723273038864135,
"step": 34250
},
{
"epoch": 0.6653476101800124,
"grad_norm": 0.04904630407691002,
"learning_rate": 1.4829782418932464e-05,
"loss": 0.017695943117141722,
"step": 34300
},
{
"epoch": 0.6663175046554934,
"grad_norm": 1.3607197999954224,
"learning_rate": 1.4819005743967759e-05,
"loss": 0.012956175804138183,
"step": 34350
},
{
"epoch": 0.6672873991309746,
"grad_norm": 2.743605136871338,
"learning_rate": 1.480822906900305e-05,
"loss": 0.013907837867736816,
"step": 34400
},
{
"epoch": 0.6682572936064556,
"grad_norm": 0.043786123394966125,
"learning_rate": 1.4797452394038345e-05,
"loss": 0.012555279731750489,
"step": 34450
},
{
"epoch": 0.6692271880819367,
"grad_norm": 5.655832767486572,
"learning_rate": 1.4786675719073638e-05,
"loss": 0.009299813508987427,
"step": 34500
},
{
"epoch": 0.6701970825574177,
"grad_norm": 0.056478291749954224,
"learning_rate": 1.4775899044108932e-05,
"loss": 0.009523324966430664,
"step": 34550
},
{
"epoch": 0.6711669770328988,
"grad_norm": 0.001088612829335034,
"learning_rate": 1.4765122369144225e-05,
"loss": 0.011521224975585937,
"step": 34600
},
{
"epoch": 0.6721368715083799,
"grad_norm": 0.11827573925256729,
"learning_rate": 1.475434569417952e-05,
"loss": 0.012491198778152466,
"step": 34650
},
{
"epoch": 0.673106765983861,
"grad_norm": 0.1949751079082489,
"learning_rate": 1.4743569019214811e-05,
"loss": 0.011371943950653076,
"step": 34700
},
{
"epoch": 0.674076660459342,
"grad_norm": 0.4207797646522522,
"learning_rate": 1.4732792344250106e-05,
"loss": 0.011109905242919922,
"step": 34750
},
{
"epoch": 0.6750465549348231,
"grad_norm": 0.1514696627855301,
"learning_rate": 1.4722015669285399e-05,
"loss": 0.0088551265001297,
"step": 34800
},
{
"epoch": 0.6760164494103041,
"grad_norm": 0.14414481818675995,
"learning_rate": 1.4711238994320694e-05,
"loss": 0.012893254756927491,
"step": 34850
},
{
"epoch": 0.6769863438857853,
"grad_norm": 0.026235194876790047,
"learning_rate": 1.4700462319355989e-05,
"loss": 0.015163114070892334,
"step": 34900
},
{
"epoch": 0.6779562383612663,
"grad_norm": 0.2076328694820404,
"learning_rate": 1.468968564439128e-05,
"loss": 0.009133043885231017,
"step": 34950
},
{
"epoch": 0.6789261328367474,
"grad_norm": 4.177162170410156,
"learning_rate": 1.4678908969426575e-05,
"loss": 0.015005598068237305,
"step": 35000
},
{
"epoch": 0.6798960273122284,
"grad_norm": 0.8248412609100342,
"learning_rate": 1.4668132294461868e-05,
"loss": 0.013199621438980102,
"step": 35050
},
{
"epoch": 0.6808659217877095,
"grad_norm": 0.10502996295690536,
"learning_rate": 1.4657355619497161e-05,
"loss": 0.016558338403701783,
"step": 35100
},
{
"epoch": 0.6818358162631906,
"grad_norm": 0.14582107961177826,
"learning_rate": 1.4646578944532455e-05,
"loss": 0.010456066131591796,
"step": 35150
},
{
"epoch": 0.6828057107386716,
"grad_norm": 0.5859802961349487,
"learning_rate": 1.463580226956775e-05,
"loss": 0.014589511156082154,
"step": 35200
},
{
"epoch": 0.6837756052141527,
"grad_norm": 0.1268833875656128,
"learning_rate": 1.4625025594603041e-05,
"loss": 0.008093025088310241,
"step": 35250
},
{
"epoch": 0.6847454996896337,
"grad_norm": 0.5912744998931885,
"learning_rate": 1.4614248919638336e-05,
"loss": 0.01298332929611206,
"step": 35300
},
{
"epoch": 0.6857153941651148,
"grad_norm": 0.276562362909317,
"learning_rate": 1.4603472244673629e-05,
"loss": 0.020549101829528807,
"step": 35350
},
{
"epoch": 0.6866852886405959,
"grad_norm": 1.4598941802978516,
"learning_rate": 1.4592695569708924e-05,
"loss": 0.0147046160697937,
"step": 35400
},
{
"epoch": 0.687655183116077,
"grad_norm": 0.2364005148410797,
"learning_rate": 1.4581918894744215e-05,
"loss": 0.021506693363189697,
"step": 35450
},
{
"epoch": 0.688625077591558,
"grad_norm": 0.12074688822031021,
"learning_rate": 1.457114221977951e-05,
"loss": 0.012114797830581664,
"step": 35500
},
{
"epoch": 0.6895949720670391,
"grad_norm": 3.5380935668945312,
"learning_rate": 1.4560365544814805e-05,
"loss": 0.013411445617675781,
"step": 35550
},
{
"epoch": 0.6905648665425201,
"grad_norm": 0.12523096799850464,
"learning_rate": 1.4549588869850098e-05,
"loss": 0.014187207221984863,
"step": 35600
},
{
"epoch": 0.6915347610180013,
"grad_norm": 0.3893623650074005,
"learning_rate": 1.4538812194885393e-05,
"loss": 0.013139295578002929,
"step": 35650
},
{
"epoch": 0.6925046554934823,
"grad_norm": 0.05752887576818466,
"learning_rate": 1.4528035519920684e-05,
"loss": 0.014474309682846069,
"step": 35700
},
{
"epoch": 0.6934745499689634,
"grad_norm": 0.1142883151769638,
"learning_rate": 1.451725884495598e-05,
"loss": 0.012465753555297852,
"step": 35750
},
{
"epoch": 0.6944444444444444,
"grad_norm": 0.0022766422480344772,
"learning_rate": 1.4506482169991272e-05,
"loss": 0.014360055923461915,
"step": 35800
},
{
"epoch": 0.6954143389199255,
"grad_norm": 6.676549911499023,
"learning_rate": 1.4495705495026566e-05,
"loss": 0.01366490125656128,
"step": 35850
},
{
"epoch": 0.6963842333954066,
"grad_norm": 4.237033367156982,
"learning_rate": 1.4484928820061859e-05,
"loss": 0.013136000633239745,
"step": 35900
},
{
"epoch": 0.6973541278708877,
"grad_norm": 0.22524772584438324,
"learning_rate": 1.4474152145097154e-05,
"loss": 0.013059021234512328,
"step": 35950
},
{
"epoch": 0.6983240223463687,
"grad_norm": 0.15251073241233826,
"learning_rate": 1.4463375470132445e-05,
"loss": 0.014076312780380249,
"step": 36000
},
{
"epoch": 0.6992939168218498,
"grad_norm": 0.013344640843570232,
"learning_rate": 1.445259879516774e-05,
"loss": 0.01529719352722168,
"step": 36050
},
{
"epoch": 0.7002638112973308,
"grad_norm": 0.2067405730485916,
"learning_rate": 1.4441822120203033e-05,
"loss": 0.019111555814743043,
"step": 36100
},
{
"epoch": 0.701233705772812,
"grad_norm": 0.23689009249210358,
"learning_rate": 1.4431045445238328e-05,
"loss": 0.014478622674942017,
"step": 36150
},
{
"epoch": 0.702203600248293,
"grad_norm": 0.10999093949794769,
"learning_rate": 1.442026877027362e-05,
"loss": 0.012930406332015991,
"step": 36200
},
{
"epoch": 0.7031734947237741,
"grad_norm": 1.099181890487671,
"learning_rate": 1.4409492095308914e-05,
"loss": 0.01202729344367981,
"step": 36250
},
{
"epoch": 0.7041433891992551,
"grad_norm": 0.3073531687259674,
"learning_rate": 1.4398715420344209e-05,
"loss": 0.011597330570220948,
"step": 36300
},
{
"epoch": 0.7051132836747361,
"grad_norm": 0.018076347187161446,
"learning_rate": 1.4387938745379502e-05,
"loss": 0.012407386302947998,
"step": 36350
},
{
"epoch": 0.7060831781502173,
"grad_norm": 0.0943964421749115,
"learning_rate": 1.4377162070414795e-05,
"loss": 0.015273030996322632,
"step": 36400
},
{
"epoch": 0.7070530726256983,
"grad_norm": 1.5073939561843872,
"learning_rate": 1.4366385395450089e-05,
"loss": 0.011541241407394409,
"step": 36450
},
{
"epoch": 0.7080229671011794,
"grad_norm": 0.2898995578289032,
"learning_rate": 1.4355608720485383e-05,
"loss": 0.012251074314117432,
"step": 36500
},
{
"epoch": 0.7089928615766604,
"grad_norm": 0.1258097141981125,
"learning_rate": 1.4344832045520675e-05,
"loss": 0.015437039136886597,
"step": 36550
},
{
"epoch": 0.7099627560521415,
"grad_norm": 0.950066089630127,
"learning_rate": 1.433405537055597e-05,
"loss": 0.0104659104347229,
"step": 36600
},
{
"epoch": 0.7109326505276226,
"grad_norm": 0.016435345634818077,
"learning_rate": 1.4323278695591263e-05,
"loss": 0.007423868775367737,
"step": 36650
},
{
"epoch": 0.7119025450031037,
"grad_norm": 0.10472806543111801,
"learning_rate": 1.4312502020626558e-05,
"loss": 0.015407694578170776,
"step": 36700
},
{
"epoch": 0.7128724394785847,
"grad_norm": 0.3551085889339447,
"learning_rate": 1.430172534566185e-05,
"loss": 0.015557928085327149,
"step": 36750
},
{
"epoch": 0.7138423339540658,
"grad_norm": 0.006703079212456942,
"learning_rate": 1.4290948670697144e-05,
"loss": 0.010955332517623902,
"step": 36800
},
{
"epoch": 0.7148122284295468,
"grad_norm": 0.2679126560688019,
"learning_rate": 1.4280171995732437e-05,
"loss": 0.015475200414657593,
"step": 36850
},
{
"epoch": 0.715782122905028,
"grad_norm": 0.5011937618255615,
"learning_rate": 1.4269395320767732e-05,
"loss": 0.008676909804344178,
"step": 36900
},
{
"epoch": 0.716752017380509,
"grad_norm": 1.3366972208023071,
"learning_rate": 1.4258618645803024e-05,
"loss": 0.01010279417037964,
"step": 36950
},
{
"epoch": 0.7177219118559901,
"grad_norm": 0.044449131935834885,
"learning_rate": 1.4247841970838318e-05,
"loss": 0.01210085153579712,
"step": 37000
},
{
"epoch": 0.7186918063314711,
"grad_norm": 0.08509023487567902,
"learning_rate": 1.4237065295873613e-05,
"loss": 0.016982359886169432,
"step": 37050
},
{
"epoch": 0.7196617008069522,
"grad_norm": 0.06214037910103798,
"learning_rate": 1.4226288620908906e-05,
"loss": 0.009754824638366699,
"step": 37100
},
{
"epoch": 0.7206315952824333,
"grad_norm": 0.5136099457740784,
"learning_rate": 1.42155119459442e-05,
"loss": 0.015826308727264406,
"step": 37150
},
{
"epoch": 0.7216014897579144,
"grad_norm": 0.01976783573627472,
"learning_rate": 1.4204735270979493e-05,
"loss": 0.01208235502243042,
"step": 37200
},
{
"epoch": 0.7225713842333954,
"grad_norm": 0.23433952033519745,
"learning_rate": 1.4193958596014788e-05,
"loss": 0.013588857650756837,
"step": 37250
},
{
"epoch": 0.7235412787088765,
"grad_norm": 0.021586596965789795,
"learning_rate": 1.4183181921050079e-05,
"loss": 0.00997486650943756,
"step": 37300
},
{
"epoch": 0.7245111731843575,
"grad_norm": 0.07946539670228958,
"learning_rate": 1.4172405246085374e-05,
"loss": 0.016497819423675536,
"step": 37350
},
{
"epoch": 0.7254810676598387,
"grad_norm": 0.004740071017295122,
"learning_rate": 1.4161628571120667e-05,
"loss": 0.010620299577713013,
"step": 37400
},
{
"epoch": 0.7264509621353197,
"grad_norm": 0.0024637789465487003,
"learning_rate": 1.4150851896155962e-05,
"loss": 0.011473933458328247,
"step": 37450
},
{
"epoch": 0.7274208566108007,
"grad_norm": 0.0008725834195502102,
"learning_rate": 1.4140075221191253e-05,
"loss": 0.011359138488769531,
"step": 37500
},
{
"epoch": 0.7283907510862818,
"grad_norm": 1.1745893955230713,
"learning_rate": 1.4129298546226548e-05,
"loss": 0.00870596468448639,
"step": 37550
},
{
"epoch": 0.7293606455617628,
"grad_norm": 0.1606069803237915,
"learning_rate": 1.4118521871261841e-05,
"loss": 0.01373355746269226,
"step": 37600
},
{
"epoch": 0.730330540037244,
"grad_norm": 0.29009297490119934,
"learning_rate": 1.4107745196297136e-05,
"loss": 0.021933481693267823,
"step": 37650
},
{
"epoch": 0.731300434512725,
"grad_norm": 0.022402843460440636,
"learning_rate": 1.4096968521332431e-05,
"loss": 0.014331257343292237,
"step": 37700
},
{
"epoch": 0.7322703289882061,
"grad_norm": 0.10947322845458984,
"learning_rate": 1.4086191846367723e-05,
"loss": 0.018014949560165406,
"step": 37750
},
{
"epoch": 0.7332402234636871,
"grad_norm": 0.2009362429380417,
"learning_rate": 1.4075415171403017e-05,
"loss": 0.010898072719573975,
"step": 37800
},
{
"epoch": 0.7342101179391682,
"grad_norm": 2.548172950744629,
"learning_rate": 1.406463849643831e-05,
"loss": 0.014581743478775024,
"step": 37850
},
{
"epoch": 0.7351800124146493,
"grad_norm": 0.10922932624816895,
"learning_rate": 1.4053861821473604e-05,
"loss": 0.010737295150756837,
"step": 37900
},
{
"epoch": 0.7361499068901304,
"grad_norm": 0.5675318241119385,
"learning_rate": 1.4043085146508897e-05,
"loss": 0.008139224648475647,
"step": 37950
},
{
"epoch": 0.7371198013656114,
"grad_norm": 0.05570976808667183,
"learning_rate": 1.4032308471544192e-05,
"loss": 0.014994860887527465,
"step": 38000
},
{
"epoch": 0.7380896958410925,
"grad_norm": 0.014520283788442612,
"learning_rate": 1.4021531796579483e-05,
"loss": 0.013610669374465943,
"step": 38050
},
{
"epoch": 0.7390595903165735,
"grad_norm": 0.2976853549480438,
"learning_rate": 1.4010755121614778e-05,
"loss": 0.011685310602188111,
"step": 38100
},
{
"epoch": 0.7400294847920547,
"grad_norm": 0.3547203838825226,
"learning_rate": 1.3999978446650071e-05,
"loss": 0.012103561162948608,
"step": 38150
},
{
"epoch": 0.7409993792675357,
"grad_norm": 0.05541533604264259,
"learning_rate": 1.3989201771685366e-05,
"loss": 0.008058667182922363,
"step": 38200
},
{
"epoch": 0.7419692737430168,
"grad_norm": 0.06044553965330124,
"learning_rate": 1.3978425096720658e-05,
"loss": 0.010827337503433227,
"step": 38250
},
{
"epoch": 0.7429391682184978,
"grad_norm": 0.014662938192486763,
"learning_rate": 1.3967648421755952e-05,
"loss": 0.009780243039131165,
"step": 38300
},
{
"epoch": 0.7439090626939789,
"grad_norm": 0.07062532752752304,
"learning_rate": 1.3956871746791246e-05,
"loss": 0.012654238939285278,
"step": 38350
},
{
"epoch": 0.74487895716946,
"grad_norm": 13.988639831542969,
"learning_rate": 1.394609507182654e-05,
"loss": 0.019572280645370484,
"step": 38400
},
{
"epoch": 0.7458488516449411,
"grad_norm": 0.016505658626556396,
"learning_rate": 1.3935318396861834e-05,
"loss": 0.01444116473197937,
"step": 38450
},
{
"epoch": 0.7468187461204221,
"grad_norm": 3.8108930587768555,
"learning_rate": 1.3924541721897127e-05,
"loss": 0.014981470108032226,
"step": 38500
},
{
"epoch": 0.7477886405959032,
"grad_norm": 0.03615260124206543,
"learning_rate": 1.3913765046932422e-05,
"loss": 0.005777102112770081,
"step": 38550
},
{
"epoch": 0.7487585350713842,
"grad_norm": 0.4778033196926117,
"learning_rate": 1.3902988371967713e-05,
"loss": 0.010507502555847169,
"step": 38600
},
{
"epoch": 0.7497284295468652,
"grad_norm": 4.015876293182373,
"learning_rate": 1.3892211697003008e-05,
"loss": 0.013955295085906982,
"step": 38650
},
{
"epoch": 0.7506983240223464,
"grad_norm": 0.028223223984241486,
"learning_rate": 1.3881435022038301e-05,
"loss": 0.011329338550567628,
"step": 38700
},
{
"epoch": 0.7516682184978274,
"grad_norm": 4.328547477722168,
"learning_rate": 1.3870658347073596e-05,
"loss": 0.011585406064987182,
"step": 38750
},
{
"epoch": 0.7526381129733085,
"grad_norm": 0.19761937856674194,
"learning_rate": 1.3859881672108887e-05,
"loss": 0.011942057609558106,
"step": 38800
},
{
"epoch": 0.7536080074487895,
"grad_norm": 0.08271001279354095,
"learning_rate": 1.3849104997144182e-05,
"loss": 0.013433243036270142,
"step": 38850
},
{
"epoch": 0.7545779019242707,
"grad_norm": 0.6135267615318298,
"learning_rate": 1.3838328322179475e-05,
"loss": 0.016457103490829468,
"step": 38900
},
{
"epoch": 0.7555477963997517,
"grad_norm": 0.017893707379698753,
"learning_rate": 1.382755164721477e-05,
"loss": 0.01173388123512268,
"step": 38950
},
{
"epoch": 0.7565176908752328,
"grad_norm": 0.07176350057125092,
"learning_rate": 1.3816774972250062e-05,
"loss": 0.012148175239562988,
"step": 39000
},
{
"epoch": 0.7574875853507138,
"grad_norm": 0.008767183870077133,
"learning_rate": 1.3805998297285357e-05,
"loss": 0.013899903297424316,
"step": 39050
},
{
"epoch": 0.7584574798261949,
"grad_norm": 1.3371655941009521,
"learning_rate": 1.379522162232065e-05,
"loss": 0.02062655210494995,
"step": 39100
},
{
"epoch": 0.759427374301676,
"grad_norm": 0.3648926615715027,
"learning_rate": 1.3784444947355945e-05,
"loss": 0.014823806285858155,
"step": 39150
},
{
"epoch": 0.7603972687771571,
"grad_norm": 0.13768941164016724,
"learning_rate": 1.3773668272391238e-05,
"loss": 0.008001767992973328,
"step": 39200
},
{
"epoch": 0.7613671632526381,
"grad_norm": 0.0176136065274477,
"learning_rate": 1.3762891597426531e-05,
"loss": 0.013494634628295898,
"step": 39250
},
{
"epoch": 0.7623370577281192,
"grad_norm": 0.4217137098312378,
"learning_rate": 1.3752114922461826e-05,
"loss": 0.01678643226623535,
"step": 39300
},
{
"epoch": 0.7633069522036002,
"grad_norm": 0.011708113364875317,
"learning_rate": 1.3741338247497117e-05,
"loss": 0.013513956069946289,
"step": 39350
},
{
"epoch": 0.7642768466790814,
"grad_norm": 0.18896573781967163,
"learning_rate": 1.3730561572532412e-05,
"loss": 0.012324445247650147,
"step": 39400
},
{
"epoch": 0.7652467411545624,
"grad_norm": 0.05553916469216347,
"learning_rate": 1.3719784897567705e-05,
"loss": 0.010808863639831544,
"step": 39450
},
{
"epoch": 0.7662166356300435,
"grad_norm": 0.03677207604050636,
"learning_rate": 1.3709008222603e-05,
"loss": 0.012216088771820068,
"step": 39500
},
{
"epoch": 0.7671865301055245,
"grad_norm": 0.19025000929832458,
"learning_rate": 1.3698231547638292e-05,
"loss": 0.01643786072731018,
"step": 39550
},
{
"epoch": 0.7681564245810056,
"grad_norm": 0.10397391021251678,
"learning_rate": 1.3687454872673586e-05,
"loss": 0.008824735283851623,
"step": 39600
},
{
"epoch": 0.7691263190564867,
"grad_norm": 2.047147035598755,
"learning_rate": 1.367667819770888e-05,
"loss": 0.013945077657699584,
"step": 39650
},
{
"epoch": 0.7700962135319678,
"grad_norm": 0.09775671362876892,
"learning_rate": 1.3665901522744174e-05,
"loss": 0.015231665372848511,
"step": 39700
},
{
"epoch": 0.7710661080074488,
"grad_norm": 3.5340893268585205,
"learning_rate": 1.3655124847779466e-05,
"loss": 0.008505199551582337,
"step": 39750
},
{
"epoch": 0.7720360024829298,
"grad_norm": 1.7976154088974,
"learning_rate": 1.364434817281476e-05,
"loss": 0.01221422791481018,
"step": 39800
},
{
"epoch": 0.7730058969584109,
"grad_norm": 0.22356222569942474,
"learning_rate": 1.3633571497850056e-05,
"loss": 0.011724320650100708,
"step": 39850
},
{
"epoch": 0.773975791433892,
"grad_norm": 0.15518562495708466,
"learning_rate": 1.3622794822885347e-05,
"loss": 0.010762434005737304,
"step": 39900
},
{
"epoch": 0.7749456859093731,
"grad_norm": 0.06828810274600983,
"learning_rate": 1.3612018147920642e-05,
"loss": 0.016326544284820558,
"step": 39950
},
{
"epoch": 0.7759155803848541,
"grad_norm": 0.0030448208563029766,
"learning_rate": 1.3601241472955935e-05,
"loss": 0.007850555181503295,
"step": 40000
},
{
"epoch": 0.7768854748603352,
"grad_norm": 0.016075054183602333,
"learning_rate": 1.359046479799123e-05,
"loss": 0.012592874765396118,
"step": 40050
},
{
"epoch": 0.7778553693358162,
"grad_norm": 0.23300747573375702,
"learning_rate": 1.3579688123026521e-05,
"loss": 0.015246310234069825,
"step": 40100
},
{
"epoch": 0.7788252638112974,
"grad_norm": 0.8599169254302979,
"learning_rate": 1.3568911448061816e-05,
"loss": 0.015351239442825317,
"step": 40150
},
{
"epoch": 0.7797951582867784,
"grad_norm": 0.06253393739461899,
"learning_rate": 1.355813477309711e-05,
"loss": 0.01192333221435547,
"step": 40200
},
{
"epoch": 0.7807650527622595,
"grad_norm": 0.6460727453231812,
"learning_rate": 1.3547358098132404e-05,
"loss": 0.01222865104675293,
"step": 40250
},
{
"epoch": 0.7817349472377405,
"grad_norm": 0.09370553493499756,
"learning_rate": 1.3536581423167696e-05,
"loss": 0.015481972694396972,
"step": 40300
},
{
"epoch": 0.7827048417132216,
"grad_norm": 0.06581387668848038,
"learning_rate": 1.352580474820299e-05,
"loss": 0.01683463454246521,
"step": 40350
},
{
"epoch": 0.7836747361887026,
"grad_norm": 1.4090982675552368,
"learning_rate": 1.3515028073238284e-05,
"loss": 0.01102339506149292,
"step": 40400
},
{
"epoch": 0.7846446306641838,
"grad_norm": 0.1275385171175003,
"learning_rate": 1.3504251398273579e-05,
"loss": 0.01333919882774353,
"step": 40450
},
{
"epoch": 0.7856145251396648,
"grad_norm": 0.1472180187702179,
"learning_rate": 1.349347472330887e-05,
"loss": 0.01098235249519348,
"step": 40500
},
{
"epoch": 0.7865844196151459,
"grad_norm": 1.9831448793411255,
"learning_rate": 1.3482698048344165e-05,
"loss": 0.016803568601608275,
"step": 40550
},
{
"epoch": 0.7875543140906269,
"grad_norm": 1.6919479370117188,
"learning_rate": 1.347192137337946e-05,
"loss": 0.015242969989776612,
"step": 40600
},
{
"epoch": 0.788524208566108,
"grad_norm": 2.0834360122680664,
"learning_rate": 1.3461144698414751e-05,
"loss": 0.015859192609786986,
"step": 40650
},
{
"epoch": 0.7894941030415891,
"grad_norm": 0.5180131196975708,
"learning_rate": 1.3450368023450046e-05,
"loss": 0.012452244758605957,
"step": 40700
},
{
"epoch": 0.7904639975170702,
"grad_norm": 1.7989636659622192,
"learning_rate": 1.343959134848534e-05,
"loss": 0.014033384323120117,
"step": 40750
},
{
"epoch": 0.7914338919925512,
"grad_norm": 0.3269071578979492,
"learning_rate": 1.3428814673520634e-05,
"loss": 0.009815393090248108,
"step": 40800
},
{
"epoch": 0.7924037864680322,
"grad_norm": 0.10110092163085938,
"learning_rate": 1.3418037998555926e-05,
"loss": 0.013134450912475585,
"step": 40850
},
{
"epoch": 0.7933736809435133,
"grad_norm": 0.24154163897037506,
"learning_rate": 1.340726132359122e-05,
"loss": 0.015682837963104247,
"step": 40900
},
{
"epoch": 0.7943435754189944,
"grad_norm": 0.004043806344270706,
"learning_rate": 1.3396484648626514e-05,
"loss": 0.01079138159751892,
"step": 40950
},
{
"epoch": 0.7953134698944755,
"grad_norm": 0.09349637478590012,
"learning_rate": 1.3385707973661808e-05,
"loss": 0.009041820764541625,
"step": 41000
},
{
"epoch": 0.7962833643699565,
"grad_norm": 0.05167528614401817,
"learning_rate": 1.33749312986971e-05,
"loss": 0.013015373945236205,
"step": 41050
},
{
"epoch": 0.7972532588454376,
"grad_norm": 0.08983754366636276,
"learning_rate": 1.3364154623732395e-05,
"loss": 0.008832918405532837,
"step": 41100
},
{
"epoch": 0.7982231533209186,
"grad_norm": 0.023780521005392075,
"learning_rate": 1.3353377948767688e-05,
"loss": 0.005831308364868164,
"step": 41150
},
{
"epoch": 0.7991930477963998,
"grad_norm": 0.01728859171271324,
"learning_rate": 1.3342601273802983e-05,
"loss": 0.008154717683792114,
"step": 41200
},
{
"epoch": 0.8001629422718808,
"grad_norm": 0.015846790745854378,
"learning_rate": 1.3331824598838276e-05,
"loss": 0.015815972089767455,
"step": 41250
},
{
"epoch": 0.8011328367473619,
"grad_norm": 1.6027042865753174,
"learning_rate": 1.3321047923873569e-05,
"loss": 0.01756986141204834,
"step": 41300
},
{
"epoch": 0.8021027312228429,
"grad_norm": 0.014624303206801414,
"learning_rate": 1.3310271248908864e-05,
"loss": 0.015521323680877686,
"step": 41350
},
{
"epoch": 0.803072625698324,
"grad_norm": 0.6044580340385437,
"learning_rate": 1.3299494573944155e-05,
"loss": 0.01573086380958557,
"step": 41400
},
{
"epoch": 0.8040425201738051,
"grad_norm": 0.11641865223646164,
"learning_rate": 1.328871789897945e-05,
"loss": 0.01016334891319275,
"step": 41450
},
{
"epoch": 0.8050124146492862,
"grad_norm": 0.02770138345658779,
"learning_rate": 1.3277941224014743e-05,
"loss": 0.012869092226028443,
"step": 41500
},
{
"epoch": 0.8059823091247672,
"grad_norm": 0.33884403109550476,
"learning_rate": 1.3267164549050038e-05,
"loss": 0.018186157941818236,
"step": 41550
},
{
"epoch": 0.8069522036002483,
"grad_norm": 0.7148659229278564,
"learning_rate": 1.325638787408533e-05,
"loss": 0.01030908226966858,
"step": 41600
},
{
"epoch": 0.8079220980757293,
"grad_norm": 1.3514515161514282,
"learning_rate": 1.3245611199120625e-05,
"loss": 0.011520078182220459,
"step": 41650
},
{
"epoch": 0.8088919925512105,
"grad_norm": 0.23971155285835266,
"learning_rate": 1.3234834524155918e-05,
"loss": 0.011489888429641723,
"step": 41700
},
{
"epoch": 0.8098618870266915,
"grad_norm": 0.10502474009990692,
"learning_rate": 1.3224057849191213e-05,
"loss": 0.015566623210906983,
"step": 41750
},
{
"epoch": 0.8108317815021726,
"grad_norm": 0.19168072938919067,
"learning_rate": 1.3213281174226504e-05,
"loss": 0.013079316616058349,
"step": 41800
},
{
"epoch": 0.8118016759776536,
"grad_norm": 0.0014026749413460493,
"learning_rate": 1.3202504499261799e-05,
"loss": 0.010559054613113404,
"step": 41850
},
{
"epoch": 0.8127715704531348,
"grad_norm": 0.02225392684340477,
"learning_rate": 1.3191727824297092e-05,
"loss": 0.014253293275833129,
"step": 41900
},
{
"epoch": 0.8137414649286158,
"grad_norm": 0.3816450238227844,
"learning_rate": 1.3180951149332385e-05,
"loss": 0.013925464153289794,
"step": 41950
},
{
"epoch": 0.8147113594040968,
"grad_norm": 0.5388890504837036,
"learning_rate": 1.317017447436768e-05,
"loss": 0.01136979341506958,
"step": 42000
},
{
"epoch": 0.8156812538795779,
"grad_norm": 0.13996274769306183,
"learning_rate": 1.3159397799402973e-05,
"loss": 0.014642359018325805,
"step": 42050
},
{
"epoch": 0.8166511483550589,
"grad_norm": 0.15901131927967072,
"learning_rate": 1.3148621124438268e-05,
"loss": 0.01507319688796997,
"step": 42100
},
{
"epoch": 0.81762104283054,
"grad_norm": 0.024100415408611298,
"learning_rate": 1.313784444947356e-05,
"loss": 0.009771895408630372,
"step": 42150
},
{
"epoch": 0.8185909373060211,
"grad_norm": 0.022184448316693306,
"learning_rate": 1.3127067774508854e-05,
"loss": 0.010653589963912963,
"step": 42200
},
{
"epoch": 0.8195608317815022,
"grad_norm": 0.036993805319070816,
"learning_rate": 1.3116291099544148e-05,
"loss": 0.01122673749923706,
"step": 42250
},
{
"epoch": 0.8205307262569832,
"grad_norm": 2.7773025035858154,
"learning_rate": 1.3105514424579442e-05,
"loss": 0.010721347332000732,
"step": 42300
},
{
"epoch": 0.8215006207324643,
"grad_norm": 0.3182041645050049,
"learning_rate": 1.3094737749614734e-05,
"loss": 0.01731890320777893,
"step": 42350
},
{
"epoch": 0.8224705152079453,
"grad_norm": 0.08783317357301712,
"learning_rate": 1.3083961074650029e-05,
"loss": 0.008271402716636657,
"step": 42400
},
{
"epoch": 0.8234404096834265,
"grad_norm": 0.04280201718211174,
"learning_rate": 1.3073184399685322e-05,
"loss": 0.01800162434577942,
"step": 42450
},
{
"epoch": 0.8244103041589075,
"grad_norm": 0.07678304612636566,
"learning_rate": 1.3062407724720617e-05,
"loss": 0.008258128762245178,
"step": 42500
},
{
"epoch": 0.8253801986343886,
"grad_norm": 0.043400838971138,
"learning_rate": 1.3051631049755908e-05,
"loss": 0.012848950624465942,
"step": 42550
},
{
"epoch": 0.8263500931098696,
"grad_norm": 0.013833164237439632,
"learning_rate": 1.3040854374791203e-05,
"loss": 0.0122756028175354,
"step": 42600
},
{
"epoch": 0.8273199875853507,
"grad_norm": 0.09889528155326843,
"learning_rate": 1.3030077699826496e-05,
"loss": 0.01705217480659485,
"step": 42650
},
{
"epoch": 0.8282898820608318,
"grad_norm": 0.06928832828998566,
"learning_rate": 1.301930102486179e-05,
"loss": 0.016627423763275147,
"step": 42700
},
{
"epoch": 0.8292597765363129,
"grad_norm": 0.21198605000972748,
"learning_rate": 1.3008524349897084e-05,
"loss": 0.008090019226074219,
"step": 42750
},
{
"epoch": 0.8302296710117939,
"grad_norm": 1.5199859142303467,
"learning_rate": 1.2997747674932377e-05,
"loss": 0.014316335916519166,
"step": 42800
},
{
"epoch": 0.831199565487275,
"grad_norm": 0.0984615907073021,
"learning_rate": 1.2986970999967672e-05,
"loss": 0.012811027765274048,
"step": 42850
},
{
"epoch": 0.832169459962756,
"grad_norm": 3.242669105529785,
"learning_rate": 1.2976194325002964e-05,
"loss": 0.01454485297203064,
"step": 42900
},
{
"epoch": 0.8331393544382372,
"grad_norm": 0.10325007140636444,
"learning_rate": 1.2965417650038259e-05,
"loss": 0.015784757137298586,
"step": 42950
},
{
"epoch": 0.8341092489137182,
"grad_norm": 0.11286969482898712,
"learning_rate": 1.2954640975073552e-05,
"loss": 0.009359683990478516,
"step": 43000
},
{
"epoch": 0.8350791433891993,
"grad_norm": 0.41477829217910767,
"learning_rate": 1.2943864300108847e-05,
"loss": 0.011049963235855102,
"step": 43050
},
{
"epoch": 0.8360490378646803,
"grad_norm": 0.3554072380065918,
"learning_rate": 1.2933087625144138e-05,
"loss": 0.019100894927978517,
"step": 43100
},
{
"epoch": 0.8370189323401613,
"grad_norm": 0.18765778839588165,
"learning_rate": 1.2922310950179433e-05,
"loss": 0.011800243854522705,
"step": 43150
},
{
"epoch": 0.8379888268156425,
"grad_norm": 0.019295310601592064,
"learning_rate": 1.2911534275214726e-05,
"loss": 0.011093273162841796,
"step": 43200
},
{
"epoch": 0.8389587212911235,
"grad_norm": 0.04280606657266617,
"learning_rate": 1.290075760025002e-05,
"loss": 0.008412720561027527,
"step": 43250
},
{
"epoch": 0.8399286157666046,
"grad_norm": 0.0026010742876678705,
"learning_rate": 1.2889980925285312e-05,
"loss": 0.014492534399032593,
"step": 43300
},
{
"epoch": 0.8408985102420856,
"grad_norm": 0.059462808072566986,
"learning_rate": 1.2879204250320607e-05,
"loss": 0.011947336196899415,
"step": 43350
},
{
"epoch": 0.8418684047175667,
"grad_norm": 0.1917988657951355,
"learning_rate": 1.2868427575355902e-05,
"loss": 0.016186723709106444,
"step": 43400
},
{
"epoch": 0.8428382991930478,
"grad_norm": 0.07792427390813828,
"learning_rate": 1.2857650900391194e-05,
"loss": 0.015511668920516967,
"step": 43450
},
{
"epoch": 0.8438081936685289,
"grad_norm": 0.0658286064863205,
"learning_rate": 1.2846874225426488e-05,
"loss": 0.012388651371002197,
"step": 43500
},
{
"epoch": 0.8447780881440099,
"grad_norm": 0.023095980286598206,
"learning_rate": 1.2836097550461782e-05,
"loss": 0.011878821849822998,
"step": 43550
},
{
"epoch": 0.845747982619491,
"grad_norm": 0.012894881889224052,
"learning_rate": 1.2825320875497076e-05,
"loss": 0.014157233238220214,
"step": 43600
},
{
"epoch": 0.846717877094972,
"grad_norm": 0.9882586002349854,
"learning_rate": 1.2814544200532368e-05,
"loss": 0.011661651134490967,
"step": 43650
},
{
"epoch": 0.8476877715704532,
"grad_norm": 0.8191036581993103,
"learning_rate": 1.2803767525567663e-05,
"loss": 0.011341378688812256,
"step": 43700
},
{
"epoch": 0.8486576660459342,
"grad_norm": 0.08110763877630234,
"learning_rate": 1.2792990850602956e-05,
"loss": 0.014700998067855835,
"step": 43750
},
{
"epoch": 0.8496275605214153,
"grad_norm": 0.03943950682878494,
"learning_rate": 1.278221417563825e-05,
"loss": 0.01113856554031372,
"step": 43800
},
{
"epoch": 0.8505974549968963,
"grad_norm": 0.157288059592247,
"learning_rate": 1.2771437500673542e-05,
"loss": 0.010624353885650634,
"step": 43850
},
{
"epoch": 0.8515673494723774,
"grad_norm": 0.510759174823761,
"learning_rate": 1.2760660825708837e-05,
"loss": 0.013242442607879639,
"step": 43900
},
{
"epoch": 0.8525372439478585,
"grad_norm": 6.026455879211426,
"learning_rate": 1.274988415074413e-05,
"loss": 0.017618075609207154,
"step": 43950
},
{
"epoch": 0.8535071384233396,
"grad_norm": 2.444929361343384,
"learning_rate": 1.2739107475779423e-05,
"loss": 0.011918938159942627,
"step": 44000
},
{
"epoch": 0.8544770328988206,
"grad_norm": 0.4202197194099426,
"learning_rate": 1.2728330800814717e-05,
"loss": 0.014063483476638794,
"step": 44050
},
{
"epoch": 0.8554469273743017,
"grad_norm": 0.857700526714325,
"learning_rate": 1.2717554125850011e-05,
"loss": 0.014173794984817505,
"step": 44100
},
{
"epoch": 0.8564168218497827,
"grad_norm": 0.0014903662959113717,
"learning_rate": 1.2706777450885306e-05,
"loss": 0.009822830557823181,
"step": 44150
},
{
"epoch": 0.8573867163252639,
"grad_norm": 0.05318353325128555,
"learning_rate": 1.2696000775920598e-05,
"loss": 0.015525661706924439,
"step": 44200
},
{
"epoch": 0.8583566108007449,
"grad_norm": 0.015561738051474094,
"learning_rate": 1.2685224100955893e-05,
"loss": 0.015870593786239624,
"step": 44250
},
{
"epoch": 0.8593265052762259,
"grad_norm": 0.0346713587641716,
"learning_rate": 1.2674447425991186e-05,
"loss": 0.010679446458816529,
"step": 44300
},
{
"epoch": 0.860296399751707,
"grad_norm": 0.07912234961986542,
"learning_rate": 1.266367075102648e-05,
"loss": 0.010605388879776,
"step": 44350
},
{
"epoch": 0.861266294227188,
"grad_norm": 0.278957724571228,
"learning_rate": 1.2652894076061772e-05,
"loss": 0.009512117505073548,
"step": 44400
},
{
"epoch": 0.8622361887026692,
"grad_norm": 0.4787423014640808,
"learning_rate": 1.2642117401097067e-05,
"loss": 0.010703068971633912,
"step": 44450
},
{
"epoch": 0.8632060831781502,
"grad_norm": 0.936195433139801,
"learning_rate": 1.263134072613236e-05,
"loss": 0.01435288667678833,
"step": 44500
},
{
"epoch": 0.8641759776536313,
"grad_norm": 0.005316342692822218,
"learning_rate": 1.2620564051167655e-05,
"loss": 0.010985485315322875,
"step": 44550
},
{
"epoch": 0.8651458721291123,
"grad_norm": 0.05086016282439232,
"learning_rate": 1.2609787376202946e-05,
"loss": 0.009080511331558228,
"step": 44600
},
{
"epoch": 0.8661157666045934,
"grad_norm": 0.011622115969657898,
"learning_rate": 1.2599010701238241e-05,
"loss": 0.01460861086845398,
"step": 44650
},
{
"epoch": 0.8670856610800745,
"grad_norm": 0.11308765411376953,
"learning_rate": 1.2588234026273534e-05,
"loss": 0.014991286993026733,
"step": 44700
},
{
"epoch": 0.8680555555555556,
"grad_norm": 0.002706480911001563,
"learning_rate": 1.2577457351308828e-05,
"loss": 0.011885005235671996,
"step": 44750
},
{
"epoch": 0.8690254500310366,
"grad_norm": 0.011361288838088512,
"learning_rate": 1.2566680676344122e-05,
"loss": 0.009441599845886231,
"step": 44800
},
{
"epoch": 0.8699953445065177,
"grad_norm": 0.26590201258659363,
"learning_rate": 1.2555904001379416e-05,
"loss": 0.020054712295532226,
"step": 44850
},
{
"epoch": 0.8709652389819987,
"grad_norm": 0.07085949182510376,
"learning_rate": 1.254512732641471e-05,
"loss": 0.011335647106170655,
"step": 44900
},
{
"epoch": 0.8719351334574799,
"grad_norm": 0.049661703407764435,
"learning_rate": 1.2534350651450002e-05,
"loss": 0.01155774474143982,
"step": 44950
},
{
"epoch": 0.8729050279329609,
"grad_norm": 0.0034034871496260166,
"learning_rate": 1.2523573976485297e-05,
"loss": 0.01653545379638672,
"step": 45000
},
{
"epoch": 0.873874922408442,
"grad_norm": 0.19574588537216187,
"learning_rate": 1.251279730152059e-05,
"loss": 0.011356918811798096,
"step": 45050
},
{
"epoch": 0.874844816883923,
"grad_norm": 0.013320011086761951,
"learning_rate": 1.2502020626555885e-05,
"loss": 0.010129574537277222,
"step": 45100
},
{
"epoch": 0.8758147113594041,
"grad_norm": 0.004861229099333286,
"learning_rate": 1.2491243951591176e-05,
"loss": 0.008824251890182495,
"step": 45150
},
{
"epoch": 0.8767846058348852,
"grad_norm": 0.08096335083246231,
"learning_rate": 1.2480467276626471e-05,
"loss": 0.008589981198310852,
"step": 45200
},
{
"epoch": 0.8777545003103663,
"grad_norm": 0.2224770337343216,
"learning_rate": 1.2469690601661764e-05,
"loss": 0.013754096031188965,
"step": 45250
},
{
"epoch": 0.8787243947858473,
"grad_norm": 0.15572308003902435,
"learning_rate": 1.2458913926697057e-05,
"loss": 0.00870846450328827,
"step": 45300
},
{
"epoch": 0.8796942892613284,
"grad_norm": 3.400616407394409,
"learning_rate": 1.244813725173235e-05,
"loss": 0.012854855060577392,
"step": 45350
},
{
"epoch": 0.8806641837368094,
"grad_norm": 0.10432376712560654,
"learning_rate": 1.2437360576767645e-05,
"loss": 0.017504284381866454,
"step": 45400
},
{
"epoch": 0.8816340782122905,
"grad_norm": 0.2897157073020935,
"learning_rate": 1.2426583901802937e-05,
"loss": 0.012574498653411864,
"step": 45450
},
{
"epoch": 0.8826039726877716,
"grad_norm": 1.3505805730819702,
"learning_rate": 1.2415807226838232e-05,
"loss": 0.012126814126968383,
"step": 45500
},
{
"epoch": 0.8835738671632526,
"grad_norm": 0.04246855527162552,
"learning_rate": 1.2405030551873527e-05,
"loss": 0.005932779908180237,
"step": 45550
},
{
"epoch": 0.8845437616387337,
"grad_norm": 0.38173991441726685,
"learning_rate": 1.239425387690882e-05,
"loss": 0.012164273262023927,
"step": 45600
},
{
"epoch": 0.8855136561142147,
"grad_norm": 2.185842752456665,
"learning_rate": 1.2383477201944115e-05,
"loss": 0.013052973747253418,
"step": 45650
},
{
"epoch": 0.8864835505896959,
"grad_norm": 0.15116983652114868,
"learning_rate": 1.2372700526979406e-05,
"loss": 0.011425377130508422,
"step": 45700
},
{
"epoch": 0.8874534450651769,
"grad_norm": 3.008242130279541,
"learning_rate": 1.2361923852014701e-05,
"loss": 0.011115719079971314,
"step": 45750
},
{
"epoch": 0.888423339540658,
"grad_norm": 0.1853034347295761,
"learning_rate": 1.2351147177049994e-05,
"loss": 0.014913790225982666,
"step": 45800
},
{
"epoch": 0.889393234016139,
"grad_norm": 0.30181679129600525,
"learning_rate": 1.2340370502085289e-05,
"loss": 0.018341716527938843,
"step": 45850
},
{
"epoch": 0.8903631284916201,
"grad_norm": 0.008464576676487923,
"learning_rate": 1.232959382712058e-05,
"loss": 0.012310686111450196,
"step": 45900
},
{
"epoch": 0.8913330229671012,
"grad_norm": 0.0047547719441354275,
"learning_rate": 1.2318817152155875e-05,
"loss": 0.008971402049064636,
"step": 45950
},
{
"epoch": 0.8923029174425823,
"grad_norm": 0.2215203195810318,
"learning_rate": 1.2308040477191168e-05,
"loss": 0.01199519157409668,
"step": 46000
},
{
"epoch": 0.8932728119180633,
"grad_norm": 0.11978214979171753,
"learning_rate": 1.2297263802226462e-05,
"loss": 0.010180273056030274,
"step": 46050
},
{
"epoch": 0.8942427063935444,
"grad_norm": 0.21645428240299225,
"learning_rate": 1.2286487127261755e-05,
"loss": 0.010190773010253906,
"step": 46100
},
{
"epoch": 0.8952126008690254,
"grad_norm": 0.1091199517250061,
"learning_rate": 1.227571045229705e-05,
"loss": 0.010859193801879883,
"step": 46150
},
{
"epoch": 0.8961824953445066,
"grad_norm": 0.20174814760684967,
"learning_rate": 1.2264933777332341e-05,
"loss": 0.01193704605102539,
"step": 46200
},
{
"epoch": 0.8971523898199876,
"grad_norm": 0.06499738991260529,
"learning_rate": 1.2254157102367636e-05,
"loss": 0.018965041637420653,
"step": 46250
},
{
"epoch": 0.8981222842954687,
"grad_norm": 0.05170740187168121,
"learning_rate": 1.224338042740293e-05,
"loss": 0.01314793825149536,
"step": 46300
},
{
"epoch": 0.8990921787709497,
"grad_norm": 0.6977347731590271,
"learning_rate": 1.2232603752438224e-05,
"loss": 0.012125167846679687,
"step": 46350
},
{
"epoch": 0.9000620732464308,
"grad_norm": 0.1110471859574318,
"learning_rate": 1.2221827077473519e-05,
"loss": 0.010592982769012452,
"step": 46400
},
{
"epoch": 0.9010319677219119,
"grad_norm": 0.43168240785598755,
"learning_rate": 1.221105040250881e-05,
"loss": 0.014600886106491089,
"step": 46450
},
{
"epoch": 0.902001862197393,
"grad_norm": 0.08322225511074066,
"learning_rate": 1.2200273727544105e-05,
"loss": 0.013785034418106079,
"step": 46500
},
{
"epoch": 0.902971756672874,
"grad_norm": 0.0700334757566452,
"learning_rate": 1.2189497052579398e-05,
"loss": 0.012823190689086914,
"step": 46550
},
{
"epoch": 0.903941651148355,
"grad_norm": 0.3181156516075134,
"learning_rate": 1.2178720377614691e-05,
"loss": 0.01680142045021057,
"step": 46600
},
{
"epoch": 0.9049115456238361,
"grad_norm": 0.030726848170161247,
"learning_rate": 1.2167943702649985e-05,
"loss": 0.012376610040664673,
"step": 46650
},
{
"epoch": 0.9058814400993171,
"grad_norm": 0.017003104090690613,
"learning_rate": 1.215716702768528e-05,
"loss": 0.014558795690536499,
"step": 46700
},
{
"epoch": 0.9068513345747983,
"grad_norm": 0.3867703676223755,
"learning_rate": 1.2146390352720571e-05,
"loss": 0.010922759771347046,
"step": 46750
},
{
"epoch": 0.9078212290502793,
"grad_norm": 0.07643905282020569,
"learning_rate": 1.2135613677755866e-05,
"loss": 0.012134585380554199,
"step": 46800
},
{
"epoch": 0.9087911235257604,
"grad_norm": 0.10682686418294907,
"learning_rate": 1.2124837002791159e-05,
"loss": 0.010475543737411499,
"step": 46850
},
{
"epoch": 0.9097610180012414,
"grad_norm": 0.4514603316783905,
"learning_rate": 1.2114060327826454e-05,
"loss": 0.008329501152038574,
"step": 46900
},
{
"epoch": 0.9107309124767226,
"grad_norm": 1.0294262170791626,
"learning_rate": 1.2103283652861749e-05,
"loss": 0.013690569400787354,
"step": 46950
},
{
"epoch": 0.9117008069522036,
"grad_norm": 0.03038082830607891,
"learning_rate": 1.209250697789704e-05,
"loss": 0.0099516499042511,
"step": 47000
},
{
"epoch": 0.9126707014276847,
"grad_norm": 0.09731799364089966,
"learning_rate": 1.2081730302932335e-05,
"loss": 0.012321420907974244,
"step": 47050
},
{
"epoch": 0.9136405959031657,
"grad_norm": 0.6038825511932373,
"learning_rate": 1.2070953627967628e-05,
"loss": 0.009519893527030945,
"step": 47100
},
{
"epoch": 0.9146104903786468,
"grad_norm": 0.5427056550979614,
"learning_rate": 1.2060176953002923e-05,
"loss": 0.01609891414642334,
"step": 47150
},
{
"epoch": 0.9155803848541278,
"grad_norm": 0.01235037948936224,
"learning_rate": 1.2049400278038214e-05,
"loss": 0.008343427181243897,
"step": 47200
},
{
"epoch": 0.916550279329609,
"grad_norm": 0.045149948447942734,
"learning_rate": 1.203862360307351e-05,
"loss": 0.0128147292137146,
"step": 47250
},
{
"epoch": 0.91752017380509,
"grad_norm": 0.002297518542036414,
"learning_rate": 1.2027846928108802e-05,
"loss": 0.008525106906890869,
"step": 47300
},
{
"epoch": 0.9184900682805711,
"grad_norm": 0.5789989233016968,
"learning_rate": 1.2017070253144096e-05,
"loss": 0.021048855781555176,
"step": 47350
},
{
"epoch": 0.9194599627560521,
"grad_norm": 0.19051803648471832,
"learning_rate": 1.2006293578179389e-05,
"loss": 0.01228331208229065,
"step": 47400
},
{
"epoch": 0.9204298572315333,
"grad_norm": 0.1684299260377884,
"learning_rate": 1.1995516903214684e-05,
"loss": 0.009318812489509583,
"step": 47450
},
{
"epoch": 0.9213997517070143,
"grad_norm": 0.007992559112608433,
"learning_rate": 1.1984740228249975e-05,
"loss": 0.014864132404327393,
"step": 47500
},
{
"epoch": 0.9223696461824954,
"grad_norm": 0.08090420067310333,
"learning_rate": 1.197396355328527e-05,
"loss": 0.016050975322723388,
"step": 47550
},
{
"epoch": 0.9233395406579764,
"grad_norm": 0.14604642987251282,
"learning_rate": 1.1963186878320563e-05,
"loss": 0.01075518250465393,
"step": 47600
},
{
"epoch": 0.9243094351334574,
"grad_norm": 0.047446660697460175,
"learning_rate": 1.1952410203355858e-05,
"loss": 0.00794364869594574,
"step": 47650
},
{
"epoch": 0.9252793296089385,
"grad_norm": 0.0555587075650692,
"learning_rate": 1.1941633528391153e-05,
"loss": 0.014623422622680664,
"step": 47700
},
{
"epoch": 0.9262492240844196,
"grad_norm": 0.4052380621433258,
"learning_rate": 1.1930856853426444e-05,
"loss": 0.016972141265869142,
"step": 47750
},
{
"epoch": 0.9272191185599007,
"grad_norm": 0.9645226001739502,
"learning_rate": 1.192008017846174e-05,
"loss": 0.00830783486366272,
"step": 47800
},
{
"epoch": 0.9281890130353817,
"grad_norm": 0.2091115117073059,
"learning_rate": 1.1909303503497032e-05,
"loss": 0.015872354507446288,
"step": 47850
},
{
"epoch": 0.9291589075108628,
"grad_norm": 0.006471001543104649,
"learning_rate": 1.1898526828532325e-05,
"loss": 0.006364143490791321,
"step": 47900
},
{
"epoch": 0.9301288019863438,
"grad_norm": 0.14041072130203247,
"learning_rate": 1.1887750153567619e-05,
"loss": 0.017964128255844116,
"step": 47950
},
{
"epoch": 0.931098696461825,
"grad_norm": 0.16735051572322845,
"learning_rate": 1.1876973478602913e-05,
"loss": 0.015826584100723268,
"step": 48000
},
{
"epoch": 0.932068590937306,
"grad_norm": 0.19653594493865967,
"learning_rate": 1.1866196803638207e-05,
"loss": 0.012465453147888184,
"step": 48050
},
{
"epoch": 0.9330384854127871,
"grad_norm": 1.7739659547805786,
"learning_rate": 1.18554201286735e-05,
"loss": 0.012205030918121338,
"step": 48100
},
{
"epoch": 0.9340083798882681,
"grad_norm": 0.6405571103096008,
"learning_rate": 1.1844643453708793e-05,
"loss": 0.01083716630935669,
"step": 48150
},
{
"epoch": 0.9349782743637493,
"grad_norm": 0.2030767947435379,
"learning_rate": 1.1833866778744088e-05,
"loss": 0.01170374870300293,
"step": 48200
},
{
"epoch": 0.9359481688392303,
"grad_norm": 0.3332473039627075,
"learning_rate": 1.182309010377938e-05,
"loss": 0.010207765102386475,
"step": 48250
},
{
"epoch": 0.9369180633147114,
"grad_norm": 0.20772817730903625,
"learning_rate": 1.1812313428814674e-05,
"loss": 0.01414027452468872,
"step": 48300
},
{
"epoch": 0.9378879577901924,
"grad_norm": 0.08560888469219208,
"learning_rate": 1.1801536753849967e-05,
"loss": 0.008398468494415284,
"step": 48350
},
{
"epoch": 0.9388578522656735,
"grad_norm": 0.1442621499300003,
"learning_rate": 1.1790760078885262e-05,
"loss": 0.007548823356628418,
"step": 48400
},
{
"epoch": 0.9398277467411545,
"grad_norm": 2.340155839920044,
"learning_rate": 1.1779983403920557e-05,
"loss": 0.009548107385635376,
"step": 48450
},
{
"epoch": 0.9407976412166357,
"grad_norm": 0.014150552451610565,
"learning_rate": 1.1769206728955848e-05,
"loss": 0.010985145568847656,
"step": 48500
},
{
"epoch": 0.9417675356921167,
"grad_norm": 0.03637198358774185,
"learning_rate": 1.1758430053991143e-05,
"loss": 0.013445260524749756,
"step": 48550
},
{
"epoch": 0.9427374301675978,
"grad_norm": 0.13509546220302582,
"learning_rate": 1.1747653379026436e-05,
"loss": 0.012020930051803588,
"step": 48600
},
{
"epoch": 0.9437073246430788,
"grad_norm": 0.005104159936308861,
"learning_rate": 1.173687670406173e-05,
"loss": 0.01152000665664673,
"step": 48650
},
{
"epoch": 0.94467721911856,
"grad_norm": 0.5151373744010925,
"learning_rate": 1.1726100029097023e-05,
"loss": 0.011212599277496339,
"step": 48700
},
{
"epoch": 0.945647113594041,
"grad_norm": 1.3944995403289795,
"learning_rate": 1.1715323354132318e-05,
"loss": 0.015623736381530761,
"step": 48750
},
{
"epoch": 0.946617008069522,
"grad_norm": 1.1535437107086182,
"learning_rate": 1.1704546679167609e-05,
"loss": 0.015163114070892334,
"step": 48800
},
{
"epoch": 0.9475869025450031,
"grad_norm": 0.11572514474391937,
"learning_rate": 1.1693770004202904e-05,
"loss": 0.012287333011627197,
"step": 48850
},
{
"epoch": 0.9485567970204841,
"grad_norm": 3.7169265747070312,
"learning_rate": 1.1682993329238197e-05,
"loss": 0.013222954273223876,
"step": 48900
},
{
"epoch": 0.9495266914959652,
"grad_norm": 0.006147901061922312,
"learning_rate": 1.1672216654273492e-05,
"loss": 0.010324739217758179,
"step": 48950
},
{
"epoch": 0.9504965859714463,
"grad_norm": 2.377066135406494,
"learning_rate": 1.1661439979308783e-05,
"loss": 0.009132164120674134,
"step": 49000
},
{
"epoch": 0.9514664804469274,
"grad_norm": 0.2680869698524475,
"learning_rate": 1.1650663304344078e-05,
"loss": 0.007903059720993042,
"step": 49050
},
{
"epoch": 0.9524363749224084,
"grad_norm": 0.0036416344810277224,
"learning_rate": 1.1639886629379373e-05,
"loss": 0.010024495124816894,
"step": 49100
},
{
"epoch": 0.9534062693978895,
"grad_norm": 0.09393948316574097,
"learning_rate": 1.1629109954414666e-05,
"loss": 0.006630579233169556,
"step": 49150
},
{
"epoch": 0.9543761638733705,
"grad_norm": 0.024503029882907867,
"learning_rate": 1.1618333279449961e-05,
"loss": 0.015390309095382691,
"step": 49200
},
{
"epoch": 0.9553460583488517,
"grad_norm": 0.008736686781048775,
"learning_rate": 1.1607556604485253e-05,
"loss": 0.005021333694458008,
"step": 49250
},
{
"epoch": 0.9563159528243327,
"grad_norm": 0.016973629593849182,
"learning_rate": 1.1596779929520547e-05,
"loss": 0.01274700164794922,
"step": 49300
},
{
"epoch": 0.9572858472998138,
"grad_norm": 4.72426176071167,
"learning_rate": 1.158600325455584e-05,
"loss": 0.017660683393478392,
"step": 49350
},
{
"epoch": 0.9582557417752948,
"grad_norm": 0.004462205804884434,
"learning_rate": 1.1575226579591134e-05,
"loss": 0.008650290966033935,
"step": 49400
},
{
"epoch": 0.959225636250776,
"grad_norm": 1.2752540111541748,
"learning_rate": 1.1564449904626427e-05,
"loss": 0.008840162754058838,
"step": 49450
},
{
"epoch": 0.960195530726257,
"grad_norm": 0.04958091676235199,
"learning_rate": 1.1553673229661722e-05,
"loss": 0.014927923679351807,
"step": 49500
},
{
"epoch": 0.9611654252017381,
"grad_norm": 0.11924762278795242,
"learning_rate": 1.1542896554697013e-05,
"loss": 0.011384433507919312,
"step": 49550
},
{
"epoch": 0.9621353196772191,
"grad_norm": 1.5456665754318237,
"learning_rate": 1.1532119879732308e-05,
"loss": 0.01652174711227417,
"step": 49600
},
{
"epoch": 0.9631052141527002,
"grad_norm": 0.09029275178909302,
"learning_rate": 1.1521343204767601e-05,
"loss": 0.011674209833145141,
"step": 49650
},
{
"epoch": 0.9640751086281812,
"grad_norm": 0.08050458133220673,
"learning_rate": 1.1510566529802896e-05,
"loss": 0.013446836471557618,
"step": 49700
},
{
"epoch": 0.9650450031036624,
"grad_norm": 0.0019389111548662186,
"learning_rate": 1.1499789854838188e-05,
"loss": 0.009251207113265991,
"step": 49750
},
{
"epoch": 0.9660148975791434,
"grad_norm": 0.5366764068603516,
"learning_rate": 1.1489013179873482e-05,
"loss": 0.01728030204772949,
"step": 49800
},
{
"epoch": 0.9669847920546245,
"grad_norm": 0.2538897693157196,
"learning_rate": 1.1478236504908777e-05,
"loss": 0.010403580665588379,
"step": 49850
},
{
"epoch": 0.9679546865301055,
"grad_norm": 0.3236939609050751,
"learning_rate": 1.146745982994407e-05,
"loss": 0.011456176042556762,
"step": 49900
},
{
"epoch": 0.9689245810055865,
"grad_norm": 0.0561516247689724,
"learning_rate": 1.1456683154979364e-05,
"loss": 0.011840370893478393,
"step": 49950
},
{
"epoch": 0.9698944754810677,
"grad_norm": 0.31649890542030334,
"learning_rate": 1.1445906480014657e-05,
"loss": 0.01205718994140625,
"step": 50000
},
{
"epoch": 0.9708643699565487,
"grad_norm": 0.12749825417995453,
"learning_rate": 1.1435129805049952e-05,
"loss": 0.009861626029014588,
"step": 50050
},
{
"epoch": 0.9718342644320298,
"grad_norm": 0.04131844639778137,
"learning_rate": 1.1424353130085243e-05,
"loss": 0.009942956566810608,
"step": 50100
},
{
"epoch": 0.9728041589075108,
"grad_norm": 0.059839457273483276,
"learning_rate": 1.1413576455120538e-05,
"loss": 0.008443999290466308,
"step": 50150
},
{
"epoch": 0.9737740533829919,
"grad_norm": 0.9020216464996338,
"learning_rate": 1.1402799780155831e-05,
"loss": 0.005326822996139526,
"step": 50200
},
{
"epoch": 0.974743947858473,
"grad_norm": 0.016258511692285538,
"learning_rate": 1.1392023105191126e-05,
"loss": 0.01237388253211975,
"step": 50250
},
{
"epoch": 0.9757138423339541,
"grad_norm": 0.9826646447181702,
"learning_rate": 1.1381246430226417e-05,
"loss": 0.010429953336715697,
"step": 50300
},
{
"epoch": 0.9766837368094351,
"grad_norm": 0.8492814302444458,
"learning_rate": 1.1370469755261712e-05,
"loss": 0.013360403776168824,
"step": 50350
},
{
"epoch": 0.9776536312849162,
"grad_norm": 0.22471806406974792,
"learning_rate": 1.1359693080297005e-05,
"loss": 0.011282609701156616,
"step": 50400
},
{
"epoch": 0.9786235257603972,
"grad_norm": 0.06106237322092056,
"learning_rate": 1.13489164053323e-05,
"loss": 0.012179501056671142,
"step": 50450
},
{
"epoch": 0.9795934202358784,
"grad_norm": 0.7282668948173523,
"learning_rate": 1.1338139730367595e-05,
"loss": 0.012049946784973144,
"step": 50500
},
{
"epoch": 0.9805633147113594,
"grad_norm": 0.07935352623462677,
"learning_rate": 1.1327363055402887e-05,
"loss": 0.011173079013824463,
"step": 50550
},
{
"epoch": 0.9815332091868405,
"grad_norm": 0.049129486083984375,
"learning_rate": 1.1316586380438182e-05,
"loss": 0.014997845888137818,
"step": 50600
},
{
"epoch": 0.9825031036623215,
"grad_norm": 0.009087535552680492,
"learning_rate": 1.1305809705473475e-05,
"loss": 0.011520069837570191,
"step": 50650
},
{
"epoch": 0.9834729981378026,
"grad_norm": 0.8175909519195557,
"learning_rate": 1.1295033030508768e-05,
"loss": 0.009147930145263671,
"step": 50700
},
{
"epoch": 0.9844428926132837,
"grad_norm": 0.04308192431926727,
"learning_rate": 1.1284256355544061e-05,
"loss": 0.00793870449066162,
"step": 50750
},
{
"epoch": 0.9854127870887648,
"grad_norm": 3.0032317638397217,
"learning_rate": 1.1273479680579356e-05,
"loss": 0.016941237449645995,
"step": 50800
},
{
"epoch": 0.9863826815642458,
"grad_norm": 0.005765567533671856,
"learning_rate": 1.1262703005614647e-05,
"loss": 0.012457648515701294,
"step": 50850
},
{
"epoch": 0.9873525760397269,
"grad_norm": 1.716653823852539,
"learning_rate": 1.1251926330649942e-05,
"loss": 0.010403302907943725,
"step": 50900
},
{
"epoch": 0.9883224705152079,
"grad_norm": 3.637549638748169,
"learning_rate": 1.1241149655685235e-05,
"loss": 0.012719739675521851,
"step": 50950
},
{
"epoch": 0.9892923649906891,
"grad_norm": 1.3579350709915161,
"learning_rate": 1.123037298072053e-05,
"loss": 0.011442365646362305,
"step": 51000
},
{
"epoch": 0.9902622594661701,
"grad_norm": 0.253671795129776,
"learning_rate": 1.1219596305755822e-05,
"loss": 0.01045155644416809,
"step": 51050
},
{
"epoch": 0.9912321539416511,
"grad_norm": 0.2456391453742981,
"learning_rate": 1.1208819630791116e-05,
"loss": 0.007976468205451965,
"step": 51100
},
{
"epoch": 0.9922020484171322,
"grad_norm": 0.000770356971770525,
"learning_rate": 1.119804295582641e-05,
"loss": 0.011402822732925415,
"step": 51150
},
{
"epoch": 0.9931719428926132,
"grad_norm": 0.5046232342720032,
"learning_rate": 1.1187266280861705e-05,
"loss": 0.01546269416809082,
"step": 51200
},
{
"epoch": 0.9941418373680944,
"grad_norm": 0.01354404166340828,
"learning_rate": 1.1176489605896998e-05,
"loss": 0.013317136764526368,
"step": 51250
},
{
"epoch": 0.9951117318435754,
"grad_norm": 0.019960785284638405,
"learning_rate": 1.116571293093229e-05,
"loss": 0.009596403241157531,
"step": 51300
},
{
"epoch": 0.9960816263190565,
"grad_norm": 0.000762759183999151,
"learning_rate": 1.1154936255967586e-05,
"loss": 0.012175165414810181,
"step": 51350
},
{
"epoch": 0.9970515207945375,
"grad_norm": 0.18124772608280182,
"learning_rate": 1.1144159581002877e-05,
"loss": 0.006460702419281006,
"step": 51400
},
{
"epoch": 0.9980214152700186,
"grad_norm": 0.005583456717431545,
"learning_rate": 1.1133382906038172e-05,
"loss": 0.015311251878738403,
"step": 51450
},
{
"epoch": 0.9989913097454997,
"grad_norm": 0.0036591431125998497,
"learning_rate": 1.1122606231073465e-05,
"loss": 0.010004911422729492,
"step": 51500
},
{
"epoch": 0.9999612042209808,
"grad_norm": 0.026024863123893738,
"learning_rate": 1.111182955610876e-05,
"loss": 0.011015715599060059,
"step": 51550
},
{
"epoch": 1.0,
"eval_ce-val_accuracy": 0.9963412070194216,
"eval_ce-val_accuracy_threshold": 0.33756840229034424,
"eval_ce-val_average_precision": 0.999500331869223,
"eval_ce-val_f1": 0.9903832352519019,
"eval_ce-val_f1_threshold": 0.15013408660888672,
"eval_ce-val_precision": 0.9861791987651322,
"eval_ce-val_recall": 0.994623268418549,
"eval_loss": 0.009881999343633652,
"eval_runtime": 314.5595,
"eval_samples_per_second": 1164.298,
"eval_steps_per_second": 18.194,
"step": 51552
},
{
"epoch": 1.000931098696462,
"grad_norm": 0.04931466281414032,
"learning_rate": 1.1101052881144051e-05,
"loss": 0.011022998094558716,
"step": 51600
},
{
"epoch": 1.001900993171943,
"grad_norm": 0.13730710744857788,
"learning_rate": 1.1090276206179346e-05,
"loss": 0.012091619968414307,
"step": 51650
},
{
"epoch": 1.002870887647424,
"grad_norm": 0.9200039505958557,
"learning_rate": 1.107949953121464e-05,
"loss": 0.010339250564575195,
"step": 51700
},
{
"epoch": 1.003840782122905,
"grad_norm": 0.0477994866669178,
"learning_rate": 1.1068722856249934e-05,
"loss": 0.008963844776153564,
"step": 51750
},
{
"epoch": 1.0048106765983862,
"grad_norm": 0.035223931074142456,
"learning_rate": 1.1057946181285226e-05,
"loss": 0.01072076678276062,
"step": 51800
},
{
"epoch": 1.0057805710738672,
"grad_norm": 0.0004753206158056855,
"learning_rate": 1.104716950632052e-05,
"loss": 0.006873180270195007,
"step": 51850
},
{
"epoch": 1.0067504655493482,
"grad_norm": 0.0004571709723677486,
"learning_rate": 1.1036392831355814e-05,
"loss": 0.012962251901626587,
"step": 51900
},
{
"epoch": 1.0077203600248292,
"grad_norm": 0.006272009573876858,
"learning_rate": 1.1025616156391109e-05,
"loss": 0.010167335271835326,
"step": 51950
},
{
"epoch": 1.0086902545003105,
"grad_norm": 0.1642366647720337,
"learning_rate": 1.1014839481426402e-05,
"loss": 0.009587610960006715,
"step": 52000
},
{
"epoch": 1.0096601489757915,
"grad_norm": 8.591952323913574,
"learning_rate": 1.1004062806461695e-05,
"loss": 0.010346212387084962,
"step": 52050
},
{
"epoch": 1.0106300434512725,
"grad_norm": 0.07947248220443726,
"learning_rate": 1.099328613149699e-05,
"loss": 0.006845234036445618,
"step": 52100
},
{
"epoch": 1.0115999379267535,
"grad_norm": 0.006813167128711939,
"learning_rate": 1.0982509456532281e-05,
"loss": 0.008352599143981933,
"step": 52150
},
{
"epoch": 1.0125698324022347,
"grad_norm": 1.755228042602539,
"learning_rate": 1.0971732781567576e-05,
"loss": 0.013376086950302124,
"step": 52200
},
{
"epoch": 1.0135397268777158,
"grad_norm": 1.382287859916687,
"learning_rate": 1.096095610660287e-05,
"loss": 0.009921017289161681,
"step": 52250
},
{
"epoch": 1.0145096213531968,
"grad_norm": 0.4828871190547943,
"learning_rate": 1.0950179431638164e-05,
"loss": 0.014618127346038819,
"step": 52300
},
{
"epoch": 1.0154795158286778,
"grad_norm": 0.1663595736026764,
"learning_rate": 1.0939402756673456e-05,
"loss": 0.0068564373254776,
"step": 52350
},
{
"epoch": 1.0164494103041588,
"grad_norm": 0.9590847492218018,
"learning_rate": 1.092862608170875e-05,
"loss": 0.013661284446716309,
"step": 52400
},
{
"epoch": 1.01741930477964,
"grad_norm": 0.057969626039266586,
"learning_rate": 1.0917849406744044e-05,
"loss": 0.009425458312034608,
"step": 52450
},
{
"epoch": 1.018389199255121,
"grad_norm": 2.8329989910125732,
"learning_rate": 1.0907072731779339e-05,
"loss": 0.009828442335128784,
"step": 52500
},
{
"epoch": 1.019359093730602,
"grad_norm": 0.021858349442481995,
"learning_rate": 1.089629605681463e-05,
"loss": 0.008514501452445984,
"step": 52550
},
{
"epoch": 1.020328988206083,
"grad_norm": 0.2961184084415436,
"learning_rate": 1.0885519381849925e-05,
"loss": 0.012648154497146607,
"step": 52600
},
{
"epoch": 1.0212988826815643,
"grad_norm": 0.20600508153438568,
"learning_rate": 1.087474270688522e-05,
"loss": 0.011835060119628905,
"step": 52650
},
{
"epoch": 1.0222687771570453,
"grad_norm": 0.05543051287531853,
"learning_rate": 1.0863966031920513e-05,
"loss": 0.010202394723892212,
"step": 52700
},
{
"epoch": 1.0232386716325264,
"grad_norm": 0.09994525462388992,
"learning_rate": 1.0853189356955806e-05,
"loss": 0.016087939739227296,
"step": 52750
},
{
"epoch": 1.0242085661080074,
"grad_norm": 0.07471498101949692,
"learning_rate": 1.08424126819911e-05,
"loss": 0.006659422516822815,
"step": 52800
},
{
"epoch": 1.0251784605834886,
"grad_norm": 0.5331398248672485,
"learning_rate": 1.0831636007026394e-05,
"loss": 0.013434828519821166,
"step": 52850
},
{
"epoch": 1.0261483550589696,
"grad_norm": 0.00969494879245758,
"learning_rate": 1.0820859332061685e-05,
"loss": 0.012539198398590088,
"step": 52900
},
{
"epoch": 1.0271182495344506,
"grad_norm": 0.5283724665641785,
"learning_rate": 1.081008265709698e-05,
"loss": 0.007128288149833679,
"step": 52950
},
{
"epoch": 1.0280881440099316,
"grad_norm": 0.014323423616588116,
"learning_rate": 1.0799305982132273e-05,
"loss": 0.010339093208312989,
"step": 53000
},
{
"epoch": 1.0290580384854129,
"grad_norm": 0.10301046818494797,
"learning_rate": 1.0788529307167568e-05,
"loss": 0.00840785264968872,
"step": 53050
},
{
"epoch": 1.030027932960894,
"grad_norm": 1.626081109046936,
"learning_rate": 1.077775263220286e-05,
"loss": 0.009251479506492615,
"step": 53100
},
{
"epoch": 1.030997827436375,
"grad_norm": 0.029879838228225708,
"learning_rate": 1.0766975957238155e-05,
"loss": 0.009033103585243224,
"step": 53150
},
{
"epoch": 1.031967721911856,
"grad_norm": 0.0901498794555664,
"learning_rate": 1.0756199282273448e-05,
"loss": 0.00871380627155304,
"step": 53200
},
{
"epoch": 1.0329376163873372,
"grad_norm": 0.012171002104878426,
"learning_rate": 1.0745422607308743e-05,
"loss": 0.011328972578048706,
"step": 53250
},
{
"epoch": 1.0339075108628182,
"grad_norm": 0.03981078416109085,
"learning_rate": 1.0734645932344034e-05,
"loss": 0.0062888008356094365,
"step": 53300
},
{
"epoch": 1.0348774053382992,
"grad_norm": 0.020450090989470482,
"learning_rate": 1.0723869257379329e-05,
"loss": 0.01264844298362732,
"step": 53350
},
{
"epoch": 1.0358472998137802,
"grad_norm": 0.20390941202640533,
"learning_rate": 1.0713092582414624e-05,
"loss": 0.00810799777507782,
"step": 53400
},
{
"epoch": 1.0368171942892612,
"grad_norm": 0.01530260220170021,
"learning_rate": 1.0702315907449915e-05,
"loss": 0.008101147413253785,
"step": 53450
},
{
"epoch": 1.0377870887647425,
"grad_norm": 0.021055378019809723,
"learning_rate": 1.069153923248521e-05,
"loss": 0.01095322608947754,
"step": 53500
},
{
"epoch": 1.0387569832402235,
"grad_norm": 0.001069198944605887,
"learning_rate": 1.0680762557520503e-05,
"loss": 0.011601181030273437,
"step": 53550
},
{
"epoch": 1.0397268777157045,
"grad_norm": 0.012236926704645157,
"learning_rate": 1.0669985882555798e-05,
"loss": 0.009461253881454468,
"step": 53600
},
{
"epoch": 1.0406967721911855,
"grad_norm": 0.12837058305740356,
"learning_rate": 1.065920920759109e-05,
"loss": 0.009943958520889282,
"step": 53650
},
{
"epoch": 1.0416666666666667,
"grad_norm": 0.4656673073768616,
"learning_rate": 1.0648432532626385e-05,
"loss": 0.008786125779151917,
"step": 53700
},
{
"epoch": 1.0426365611421478,
"grad_norm": 0.016046399250626564,
"learning_rate": 1.0637655857661678e-05,
"loss": 0.010674391984939574,
"step": 53750
},
{
"epoch": 1.0436064556176288,
"grad_norm": 0.12353447824716568,
"learning_rate": 1.0626879182696973e-05,
"loss": 0.01003960132598877,
"step": 53800
},
{
"epoch": 1.0445763500931098,
"grad_norm": 0.27510106563568115,
"learning_rate": 1.0616102507732264e-05,
"loss": 0.01023959755897522,
"step": 53850
},
{
"epoch": 1.045546244568591,
"grad_norm": 0.036945659667253494,
"learning_rate": 1.0605325832767559e-05,
"loss": 0.006538587212562561,
"step": 53900
},
{
"epoch": 1.046516139044072,
"grad_norm": 0.025796489790081978,
"learning_rate": 1.0594549157802852e-05,
"loss": 0.01494037389755249,
"step": 53950
},
{
"epoch": 1.047486033519553,
"grad_norm": 0.4464012384414673,
"learning_rate": 1.0583772482838147e-05,
"loss": 0.015746129751205443,
"step": 54000
},
{
"epoch": 1.048455927995034,
"grad_norm": 0.00668703205883503,
"learning_rate": 1.0572995807873438e-05,
"loss": 0.010869842767715455,
"step": 54050
},
{
"epoch": 1.0494258224705153,
"grad_norm": 2.842585325241089,
"learning_rate": 1.0562219132908733e-05,
"loss": 0.011381297111511231,
"step": 54100
},
{
"epoch": 1.0503957169459963,
"grad_norm": 0.08307662606239319,
"learning_rate": 1.0551442457944028e-05,
"loss": 0.011251429319381714,
"step": 54150
},
{
"epoch": 1.0513656114214773,
"grad_norm": 0.0014719933969900012,
"learning_rate": 1.054066578297932e-05,
"loss": 0.009159819483757018,
"step": 54200
},
{
"epoch": 1.0523355058969583,
"grad_norm": 0.020606260746717453,
"learning_rate": 1.0529889108014614e-05,
"loss": 0.007929980754852295,
"step": 54250
},
{
"epoch": 1.0533054003724396,
"grad_norm": 0.13004493713378906,
"learning_rate": 1.0519112433049908e-05,
"loss": 0.011997201442718507,
"step": 54300
},
{
"epoch": 1.0542752948479206,
"grad_norm": 0.24599601328372955,
"learning_rate": 1.0508335758085202e-05,
"loss": 0.013199275732040406,
"step": 54350
},
{
"epoch": 1.0552451893234016,
"grad_norm": 0.008096962235867977,
"learning_rate": 1.0497559083120494e-05,
"loss": 0.016311873197555542,
"step": 54400
},
{
"epoch": 1.0562150837988826,
"grad_norm": 0.34695443511009216,
"learning_rate": 1.0486782408155789e-05,
"loss": 0.011568185091018677,
"step": 54450
},
{
"epoch": 1.0571849782743636,
"grad_norm": 0.22395293414592743,
"learning_rate": 1.0476005733191082e-05,
"loss": 0.011894687414169311,
"step": 54500
},
{
"epoch": 1.0581548727498449,
"grad_norm": 0.03260437771677971,
"learning_rate": 1.0465229058226377e-05,
"loss": 0.014444073438644409,
"step": 54550
},
{
"epoch": 1.059124767225326,
"grad_norm": 0.018030961975455284,
"learning_rate": 1.0454452383261668e-05,
"loss": 0.007700647711753845,
"step": 54600
},
{
"epoch": 1.060094661700807,
"grad_norm": 0.0006347903981804848,
"learning_rate": 1.0443675708296963e-05,
"loss": 0.007704190015792847,
"step": 54650
},
{
"epoch": 1.061064556176288,
"grad_norm": 0.006148898974061012,
"learning_rate": 1.0432899033332256e-05,
"loss": 0.007340219020843506,
"step": 54700
},
{
"epoch": 1.0620344506517692,
"grad_norm": 0.009237848222255707,
"learning_rate": 1.042212235836755e-05,
"loss": 0.014125288724899291,
"step": 54750
},
{
"epoch": 1.0630043451272502,
"grad_norm": 1.4247099161148071,
"learning_rate": 1.0411345683402844e-05,
"loss": 0.010243568420410156,
"step": 54800
},
{
"epoch": 1.0639742396027312,
"grad_norm": 0.2413947731256485,
"learning_rate": 1.0400569008438137e-05,
"loss": 0.014130769968032838,
"step": 54850
},
{
"epoch": 1.0649441340782122,
"grad_norm": 0.6918673515319824,
"learning_rate": 1.0389792333473432e-05,
"loss": 0.010264915227890015,
"step": 54900
},
{
"epoch": 1.0659140285536934,
"grad_norm": 0.12765105068683624,
"learning_rate": 1.0379015658508724e-05,
"loss": 0.008968057036399841,
"step": 54950
},
{
"epoch": 1.0668839230291745,
"grad_norm": 0.04986311122775078,
"learning_rate": 1.0368238983544019e-05,
"loss": 0.011316138505935668,
"step": 55000
},
{
"epoch": 1.0678538175046555,
"grad_norm": 0.0769861489534378,
"learning_rate": 1.0357462308579312e-05,
"loss": 0.007071847915649414,
"step": 55050
},
{
"epoch": 1.0688237119801365,
"grad_norm": 0.4210555851459503,
"learning_rate": 1.0346685633614607e-05,
"loss": 0.011663705110549927,
"step": 55100
},
{
"epoch": 1.0697936064556177,
"grad_norm": 0.6886973977088928,
"learning_rate": 1.0335908958649898e-05,
"loss": 0.008526567816734314,
"step": 55150
},
{
"epoch": 1.0707635009310987,
"grad_norm": 0.0545625314116478,
"learning_rate": 1.0325132283685193e-05,
"loss": 0.008796229362487792,
"step": 55200
},
{
"epoch": 1.0717333954065797,
"grad_norm": 0.0035746218636631966,
"learning_rate": 1.0314355608720486e-05,
"loss": 0.007686410546302796,
"step": 55250
},
{
"epoch": 1.0727032898820608,
"grad_norm": 0.007664795964956284,
"learning_rate": 1.0303578933755781e-05,
"loss": 0.011068449020385743,
"step": 55300
},
{
"epoch": 1.073673184357542,
"grad_norm": 0.07639248669147491,
"learning_rate": 1.0292802258791072e-05,
"loss": 0.011426090002059937,
"step": 55350
},
{
"epoch": 1.074643078833023,
"grad_norm": 0.03417905047535896,
"learning_rate": 1.0282025583826367e-05,
"loss": 0.006777835488319397,
"step": 55400
},
{
"epoch": 1.075612973308504,
"grad_norm": 0.0801825225353241,
"learning_rate": 1.027124890886166e-05,
"loss": 0.011024791002273559,
"step": 55450
},
{
"epoch": 1.076582867783985,
"grad_norm": 0.07644648104906082,
"learning_rate": 1.0260472233896953e-05,
"loss": 0.0118917715549469,
"step": 55500
},
{
"epoch": 1.077552762259466,
"grad_norm": 0.007192076183855534,
"learning_rate": 1.0249695558932248e-05,
"loss": 0.00884443759918213,
"step": 55550
},
{
"epoch": 1.0785226567349473,
"grad_norm": 0.8110189437866211,
"learning_rate": 1.0238918883967542e-05,
"loss": 0.011060004234313964,
"step": 55600
},
{
"epoch": 1.0794925512104283,
"grad_norm": 0.06062622368335724,
"learning_rate": 1.0228142209002836e-05,
"loss": 0.011132882833480835,
"step": 55650
},
{
"epoch": 1.0804624456859093,
"grad_norm": 0.14978434145450592,
"learning_rate": 1.0217365534038128e-05,
"loss": 0.011115623712539673,
"step": 55700
},
{
"epoch": 1.0814323401613903,
"grad_norm": 0.13496646285057068,
"learning_rate": 1.0206588859073423e-05,
"loss": 0.007979636788368225,
"step": 55750
},
{
"epoch": 1.0824022346368716,
"grad_norm": 0.11768963187932968,
"learning_rate": 1.0195812184108716e-05,
"loss": 0.010357421636581422,
"step": 55800
},
{
"epoch": 1.0833721291123526,
"grad_norm": 0.01061262283474207,
"learning_rate": 1.018503550914401e-05,
"loss": 0.008668645024299622,
"step": 55850
},
{
"epoch": 1.0843420235878336,
"grad_norm": 0.039225850254297256,
"learning_rate": 1.0174258834179302e-05,
"loss": 0.0060651302337646484,
"step": 55900
},
{
"epoch": 1.0853119180633146,
"grad_norm": 0.019172627478837967,
"learning_rate": 1.0163482159214597e-05,
"loss": 0.010994638204574586,
"step": 55950
},
{
"epoch": 1.0862818125387959,
"grad_norm": 0.0007182133849710226,
"learning_rate": 1.015270548424989e-05,
"loss": 0.01103974223136902,
"step": 56000
},
{
"epoch": 1.0872517070142769,
"grad_norm": 0.03451617434620857,
"learning_rate": 1.0141928809285185e-05,
"loss": 0.013964263200759887,
"step": 56050
},
{
"epoch": 1.0882216014897579,
"grad_norm": 0.00380208739079535,
"learning_rate": 1.0131152134320476e-05,
"loss": 0.01113158106803894,
"step": 56100
},
{
"epoch": 1.089191495965239,
"grad_norm": 0.4477499723434448,
"learning_rate": 1.0120375459355771e-05,
"loss": 0.010933184623718261,
"step": 56150
},
{
"epoch": 1.0901613904407201,
"grad_norm": 0.2706264853477478,
"learning_rate": 1.0109598784391066e-05,
"loss": 0.009425508379936219,
"step": 56200
},
{
"epoch": 1.0911312849162011,
"grad_norm": 2.2404770851135254,
"learning_rate": 1.0098822109426358e-05,
"loss": 0.008474677801132202,
"step": 56250
},
{
"epoch": 1.0921011793916822,
"grad_norm": 0.05085292458534241,
"learning_rate": 1.0088045434461653e-05,
"loss": 0.007376531958580017,
"step": 56300
},
{
"epoch": 1.0930710738671632,
"grad_norm": 0.2042039930820465,
"learning_rate": 1.0077268759496946e-05,
"loss": 0.015207431316375732,
"step": 56350
},
{
"epoch": 1.0940409683426444,
"grad_norm": 0.001400004606693983,
"learning_rate": 1.006649208453224e-05,
"loss": 0.011070315837860107,
"step": 56400
},
{
"epoch": 1.0950108628181254,
"grad_norm": 0.033941857516765594,
"learning_rate": 1.0055715409567532e-05,
"loss": 0.010366495847702027,
"step": 56450
},
{
"epoch": 1.0959807572936064,
"grad_norm": 0.04121604189276695,
"learning_rate": 1.0044938734602827e-05,
"loss": 0.01011496901512146,
"step": 56500
},
{
"epoch": 1.0969506517690875,
"grad_norm": 0.04570454731583595,
"learning_rate": 1.003416205963812e-05,
"loss": 0.011175779104232788,
"step": 56550
},
{
"epoch": 1.0979205462445685,
"grad_norm": 0.20750437676906586,
"learning_rate": 1.0023385384673415e-05,
"loss": 0.014594887495040893,
"step": 56600
},
{
"epoch": 1.0988904407200497,
"grad_norm": 0.003194875782355666,
"learning_rate": 1.0012608709708706e-05,
"loss": 0.00948288917541504,
"step": 56650
},
{
"epoch": 1.0998603351955307,
"grad_norm": 0.24496686458587646,
"learning_rate": 1.0001832034744001e-05,
"loss": 0.011073273420333863,
"step": 56700
},
{
"epoch": 1.1008302296710117,
"grad_norm": 0.26492536067962646,
"learning_rate": 9.991055359779294e-06,
"loss": 0.008801199793815613,
"step": 56750
},
{
"epoch": 1.101800124146493,
"grad_norm": 0.12051726132631302,
"learning_rate": 9.980278684814588e-06,
"loss": 0.008033130764961243,
"step": 56800
},
{
"epoch": 1.102770018621974,
"grad_norm": 0.27666184306144714,
"learning_rate": 9.969502009849882e-06,
"loss": 0.008703486323356628,
"step": 56850
},
{
"epoch": 1.103739913097455,
"grad_norm": 0.11903724819421768,
"learning_rate": 9.958725334885176e-06,
"loss": 0.013527406454086304,
"step": 56900
},
{
"epoch": 1.104709807572936,
"grad_norm": 0.42788776755332947,
"learning_rate": 9.947948659920469e-06,
"loss": 0.009564909934997558,
"step": 56950
},
{
"epoch": 1.105679702048417,
"grad_norm": 0.23189954459667206,
"learning_rate": 9.937171984955762e-06,
"loss": 0.01011852741241455,
"step": 57000
},
{
"epoch": 1.1066495965238983,
"grad_norm": 0.11286676675081253,
"learning_rate": 9.926395309991057e-06,
"loss": 0.011357702016830444,
"step": 57050
},
{
"epoch": 1.1076194909993793,
"grad_norm": 1.0212432146072388,
"learning_rate": 9.91561863502635e-06,
"loss": 0.009664784669876098,
"step": 57100
},
{
"epoch": 1.1085893854748603,
"grad_norm": 0.04047268629074097,
"learning_rate": 9.904841960061643e-06,
"loss": 0.009687536358833314,
"step": 57150
},
{
"epoch": 1.1095592799503413,
"grad_norm": 0.08294454216957092,
"learning_rate": 9.894065285096936e-06,
"loss": 0.006463578343391419,
"step": 57200
},
{
"epoch": 1.1105291744258226,
"grad_norm": 0.01614169031381607,
"learning_rate": 9.88328861013223e-06,
"loss": 0.009857819080352784,
"step": 57250
},
{
"epoch": 1.1114990689013036,
"grad_norm": 0.00470517436042428,
"learning_rate": 9.872511935167524e-06,
"loss": 0.010130856037139892,
"step": 57300
},
{
"epoch": 1.1124689633767846,
"grad_norm": 0.017413944005966187,
"learning_rate": 9.861735260202819e-06,
"loss": 0.010480610132217407,
"step": 57350
},
{
"epoch": 1.1134388578522656,
"grad_norm": 0.13971586525440216,
"learning_rate": 9.850958585238112e-06,
"loss": 0.007595556378364563,
"step": 57400
},
{
"epoch": 1.1144087523277468,
"grad_norm": 0.3776938021183014,
"learning_rate": 9.840181910273405e-06,
"loss": 0.008610534071922303,
"step": 57450
},
{
"epoch": 1.1153786468032278,
"grad_norm": 0.9494521617889404,
"learning_rate": 9.829405235308699e-06,
"loss": 0.012484655380249024,
"step": 57500
},
{
"epoch": 1.1163485412787089,
"grad_norm": 3.1289262771606445,
"learning_rate": 9.818628560343992e-06,
"loss": 0.010204870700836182,
"step": 57550
},
{
"epoch": 1.1173184357541899,
"grad_norm": 0.0060015176422894,
"learning_rate": 9.807851885379287e-06,
"loss": 0.00760654091835022,
"step": 57600
},
{
"epoch": 1.1182883302296711,
"grad_norm": 0.0010267419274896383,
"learning_rate": 9.79707521041458e-06,
"loss": 0.010920615196228027,
"step": 57650
},
{
"epoch": 1.1192582247051521,
"grad_norm": 0.04344663396477699,
"learning_rate": 9.786298535449873e-06,
"loss": 0.014959734678268433,
"step": 57700
},
{
"epoch": 1.1202281191806331,
"grad_norm": 0.03204028308391571,
"learning_rate": 9.775521860485166e-06,
"loss": 0.010675301551818847,
"step": 57750
},
{
"epoch": 1.1211980136561142,
"grad_norm": 0.24517330527305603,
"learning_rate": 9.764745185520461e-06,
"loss": 0.010562815666198731,
"step": 57800
},
{
"epoch": 1.1221679081315954,
"grad_norm": 0.22162503004074097,
"learning_rate": 9.753968510555754e-06,
"loss": 0.0045673206448554994,
"step": 57850
},
{
"epoch": 1.1231378026070764,
"grad_norm": 0.06855642795562744,
"learning_rate": 9.743191835591047e-06,
"loss": 0.008667095303535462,
"step": 57900
},
{
"epoch": 1.1241076970825574,
"grad_norm": 0.0520123615860939,
"learning_rate": 9.73241516062634e-06,
"loss": 0.012918524742126465,
"step": 57950
},
{
"epoch": 1.1250775915580384,
"grad_norm": 0.4600003957748413,
"learning_rate": 9.721638485661635e-06,
"loss": 0.011975594758987428,
"step": 58000
},
{
"epoch": 1.1260474860335195,
"grad_norm": 0.14219403266906738,
"learning_rate": 9.710861810696928e-06,
"loss": 0.011972142457962036,
"step": 58050
},
{
"epoch": 1.1270173805090007,
"grad_norm": 0.2508797347545624,
"learning_rate": 9.700085135732222e-06,
"loss": 0.009876527190208435,
"step": 58100
},
{
"epoch": 1.1279872749844817,
"grad_norm": 0.09881020337343216,
"learning_rate": 9.689308460767516e-06,
"loss": 0.011778740882873536,
"step": 58150
},
{
"epoch": 1.1289571694599627,
"grad_norm": 0.014723585918545723,
"learning_rate": 9.67853178580281e-06,
"loss": 0.011343867778778076,
"step": 58200
},
{
"epoch": 1.1299270639354437,
"grad_norm": 0.03311719372868538,
"learning_rate": 9.667755110838103e-06,
"loss": 0.008925886750221252,
"step": 58250
},
{
"epoch": 1.130896958410925,
"grad_norm": 0.09255416691303253,
"learning_rate": 9.656978435873396e-06,
"loss": 0.008377508521080016,
"step": 58300
},
{
"epoch": 1.131866852886406,
"grad_norm": 0.003130935365334153,
"learning_rate": 9.64620176090869e-06,
"loss": 0.014613392353057862,
"step": 58350
},
{
"epoch": 1.132836747361887,
"grad_norm": 0.004846623633056879,
"learning_rate": 9.635425085943984e-06,
"loss": 0.009596949219703674,
"step": 58400
},
{
"epoch": 1.133806641837368,
"grad_norm": 0.5533701181411743,
"learning_rate": 9.624648410979277e-06,
"loss": 0.01042715072631836,
"step": 58450
},
{
"epoch": 1.1347765363128492,
"grad_norm": 0.33691728115081787,
"learning_rate": 9.61387173601457e-06,
"loss": 0.007440037131309509,
"step": 58500
},
{
"epoch": 1.1357464307883303,
"grad_norm": 0.009024172089993954,
"learning_rate": 9.603095061049863e-06,
"loss": 0.01007677435874939,
"step": 58550
},
{
"epoch": 1.1367163252638113,
"grad_norm": 0.08907906711101532,
"learning_rate": 9.592318386085158e-06,
"loss": 0.012913039922714233,
"step": 58600
},
{
"epoch": 1.1376862197392923,
"grad_norm": 0.06188477948307991,
"learning_rate": 9.581541711120451e-06,
"loss": 0.009352678656578064,
"step": 58650
},
{
"epoch": 1.1386561142147735,
"grad_norm": 0.05300740897655487,
"learning_rate": 9.570765036155746e-06,
"loss": 0.014120669364929199,
"step": 58700
},
{
"epoch": 1.1396260086902545,
"grad_norm": 0.3560042977333069,
"learning_rate": 9.55998836119104e-06,
"loss": 0.015131454467773437,
"step": 58750
},
{
"epoch": 1.1405959031657356,
"grad_norm": 0.06389273703098297,
"learning_rate": 9.549211686226333e-06,
"loss": 0.010433393716812133,
"step": 58800
},
{
"epoch": 1.1415657976412166,
"grad_norm": 0.1608138531446457,
"learning_rate": 9.538435011261626e-06,
"loss": 0.009905243515968323,
"step": 58850
},
{
"epoch": 1.1425356921166978,
"grad_norm": 0.0028658995870500803,
"learning_rate": 9.52765833629692e-06,
"loss": 0.008218841552734375,
"step": 58900
},
{
"epoch": 1.1435055865921788,
"grad_norm": 1.493757963180542,
"learning_rate": 9.516881661332214e-06,
"loss": 0.014872682094573975,
"step": 58950
},
{
"epoch": 1.1444754810676598,
"grad_norm": 0.3875971734523773,
"learning_rate": 9.506104986367507e-06,
"loss": 0.011632380485534667,
"step": 59000
},
{
"epoch": 1.1454453755431409,
"grad_norm": 0.007643993943929672,
"learning_rate": 9.4953283114028e-06,
"loss": 0.009909860491752624,
"step": 59050
},
{
"epoch": 1.1464152700186219,
"grad_norm": 0.46956542134284973,
"learning_rate": 9.484551636438095e-06,
"loss": 0.01267991542816162,
"step": 59100
},
{
"epoch": 1.147385164494103,
"grad_norm": 0.08064138144254684,
"learning_rate": 9.473774961473388e-06,
"loss": 0.014412163496017457,
"step": 59150
},
{
"epoch": 1.1483550589695841,
"grad_norm": 0.11767090857028961,
"learning_rate": 9.462998286508681e-06,
"loss": 0.007747725248336792,
"step": 59200
},
{
"epoch": 1.1493249534450651,
"grad_norm": 0.06519400328397751,
"learning_rate": 9.452221611543974e-06,
"loss": 0.013978612422943116,
"step": 59250
},
{
"epoch": 1.1502948479205461,
"grad_norm": 0.03928855061531067,
"learning_rate": 9.441444936579268e-06,
"loss": 0.0053257691860198976,
"step": 59300
},
{
"epoch": 1.1512647423960274,
"grad_norm": 2.7219903469085693,
"learning_rate": 9.430668261614562e-06,
"loss": 0.008037518262863159,
"step": 59350
},
{
"epoch": 1.1522346368715084,
"grad_norm": 0.14118452370166779,
"learning_rate": 9.419891586649857e-06,
"loss": 0.00865662157535553,
"step": 59400
},
{
"epoch": 1.1532045313469894,
"grad_norm": 0.0030036678072065115,
"learning_rate": 9.40911491168515e-06,
"loss": 0.011007028818130492,
"step": 59450
},
{
"epoch": 1.1541744258224704,
"grad_norm": 0.0671989917755127,
"learning_rate": 9.398338236720444e-06,
"loss": 0.012162750959396363,
"step": 59500
},
{
"epoch": 1.1551443202979517,
"grad_norm": 0.696707010269165,
"learning_rate": 9.387561561755737e-06,
"loss": 0.01148051381111145,
"step": 59550
},
{
"epoch": 1.1561142147734327,
"grad_norm": 0.11577127873897552,
"learning_rate": 9.37678488679103e-06,
"loss": 0.00904272198677063,
"step": 59600
},
{
"epoch": 1.1570841092489137,
"grad_norm": 0.6367550492286682,
"learning_rate": 9.366008211826325e-06,
"loss": 0.0068786549568176265,
"step": 59650
},
{
"epoch": 1.1580540037243947,
"grad_norm": 0.6166356205940247,
"learning_rate": 9.355231536861618e-06,
"loss": 0.015361328125,
"step": 59700
},
{
"epoch": 1.159023898199876,
"grad_norm": 0.0012243721866980195,
"learning_rate": 9.344454861896911e-06,
"loss": 0.011840366125106812,
"step": 59750
},
{
"epoch": 1.159993792675357,
"grad_norm": 0.006356425117701292,
"learning_rate": 9.333678186932204e-06,
"loss": 0.00854178011417389,
"step": 59800
},
{
"epoch": 1.160963687150838,
"grad_norm": 2.6208572387695312,
"learning_rate": 9.322901511967497e-06,
"loss": 0.010784318447113037,
"step": 59850
},
{
"epoch": 1.161933581626319,
"grad_norm": 0.08729662001132965,
"learning_rate": 9.312124837002792e-06,
"loss": 0.007613765001296997,
"step": 59900
},
{
"epoch": 1.1629034761018002,
"grad_norm": 0.04230942949652672,
"learning_rate": 9.301348162038085e-06,
"loss": 0.006090930700302124,
"step": 59950
},
{
"epoch": 1.1638733705772812,
"grad_norm": 0.005100870970636606,
"learning_rate": 9.290571487073379e-06,
"loss": 0.016455858945846558,
"step": 60000
},
{
"epoch": 1.1648432650527623,
"grad_norm": 0.30274850130081177,
"learning_rate": 9.279794812108672e-06,
"loss": 0.01546470046043396,
"step": 60050
},
{
"epoch": 1.1658131595282433,
"grad_norm": 0.027197476476430893,
"learning_rate": 9.269018137143967e-06,
"loss": 0.011243427991867066,
"step": 60100
},
{
"epoch": 1.1667830540037243,
"grad_norm": 1.9309356212615967,
"learning_rate": 9.25824146217926e-06,
"loss": 0.015053778886795044,
"step": 60150
},
{
"epoch": 1.1677529484792055,
"grad_norm": 0.12436617165803909,
"learning_rate": 9.247464787214555e-06,
"loss": 0.006820261478424072,
"step": 60200
},
{
"epoch": 1.1687228429546865,
"grad_norm": 0.3120616674423218,
"learning_rate": 9.236688112249848e-06,
"loss": 0.008024706840515136,
"step": 60250
},
{
"epoch": 1.1696927374301676,
"grad_norm": 0.0397704616189003,
"learning_rate": 9.225911437285141e-06,
"loss": 0.007914910316467285,
"step": 60300
},
{
"epoch": 1.1706626319056488,
"grad_norm": 0.14532391726970673,
"learning_rate": 9.215134762320434e-06,
"loss": 0.010422476530075074,
"step": 60350
},
{
"epoch": 1.1716325263811298,
"grad_norm": 0.2142135351896286,
"learning_rate": 9.204358087355729e-06,
"loss": 0.011131408214569092,
"step": 60400
},
{
"epoch": 1.1726024208566108,
"grad_norm": 0.32665055990219116,
"learning_rate": 9.193581412391022e-06,
"loss": 0.007275236248970032,
"step": 60450
},
{
"epoch": 1.1735723153320918,
"grad_norm": 0.024897614493966103,
"learning_rate": 9.182804737426315e-06,
"loss": 0.007704079747200012,
"step": 60500
},
{
"epoch": 1.1745422098075728,
"grad_norm": 0.9230888485908508,
"learning_rate": 9.172028062461608e-06,
"loss": 0.01063718557357788,
"step": 60550
},
{
"epoch": 1.175512104283054,
"grad_norm": 0.007143727038055658,
"learning_rate": 9.161251387496902e-06,
"loss": 0.007228826880455017,
"step": 60600
},
{
"epoch": 1.176481998758535,
"grad_norm": 0.9118891358375549,
"learning_rate": 9.150474712532196e-06,
"loss": 0.014060221910476685,
"step": 60650
},
{
"epoch": 1.1774518932340161,
"grad_norm": 0.2737441956996918,
"learning_rate": 9.13969803756749e-06,
"loss": 0.01123984694480896,
"step": 60700
},
{
"epoch": 1.1784217877094971,
"grad_norm": 0.49069076776504517,
"learning_rate": 9.128921362602783e-06,
"loss": 0.009755523800849914,
"step": 60750
},
{
"epoch": 1.1793916821849784,
"grad_norm": 0.03177955001592636,
"learning_rate": 9.118144687638076e-06,
"loss": 0.006125523447990417,
"step": 60800
},
{
"epoch": 1.1803615766604594,
"grad_norm": 2.8208491802215576,
"learning_rate": 9.10736801267337e-06,
"loss": 0.00635598361492157,
"step": 60850
},
{
"epoch": 1.1813314711359404,
"grad_norm": 0.33748093247413635,
"learning_rate": 9.096591337708664e-06,
"loss": 0.008556110858917236,
"step": 60900
},
{
"epoch": 1.1823013656114214,
"grad_norm": 0.05229414626955986,
"learning_rate": 9.085814662743959e-06,
"loss": 0.013156896829605103,
"step": 60950
},
{
"epoch": 1.1832712600869026,
"grad_norm": 0.012228313833475113,
"learning_rate": 9.075037987779252e-06,
"loss": 0.011504954099655152,
"step": 61000
},
{
"epoch": 1.1842411545623837,
"grad_norm": 0.0007186857401393354,
"learning_rate": 9.064261312814545e-06,
"loss": 0.008254655599594117,
"step": 61050
},
{
"epoch": 1.1852110490378647,
"grad_norm": 0.37208428978919983,
"learning_rate": 9.053484637849838e-06,
"loss": 0.010634973049163818,
"step": 61100
},
{
"epoch": 1.1861809435133457,
"grad_norm": 0.047342900186777115,
"learning_rate": 9.042707962885131e-06,
"loss": 0.010593115091323853,
"step": 61150
},
{
"epoch": 1.1871508379888267,
"grad_norm": 0.02103528566658497,
"learning_rate": 9.031931287920426e-06,
"loss": 0.007150886654853821,
"step": 61200
},
{
"epoch": 1.188120732464308,
"grad_norm": 0.00199462054297328,
"learning_rate": 9.02115461295572e-06,
"loss": 0.007624160647392273,
"step": 61250
},
{
"epoch": 1.189090626939789,
"grad_norm": 3.686844825744629,
"learning_rate": 9.010377937991013e-06,
"loss": 0.012265008687973023,
"step": 61300
},
{
"epoch": 1.19006052141527,
"grad_norm": 0.0007216570666059852,
"learning_rate": 8.999601263026306e-06,
"loss": 0.009127119779586792,
"step": 61350
},
{
"epoch": 1.1910304158907512,
"grad_norm": 0.019467012956738472,
"learning_rate": 8.9888245880616e-06,
"loss": 0.010819368362426758,
"step": 61400
},
{
"epoch": 1.1920003103662322,
"grad_norm": 1.2410645484924316,
"learning_rate": 8.978047913096894e-06,
"loss": 0.009621911644935609,
"step": 61450
},
{
"epoch": 1.1929702048417132,
"grad_norm": 0.4537646472454071,
"learning_rate": 8.967271238132187e-06,
"loss": 0.014516029357910156,
"step": 61500
},
{
"epoch": 1.1939400993171942,
"grad_norm": 0.10201362520456314,
"learning_rate": 8.956494563167482e-06,
"loss": 0.010003361701965332,
"step": 61550
},
{
"epoch": 1.1949099937926753,
"grad_norm": 0.08314236998558044,
"learning_rate": 8.945717888202775e-06,
"loss": 0.007116160988807678,
"step": 61600
},
{
"epoch": 1.1958798882681565,
"grad_norm": 0.06359340995550156,
"learning_rate": 8.934941213238068e-06,
"loss": 0.009408583045005798,
"step": 61650
},
{
"epoch": 1.1968497827436375,
"grad_norm": 0.24622827768325806,
"learning_rate": 8.924164538273363e-06,
"loss": 0.009738192558288575,
"step": 61700
},
{
"epoch": 1.1978196772191185,
"grad_norm": 0.035510655492544174,
"learning_rate": 8.913387863308656e-06,
"loss": 0.009802016019821167,
"step": 61750
},
{
"epoch": 1.1987895716945995,
"grad_norm": 0.0016278914408758283,
"learning_rate": 8.90261118834395e-06,
"loss": 0.00772739589214325,
"step": 61800
},
{
"epoch": 1.1997594661700808,
"grad_norm": 0.14160658419132233,
"learning_rate": 8.891834513379242e-06,
"loss": 0.007295028567314148,
"step": 61850
},
{
"epoch": 1.2007293606455618,
"grad_norm": 4.542331218719482,
"learning_rate": 8.881057838414536e-06,
"loss": 0.012320467233657838,
"step": 61900
},
{
"epoch": 1.2016992551210428,
"grad_norm": 0.43525606393814087,
"learning_rate": 8.87028116344983e-06,
"loss": 0.012776623964309692,
"step": 61950
},
{
"epoch": 1.2026691495965238,
"grad_norm": 1.036892056465149,
"learning_rate": 8.859504488485124e-06,
"loss": 0.006265078186988831,
"step": 62000
},
{
"epoch": 1.203639044072005,
"grad_norm": 0.14913471043109894,
"learning_rate": 8.848727813520417e-06,
"loss": 0.007494709491729736,
"step": 62050
},
{
"epoch": 1.204608938547486,
"grad_norm": 0.17765958607196808,
"learning_rate": 8.83795113855571e-06,
"loss": 0.006433714628219605,
"step": 62100
},
{
"epoch": 1.205578833022967,
"grad_norm": 0.22179652750492096,
"learning_rate": 8.827174463591005e-06,
"loss": 0.010362577438354493,
"step": 62150
},
{
"epoch": 1.206548727498448,
"grad_norm": 0.07553473860025406,
"learning_rate": 8.816397788626298e-06,
"loss": 0.011763674020767213,
"step": 62200
},
{
"epoch": 1.2075186219739291,
"grad_norm": 0.010770023800432682,
"learning_rate": 8.805621113661593e-06,
"loss": 0.00894660770893097,
"step": 62250
},
{
"epoch": 1.2084885164494104,
"grad_norm": 0.013582941144704819,
"learning_rate": 8.794844438696886e-06,
"loss": 0.0064254444837570195,
"step": 62300
},
{
"epoch": 1.2094584109248914,
"grad_norm": 0.3078850209712982,
"learning_rate": 8.784067763732179e-06,
"loss": 0.009845550656318664,
"step": 62350
},
{
"epoch": 1.2104283054003724,
"grad_norm": 0.32027581334114075,
"learning_rate": 8.773291088767472e-06,
"loss": 0.005561997890472412,
"step": 62400
},
{
"epoch": 1.2113981998758536,
"grad_norm": 0.05910608544945717,
"learning_rate": 8.762514413802767e-06,
"loss": 0.006871042847633362,
"step": 62450
},
{
"epoch": 1.2123680943513346,
"grad_norm": 0.17303982377052307,
"learning_rate": 8.75173773883806e-06,
"loss": 0.013583589792251587,
"step": 62500
},
{
"epoch": 1.2133379888268156,
"grad_norm": 0.0016346350312232971,
"learning_rate": 8.740961063873353e-06,
"loss": 0.014446932077407836,
"step": 62550
},
{
"epoch": 1.2143078833022967,
"grad_norm": 0.7791438698768616,
"learning_rate": 8.730184388908647e-06,
"loss": 0.017207759618759155,
"step": 62600
},
{
"epoch": 1.2152777777777777,
"grad_norm": 0.028775129467248917,
"learning_rate": 8.71940771394394e-06,
"loss": 0.016082524061203002,
"step": 62650
},
{
"epoch": 1.216247672253259,
"grad_norm": 0.38156262040138245,
"learning_rate": 8.708631038979235e-06,
"loss": 0.01097511053085327,
"step": 62700
},
{
"epoch": 1.21721756672874,
"grad_norm": 0.09488295018672943,
"learning_rate": 8.697854364014528e-06,
"loss": 0.009183934330940247,
"step": 62750
},
{
"epoch": 1.218187461204221,
"grad_norm": 0.15825019776821136,
"learning_rate": 8.687077689049821e-06,
"loss": 0.011556028127670289,
"step": 62800
},
{
"epoch": 1.219157355679702,
"grad_norm": 0.0048272800631821156,
"learning_rate": 8.676301014085114e-06,
"loss": 0.008433547019958496,
"step": 62850
},
{
"epoch": 1.2201272501551832,
"grad_norm": 0.028802834451198578,
"learning_rate": 8.665524339120407e-06,
"loss": 0.00994191825389862,
"step": 62900
},
{
"epoch": 1.2210971446306642,
"grad_norm": 0.15266230702400208,
"learning_rate": 8.654747664155702e-06,
"loss": 0.013666635751724244,
"step": 62950
},
{
"epoch": 1.2220670391061452,
"grad_norm": 0.0729355439543724,
"learning_rate": 8.643970989190997e-06,
"loss": 0.0110776686668396,
"step": 63000
},
{
"epoch": 1.2230369335816262,
"grad_norm": 0.02645561844110489,
"learning_rate": 8.63319431422629e-06,
"loss": 0.003797396421432495,
"step": 63050
},
{
"epoch": 1.2240068280571075,
"grad_norm": 0.10959858447313309,
"learning_rate": 8.622417639261583e-06,
"loss": 0.009820239543914795,
"step": 63100
},
{
"epoch": 1.2249767225325885,
"grad_norm": 0.042795538902282715,
"learning_rate": 8.611640964296876e-06,
"loss": 0.009580033421516419,
"step": 63150
},
{
"epoch": 1.2259466170080695,
"grad_norm": 0.03435258939862251,
"learning_rate": 8.60086428933217e-06,
"loss": 0.007328465580940247,
"step": 63200
},
{
"epoch": 1.2269165114835505,
"grad_norm": 0.07538539171218872,
"learning_rate": 8.590087614367464e-06,
"loss": 0.007034791707992554,
"step": 63250
},
{
"epoch": 1.2278864059590315,
"grad_norm": 0.03712019696831703,
"learning_rate": 8.579310939402758e-06,
"loss": 0.009296980500221253,
"step": 63300
},
{
"epoch": 1.2288563004345128,
"grad_norm": 0.287894606590271,
"learning_rate": 8.56853426443805e-06,
"loss": 0.006551910042762756,
"step": 63350
},
{
"epoch": 1.2298261949099938,
"grad_norm": 1.4108225107192993,
"learning_rate": 8.557757589473344e-06,
"loss": 0.012487185001373292,
"step": 63400
},
{
"epoch": 1.2307960893854748,
"grad_norm": 0.0015749674057587981,
"learning_rate": 8.546980914508639e-06,
"loss": 0.016569389104843138,
"step": 63450
},
{
"epoch": 1.231765983860956,
"grad_norm": 0.16347652673721313,
"learning_rate": 8.536204239543932e-06,
"loss": 0.0097618967294693,
"step": 63500
},
{
"epoch": 1.232735878336437,
"grad_norm": 0.010570725426077843,
"learning_rate": 8.525427564579225e-06,
"loss": 0.010241423845291137,
"step": 63550
},
{
"epoch": 1.233705772811918,
"grad_norm": 1.418001413345337,
"learning_rate": 8.514650889614518e-06,
"loss": 0.009225535988807678,
"step": 63600
},
{
"epoch": 1.234675667287399,
"grad_norm": 5.268980503082275,
"learning_rate": 8.503874214649811e-06,
"loss": 0.013672960996627808,
"step": 63650
},
{
"epoch": 1.23564556176288,
"grad_norm": 0.08099976181983948,
"learning_rate": 8.493097539685106e-06,
"loss": 0.012385834455490113,
"step": 63700
},
{
"epoch": 1.2366154562383613,
"grad_norm": 0.6899728775024414,
"learning_rate": 8.482320864720401e-06,
"loss": 0.006716262698173523,
"step": 63750
},
{
"epoch": 1.2375853507138423,
"grad_norm": 0.20905067026615143,
"learning_rate": 8.471544189755694e-06,
"loss": 0.008634083271026612,
"step": 63800
},
{
"epoch": 1.2385552451893234,
"grad_norm": 0.008366886526346207,
"learning_rate": 8.460767514790987e-06,
"loss": 0.010955257415771484,
"step": 63850
},
{
"epoch": 1.2395251396648044,
"grad_norm": 0.24719077348709106,
"learning_rate": 8.44999083982628e-06,
"loss": 0.00804772138595581,
"step": 63900
},
{
"epoch": 1.2404950341402856,
"grad_norm": 0.18958871066570282,
"learning_rate": 8.439214164861574e-06,
"loss": 0.008615728616714478,
"step": 63950
},
{
"epoch": 1.2414649286157666,
"grad_norm": 0.04016340151429176,
"learning_rate": 8.428437489896869e-06,
"loss": 0.010039230585098266,
"step": 64000
},
{
"epoch": 1.2424348230912476,
"grad_norm": 1.3814769983291626,
"learning_rate": 8.417660814932162e-06,
"loss": 0.006842641830444336,
"step": 64050
},
{
"epoch": 1.2434047175667287,
"grad_norm": 1.3065376281738281,
"learning_rate": 8.406884139967455e-06,
"loss": 0.015112694501876831,
"step": 64100
},
{
"epoch": 1.24437461204221,
"grad_norm": 0.06508076936006546,
"learning_rate": 8.396107465002748e-06,
"loss": 0.009530337452888489,
"step": 64150
},
{
"epoch": 1.245344506517691,
"grad_norm": 0.0020154763478785753,
"learning_rate": 8.385330790038043e-06,
"loss": 0.010464962720870972,
"step": 64200
},
{
"epoch": 1.246314400993172,
"grad_norm": 1.399607539176941,
"learning_rate": 8.374554115073336e-06,
"loss": 0.006339590549468994,
"step": 64250
},
{
"epoch": 1.247284295468653,
"grad_norm": 0.772948682308197,
"learning_rate": 8.36377744010863e-06,
"loss": 0.012481971979141235,
"step": 64300
},
{
"epoch": 1.2482541899441342,
"grad_norm": 0.02977757714688778,
"learning_rate": 8.353000765143922e-06,
"loss": 0.010137701034545898,
"step": 64350
},
{
"epoch": 1.2492240844196152,
"grad_norm": 0.03336340934038162,
"learning_rate": 8.342224090179217e-06,
"loss": 0.0066069459915161135,
"step": 64400
},
{
"epoch": 1.2501939788950962,
"grad_norm": 0.06243349239230156,
"learning_rate": 8.33144741521451e-06,
"loss": 0.01119728684425354,
"step": 64450
},
{
"epoch": 1.2511638733705772,
"grad_norm": 0.004869701340794563,
"learning_rate": 8.320670740249804e-06,
"loss": 0.007395591139793396,
"step": 64500
},
{
"epoch": 1.2521337678460585,
"grad_norm": 2.503295660018921,
"learning_rate": 8.309894065285098e-06,
"loss": 0.01030503511428833,
"step": 64550
},
{
"epoch": 1.2531036623215395,
"grad_norm": 0.0004946495755575597,
"learning_rate": 8.299117390320392e-06,
"loss": 0.009273242950439454,
"step": 64600
},
{
"epoch": 1.2540735567970205,
"grad_norm": 0.28194698691368103,
"learning_rate": 8.288340715355685e-06,
"loss": 0.018097007274627687,
"step": 64650
},
{
"epoch": 1.2550434512725015,
"grad_norm": 0.028967037796974182,
"learning_rate": 8.277564040390978e-06,
"loss": 0.010817924737930298,
"step": 64700
},
{
"epoch": 1.2560133457479825,
"grad_norm": 0.24053630232810974,
"learning_rate": 8.266787365426273e-06,
"loss": 0.013611510992050171,
"step": 64750
},
{
"epoch": 1.2569832402234637,
"grad_norm": 0.05410028249025345,
"learning_rate": 8.256010690461566e-06,
"loss": 0.00991258144378662,
"step": 64800
},
{
"epoch": 1.2579531346989448,
"grad_norm": 1.8426182270050049,
"learning_rate": 8.245234015496859e-06,
"loss": 0.005527692437171936,
"step": 64850
},
{
"epoch": 1.2589230291744258,
"grad_norm": 0.022047562524676323,
"learning_rate": 8.234457340532152e-06,
"loss": 0.006785852313041687,
"step": 64900
},
{
"epoch": 1.259892923649907,
"grad_norm": 0.04178503900766373,
"learning_rate": 8.223680665567445e-06,
"loss": 0.012958492040634156,
"step": 64950
},
{
"epoch": 1.260862818125388,
"grad_norm": 0.2684505581855774,
"learning_rate": 8.21290399060274e-06,
"loss": 0.012150666713714599,
"step": 65000
},
{
"epoch": 1.261832712600869,
"grad_norm": 0.012189896777272224,
"learning_rate": 8.202127315638033e-06,
"loss": 0.005317040681838989,
"step": 65050
},
{
"epoch": 1.26280260707635,
"grad_norm": 0.008945767767727375,
"learning_rate": 8.191350640673328e-06,
"loss": 0.010611706972122192,
"step": 65100
},
{
"epoch": 1.263772501551831,
"grad_norm": 0.9385164380073547,
"learning_rate": 8.180573965708621e-06,
"loss": 0.010615568161010742,
"step": 65150
},
{
"epoch": 1.2647423960273123,
"grad_norm": 0.015549744479358196,
"learning_rate": 8.169797290743915e-06,
"loss": 0.012563055753707886,
"step": 65200
},
{
"epoch": 1.2657122905027933,
"grad_norm": 0.1653091013431549,
"learning_rate": 8.159020615779208e-06,
"loss": 0.007098273038864135,
"step": 65250
},
{
"epoch": 1.2666821849782743,
"grad_norm": 0.004874336998909712,
"learning_rate": 8.148243940814503e-06,
"loss": 0.012574076652526855,
"step": 65300
},
{
"epoch": 1.2676520794537554,
"grad_norm": 0.05352836102247238,
"learning_rate": 8.137467265849796e-06,
"loss": 0.011027153730392456,
"step": 65350
},
{
"epoch": 1.2686219739292364,
"grad_norm": 0.29458633065223694,
"learning_rate": 8.126690590885089e-06,
"loss": 0.011562031507492066,
"step": 65400
},
{
"epoch": 1.2695918684047176,
"grad_norm": 0.03348653018474579,
"learning_rate": 8.115913915920382e-06,
"loss": 0.0065821981430053715,
"step": 65450
},
{
"epoch": 1.2705617628801986,
"grad_norm": 0.08181615173816681,
"learning_rate": 8.105137240955677e-06,
"loss": 0.010553250312805176,
"step": 65500
},
{
"epoch": 1.2715316573556796,
"grad_norm": 0.022803325206041336,
"learning_rate": 8.09436056599097e-06,
"loss": 0.011514067649841309,
"step": 65550
},
{
"epoch": 1.2725015518311609,
"grad_norm": 0.14507730305194855,
"learning_rate": 8.083583891026263e-06,
"loss": 0.007614512443542481,
"step": 65600
},
{
"epoch": 1.2734714463066419,
"grad_norm": 0.09856431186199188,
"learning_rate": 8.072807216061556e-06,
"loss": 0.008254753351211548,
"step": 65650
},
{
"epoch": 1.274441340782123,
"grad_norm": 0.09062592685222626,
"learning_rate": 8.06203054109685e-06,
"loss": 0.0110701584815979,
"step": 65700
},
{
"epoch": 1.275411235257604,
"grad_norm": 0.30751603841781616,
"learning_rate": 8.051253866132144e-06,
"loss": 0.007464201450347901,
"step": 65750
},
{
"epoch": 1.276381129733085,
"grad_norm": 0.0068125370889902115,
"learning_rate": 8.04047719116744e-06,
"loss": 0.006973616480827331,
"step": 65800
},
{
"epoch": 1.2773510242085662,
"grad_norm": 0.10588119179010391,
"learning_rate": 8.029700516202732e-06,
"loss": 0.010719690322875976,
"step": 65850
},
{
"epoch": 1.2783209186840472,
"grad_norm": 0.009747414849698544,
"learning_rate": 8.018923841238026e-06,
"loss": 0.0075430721044540405,
"step": 65900
},
{
"epoch": 1.2792908131595282,
"grad_norm": 0.003475481178611517,
"learning_rate": 8.008147166273319e-06,
"loss": 0.009073325991630554,
"step": 65950
},
{
"epoch": 1.2802607076350094,
"grad_norm": 0.1000249981880188,
"learning_rate": 7.997370491308612e-06,
"loss": 0.011927846670150757,
"step": 66000
},
{
"epoch": 1.2812306021104904,
"grad_norm": 0.058050792664289474,
"learning_rate": 7.986593816343907e-06,
"loss": 0.0133299720287323,
"step": 66050
},
{
"epoch": 1.2822004965859715,
"grad_norm": 0.00040037647704593837,
"learning_rate": 7.9758171413792e-06,
"loss": 0.00852045714855194,
"step": 66100
},
{
"epoch": 1.2831703910614525,
"grad_norm": 0.0004742340533994138,
"learning_rate": 7.965040466414493e-06,
"loss": 0.005210243463516236,
"step": 66150
},
{
"epoch": 1.2841402855369335,
"grad_norm": 0.31300756335258484,
"learning_rate": 7.954263791449786e-06,
"loss": 0.0083791184425354,
"step": 66200
},
{
"epoch": 1.2851101800124147,
"grad_norm": 0.0017161185387521982,
"learning_rate": 7.94348711648508e-06,
"loss": 0.007947552800178528,
"step": 66250
},
{
"epoch": 1.2860800744878957,
"grad_norm": 0.07401105761528015,
"learning_rate": 7.932710441520374e-06,
"loss": 0.007666552066802978,
"step": 66300
},
{
"epoch": 1.2870499689633768,
"grad_norm": 0.0008379767532460392,
"learning_rate": 7.921933766555667e-06,
"loss": 0.014177801609039307,
"step": 66350
},
{
"epoch": 1.2880198634388578,
"grad_norm": 0.0009575898293405771,
"learning_rate": 7.91115709159096e-06,
"loss": 0.00780170202255249,
"step": 66400
},
{
"epoch": 1.2889897579143388,
"grad_norm": 0.040326375514268875,
"learning_rate": 7.900380416626254e-06,
"loss": 0.011748466491699219,
"step": 66450
},
{
"epoch": 1.28995965238982,
"grad_norm": 0.29700854420661926,
"learning_rate": 7.889603741661549e-06,
"loss": 0.014556131362915038,
"step": 66500
},
{
"epoch": 1.290929546865301,
"grad_norm": 0.05313584581017494,
"learning_rate": 7.878827066696842e-06,
"loss": 0.00525145947933197,
"step": 66550
},
{
"epoch": 1.291899441340782,
"grad_norm": 4.613727569580078,
"learning_rate": 7.868050391732137e-06,
"loss": 0.009507367610931397,
"step": 66600
},
{
"epoch": 1.2928693358162633,
"grad_norm": 0.014566008001565933,
"learning_rate": 7.85727371676743e-06,
"loss": 0.011751831769943237,
"step": 66650
},
{
"epoch": 1.2938392302917443,
"grad_norm": 0.0021229626145213842,
"learning_rate": 7.846497041802723e-06,
"loss": 0.009027530550956725,
"step": 66700
},
{
"epoch": 1.2948091247672253,
"grad_norm": 0.048986416310071945,
"learning_rate": 7.835720366838016e-06,
"loss": 0.00839992105960846,
"step": 66750
},
{
"epoch": 1.2957790192427063,
"grad_norm": 0.02986958809196949,
"learning_rate": 7.824943691873311e-06,
"loss": 0.007436380982398987,
"step": 66800
},
{
"epoch": 1.2967489137181873,
"grad_norm": 1.8449760675430298,
"learning_rate": 7.814167016908604e-06,
"loss": 0.007208214998245239,
"step": 66850
},
{
"epoch": 1.2977188081936686,
"grad_norm": 0.05832957476377487,
"learning_rate": 7.803390341943897e-06,
"loss": 0.0065231496095657345,
"step": 66900
},
{
"epoch": 1.2986887026691496,
"grad_norm": 0.5748994946479797,
"learning_rate": 7.79261366697919e-06,
"loss": 0.011616054773330688,
"step": 66950
},
{
"epoch": 1.2996585971446306,
"grad_norm": 0.04267243295907974,
"learning_rate": 7.781836992014484e-06,
"loss": 0.010908373594284058,
"step": 67000
},
{
"epoch": 1.3006284916201118,
"grad_norm": 0.0062993974424898624,
"learning_rate": 7.771060317049778e-06,
"loss": 0.010806225538253784,
"step": 67050
},
{
"epoch": 1.3015983860955929,
"grad_norm": 0.1283165067434311,
"learning_rate": 7.760283642085072e-06,
"loss": 0.008584854006767274,
"step": 67100
},
{
"epoch": 1.3025682805710739,
"grad_norm": 0.0026348961982876062,
"learning_rate": 7.749506967120365e-06,
"loss": 0.009924039244651794,
"step": 67150
},
{
"epoch": 1.303538175046555,
"grad_norm": 0.03382959961891174,
"learning_rate": 7.738730292155658e-06,
"loss": 0.010448306798934937,
"step": 67200
},
{
"epoch": 1.304508069522036,
"grad_norm": 0.059476085007190704,
"learning_rate": 7.727953617190953e-06,
"loss": 0.011789822578430175,
"step": 67250
},
{
"epoch": 1.3054779639975171,
"grad_norm": 5.338925361633301,
"learning_rate": 7.717176942226246e-06,
"loss": 0.010148279666900635,
"step": 67300
},
{
"epoch": 1.3064478584729982,
"grad_norm": 0.040930021554231644,
"learning_rate": 7.70640026726154e-06,
"loss": 0.012623599767684936,
"step": 67350
},
{
"epoch": 1.3074177529484792,
"grad_norm": 0.07057181745767593,
"learning_rate": 7.695623592296834e-06,
"loss": 0.007744224667549133,
"step": 67400
},
{
"epoch": 1.3083876474239604,
"grad_norm": 0.5895511507987976,
"learning_rate": 7.684846917332127e-06,
"loss": 0.00825598180294037,
"step": 67450
},
{
"epoch": 1.3093575418994414,
"grad_norm": 0.0006509998929686844,
"learning_rate": 7.67407024236742e-06,
"loss": 0.010783611536026002,
"step": 67500
},
{
"epoch": 1.3103274363749224,
"grad_norm": 0.23836155235767365,
"learning_rate": 7.663293567402715e-06,
"loss": 0.009960966110229492,
"step": 67550
},
{
"epoch": 1.3112973308504035,
"grad_norm": 0.0004456629976630211,
"learning_rate": 7.652516892438008e-06,
"loss": 0.011315641403198242,
"step": 67600
},
{
"epoch": 1.3122672253258845,
"grad_norm": 0.07909763604402542,
"learning_rate": 7.641740217473301e-06,
"loss": 0.011128449440002441,
"step": 67650
},
{
"epoch": 1.3132371198013657,
"grad_norm": 0.01849117875099182,
"learning_rate": 7.630963542508595e-06,
"loss": 0.010902687311172485,
"step": 67700
},
{
"epoch": 1.3142070142768467,
"grad_norm": 0.0009929228108376265,
"learning_rate": 7.6201868675438886e-06,
"loss": 0.008335170149803162,
"step": 67750
},
{
"epoch": 1.3151769087523277,
"grad_norm": 0.08210676908493042,
"learning_rate": 7.609410192579182e-06,
"loss": 0.011013150215148926,
"step": 67800
},
{
"epoch": 1.3161468032278087,
"grad_norm": 0.057643286883831024,
"learning_rate": 7.598633517614476e-06,
"loss": 0.005527942180633545,
"step": 67850
},
{
"epoch": 1.3171166977032898,
"grad_norm": 2.968712568283081,
"learning_rate": 7.587856842649769e-06,
"loss": 0.011944565773010254,
"step": 67900
},
{
"epoch": 1.318086592178771,
"grad_norm": 0.00513883214443922,
"learning_rate": 7.577080167685064e-06,
"loss": 0.007082827091217041,
"step": 67950
},
{
"epoch": 1.319056486654252,
"grad_norm": 0.0053660087287425995,
"learning_rate": 7.566303492720357e-06,
"loss": 0.015438473224639893,
"step": 68000
},
{
"epoch": 1.320026381129733,
"grad_norm": 0.006405562162399292,
"learning_rate": 7.555526817755651e-06,
"loss": 0.007593492269515991,
"step": 68050
},
{
"epoch": 1.3209962756052143,
"grad_norm": 0.012665732763707638,
"learning_rate": 7.544750142790944e-06,
"loss": 0.010739994049072266,
"step": 68100
},
{
"epoch": 1.3219661700806953,
"grad_norm": 0.24923312664031982,
"learning_rate": 7.533973467826238e-06,
"loss": 0.005363255143165589,
"step": 68150
},
{
"epoch": 1.3229360645561763,
"grad_norm": 0.16752086579799652,
"learning_rate": 7.523196792861531e-06,
"loss": 0.006659956574440002,
"step": 68200
},
{
"epoch": 1.3239059590316573,
"grad_norm": 0.47215214371681213,
"learning_rate": 7.512420117896824e-06,
"loss": 0.011202459335327148,
"step": 68250
},
{
"epoch": 1.3248758535071383,
"grad_norm": 0.03683118149638176,
"learning_rate": 7.501643442932118e-06,
"loss": 0.012623052597045898,
"step": 68300
},
{
"epoch": 1.3258457479826196,
"grad_norm": 0.13501344621181488,
"learning_rate": 7.4908667679674115e-06,
"loss": 0.00867580235004425,
"step": 68350
},
{
"epoch": 1.3268156424581006,
"grad_norm": 0.0020918198861181736,
"learning_rate": 7.4800900930027056e-06,
"loss": 0.010539799928665161,
"step": 68400
},
{
"epoch": 1.3277855369335816,
"grad_norm": 3.7614617347717285,
"learning_rate": 7.469313418037999e-06,
"loss": 0.014722129106521606,
"step": 68450
},
{
"epoch": 1.3287554314090628,
"grad_norm": 0.04981841892004013,
"learning_rate": 7.458536743073293e-06,
"loss": 0.005464362502098083,
"step": 68500
},
{
"epoch": 1.3297253258845438,
"grad_norm": 0.1246306449174881,
"learning_rate": 7.447760068108586e-06,
"loss": 0.006870024204254151,
"step": 68550
},
{
"epoch": 1.3306952203600249,
"grad_norm": 0.043045494705438614,
"learning_rate": 7.43698339314388e-06,
"loss": 0.0071089941263198855,
"step": 68600
},
{
"epoch": 1.3316651148355059,
"grad_norm": 0.45218825340270996,
"learning_rate": 7.426206718179174e-06,
"loss": 0.011919317245483398,
"step": 68650
},
{
"epoch": 1.3326350093109869,
"grad_norm": 0.08645722270011902,
"learning_rate": 7.415430043214468e-06,
"loss": 0.011730085611343383,
"step": 68700
},
{
"epoch": 1.3336049037864681,
"grad_norm": 0.0038552088662981987,
"learning_rate": 7.404653368249761e-06,
"loss": 0.01108281135559082,
"step": 68750
},
{
"epoch": 1.3345747982619491,
"grad_norm": 0.19621869921684265,
"learning_rate": 7.393876693285055e-06,
"loss": 0.010444782972335816,
"step": 68800
},
{
"epoch": 1.3355446927374302,
"grad_norm": 0.39344483613967896,
"learning_rate": 7.383100018320348e-06,
"loss": 0.009559204578399658,
"step": 68850
},
{
"epoch": 1.3365145872129112,
"grad_norm": 0.054704174399375916,
"learning_rate": 7.372323343355642e-06,
"loss": 0.009251146316528321,
"step": 68900
},
{
"epoch": 1.3374844816883922,
"grad_norm": 0.04725747928023338,
"learning_rate": 7.361546668390935e-06,
"loss": 0.01059167981147766,
"step": 68950
},
{
"epoch": 1.3384543761638734,
"grad_norm": 0.002974151400849223,
"learning_rate": 7.3507699934262286e-06,
"loss": 0.009108521342277527,
"step": 69000
},
{
"epoch": 1.3394242706393544,
"grad_norm": 0.0012344892602413893,
"learning_rate": 7.3399933184615226e-06,
"loss": 0.01135763645172119,
"step": 69050
},
{
"epoch": 1.3403941651148354,
"grad_norm": 0.08642445504665375,
"learning_rate": 7.329216643496816e-06,
"loss": 0.010228137969970703,
"step": 69100
},
{
"epoch": 1.3413640595903167,
"grad_norm": 0.44265326857566833,
"learning_rate": 7.31843996853211e-06,
"loss": 0.008213025331497193,
"step": 69150
},
{
"epoch": 1.3423339540657977,
"grad_norm": 0.18414172530174255,
"learning_rate": 7.307663293567403e-06,
"loss": 0.012606470584869385,
"step": 69200
},
{
"epoch": 1.3433038485412787,
"grad_norm": 0.052248504012823105,
"learning_rate": 7.296886618602697e-06,
"loss": 0.012610722780227662,
"step": 69250
},
{
"epoch": 1.3442737430167597,
"grad_norm": 0.001294712070375681,
"learning_rate": 7.28610994363799e-06,
"loss": 0.006971522569656372,
"step": 69300
},
{
"epoch": 1.3452436374922407,
"grad_norm": 0.0767633318901062,
"learning_rate": 7.275333268673283e-06,
"loss": 0.0096689373254776,
"step": 69350
},
{
"epoch": 1.346213531967722,
"grad_norm": 0.13307639956474304,
"learning_rate": 7.264556593708578e-06,
"loss": 0.009466250538825988,
"step": 69400
},
{
"epoch": 1.347183426443203,
"grad_norm": 0.0224497988820076,
"learning_rate": 7.253779918743872e-06,
"loss": 0.012855775356292724,
"step": 69450
},
{
"epoch": 1.348153320918684,
"grad_norm": 0.0029541756957769394,
"learning_rate": 7.243003243779165e-06,
"loss": 0.012144321203231811,
"step": 69500
},
{
"epoch": 1.3491232153941652,
"grad_norm": 0.12903974950313568,
"learning_rate": 7.232226568814459e-06,
"loss": 0.014945061206817626,
"step": 69550
},
{
"epoch": 1.3500931098696463,
"grad_norm": 0.0015178063185885549,
"learning_rate": 7.221449893849752e-06,
"loss": 0.007364567518234253,
"step": 69600
},
{
"epoch": 1.3510630043451273,
"grad_norm": 0.08197904378175735,
"learning_rate": 7.2106732188850456e-06,
"loss": 0.00746238112449646,
"step": 69650
},
{
"epoch": 1.3520328988206083,
"grad_norm": 0.04455501213669777,
"learning_rate": 7.1998965439203396e-06,
"loss": 0.006308745741844177,
"step": 69700
},
{
"epoch": 1.3530027932960893,
"grad_norm": 0.14569249749183655,
"learning_rate": 7.189119868955633e-06,
"loss": 0.009471026062965394,
"step": 69750
},
{
"epoch": 1.3539726877715705,
"grad_norm": 0.05018019303679466,
"learning_rate": 7.178343193990927e-06,
"loss": 0.008512671589851379,
"step": 69800
},
{
"epoch": 1.3549425822470516,
"grad_norm": 0.04129014536738396,
"learning_rate": 7.16756651902622e-06,
"loss": 0.010102065801620484,
"step": 69850
},
{
"epoch": 1.3559124767225326,
"grad_norm": 2.4785730838775635,
"learning_rate": 7.156789844061514e-06,
"loss": 0.012344313859939575,
"step": 69900
},
{
"epoch": 1.3568823711980136,
"grad_norm": 0.3468509018421173,
"learning_rate": 7.146013169096807e-06,
"loss": 0.005959228277206421,
"step": 69950
},
{
"epoch": 1.3578522656734946,
"grad_norm": 0.7895374894142151,
"learning_rate": 7.1352364941321e-06,
"loss": 0.010256036520004272,
"step": 70000
},
{
"epoch": 1.3588221601489758,
"grad_norm": 0.025011079385876656,
"learning_rate": 7.124459819167394e-06,
"loss": 0.010330511331558228,
"step": 70050
},
{
"epoch": 1.3597920546244568,
"grad_norm": 0.21916735172271729,
"learning_rate": 7.113683144202689e-06,
"loss": 0.010679669380187988,
"step": 70100
},
{
"epoch": 1.3607619490999379,
"grad_norm": 0.02261362597346306,
"learning_rate": 7.102906469237982e-06,
"loss": 0.005253351926803589,
"step": 70150
},
{
"epoch": 1.361731843575419,
"grad_norm": 0.06243106350302696,
"learning_rate": 7.092129794273276e-06,
"loss": 0.009046060442924499,
"step": 70200
},
{
"epoch": 1.3627017380509001,
"grad_norm": 0.06883809715509415,
"learning_rate": 7.081353119308569e-06,
"loss": 0.00660774290561676,
"step": 70250
},
{
"epoch": 1.3636716325263811,
"grad_norm": 0.139311745762825,
"learning_rate": 7.0705764443438626e-06,
"loss": 0.005113978385925293,
"step": 70300
},
{
"epoch": 1.3646415270018621,
"grad_norm": 0.09924070537090302,
"learning_rate": 7.0597997693791566e-06,
"loss": 0.010460654497146607,
"step": 70350
},
{
"epoch": 1.3656114214773432,
"grad_norm": 0.00047027209075167775,
"learning_rate": 7.04902309441445e-06,
"loss": 0.0073311817646026615,
"step": 70400
},
{
"epoch": 1.3665813159528244,
"grad_norm": 0.005119858309626579,
"learning_rate": 7.038246419449744e-06,
"loss": 0.008369648456573486,
"step": 70450
},
{
"epoch": 1.3675512104283054,
"grad_norm": 0.0019964384846389294,
"learning_rate": 7.027469744485037e-06,
"loss": 0.007828481197357177,
"step": 70500
},
{
"epoch": 1.3685211049037864,
"grad_norm": 0.031869687139987946,
"learning_rate": 7.016693069520331e-06,
"loss": 0.013325400352478027,
"step": 70550
},
{
"epoch": 1.3694909993792677,
"grad_norm": 0.35334184765815735,
"learning_rate": 7.005916394555624e-06,
"loss": 0.00903805136680603,
"step": 70600
},
{
"epoch": 1.3704608938547487,
"grad_norm": 0.6638148427009583,
"learning_rate": 6.995139719590918e-06,
"loss": 0.010333757400512695,
"step": 70650
},
{
"epoch": 1.3714307883302297,
"grad_norm": 0.11783026903867722,
"learning_rate": 6.984363044626211e-06,
"loss": 0.00865904450416565,
"step": 70700
},
{
"epoch": 1.3724006828057107,
"grad_norm": 0.2931656241416931,
"learning_rate": 6.973586369661504e-06,
"loss": 0.006304901838302612,
"step": 70750
},
{
"epoch": 1.3733705772811917,
"grad_norm": 0.35728830099105835,
"learning_rate": 6.962809694696799e-06,
"loss": 0.0071556586027145385,
"step": 70800
},
{
"epoch": 1.374340471756673,
"grad_norm": 0.11258313804864883,
"learning_rate": 6.952033019732093e-06,
"loss": 0.01360297679901123,
"step": 70850
},
{
"epoch": 1.375310366232154,
"grad_norm": 0.062232475727796555,
"learning_rate": 6.941256344767386e-06,
"loss": 0.010311110019683838,
"step": 70900
},
{
"epoch": 1.376280260707635,
"grad_norm": 0.10871279984712601,
"learning_rate": 6.9304796698026796e-06,
"loss": 0.009295150637626648,
"step": 70950
},
{
"epoch": 1.377250155183116,
"grad_norm": 0.020369572564959526,
"learning_rate": 6.9197029948379736e-06,
"loss": 0.011311392784118652,
"step": 71000
},
{
"epoch": 1.378220049658597,
"grad_norm": 0.0682191401720047,
"learning_rate": 6.908926319873267e-06,
"loss": 0.012672734260559083,
"step": 71050
},
{
"epoch": 1.3791899441340782,
"grad_norm": 0.0025100938510149717,
"learning_rate": 6.898149644908561e-06,
"loss": 0.010562305450439452,
"step": 71100
},
{
"epoch": 1.3801598386095593,
"grad_norm": 0.25089579820632935,
"learning_rate": 6.887372969943854e-06,
"loss": 0.009273111820220947,
"step": 71150
},
{
"epoch": 1.3811297330850403,
"grad_norm": 0.16741563379764557,
"learning_rate": 6.876596294979148e-06,
"loss": 0.0141736900806427,
"step": 71200
},
{
"epoch": 1.3820996275605215,
"grad_norm": 0.13076601922512054,
"learning_rate": 6.865819620014441e-06,
"loss": 0.011469494104385376,
"step": 71250
},
{
"epoch": 1.3830695220360025,
"grad_norm": 0.006428553722798824,
"learning_rate": 6.855042945049735e-06,
"loss": 0.009037972688674926,
"step": 71300
},
{
"epoch": 1.3840394165114835,
"grad_norm": 5.386749267578125,
"learning_rate": 6.844266270085028e-06,
"loss": 0.007603196501731873,
"step": 71350
},
{
"epoch": 1.3850093109869646,
"grad_norm": 0.273266077041626,
"learning_rate": 6.833489595120321e-06,
"loss": 0.014796878099441529,
"step": 71400
},
{
"epoch": 1.3859792054624456,
"grad_norm": 0.06695589423179626,
"learning_rate": 6.822712920155615e-06,
"loss": 0.009056371450424195,
"step": 71450
},
{
"epoch": 1.3869490999379268,
"grad_norm": 0.20413890480995178,
"learning_rate": 6.81193624519091e-06,
"loss": 0.010044739246368409,
"step": 71500
},
{
"epoch": 1.3879189944134078,
"grad_norm": 0.8267840147018433,
"learning_rate": 6.801159570226203e-06,
"loss": 0.009756130576133728,
"step": 71550
},
{
"epoch": 1.3888888888888888,
"grad_norm": 0.9581722021102905,
"learning_rate": 6.7903828952614966e-06,
"loss": 0.013105784654617309,
"step": 71600
},
{
"epoch": 1.38985878336437,
"grad_norm": 0.19870594143867493,
"learning_rate": 6.7796062202967906e-06,
"loss": 0.008386635184288026,
"step": 71650
},
{
"epoch": 1.390828677839851,
"grad_norm": 0.017886757850646973,
"learning_rate": 6.768829545332084e-06,
"loss": 0.012178926467895508,
"step": 71700
},
{
"epoch": 1.391798572315332,
"grad_norm": 0.5423800349235535,
"learning_rate": 6.758052870367378e-06,
"loss": 0.011919753551483154,
"step": 71750
},
{
"epoch": 1.3927684667908131,
"grad_norm": 0.5293105840682983,
"learning_rate": 6.747276195402671e-06,
"loss": 0.011678470373153687,
"step": 71800
},
{
"epoch": 1.3937383612662941,
"grad_norm": 0.05110400170087814,
"learning_rate": 6.736499520437965e-06,
"loss": 0.01019969344139099,
"step": 71850
},
{
"epoch": 1.3947082557417754,
"grad_norm": 0.17633084952831268,
"learning_rate": 6.725722845473258e-06,
"loss": 0.01095625877380371,
"step": 71900
},
{
"epoch": 1.3956781502172564,
"grad_norm": 0.257960706949234,
"learning_rate": 6.714946170508552e-06,
"loss": 0.010424728393554688,
"step": 71950
},
{
"epoch": 1.3966480446927374,
"grad_norm": 0.005542585160583258,
"learning_rate": 6.704169495543845e-06,
"loss": 0.010948737859725952,
"step": 72000
},
{
"epoch": 1.3976179391682184,
"grad_norm": 1.1198784112930298,
"learning_rate": 6.693392820579138e-06,
"loss": 0.012689379453659057,
"step": 72050
},
{
"epoch": 1.3985878336436994,
"grad_norm": 0.002227506833150983,
"learning_rate": 6.682616145614432e-06,
"loss": 0.011022214889526366,
"step": 72100
},
{
"epoch": 1.3995577281191807,
"grad_norm": 0.00536131439730525,
"learning_rate": 6.6718394706497256e-06,
"loss": 0.007920385003089905,
"step": 72150
},
{
"epoch": 1.4005276225946617,
"grad_norm": 0.24517302215099335,
"learning_rate": 6.6610627956850196e-06,
"loss": 0.01011483907699585,
"step": 72200
},
{
"epoch": 1.4014975170701427,
"grad_norm": 0.14677758514881134,
"learning_rate": 6.6502861207203136e-06,
"loss": 0.013012501001358033,
"step": 72250
},
{
"epoch": 1.402467411545624,
"grad_norm": 0.17985038459300995,
"learning_rate": 6.639509445755608e-06,
"loss": 0.008549560308456421,
"step": 72300
},
{
"epoch": 1.403437306021105,
"grad_norm": 0.08253972232341766,
"learning_rate": 6.628732770790901e-06,
"loss": 0.005260766148567199,
"step": 72350
},
{
"epoch": 1.404407200496586,
"grad_norm": 0.0006470762309618294,
"learning_rate": 6.617956095826195e-06,
"loss": 0.009364647269248962,
"step": 72400
},
{
"epoch": 1.405377094972067,
"grad_norm": 0.11058472841978073,
"learning_rate": 6.607179420861488e-06,
"loss": 0.010132018327713013,
"step": 72450
},
{
"epoch": 1.406346989447548,
"grad_norm": 0.10118259489536285,
"learning_rate": 6.596402745896782e-06,
"loss": 0.013946386575698853,
"step": 72500
},
{
"epoch": 1.4073168839230292,
"grad_norm": 0.1052485778927803,
"learning_rate": 6.585626070932075e-06,
"loss": 0.008086941242218017,
"step": 72550
},
{
"epoch": 1.4082867783985102,
"grad_norm": 0.0006936773424968123,
"learning_rate": 6.574849395967369e-06,
"loss": 0.009705582857131958,
"step": 72600
},
{
"epoch": 1.4092566728739913,
"grad_norm": 0.0038152984343469143,
"learning_rate": 6.564072721002662e-06,
"loss": 0.010051922798156738,
"step": 72650
},
{
"epoch": 1.4102265673494725,
"grad_norm": 0.3321658968925476,
"learning_rate": 6.553296046037955e-06,
"loss": 0.00870543897151947,
"step": 72700
},
{
"epoch": 1.4111964618249535,
"grad_norm": 0.3176027238368988,
"learning_rate": 6.542519371073249e-06,
"loss": 0.010105626583099365,
"step": 72750
},
{
"epoch": 1.4121663563004345,
"grad_norm": 0.09649870544672012,
"learning_rate": 6.5317426961085426e-06,
"loss": 0.004979972243309021,
"step": 72800
},
{
"epoch": 1.4131362507759155,
"grad_norm": 0.11178776621818542,
"learning_rate": 6.5209660211438366e-06,
"loss": 0.006694594621658325,
"step": 72850
},
{
"epoch": 1.4141061452513966,
"grad_norm": 0.005296346265822649,
"learning_rate": 6.51018934617913e-06,
"loss": 0.007952567338943482,
"step": 72900
},
{
"epoch": 1.4150760397268778,
"grad_norm": 0.07025102525949478,
"learning_rate": 6.499412671214425e-06,
"loss": 0.01599743962287903,
"step": 72950
},
{
"epoch": 1.4160459342023588,
"grad_norm": 0.025382163003087044,
"learning_rate": 6.488635996249718e-06,
"loss": 0.012523999214172363,
"step": 73000
},
{
"epoch": 1.4170158286778398,
"grad_norm": 0.007127868011593819,
"learning_rate": 6.477859321285012e-06,
"loss": 0.011725735664367676,
"step": 73050
},
{
"epoch": 1.417985723153321,
"grad_norm": 0.002921838779002428,
"learning_rate": 6.467082646320305e-06,
"loss": 0.009494543075561523,
"step": 73100
},
{
"epoch": 1.418955617628802,
"grad_norm": 0.05626332759857178,
"learning_rate": 6.456305971355599e-06,
"loss": 0.011280207633972169,
"step": 73150
},
{
"epoch": 1.419925512104283,
"grad_norm": 0.2256452739238739,
"learning_rate": 6.445529296390892e-06,
"loss": 0.006812911033630371,
"step": 73200
},
{
"epoch": 1.420895406579764,
"grad_norm": 1.1547918319702148,
"learning_rate": 6.434752621426186e-06,
"loss": 0.01126556634902954,
"step": 73250
},
{
"epoch": 1.4218653010552451,
"grad_norm": 0.6369937062263489,
"learning_rate": 6.423975946461479e-06,
"loss": 0.010319757461547851,
"step": 73300
},
{
"epoch": 1.4228351955307263,
"grad_norm": 0.045710254460573196,
"learning_rate": 6.413199271496772e-06,
"loss": 0.010197668075561524,
"step": 73350
},
{
"epoch": 1.4238050900062074,
"grad_norm": 1.460798740386963,
"learning_rate": 6.402422596532066e-06,
"loss": 0.009620085954666138,
"step": 73400
},
{
"epoch": 1.4247749844816884,
"grad_norm": 0.2446492463350296,
"learning_rate": 6.3916459215673596e-06,
"loss": 0.009981805086135864,
"step": 73450
},
{
"epoch": 1.4257448789571694,
"grad_norm": 0.30026975274086,
"learning_rate": 6.3808692466026536e-06,
"loss": 0.009454054832458496,
"step": 73500
},
{
"epoch": 1.4267147734326504,
"grad_norm": 3.976351022720337,
"learning_rate": 6.370092571637947e-06,
"loss": 0.013267358541488647,
"step": 73550
},
{
"epoch": 1.4276846679081316,
"grad_norm": 0.021770723164081573,
"learning_rate": 6.359315896673241e-06,
"loss": 0.0075926375389099125,
"step": 73600
},
{
"epoch": 1.4286545623836127,
"grad_norm": 0.2622995376586914,
"learning_rate": 6.348539221708535e-06,
"loss": 0.01376733899116516,
"step": 73650
},
{
"epoch": 1.4296244568590937,
"grad_norm": 1.6980684995651245,
"learning_rate": 6.337762546743829e-06,
"loss": 0.008843038082122803,
"step": 73700
},
{
"epoch": 1.430594351334575,
"grad_norm": 0.05233754962682724,
"learning_rate": 6.326985871779122e-06,
"loss": 0.007890216708183288,
"step": 73750
},
{
"epoch": 1.431564245810056,
"grad_norm": 0.19410008192062378,
"learning_rate": 6.316209196814416e-06,
"loss": 0.01182650089263916,
"step": 73800
},
{
"epoch": 1.432534140285537,
"grad_norm": 0.21509350836277008,
"learning_rate": 6.305432521849709e-06,
"loss": 0.016488548517227173,
"step": 73850
},
{
"epoch": 1.433504034761018,
"grad_norm": 0.10234174132347107,
"learning_rate": 6.294655846885003e-06,
"loss": 0.007265148162841797,
"step": 73900
},
{
"epoch": 1.434473929236499,
"grad_norm": 0.1807388812303543,
"learning_rate": 6.283879171920296e-06,
"loss": 0.00784001052379608,
"step": 73950
},
{
"epoch": 1.4354438237119802,
"grad_norm": 0.0774458795785904,
"learning_rate": 6.273102496955589e-06,
"loss": 0.00918035328388214,
"step": 74000
},
{
"epoch": 1.4364137181874612,
"grad_norm": 0.22661976516246796,
"learning_rate": 6.262325821990883e-06,
"loss": 0.012511824369430541,
"step": 74050
},
{
"epoch": 1.4373836126629422,
"grad_norm": 0.029347628355026245,
"learning_rate": 6.2515491470261766e-06,
"loss": 0.008441939949989319,
"step": 74100
},
{
"epoch": 1.4383535071384235,
"grad_norm": 0.0295832771807909,
"learning_rate": 6.2407724720614706e-06,
"loss": 0.006935799121856689,
"step": 74150
},
{
"epoch": 1.4393234016139045,
"grad_norm": 0.8323389291763306,
"learning_rate": 6.229995797096764e-06,
"loss": 0.008999732732772827,
"step": 74200
},
{
"epoch": 1.4402932960893855,
"grad_norm": 1.81358003616333,
"learning_rate": 6.219219122132058e-06,
"loss": 0.008964329361915588,
"step": 74250
},
{
"epoch": 1.4412631905648665,
"grad_norm": 0.24141357839107513,
"learning_rate": 6.208442447167351e-06,
"loss": 0.01323567509651184,
"step": 74300
},
{
"epoch": 1.4422330850403475,
"grad_norm": 0.002131373854354024,
"learning_rate": 6.197665772202646e-06,
"loss": 0.0074788802862167355,
"step": 74350
},
{
"epoch": 1.4432029795158288,
"grad_norm": 0.001160091022029519,
"learning_rate": 6.186889097237939e-06,
"loss": 0.012600805759429932,
"step": 74400
},
{
"epoch": 1.4441728739913098,
"grad_norm": 0.0450500063598156,
"learning_rate": 6.176112422273233e-06,
"loss": 0.009565934538841248,
"step": 74450
},
{
"epoch": 1.4451427684667908,
"grad_norm": 0.011452367529273033,
"learning_rate": 6.165335747308526e-06,
"loss": 0.008065120577812194,
"step": 74500
},
{
"epoch": 1.4461126629422718,
"grad_norm": 0.07390701025724411,
"learning_rate": 6.15455907234382e-06,
"loss": 0.012878499031066894,
"step": 74550
},
{
"epoch": 1.4470825574177528,
"grad_norm": 0.03945234790444374,
"learning_rate": 6.143782397379113e-06,
"loss": 0.007512972354888916,
"step": 74600
},
{
"epoch": 1.448052451893234,
"grad_norm": 0.15444202721118927,
"learning_rate": 6.133005722414407e-06,
"loss": 0.008061243891716004,
"step": 74650
},
{
"epoch": 1.449022346368715,
"grad_norm": 0.017603769898414612,
"learning_rate": 6.1222290474497e-06,
"loss": 0.009594188332557678,
"step": 74700
},
{
"epoch": 1.449992240844196,
"grad_norm": 0.30013507604599,
"learning_rate": 6.1114523724849936e-06,
"loss": 0.009222349524497986,
"step": 74750
},
{
"epoch": 1.4509621353196773,
"grad_norm": 0.1748836636543274,
"learning_rate": 6.100675697520288e-06,
"loss": 0.011984682083129883,
"step": 74800
},
{
"epoch": 1.4519320297951583,
"grad_norm": 0.04227448254823685,
"learning_rate": 6.089899022555581e-06,
"loss": 0.008730010390281677,
"step": 74850
},
{
"epoch": 1.4529019242706394,
"grad_norm": 0.2049311399459839,
"learning_rate": 6.079122347590875e-06,
"loss": 0.008296215534210205,
"step": 74900
},
{
"epoch": 1.4538718187461204,
"grad_norm": 0.08997488021850586,
"learning_rate": 6.068345672626168e-06,
"loss": 0.008181063532829285,
"step": 74950
},
{
"epoch": 1.4548417132216014,
"grad_norm": 0.45327746868133545,
"learning_rate": 6.057568997661462e-06,
"loss": 0.011192800998687745,
"step": 75000
},
{
"epoch": 1.4558116076970826,
"grad_norm": 0.04781936854124069,
"learning_rate": 6.046792322696755e-06,
"loss": 0.01098185896873474,
"step": 75050
},
{
"epoch": 1.4567815021725636,
"grad_norm": 0.08918268233537674,
"learning_rate": 6.03601564773205e-06,
"loss": 0.009501221776008605,
"step": 75100
},
{
"epoch": 1.4577513966480447,
"grad_norm": 0.003299060510471463,
"learning_rate": 6.025238972767343e-06,
"loss": 0.010712462663650512,
"step": 75150
},
{
"epoch": 1.4587212911235259,
"grad_norm": 3.91145658493042,
"learning_rate": 6.014462297802637e-06,
"loss": 0.010288523435592651,
"step": 75200
},
{
"epoch": 1.459691185599007,
"grad_norm": 0.04555096477270126,
"learning_rate": 6.00368562283793e-06,
"loss": 0.006607139110565185,
"step": 75250
},
{
"epoch": 1.460661080074488,
"grad_norm": 0.035806361585855484,
"learning_rate": 5.992908947873224e-06,
"loss": 0.009760237336158752,
"step": 75300
},
{
"epoch": 1.461630974549969,
"grad_norm": 0.4285770654678345,
"learning_rate": 5.982132272908517e-06,
"loss": 0.01283442735671997,
"step": 75350
},
{
"epoch": 1.46260086902545,
"grad_norm": 1.976847767829895,
"learning_rate": 5.9713555979438106e-06,
"loss": 0.00793339967727661,
"step": 75400
},
{
"epoch": 1.4635707635009312,
"grad_norm": 0.005553924012929201,
"learning_rate": 5.960578922979105e-06,
"loss": 0.013299705982208252,
"step": 75450
},
{
"epoch": 1.4645406579764122,
"grad_norm": 0.09472525119781494,
"learning_rate": 5.949802248014398e-06,
"loss": 0.006723220944404602,
"step": 75500
},
{
"epoch": 1.4655105524518932,
"grad_norm": 0.3559606671333313,
"learning_rate": 5.939025573049692e-06,
"loss": 0.012488783597946166,
"step": 75550
},
{
"epoch": 1.4664804469273742,
"grad_norm": 0.13156059384346008,
"learning_rate": 5.928248898084985e-06,
"loss": 0.006043215394020081,
"step": 75600
},
{
"epoch": 1.4674503414028552,
"grad_norm": 0.3109842836856842,
"learning_rate": 5.917472223120279e-06,
"loss": 0.0075871914625167845,
"step": 75650
},
{
"epoch": 1.4684202358783365,
"grad_norm": 0.017117930576205254,
"learning_rate": 5.906695548155572e-06,
"loss": 0.010498493909835815,
"step": 75700
},
{
"epoch": 1.4693901303538175,
"grad_norm": 0.5605753660202026,
"learning_rate": 5.895918873190865e-06,
"loss": 0.010050270557403564,
"step": 75750
},
{
"epoch": 1.4703600248292985,
"grad_norm": 0.019117014482617378,
"learning_rate": 5.88514219822616e-06,
"loss": 0.009482232928276062,
"step": 75800
},
{
"epoch": 1.4713299193047797,
"grad_norm": 0.0051208860240876675,
"learning_rate": 5.874365523261454e-06,
"loss": 0.0072123616933822635,
"step": 75850
},
{
"epoch": 1.4722998137802608,
"grad_norm": 5.851313591003418,
"learning_rate": 5.863588848296747e-06,
"loss": 0.009033583998680116,
"step": 75900
},
{
"epoch": 1.4732697082557418,
"grad_norm": 0.04122700169682503,
"learning_rate": 5.852812173332041e-06,
"loss": 0.01291088104248047,
"step": 75950
},
{
"epoch": 1.4742396027312228,
"grad_norm": 0.21355517208576202,
"learning_rate": 5.842035498367334e-06,
"loss": 0.00955152690410614,
"step": 76000
},
{
"epoch": 1.4752094972067038,
"grad_norm": 0.00910529401153326,
"learning_rate": 5.831258823402628e-06,
"loss": 0.009005100727081298,
"step": 76050
},
{
"epoch": 1.476179391682185,
"grad_norm": 0.04961603134870529,
"learning_rate": 5.820482148437922e-06,
"loss": 0.007267175316810608,
"step": 76100
},
{
"epoch": 1.477149286157666,
"grad_norm": 0.004128265660256147,
"learning_rate": 5.809705473473215e-06,
"loss": 0.013259943723678589,
"step": 76150
},
{
"epoch": 1.478119180633147,
"grad_norm": 0.01046920195221901,
"learning_rate": 5.798928798508509e-06,
"loss": 0.006970714330673217,
"step": 76200
},
{
"epoch": 1.4790890751086283,
"grad_norm": 0.0037826518528163433,
"learning_rate": 5.788152123543802e-06,
"loss": 0.008893852233886718,
"step": 76250
},
{
"epoch": 1.4800589695841093,
"grad_norm": 0.30438464879989624,
"learning_rate": 5.777375448579096e-06,
"loss": 0.009275071620941162,
"step": 76300
},
{
"epoch": 1.4810288640595903,
"grad_norm": 0.12030501663684845,
"learning_rate": 5.766598773614389e-06,
"loss": 0.008245822787284852,
"step": 76350
},
{
"epoch": 1.4819987585350713,
"grad_norm": 0.008981515653431416,
"learning_rate": 5.755822098649683e-06,
"loss": 0.007870566248893738,
"step": 76400
},
{
"epoch": 1.4829686530105524,
"grad_norm": 0.009698296897113323,
"learning_rate": 5.745045423684976e-06,
"loss": 0.00824795663356781,
"step": 76450
},
{
"epoch": 1.4839385474860336,
"grad_norm": 0.049621641635894775,
"learning_rate": 5.734268748720271e-06,
"loss": 0.013801581859588623,
"step": 76500
},
{
"epoch": 1.4849084419615146,
"grad_norm": 0.0017502781702205539,
"learning_rate": 5.723492073755564e-06,
"loss": 0.008995899558067321,
"step": 76550
},
{
"epoch": 1.4858783364369956,
"grad_norm": 0.4935099184513092,
"learning_rate": 5.712715398790858e-06,
"loss": 0.00963567316532135,
"step": 76600
},
{
"epoch": 1.4868482309124766,
"grad_norm": 0.058969646692276,
"learning_rate": 5.7019387238261514e-06,
"loss": 0.010586849451065063,
"step": 76650
},
{
"epoch": 1.4878181253879577,
"grad_norm": 0.15165823698043823,
"learning_rate": 5.691162048861445e-06,
"loss": 0.012565140724182128,
"step": 76700
},
{
"epoch": 1.488788019863439,
"grad_norm": 4.187343120574951,
"learning_rate": 5.680385373896739e-06,
"loss": 0.006915196776390076,
"step": 76750
},
{
"epoch": 1.48975791433892,
"grad_norm": 0.030921652913093567,
"learning_rate": 5.669608698932032e-06,
"loss": 0.006324300169944763,
"step": 76800
},
{
"epoch": 1.490727808814401,
"grad_norm": 0.027608945965766907,
"learning_rate": 5.658832023967326e-06,
"loss": 0.008555254340171814,
"step": 76850
},
{
"epoch": 1.4916977032898822,
"grad_norm": 0.02391733042895794,
"learning_rate": 5.648055349002619e-06,
"loss": 0.007927553653717041,
"step": 76900
},
{
"epoch": 1.4926675977653632,
"grad_norm": 0.0026172336656600237,
"learning_rate": 5.637278674037913e-06,
"loss": 0.008674402236938477,
"step": 76950
},
{
"epoch": 1.4936374922408442,
"grad_norm": 0.08146928250789642,
"learning_rate": 5.626501999073206e-06,
"loss": 0.00879134178161621,
"step": 77000
},
{
"epoch": 1.4946073867163252,
"grad_norm": 1.4192004203796387,
"learning_rate": 5.6157253241085e-06,
"loss": 0.01095924735069275,
"step": 77050
},
{
"epoch": 1.4955772811918062,
"grad_norm": 0.024049248546361923,
"learning_rate": 5.604948649143793e-06,
"loss": 0.009870456457138061,
"step": 77100
},
{
"epoch": 1.4965471756672875,
"grad_norm": 0.2836850583553314,
"learning_rate": 5.594171974179086e-06,
"loss": 0.0125683331489563,
"step": 77150
},
{
"epoch": 1.4975170701427685,
"grad_norm": 0.10161936283111572,
"learning_rate": 5.583395299214381e-06,
"loss": 0.011144528388977051,
"step": 77200
},
{
"epoch": 1.4984869646182495,
"grad_norm": 2.358774185180664,
"learning_rate": 5.572618624249675e-06,
"loss": 0.010686475038528442,
"step": 77250
},
{
"epoch": 1.4994568590937307,
"grad_norm": 0.34739387035369873,
"learning_rate": 5.5618419492849684e-06,
"loss": 0.008340981006622315,
"step": 77300
},
{
"epoch": 1.5004267535692115,
"grad_norm": 0.013450550846755505,
"learning_rate": 5.551065274320262e-06,
"loss": 0.011444764137268066,
"step": 77350
},
{
"epoch": 1.5013966480446927,
"grad_norm": 0.3263842463493347,
"learning_rate": 5.540288599355556e-06,
"loss": 0.009913877248764039,
"step": 77400
},
{
"epoch": 1.5023665425201738,
"grad_norm": 0.005623707547783852,
"learning_rate": 5.529511924390849e-06,
"loss": 0.01237200379371643,
"step": 77450
},
{
"epoch": 1.5033364369956548,
"grad_norm": 0.015493443235754967,
"learning_rate": 5.518735249426143e-06,
"loss": 0.007900894284248353,
"step": 77500
},
{
"epoch": 1.504306331471136,
"grad_norm": 1.4616572856903076,
"learning_rate": 5.507958574461436e-06,
"loss": 0.012082492113113403,
"step": 77550
},
{
"epoch": 1.505276225946617,
"grad_norm": 0.3165079355239868,
"learning_rate": 5.49718189949673e-06,
"loss": 0.0057794547080993654,
"step": 77600
},
{
"epoch": 1.506246120422098,
"grad_norm": 0.882084846496582,
"learning_rate": 5.486405224532023e-06,
"loss": 0.008174406290054321,
"step": 77650
},
{
"epoch": 1.5072160148975793,
"grad_norm": 0.14873750507831573,
"learning_rate": 5.475628549567317e-06,
"loss": 0.009051690101623535,
"step": 77700
},
{
"epoch": 1.50818590937306,
"grad_norm": 0.014967982657253742,
"learning_rate": 5.46485187460261e-06,
"loss": 0.00808133840560913,
"step": 77750
},
{
"epoch": 1.5091558038485413,
"grad_norm": 0.01077311672270298,
"learning_rate": 5.454075199637903e-06,
"loss": 0.006903171539306641,
"step": 77800
},
{
"epoch": 1.5101256983240223,
"grad_norm": 0.8390721082687378,
"learning_rate": 5.443298524673197e-06,
"loss": 0.007945066690444947,
"step": 77850
},
{
"epoch": 1.5110955927995033,
"grad_norm": 0.043719176203012466,
"learning_rate": 5.432521849708491e-06,
"loss": 0.01224572777748108,
"step": 77900
},
{
"epoch": 1.5120654872749846,
"grad_norm": 0.17702528834342957,
"learning_rate": 5.4217451747437854e-06,
"loss": 0.0071883141994476315,
"step": 77950
},
{
"epoch": 1.5130353817504656,
"grad_norm": 0.5912231802940369,
"learning_rate": 5.410968499779079e-06,
"loss": 0.011090841293334961,
"step": 78000
},
{
"epoch": 1.5140052762259466,
"grad_norm": 0.11702877283096313,
"learning_rate": 5.400191824814373e-06,
"loss": 0.00891302764415741,
"step": 78050
},
{
"epoch": 1.5149751707014278,
"grad_norm": 0.02083415724337101,
"learning_rate": 5.389415149849666e-06,
"loss": 0.006583401560783386,
"step": 78100
},
{
"epoch": 1.5159450651769086,
"grad_norm": 0.07226935029029846,
"learning_rate": 5.37863847488496e-06,
"loss": 0.008914055824279786,
"step": 78150
},
{
"epoch": 1.5169149596523899,
"grad_norm": 0.054130785167217255,
"learning_rate": 5.367861799920253e-06,
"loss": 0.007104674577713013,
"step": 78200
},
{
"epoch": 1.5178848541278709,
"grad_norm": 0.3777521848678589,
"learning_rate": 5.357085124955547e-06,
"loss": 0.008360546827316285,
"step": 78250
},
{
"epoch": 1.518854748603352,
"grad_norm": 0.06439903378486633,
"learning_rate": 5.34630844999084e-06,
"loss": 0.008698814511299134,
"step": 78300
},
{
"epoch": 1.5198246430788331,
"grad_norm": 0.25060492753982544,
"learning_rate": 5.335531775026134e-06,
"loss": 0.011243381500244141,
"step": 78350
},
{
"epoch": 1.520794537554314,
"grad_norm": 0.2823356091976166,
"learning_rate": 5.324755100061427e-06,
"loss": 0.008227289319038392,
"step": 78400
},
{
"epoch": 1.5217644320297952,
"grad_norm": 0.03710300475358963,
"learning_rate": 5.31397842509672e-06,
"loss": 0.00855037808418274,
"step": 78450
},
{
"epoch": 1.5227343265052762,
"grad_norm": 0.055223166942596436,
"learning_rate": 5.303201750132014e-06,
"loss": 0.006666557788848877,
"step": 78500
},
{
"epoch": 1.5237042209807572,
"grad_norm": 2.735996723175049,
"learning_rate": 5.292425075167308e-06,
"loss": 0.008281652331352234,
"step": 78550
},
{
"epoch": 1.5246741154562384,
"grad_norm": 0.0023074927739799023,
"learning_rate": 5.281648400202602e-06,
"loss": 0.0062324595451354985,
"step": 78600
},
{
"epoch": 1.5256440099317194,
"grad_norm": 0.34287092089653015,
"learning_rate": 5.2708717252378964e-06,
"loss": 0.010197151899337769,
"step": 78650
},
{
"epoch": 1.5266139044072005,
"grad_norm": 0.2212245613336563,
"learning_rate": 5.26009505027319e-06,
"loss": 0.0072429358959198,
"step": 78700
},
{
"epoch": 1.5275837988826817,
"grad_norm": 0.3682020604610443,
"learning_rate": 5.249318375308483e-06,
"loss": 0.016049787998199463,
"step": 78750
},
{
"epoch": 1.5285536933581625,
"grad_norm": 0.08261451125144958,
"learning_rate": 5.238541700343777e-06,
"loss": 0.010055978298187256,
"step": 78800
},
{
"epoch": 1.5295235878336437,
"grad_norm": 0.039223406463861465,
"learning_rate": 5.22776502537907e-06,
"loss": 0.005088759064674378,
"step": 78850
},
{
"epoch": 1.5304934823091247,
"grad_norm": 0.2278864085674286,
"learning_rate": 5.216988350414364e-06,
"loss": 0.007845354080200196,
"step": 78900
},
{
"epoch": 1.5314633767846058,
"grad_norm": 4.909081935882568,
"learning_rate": 5.206211675449657e-06,
"loss": 0.008187551498413086,
"step": 78950
},
{
"epoch": 1.532433271260087,
"grad_norm": 0.48308756947517395,
"learning_rate": 5.195435000484951e-06,
"loss": 0.010789599418640137,
"step": 79000
},
{
"epoch": 1.533403165735568,
"grad_norm": 0.031995199620723724,
"learning_rate": 5.184658325520244e-06,
"loss": 0.009174128174781799,
"step": 79050
},
{
"epoch": 1.534373060211049,
"grad_norm": 0.07204180210828781,
"learning_rate": 5.173881650555537e-06,
"loss": 0.013010393381118774,
"step": 79100
},
{
"epoch": 1.5353429546865303,
"grad_norm": 0.12008102238178253,
"learning_rate": 5.1631049755908314e-06,
"loss": 0.012241389751434326,
"step": 79150
},
{
"epoch": 1.536312849162011,
"grad_norm": 0.08391275256872177,
"learning_rate": 5.152328300626125e-06,
"loss": 0.007093976140022278,
"step": 79200
},
{
"epoch": 1.5372827436374923,
"grad_norm": 0.29354989528656006,
"learning_rate": 5.141551625661419e-06,
"loss": 0.008549766540527344,
"step": 79250
},
{
"epoch": 1.5382526381129733,
"grad_norm": 0.2891119122505188,
"learning_rate": 5.130774950696712e-06,
"loss": 0.007381114959716797,
"step": 79300
},
{
"epoch": 1.5392225325884543,
"grad_norm": 0.13021059334278107,
"learning_rate": 5.119998275732007e-06,
"loss": 0.009071699976921081,
"step": 79350
},
{
"epoch": 1.5401924270639356,
"grad_norm": 0.08805567026138306,
"learning_rate": 5.1092216007673e-06,
"loss": 0.013956053256988525,
"step": 79400
},
{
"epoch": 1.5411623215394163,
"grad_norm": 0.0008855572086758912,
"learning_rate": 5.098444925802594e-06,
"loss": 0.009428901672363281,
"step": 79450
},
{
"epoch": 1.5421322160148976,
"grad_norm": 0.07261622697114944,
"learning_rate": 5.087668250837887e-06,
"loss": 0.006743089556694031,
"step": 79500
},
{
"epoch": 1.5431021104903786,
"grad_norm": 0.4095045030117035,
"learning_rate": 5.076891575873181e-06,
"loss": 0.010814261436462403,
"step": 79550
},
{
"epoch": 1.5440720049658596,
"grad_norm": 0.08352615684270859,
"learning_rate": 5.066114900908474e-06,
"loss": 0.008777419924736024,
"step": 79600
},
{
"epoch": 1.5450418994413408,
"grad_norm": 0.14103566110134125,
"learning_rate": 5.055338225943768e-06,
"loss": 0.009997891783714295,
"step": 79650
},
{
"epoch": 1.5460117939168219,
"grad_norm": 0.005672645289450884,
"learning_rate": 5.044561550979061e-06,
"loss": 0.010445317029953003,
"step": 79700
},
{
"epoch": 1.5469816883923029,
"grad_norm": 0.06726996600627899,
"learning_rate": 5.033784876014354e-06,
"loss": 0.006812725067138672,
"step": 79750
},
{
"epoch": 1.5479515828677841,
"grad_norm": 0.05187563598155975,
"learning_rate": 5.0230082010496484e-06,
"loss": 0.008955934643745422,
"step": 79800
},
{
"epoch": 1.548921477343265,
"grad_norm": 2.5110349655151367,
"learning_rate": 5.012231526084942e-06,
"loss": 0.01137693166732788,
"step": 79850
},
{
"epoch": 1.5498913718187461,
"grad_norm": 0.4152529239654541,
"learning_rate": 5.001454851120236e-06,
"loss": 0.010278736352920531,
"step": 79900
},
{
"epoch": 1.5508612662942272,
"grad_norm": 0.07518861442804337,
"learning_rate": 4.99067817615553e-06,
"loss": 0.0150099515914917,
"step": 79950
},
{
"epoch": 1.5518311607697082,
"grad_norm": 0.001457787468098104,
"learning_rate": 4.979901501190823e-06,
"loss": 0.007580052018165588,
"step": 80000
},
{
"epoch": 1.5528010552451894,
"grad_norm": 0.06464844197034836,
"learning_rate": 4.969124826226117e-06,
"loss": 0.00509932816028595,
"step": 80050
},
{
"epoch": 1.5537709497206704,
"grad_norm": 0.06645611673593521,
"learning_rate": 4.95834815126141e-06,
"loss": 0.005640425086021424,
"step": 80100
},
{
"epoch": 1.5547408441961514,
"grad_norm": 0.015417337417602539,
"learning_rate": 4.947571476296704e-06,
"loss": 0.008679096698760986,
"step": 80150
},
{
"epoch": 1.5557107386716327,
"grad_norm": 0.6081642508506775,
"learning_rate": 4.936794801331997e-06,
"loss": 0.013196330070495605,
"step": 80200
},
{
"epoch": 1.5566806331471135,
"grad_norm": 0.0005010347231291234,
"learning_rate": 4.926018126367291e-06,
"loss": 0.009312012791633606,
"step": 80250
},
{
"epoch": 1.5576505276225947,
"grad_norm": 0.07565834373235703,
"learning_rate": 4.915241451402585e-06,
"loss": 0.008880857229232788,
"step": 80300
},
{
"epoch": 1.5586204220980757,
"grad_norm": 0.0812416672706604,
"learning_rate": 4.904464776437878e-06,
"loss": 0.007441931962966919,
"step": 80350
},
{
"epoch": 1.5595903165735567,
"grad_norm": 0.002417659154161811,
"learning_rate": 4.893688101473172e-06,
"loss": 0.008134024739265442,
"step": 80400
},
{
"epoch": 1.560560211049038,
"grad_norm": 0.05843966826796532,
"learning_rate": 4.8829114265084654e-06,
"loss": 0.007802451252937317,
"step": 80450
},
{
"epoch": 1.561530105524519,
"grad_norm": 0.0016281628049910069,
"learning_rate": 4.872134751543759e-06,
"loss": 0.007473285794258117,
"step": 80500
},
{
"epoch": 1.5625,
"grad_norm": 2.1438279151916504,
"learning_rate": 4.861358076579053e-06,
"loss": 0.011270670890808106,
"step": 80550
},
{
"epoch": 1.563469894475481,
"grad_norm": 0.014392624609172344,
"learning_rate": 4.850581401614347e-06,
"loss": 0.009168186187744141,
"step": 80600
},
{
"epoch": 1.564439788950962,
"grad_norm": 0.0008996193064376712,
"learning_rate": 4.83980472664964e-06,
"loss": 0.0084586501121521,
"step": 80650
},
{
"epoch": 1.5654096834264433,
"grad_norm": 0.005586319603025913,
"learning_rate": 4.829028051684934e-06,
"loss": 0.008385965824127197,
"step": 80700
},
{
"epoch": 1.5663795779019243,
"grad_norm": 0.07212182134389877,
"learning_rate": 4.818251376720227e-06,
"loss": 0.0046138232946395874,
"step": 80750
},
{
"epoch": 1.5673494723774053,
"grad_norm": 0.05553540587425232,
"learning_rate": 4.807474701755521e-06,
"loss": 0.013522316217422486,
"step": 80800
},
{
"epoch": 1.5683193668528865,
"grad_norm": 0.008243289776146412,
"learning_rate": 4.796698026790814e-06,
"loss": 0.011394627094268799,
"step": 80850
},
{
"epoch": 1.5692892613283673,
"grad_norm": 0.13433493673801422,
"learning_rate": 4.785921351826108e-06,
"loss": 0.010846952199935913,
"step": 80900
},
{
"epoch": 1.5702591558038486,
"grad_norm": 0.5418771505355835,
"learning_rate": 4.775144676861402e-06,
"loss": 0.009304192662239075,
"step": 80950
},
{
"epoch": 1.5712290502793296,
"grad_norm": 0.018216246739029884,
"learning_rate": 4.764368001896695e-06,
"loss": 0.009548236131668091,
"step": 81000
},
{
"epoch": 1.5721989447548106,
"grad_norm": 0.003914991859346628,
"learning_rate": 4.753591326931989e-06,
"loss": 0.009060502648353577,
"step": 81050
},
{
"epoch": 1.5731688392302918,
"grad_norm": 0.030777152627706528,
"learning_rate": 4.7428146519672824e-06,
"loss": 0.007650317549705505,
"step": 81100
},
{
"epoch": 1.5741387337057728,
"grad_norm": 0.20505507290363312,
"learning_rate": 4.732037977002576e-06,
"loss": 0.00566028356552124,
"step": 81150
},
{
"epoch": 1.5751086281812539,
"grad_norm": 0.2508404552936554,
"learning_rate": 4.72126130203787e-06,
"loss": 0.006037653684616089,
"step": 81200
},
{
"epoch": 1.576078522656735,
"grad_norm": 0.001980806700885296,
"learning_rate": 4.710484627073163e-06,
"loss": 0.011073638200759888,
"step": 81250
},
{
"epoch": 1.5770484171322159,
"grad_norm": 0.0010654466459527612,
"learning_rate": 4.699707952108457e-06,
"loss": 0.013449044227600097,
"step": 81300
},
{
"epoch": 1.5780183116076971,
"grad_norm": 0.9233915209770203,
"learning_rate": 4.688931277143751e-06,
"loss": 0.008911912441253661,
"step": 81350
},
{
"epoch": 1.5789882060831781,
"grad_norm": 0.11541356146335602,
"learning_rate": 4.678154602179044e-06,
"loss": 0.010942102670669555,
"step": 81400
},
{
"epoch": 1.5799581005586592,
"grad_norm": 0.0042203618213534355,
"learning_rate": 4.667377927214338e-06,
"loss": 0.008917818069458008,
"step": 81450
},
{
"epoch": 1.5809279950341404,
"grad_norm": 0.21217826008796692,
"learning_rate": 4.656601252249631e-06,
"loss": 0.010758384466171264,
"step": 81500
},
{
"epoch": 1.5818978895096214,
"grad_norm": 0.008385684341192245,
"learning_rate": 4.645824577284925e-06,
"loss": 0.009808256030082703,
"step": 81550
},
{
"epoch": 1.5828677839851024,
"grad_norm": 0.030433371663093567,
"learning_rate": 4.635047902320218e-06,
"loss": 0.014509979486465454,
"step": 81600
},
{
"epoch": 1.5838376784605837,
"grad_norm": 1.3596315383911133,
"learning_rate": 4.624271227355512e-06,
"loss": 0.00957200288772583,
"step": 81650
},
{
"epoch": 1.5848075729360644,
"grad_norm": 0.21091458201408386,
"learning_rate": 4.613494552390806e-06,
"loss": 0.009889385104179383,
"step": 81700
},
{
"epoch": 1.5857774674115457,
"grad_norm": 0.16576290130615234,
"learning_rate": 4.6027178774260994e-06,
"loss": 0.01096264123916626,
"step": 81750
},
{
"epoch": 1.5867473618870267,
"grad_norm": 0.07737179100513458,
"learning_rate": 4.591941202461393e-06,
"loss": 0.007406131029129028,
"step": 81800
},
{
"epoch": 1.5877172563625077,
"grad_norm": 0.056661829352378845,
"learning_rate": 4.581164527496687e-06,
"loss": 0.006993125677108764,
"step": 81850
},
{
"epoch": 1.588687150837989,
"grad_norm": 1.9635424613952637,
"learning_rate": 4.57038785253198e-06,
"loss": 0.013776564598083496,
"step": 81900
},
{
"epoch": 1.5896570453134697,
"grad_norm": 0.0007083421805873513,
"learning_rate": 4.559611177567274e-06,
"loss": 0.009855226278305054,
"step": 81950
},
{
"epoch": 1.590626939788951,
"grad_norm": 0.004925829358398914,
"learning_rate": 4.548834502602568e-06,
"loss": 0.011289727687835694,
"step": 82000
},
{
"epoch": 1.591596834264432,
"grad_norm": 0.0016559486975893378,
"learning_rate": 4.538057827637861e-06,
"loss": 0.010576225519180298,
"step": 82050
},
{
"epoch": 1.592566728739913,
"grad_norm": 0.9126780033111572,
"learning_rate": 4.527281152673155e-06,
"loss": 0.007786025404930115,
"step": 82100
},
{
"epoch": 1.5935366232153942,
"grad_norm": 0.0006675008335150778,
"learning_rate": 4.516504477708448e-06,
"loss": 0.0060846364498138425,
"step": 82150
},
{
"epoch": 1.5945065176908753,
"grad_norm": 0.01810169406235218,
"learning_rate": 4.505727802743742e-06,
"loss": 0.004783752262592316,
"step": 82200
},
{
"epoch": 1.5954764121663563,
"grad_norm": 0.004018501378595829,
"learning_rate": 4.494951127779035e-06,
"loss": 0.009719109535217286,
"step": 82250
},
{
"epoch": 1.5964463066418375,
"grad_norm": 0.04265689104795456,
"learning_rate": 4.4841744528143284e-06,
"loss": 0.009485295414924622,
"step": 82300
},
{
"epoch": 1.5974162011173183,
"grad_norm": 0.05259282514452934,
"learning_rate": 4.4733977778496224e-06,
"loss": 0.014399540424346925,
"step": 82350
},
{
"epoch": 1.5983860955927995,
"grad_norm": 0.09210766851902008,
"learning_rate": 4.4626211028849164e-06,
"loss": 0.008068094849586488,
"step": 82400
},
{
"epoch": 1.5993559900682806,
"grad_norm": 4.598770618438721,
"learning_rate": 4.45184442792021e-06,
"loss": 0.011852128505706787,
"step": 82450
},
{
"epoch": 1.6003258845437616,
"grad_norm": 1.536316156387329,
"learning_rate": 4.441067752955504e-06,
"loss": 0.011672877073287964,
"step": 82500
},
{
"epoch": 1.6012957790192428,
"grad_norm": 0.0015808655880391598,
"learning_rate": 4.430291077990797e-06,
"loss": 0.011356513500213623,
"step": 82550
},
{
"epoch": 1.6022656734947238,
"grad_norm": 0.02019406668841839,
"learning_rate": 4.419514403026091e-06,
"loss": 0.008576840162277222,
"step": 82600
},
{
"epoch": 1.6032355679702048,
"grad_norm": 0.05721912533044815,
"learning_rate": 4.408737728061384e-06,
"loss": 0.009753488898277284,
"step": 82650
},
{
"epoch": 1.604205462445686,
"grad_norm": 0.2310655266046524,
"learning_rate": 4.397961053096678e-06,
"loss": 0.011764360666275024,
"step": 82700
},
{
"epoch": 1.6051753569211669,
"grad_norm": 0.11740949749946594,
"learning_rate": 4.387184378131972e-06,
"loss": 0.00880060076713562,
"step": 82750
},
{
"epoch": 1.606145251396648,
"grad_norm": 0.010400653816759586,
"learning_rate": 4.376407703167265e-06,
"loss": 0.006225829720497131,
"step": 82800
},
{
"epoch": 1.6071151458721291,
"grad_norm": 0.038225915282964706,
"learning_rate": 4.365631028202559e-06,
"loss": 0.012793362140655518,
"step": 82850
},
{
"epoch": 1.6080850403476101,
"grad_norm": 5.7360711097717285,
"learning_rate": 4.354854353237852e-06,
"loss": 0.01000747799873352,
"step": 82900
},
{
"epoch": 1.6090549348230914,
"grad_norm": 0.2608398497104645,
"learning_rate": 4.344077678273146e-06,
"loss": 0.009328984022140503,
"step": 82950
},
{
"epoch": 1.6100248292985722,
"grad_norm": 0.4575970470905304,
"learning_rate": 4.3333010033084394e-06,
"loss": 0.011898938417434692,
"step": 83000
},
{
"epoch": 1.6109947237740534,
"grad_norm": 0.00031453280826099217,
"learning_rate": 4.322524328343733e-06,
"loss": 0.011416503190994264,
"step": 83050
},
{
"epoch": 1.6119646182495344,
"grad_norm": 0.5609812140464783,
"learning_rate": 4.311747653379027e-06,
"loss": 0.006319191455841065,
"step": 83100
},
{
"epoch": 1.6129345127250154,
"grad_norm": 0.0006625772221013904,
"learning_rate": 4.300970978414321e-06,
"loss": 0.008569391369819641,
"step": 83150
},
{
"epoch": 1.6139044072004967,
"grad_norm": 0.0021445793099701405,
"learning_rate": 4.290194303449614e-06,
"loss": 0.008201799988746643,
"step": 83200
},
{
"epoch": 1.6148743016759777,
"grad_norm": 0.005074641667306423,
"learning_rate": 4.279417628484908e-06,
"loss": 0.00903885543346405,
"step": 83250
},
{
"epoch": 1.6158441961514587,
"grad_norm": 4.3482184410095215,
"learning_rate": 4.268640953520201e-06,
"loss": 0.008718165159225464,
"step": 83300
},
{
"epoch": 1.61681409062694,
"grad_norm": 0.06287491321563721,
"learning_rate": 4.257864278555495e-06,
"loss": 0.009286146759986878,
"step": 83350
},
{
"epoch": 1.6177839851024207,
"grad_norm": 0.10183337330818176,
"learning_rate": 4.247087603590788e-06,
"loss": 0.007895967364311219,
"step": 83400
},
{
"epoch": 1.618753879577902,
"grad_norm": 0.0018325279233977199,
"learning_rate": 4.236310928626082e-06,
"loss": 0.008061986565589905,
"step": 83450
},
{
"epoch": 1.619723774053383,
"grad_norm": 0.023283787071704865,
"learning_rate": 4.225534253661376e-06,
"loss": 0.007292163968086243,
"step": 83500
},
{
"epoch": 1.620693668528864,
"grad_norm": 0.01891452446579933,
"learning_rate": 4.214757578696669e-06,
"loss": 0.008070356249809264,
"step": 83550
},
{
"epoch": 1.6216635630043452,
"grad_norm": 0.02876999042928219,
"learning_rate": 4.203980903731963e-06,
"loss": 0.010207071304321289,
"step": 83600
},
{
"epoch": 1.6226334574798262,
"grad_norm": 0.10315004736185074,
"learning_rate": 4.1932042287672564e-06,
"loss": 0.010214983224868775,
"step": 83650
},
{
"epoch": 1.6236033519553073,
"grad_norm": 0.03388327360153198,
"learning_rate": 4.18242755380255e-06,
"loss": 0.010612834692001343,
"step": 83700
},
{
"epoch": 1.6245732464307885,
"grad_norm": 2.3742759227752686,
"learning_rate": 4.171650878837844e-06,
"loss": 0.008306831121444702,
"step": 83750
},
{
"epoch": 1.6255431409062693,
"grad_norm": 0.002917750971391797,
"learning_rate": 4.160874203873138e-06,
"loss": 0.012157180309295655,
"step": 83800
},
{
"epoch": 1.6265130353817505,
"grad_norm": 0.1379164308309555,
"learning_rate": 4.150097528908431e-06,
"loss": 0.007811589241027832,
"step": 83850
},
{
"epoch": 1.6274829298572315,
"grad_norm": 0.011732243001461029,
"learning_rate": 4.139320853943725e-06,
"loss": 0.008624168634414673,
"step": 83900
},
{
"epoch": 1.6284528243327125,
"grad_norm": 0.37736082077026367,
"learning_rate": 4.128544178979018e-06,
"loss": 0.009813822507858276,
"step": 83950
},
{
"epoch": 1.6294227188081938,
"grad_norm": 0.0037959502078592777,
"learning_rate": 4.117767504014312e-06,
"loss": 0.007070537805557251,
"step": 84000
},
{
"epoch": 1.6303926132836746,
"grad_norm": 0.11204313486814499,
"learning_rate": 4.106990829049605e-06,
"loss": 0.006803990602493286,
"step": 84050
},
{
"epoch": 1.6313625077591558,
"grad_norm": 0.3679674565792084,
"learning_rate": 4.096214154084899e-06,
"loss": 0.007433233261108399,
"step": 84100
},
{
"epoch": 1.6323324022346368,
"grad_norm": 0.16110078990459442,
"learning_rate": 4.085437479120193e-06,
"loss": 0.007038601040840149,
"step": 84150
},
{
"epoch": 1.6333022967101178,
"grad_norm": 0.006069250404834747,
"learning_rate": 4.074660804155486e-06,
"loss": 0.014864935874938964,
"step": 84200
},
{
"epoch": 1.634272191185599,
"grad_norm": 0.030351970344781876,
"learning_rate": 4.06388412919078e-06,
"loss": 0.010292413234710694,
"step": 84250
},
{
"epoch": 1.63524208566108,
"grad_norm": 0.24867458641529083,
"learning_rate": 4.0531074542260734e-06,
"loss": 0.008270949721336365,
"step": 84300
},
{
"epoch": 1.636211980136561,
"grad_norm": 0.33111244440078735,
"learning_rate": 4.042330779261367e-06,
"loss": 0.008354700803756714,
"step": 84350
},
{
"epoch": 1.6371818746120423,
"grad_norm": 2.440460681915283,
"learning_rate": 4.031554104296661e-06,
"loss": 0.011723284721374511,
"step": 84400
},
{
"epoch": 1.6381517690875231,
"grad_norm": 0.0034247864969074726,
"learning_rate": 4.020777429331954e-06,
"loss": 0.008850112557411194,
"step": 84450
},
{
"epoch": 1.6391216635630044,
"grad_norm": 0.9127389192581177,
"learning_rate": 4.010000754367248e-06,
"loss": 0.01198221206665039,
"step": 84500
},
{
"epoch": 1.6400915580384854,
"grad_norm": 0.07526463270187378,
"learning_rate": 3.999224079402542e-06,
"loss": 0.011685384511947632,
"step": 84550
},
{
"epoch": 1.6410614525139664,
"grad_norm": 0.06160460785031319,
"learning_rate": 3.988447404437835e-06,
"loss": 0.008528870940208435,
"step": 84600
},
{
"epoch": 1.6420313469894476,
"grad_norm": 0.6712585091590881,
"learning_rate": 3.977670729473129e-06,
"loss": 0.006670400500297546,
"step": 84650
},
{
"epoch": 1.6430012414649287,
"grad_norm": 0.06789831072092056,
"learning_rate": 3.966894054508422e-06,
"loss": 0.010675745010375976,
"step": 84700
},
{
"epoch": 1.6439711359404097,
"grad_norm": 0.0016161116072908044,
"learning_rate": 3.956117379543716e-06,
"loss": 0.01006313920021057,
"step": 84750
},
{
"epoch": 1.644941030415891,
"grad_norm": 0.04738567769527435,
"learning_rate": 3.945340704579009e-06,
"loss": 0.008681559562683105,
"step": 84800
},
{
"epoch": 1.6459109248913717,
"grad_norm": 0.06322205811738968,
"learning_rate": 3.934564029614303e-06,
"loss": 0.008469580411911011,
"step": 84850
},
{
"epoch": 1.646880819366853,
"grad_norm": 0.00288754072971642,
"learning_rate": 3.923787354649597e-06,
"loss": 0.007800232172012329,
"step": 84900
},
{
"epoch": 1.647850713842334,
"grad_norm": 0.0016231848858296871,
"learning_rate": 3.9130106796848905e-06,
"loss": 0.007356765866279602,
"step": 84950
},
{
"epoch": 1.648820608317815,
"grad_norm": 0.4526185393333435,
"learning_rate": 3.902234004720184e-06,
"loss": 0.01149672269821167,
"step": 85000
},
{
"epoch": 1.6497905027932962,
"grad_norm": 0.21585942804813385,
"learning_rate": 3.891457329755478e-06,
"loss": 0.007406249046325684,
"step": 85050
},
{
"epoch": 1.650760397268777,
"grad_norm": 0.001526480307802558,
"learning_rate": 3.880680654790771e-06,
"loss": 0.008054419755935668,
"step": 85100
},
{
"epoch": 1.6517302917442582,
"grad_norm": 0.03151676803827286,
"learning_rate": 3.869903979826065e-06,
"loss": 0.010330653190612793,
"step": 85150
},
{
"epoch": 1.6527001862197392,
"grad_norm": 0.12063322216272354,
"learning_rate": 3.859127304861359e-06,
"loss": 0.00837298333644867,
"step": 85200
},
{
"epoch": 1.6536700806952203,
"grad_norm": 0.0403924286365509,
"learning_rate": 3.848350629896652e-06,
"loss": 0.009492009282112121,
"step": 85250
},
{
"epoch": 1.6546399751707015,
"grad_norm": 0.001515073236078024,
"learning_rate": 3.837573954931946e-06,
"loss": 0.009984200596809387,
"step": 85300
},
{
"epoch": 1.6556098696461825,
"grad_norm": 0.002673125360161066,
"learning_rate": 3.826797279967239e-06,
"loss": 0.007555482387542725,
"step": 85350
},
{
"epoch": 1.6565797641216635,
"grad_norm": 0.6092742681503296,
"learning_rate": 3.816020605002533e-06,
"loss": 0.006873829960823059,
"step": 85400
},
{
"epoch": 1.6575496585971448,
"grad_norm": 4.833026885986328,
"learning_rate": 3.8052439300378263e-06,
"loss": 0.009672647714614869,
"step": 85450
},
{
"epoch": 1.6585195530726256,
"grad_norm": 2.7416746616363525,
"learning_rate": 3.79446725507312e-06,
"loss": 0.008716377019882203,
"step": 85500
},
{
"epoch": 1.6594894475481068,
"grad_norm": 0.01382538489997387,
"learning_rate": 3.7836905801084134e-06,
"loss": 0.010225625038146972,
"step": 85550
},
{
"epoch": 1.6604593420235878,
"grad_norm": 0.29831594228744507,
"learning_rate": 3.7729139051437075e-06,
"loss": 0.009470062851905823,
"step": 85600
},
{
"epoch": 1.6614292364990688,
"grad_norm": 0.06965691596269608,
"learning_rate": 3.762137230179001e-06,
"loss": 0.007705588340759278,
"step": 85650
},
{
"epoch": 1.66239913097455,
"grad_norm": 0.0029076000209897757,
"learning_rate": 3.7513605552142946e-06,
"loss": 0.006257765889167786,
"step": 85700
},
{
"epoch": 1.663369025450031,
"grad_norm": 0.1859055757522583,
"learning_rate": 3.740583880249588e-06,
"loss": 0.007536585927009582,
"step": 85750
},
{
"epoch": 1.664338919925512,
"grad_norm": 0.6258267164230347,
"learning_rate": 3.7298072052848818e-06,
"loss": 0.005221771597862243,
"step": 85800
},
{
"epoch": 1.6653088144009933,
"grad_norm": 0.6869074702262878,
"learning_rate": 3.719030530320175e-06,
"loss": 0.008508050441741943,
"step": 85850
},
{
"epoch": 1.6662787088764741,
"grad_norm": 0.0021780699025839567,
"learning_rate": 3.7082538553554685e-06,
"loss": 0.008199980854988098,
"step": 85900
},
{
"epoch": 1.6672486033519553,
"grad_norm": 0.24278181791305542,
"learning_rate": 3.697477180390763e-06,
"loss": 0.011528396606445312,
"step": 85950
},
{
"epoch": 1.6682184978274364,
"grad_norm": 0.14358291029930115,
"learning_rate": 3.686700505426056e-06,
"loss": 0.005857929587364197,
"step": 86000
},
{
"epoch": 1.6691883923029174,
"grad_norm": 0.0946425199508667,
"learning_rate": 3.6759238304613497e-06,
"loss": 0.005703723430633545,
"step": 86050
},
{
"epoch": 1.6701582867783986,
"grad_norm": 0.0018080163281410933,
"learning_rate": 3.6651471554966433e-06,
"loss": 0.009026229977607728,
"step": 86100
},
{
"epoch": 1.6711281812538796,
"grad_norm": 0.001699203858152032,
"learning_rate": 3.654370480531937e-06,
"loss": 0.010489603281021118,
"step": 86150
},
{
"epoch": 1.6720980757293606,
"grad_norm": 0.1369491070508957,
"learning_rate": 3.6435938055672305e-06,
"loss": 0.010380499362945557,
"step": 86200
},
{
"epoch": 1.6730679702048417,
"grad_norm": 0.003383740084245801,
"learning_rate": 3.632817130602524e-06,
"loss": 0.007680900692939759,
"step": 86250
},
{
"epoch": 1.6740378646803227,
"grad_norm": 0.16986218094825745,
"learning_rate": 3.622040455637818e-06,
"loss": 0.011866482496261597,
"step": 86300
},
{
"epoch": 1.675007759155804,
"grad_norm": 0.18751446902751923,
"learning_rate": 3.6112637806731116e-06,
"loss": 0.005128522515296936,
"step": 86350
},
{
"epoch": 1.675977653631285,
"grad_norm": 0.014115598052740097,
"learning_rate": 3.600487105708405e-06,
"loss": 0.007044323682785034,
"step": 86400
},
{
"epoch": 1.676947548106766,
"grad_norm": 0.05365321785211563,
"learning_rate": 3.589710430743699e-06,
"loss": 0.006545979380607605,
"step": 86450
},
{
"epoch": 1.6779174425822472,
"grad_norm": 0.04420368745923042,
"learning_rate": 3.578933755778992e-06,
"loss": 0.010442807674407958,
"step": 86500
},
{
"epoch": 1.678887337057728,
"grad_norm": 0.08096443861722946,
"learning_rate": 3.5681570808142855e-06,
"loss": 0.009220988154411316,
"step": 86550
},
{
"epoch": 1.6798572315332092,
"grad_norm": 0.8439462780952454,
"learning_rate": 3.557380405849579e-06,
"loss": 0.007643036842346192,
"step": 86600
},
{
"epoch": 1.6808271260086902,
"grad_norm": 0.8949611783027649,
"learning_rate": 3.546603730884873e-06,
"loss": 0.009003739356994628,
"step": 86650
},
{
"epoch": 1.6817970204841712,
"grad_norm": 0.09969016909599304,
"learning_rate": 3.5358270559201667e-06,
"loss": 0.011510480642318726,
"step": 86700
},
{
"epoch": 1.6827669149596525,
"grad_norm": 0.08829014003276825,
"learning_rate": 3.5250503809554603e-06,
"loss": 0.013323155641555786,
"step": 86750
},
{
"epoch": 1.6837368094351335,
"grad_norm": 0.12339182198047638,
"learning_rate": 3.514273705990754e-06,
"loss": 0.007389287352561951,
"step": 86800
},
{
"epoch": 1.6847067039106145,
"grad_norm": 0.043362878262996674,
"learning_rate": 3.5034970310260475e-06,
"loss": 0.008098645806312562,
"step": 86850
},
{
"epoch": 1.6856765983860957,
"grad_norm": 0.05678665637969971,
"learning_rate": 3.492720356061341e-06,
"loss": 0.006841681003570557,
"step": 86900
},
{
"epoch": 1.6866464928615765,
"grad_norm": 0.04954465478658676,
"learning_rate": 3.4819436810966346e-06,
"loss": 0.006019186973571777,
"step": 86950
},
{
"epoch": 1.6876163873370578,
"grad_norm": 0.04623192548751831,
"learning_rate": 3.4711670061319286e-06,
"loss": 0.0063313072919845585,
"step": 87000
},
{
"epoch": 1.6885862818125388,
"grad_norm": 0.05225834622979164,
"learning_rate": 3.4603903311672222e-06,
"loss": 0.00972770094871521,
"step": 87050
},
{
"epoch": 1.6895561762880198,
"grad_norm": 0.08629638701677322,
"learning_rate": 3.449613656202516e-06,
"loss": 0.01352167248725891,
"step": 87100
},
{
"epoch": 1.690526070763501,
"grad_norm": 0.3092690706253052,
"learning_rate": 3.4388369812378094e-06,
"loss": 0.007848586440086364,
"step": 87150
},
{
"epoch": 1.691495965238982,
"grad_norm": 0.2737095355987549,
"learning_rate": 3.4280603062731025e-06,
"loss": 0.009379087686538697,
"step": 87200
},
{
"epoch": 1.692465859714463,
"grad_norm": 0.2719198167324066,
"learning_rate": 3.417283631308396e-06,
"loss": 0.011072784662246704,
"step": 87250
},
{
"epoch": 1.6934357541899443,
"grad_norm": 0.1617434322834015,
"learning_rate": 3.4065069563436897e-06,
"loss": 0.009463454484939576,
"step": 87300
},
{
"epoch": 1.694405648665425,
"grad_norm": 0.11131928861141205,
"learning_rate": 3.3957302813789837e-06,
"loss": 0.010085492134094239,
"step": 87350
},
{
"epoch": 1.6953755431409063,
"grad_norm": 0.025582876056432724,
"learning_rate": 3.3849536064142773e-06,
"loss": 0.007589172720909119,
"step": 87400
},
{
"epoch": 1.6963454376163873,
"grad_norm": 0.22788487374782562,
"learning_rate": 3.374176931449571e-06,
"loss": 0.008108940720558167,
"step": 87450
},
{
"epoch": 1.6973153320918684,
"grad_norm": 0.0009850695496425033,
"learning_rate": 3.3634002564848645e-06,
"loss": 0.009013856649398804,
"step": 87500
},
{
"epoch": 1.6982852265673496,
"grad_norm": 4.196169853210449,
"learning_rate": 3.352623581520158e-06,
"loss": 0.00903675377368927,
"step": 87550
},
{
"epoch": 1.6992551210428304,
"grad_norm": 0.09591202437877655,
"learning_rate": 3.3418469065554516e-06,
"loss": 0.008793619871139526,
"step": 87600
},
{
"epoch": 1.7002250155183116,
"grad_norm": 0.5763933658599854,
"learning_rate": 3.331070231590745e-06,
"loss": 0.007580864429473877,
"step": 87650
},
{
"epoch": 1.7011949099937926,
"grad_norm": 0.28002429008483887,
"learning_rate": 3.3202935566260392e-06,
"loss": 0.006398821473121643,
"step": 87700
},
{
"epoch": 1.7021648044692737,
"grad_norm": 0.15343664586544037,
"learning_rate": 3.309516881661333e-06,
"loss": 0.012194329500198364,
"step": 87750
},
{
"epoch": 1.7031346989447549,
"grad_norm": 0.2133241444826126,
"learning_rate": 3.2987402066966264e-06,
"loss": 0.010830810070037842,
"step": 87800
},
{
"epoch": 1.704104593420236,
"grad_norm": 0.20241476595401764,
"learning_rate": 3.2879635317319195e-06,
"loss": 0.008619242906570434,
"step": 87850
},
{
"epoch": 1.705074487895717,
"grad_norm": 1.7081472873687744,
"learning_rate": 3.277186856767213e-06,
"loss": 0.011400961875915527,
"step": 87900
},
{
"epoch": 1.7060443823711982,
"grad_norm": 0.03838106244802475,
"learning_rate": 3.2664101818025067e-06,
"loss": 0.01188423752784729,
"step": 87950
},
{
"epoch": 1.707014276846679,
"grad_norm": 0.05929063260555267,
"learning_rate": 3.2556335068378003e-06,
"loss": 0.006974325180053711,
"step": 88000
},
{
"epoch": 1.7079841713221602,
"grad_norm": 0.3601554334163666,
"learning_rate": 3.2448568318730943e-06,
"loss": 0.010480151176452637,
"step": 88050
},
{
"epoch": 1.7089540657976412,
"grad_norm": 0.14530307054519653,
"learning_rate": 3.234080156908388e-06,
"loss": 0.010269461870193482,
"step": 88100
},
{
"epoch": 1.7099239602731222,
"grad_norm": 0.008624673821032047,
"learning_rate": 3.2233034819436815e-06,
"loss": 0.008001194596290588,
"step": 88150
},
{
"epoch": 1.7108938547486034,
"grad_norm": 0.24361449480056763,
"learning_rate": 3.212526806978975e-06,
"loss": 0.0121296226978302,
"step": 88200
},
{
"epoch": 1.7118637492240845,
"grad_norm": 0.0025675008073449135,
"learning_rate": 3.2017501320142686e-06,
"loss": 0.006839089393615723,
"step": 88250
},
{
"epoch": 1.7128336436995655,
"grad_norm": 0.09138184785842896,
"learning_rate": 3.1909734570495622e-06,
"loss": 0.008672207593917847,
"step": 88300
},
{
"epoch": 1.7138035381750467,
"grad_norm": 0.5455693006515503,
"learning_rate": 3.180196782084856e-06,
"loss": 0.00803396999835968,
"step": 88350
},
{
"epoch": 1.7147734326505275,
"grad_norm": 0.27769994735717773,
"learning_rate": 3.169420107120149e-06,
"loss": 0.009091012477874756,
"step": 88400
},
{
"epoch": 1.7157433271260087,
"grad_norm": 0.009906435385346413,
"learning_rate": 3.1586434321554434e-06,
"loss": 0.00844333291053772,
"step": 88450
},
{
"epoch": 1.7167132216014898,
"grad_norm": 0.051845960319042206,
"learning_rate": 3.1478667571907365e-06,
"loss": 0.007592194676399231,
"step": 88500
},
{
"epoch": 1.7176831160769708,
"grad_norm": 0.03090464510023594,
"learning_rate": 3.13709008222603e-06,
"loss": 0.0046571075916290285,
"step": 88550
},
{
"epoch": 1.718653010552452,
"grad_norm": 0.04890689626336098,
"learning_rate": 3.1263134072613237e-06,
"loss": 0.018171044588088988,
"step": 88600
},
{
"epoch": 1.7196229050279328,
"grad_norm": 0.05086488276720047,
"learning_rate": 3.1155367322966173e-06,
"loss": 0.012567259073257446,
"step": 88650
},
{
"epoch": 1.720592799503414,
"grad_norm": 0.06687403470277786,
"learning_rate": 3.104760057331911e-06,
"loss": 0.012102587223052979,
"step": 88700
},
{
"epoch": 1.721562693978895,
"grad_norm": 0.3909635841846466,
"learning_rate": 3.0939833823672045e-06,
"loss": 0.008089342117309571,
"step": 88750
},
{
"epoch": 1.722532588454376,
"grad_norm": 0.7422420382499695,
"learning_rate": 3.0832067074024985e-06,
"loss": 0.009202653169631958,
"step": 88800
},
{
"epoch": 1.7235024829298573,
"grad_norm": 0.0938820019364357,
"learning_rate": 3.072430032437792e-06,
"loss": 0.008810549974441528,
"step": 88850
},
{
"epoch": 1.7244723774053383,
"grad_norm": 0.0027660122141242027,
"learning_rate": 3.0616533574730856e-06,
"loss": 0.007855842113494873,
"step": 88900
},
{
"epoch": 1.7254422718808193,
"grad_norm": 0.03197028487920761,
"learning_rate": 3.0508766825083792e-06,
"loss": 0.009288341999053954,
"step": 88950
},
{
"epoch": 1.7264121663563006,
"grad_norm": 0.6662591695785522,
"learning_rate": 3.040100007543673e-06,
"loss": 0.009328324794769288,
"step": 89000
},
{
"epoch": 1.7273820608317814,
"grad_norm": 0.24134400486946106,
"learning_rate": 3.029323332578966e-06,
"loss": 0.00811453938484192,
"step": 89050
},
{
"epoch": 1.7283519553072626,
"grad_norm": 0.24863764643669128,
"learning_rate": 3.0185466576142595e-06,
"loss": 0.0072233319282531735,
"step": 89100
},
{
"epoch": 1.7293218497827436,
"grad_norm": 0.11773120611906052,
"learning_rate": 3.007769982649554e-06,
"loss": 0.006629384160041809,
"step": 89150
},
{
"epoch": 1.7302917442582246,
"grad_norm": 0.16357240080833435,
"learning_rate": 2.996993307684847e-06,
"loss": 0.00780647099018097,
"step": 89200
},
{
"epoch": 1.7312616387337059,
"grad_norm": 0.0005478064413182437,
"learning_rate": 2.9862166327201407e-06,
"loss": 0.008727927803993224,
"step": 89250
},
{
"epoch": 1.7322315332091869,
"grad_norm": 0.020349636673927307,
"learning_rate": 2.9754399577554343e-06,
"loss": 0.006670922040939331,
"step": 89300
},
{
"epoch": 1.733201427684668,
"grad_norm": 0.004095559008419514,
"learning_rate": 2.964663282790728e-06,
"loss": 0.008210720419883728,
"step": 89350
},
{
"epoch": 1.7341713221601491,
"grad_norm": 0.021984025835990906,
"learning_rate": 2.9538866078260215e-06,
"loss": 0.006560840606689453,
"step": 89400
},
{
"epoch": 1.73514121663563,
"grad_norm": 0.0017808245029300451,
"learning_rate": 2.943109932861315e-06,
"loss": 0.00939340889453888,
"step": 89450
},
{
"epoch": 1.7361111111111112,
"grad_norm": 0.05358776077628136,
"learning_rate": 2.932333257896609e-06,
"loss": 0.008276232481002808,
"step": 89500
},
{
"epoch": 1.7370810055865922,
"grad_norm": 0.047172993421554565,
"learning_rate": 2.9215565829319026e-06,
"loss": 0.005374414920806885,
"step": 89550
},
{
"epoch": 1.7380509000620732,
"grad_norm": 9.783132553100586,
"learning_rate": 2.9107799079671962e-06,
"loss": 0.0106581711769104,
"step": 89600
},
{
"epoch": 1.7390207945375544,
"grad_norm": 0.002196686575189233,
"learning_rate": 2.90000323300249e-06,
"loss": 0.008630460500717163,
"step": 89650
},
{
"epoch": 1.7399906890130352,
"grad_norm": 0.003969006240367889,
"learning_rate": 2.8892265580377834e-06,
"loss": 0.010060198307037353,
"step": 89700
},
{
"epoch": 1.7409605834885165,
"grad_norm": 0.052971821278333664,
"learning_rate": 2.8784498830730765e-06,
"loss": 0.006120312213897705,
"step": 89750
},
{
"epoch": 1.7419304779639975,
"grad_norm": 0.7778472304344177,
"learning_rate": 2.86767320810837e-06,
"loss": 0.00792294681072235,
"step": 89800
},
{
"epoch": 1.7429003724394785,
"grad_norm": 0.05869757756590843,
"learning_rate": 2.856896533143664e-06,
"loss": 0.005748578310012817,
"step": 89850
},
{
"epoch": 1.7438702669149597,
"grad_norm": 0.000320307444781065,
"learning_rate": 2.8461198581789577e-06,
"loss": 0.009468759894371032,
"step": 89900
},
{
"epoch": 1.7448401613904407,
"grad_norm": 0.006394130643457174,
"learning_rate": 2.8353431832142513e-06,
"loss": 0.0055860793590545655,
"step": 89950
},
{
"epoch": 1.7458100558659218,
"grad_norm": 0.766309916973114,
"learning_rate": 2.824566508249545e-06,
"loss": 0.006365313529968262,
"step": 90000
},
{
"epoch": 1.746779950341403,
"grad_norm": 0.944915235042572,
"learning_rate": 2.8137898332848385e-06,
"loss": 0.010649431943893433,
"step": 90050
},
{
"epoch": 1.7477498448168838,
"grad_norm": 0.04433741420507431,
"learning_rate": 2.803013158320132e-06,
"loss": 0.012598178386688232,
"step": 90100
},
{
"epoch": 1.748719739292365,
"grad_norm": 0.8978362083435059,
"learning_rate": 2.7922364833554256e-06,
"loss": 0.005295076370239258,
"step": 90150
},
{
"epoch": 1.749689633767846,
"grad_norm": 0.021818457171320915,
"learning_rate": 2.7814598083907196e-06,
"loss": 0.004375053644180298,
"step": 90200
},
{
"epoch": 1.750659528243327,
"grad_norm": 0.1129290983080864,
"learning_rate": 2.7706831334260132e-06,
"loss": 0.005446012616157531,
"step": 90250
},
{
"epoch": 1.7516294227188083,
"grad_norm": 2.1925201416015625,
"learning_rate": 2.759906458461307e-06,
"loss": 0.010336420536041259,
"step": 90300
},
{
"epoch": 1.7525993171942893,
"grad_norm": 0.1148567646741867,
"learning_rate": 2.7491297834966004e-06,
"loss": 0.01506085753440857,
"step": 90350
},
{
"epoch": 1.7535692116697703,
"grad_norm": 0.0007074130699038506,
"learning_rate": 2.7383531085318935e-06,
"loss": 0.007875357270240784,
"step": 90400
},
{
"epoch": 1.7545391061452515,
"grad_norm": 0.0010888195829465985,
"learning_rate": 2.727576433567187e-06,
"loss": 0.00712762713432312,
"step": 90450
},
{
"epoch": 1.7555090006207323,
"grad_norm": 0.0002946695894934237,
"learning_rate": 2.7167997586024807e-06,
"loss": 0.007263663411140442,
"step": 90500
},
{
"epoch": 1.7564788950962136,
"grad_norm": 0.12567798793315887,
"learning_rate": 2.7060230836377747e-06,
"loss": 0.01183547019958496,
"step": 90550
},
{
"epoch": 1.7574487895716946,
"grad_norm": 0.020071256905794144,
"learning_rate": 2.6952464086730683e-06,
"loss": 0.005407539010047912,
"step": 90600
},
{
"epoch": 1.7584186840471756,
"grad_norm": 0.5823246240615845,
"learning_rate": 2.684469733708362e-06,
"loss": 0.008287495374679566,
"step": 90650
},
{
"epoch": 1.7593885785226568,
"grad_norm": 0.0009665252291597426,
"learning_rate": 2.6736930587436555e-06,
"loss": 0.01084843635559082,
"step": 90700
},
{
"epoch": 1.7603584729981379,
"grad_norm": 0.0024437778629362583,
"learning_rate": 2.662916383778949e-06,
"loss": 0.013194255828857422,
"step": 90750
},
{
"epoch": 1.7613283674736189,
"grad_norm": 0.06734281778335571,
"learning_rate": 2.6521397088142426e-06,
"loss": 0.006578063368797302,
"step": 90800
},
{
"epoch": 1.7622982619490999,
"grad_norm": 0.19234754145145416,
"learning_rate": 2.6413630338495362e-06,
"loss": 0.010411605834960938,
"step": 90850
},
{
"epoch": 1.763268156424581,
"grad_norm": 0.0028453185223042965,
"learning_rate": 2.6305863588848302e-06,
"loss": 0.006344481110572815,
"step": 90900
},
{
"epoch": 1.7642380509000621,
"grad_norm": 0.06246383860707283,
"learning_rate": 2.619809683920124e-06,
"loss": 0.0073875969648361205,
"step": 90950
},
{
"epoch": 1.7652079453755432,
"grad_norm": 0.15871261060237885,
"learning_rate": 2.6090330089554174e-06,
"loss": 0.009121181964874268,
"step": 91000
},
{
"epoch": 1.7661778398510242,
"grad_norm": 0.310638427734375,
"learning_rate": 2.5982563339907106e-06,
"loss": 0.012160261869430542,
"step": 91050
},
{
"epoch": 1.7671477343265054,
"grad_norm": 0.5135254263877869,
"learning_rate": 2.587479659026004e-06,
"loss": 0.009824393391609192,
"step": 91100
},
{
"epoch": 1.7681176288019862,
"grad_norm": 0.013066467829048634,
"learning_rate": 2.5767029840612977e-06,
"loss": 0.005140823125839233,
"step": 91150
},
{
"epoch": 1.7690875232774674,
"grad_norm": 0.001124839880503714,
"learning_rate": 2.5659263090965913e-06,
"loss": 0.007855725288391114,
"step": 91200
},
{
"epoch": 1.7700574177529484,
"grad_norm": 17.3564395904541,
"learning_rate": 2.555149634131885e-06,
"loss": 0.006035804152488709,
"step": 91250
},
{
"epoch": 1.7710273122284295,
"grad_norm": 0.004674045834690332,
"learning_rate": 2.544372959167179e-06,
"loss": 0.008808953762054443,
"step": 91300
},
{
"epoch": 1.7719972067039107,
"grad_norm": 0.0023436432238668203,
"learning_rate": 2.5335962842024725e-06,
"loss": 0.005589414834976197,
"step": 91350
},
{
"epoch": 1.7729671011793917,
"grad_norm": 0.0002765074896160513,
"learning_rate": 2.522819609237766e-06,
"loss": 0.011621966361999511,
"step": 91400
},
{
"epoch": 1.7739369956548727,
"grad_norm": 0.00549640879034996,
"learning_rate": 2.5120429342730596e-06,
"loss": 0.00925673484802246,
"step": 91450
},
{
"epoch": 1.774906890130354,
"grad_norm": 0.26289626955986023,
"learning_rate": 2.5012662593083532e-06,
"loss": 0.013222092390060424,
"step": 91500
},
{
"epoch": 1.7758767846058348,
"grad_norm": 0.09443680942058563,
"learning_rate": 2.490489584343647e-06,
"loss": 0.007493534684181213,
"step": 91550
},
{
"epoch": 1.776846679081316,
"grad_norm": 0.02722826972603798,
"learning_rate": 2.4797129093789404e-06,
"loss": 0.006481832265853882,
"step": 91600
},
{
"epoch": 1.777816573556797,
"grad_norm": 0.04426329955458641,
"learning_rate": 2.468936234414234e-06,
"loss": 0.016478172540664672,
"step": 91650
},
{
"epoch": 1.778786468032278,
"grad_norm": 0.0772467628121376,
"learning_rate": 2.458159559449528e-06,
"loss": 0.009176092743873597,
"step": 91700
},
{
"epoch": 1.7797563625077593,
"grad_norm": 0.09490852802991867,
"learning_rate": 2.447382884484821e-06,
"loss": 0.006695954799652099,
"step": 91750
},
{
"epoch": 1.7807262569832403,
"grad_norm": 0.19740049540996552,
"learning_rate": 2.4366062095201147e-06,
"loss": 0.003776761889457703,
"step": 91800
},
{
"epoch": 1.7816961514587213,
"grad_norm": 0.13821597397327423,
"learning_rate": 2.4258295345554083e-06,
"loss": 0.00864011287689209,
"step": 91850
},
{
"epoch": 1.7826660459342023,
"grad_norm": 0.08505205810070038,
"learning_rate": 2.4150528595907023e-06,
"loss": 0.01047637939453125,
"step": 91900
},
{
"epoch": 1.7836359404096833,
"grad_norm": 0.27133995294570923,
"learning_rate": 2.404276184625996e-06,
"loss": 0.007478057742118836,
"step": 91950
},
{
"epoch": 1.7846058348851646,
"grad_norm": 0.4542582631111145,
"learning_rate": 2.393499509661289e-06,
"loss": 0.010361738204956054,
"step": 92000
},
{
"epoch": 1.7855757293606456,
"grad_norm": 0.02163510024547577,
"learning_rate": 2.3827228346965826e-06,
"loss": 0.006302187442779541,
"step": 92050
},
{
"epoch": 1.7865456238361266,
"grad_norm": 0.031280796974897385,
"learning_rate": 2.3719461597318766e-06,
"loss": 0.009579989314079284,
"step": 92100
},
{
"epoch": 1.7875155183116078,
"grad_norm": 0.08608546108007431,
"learning_rate": 2.3611694847671702e-06,
"loss": 0.010460764169692993,
"step": 92150
},
{
"epoch": 1.7884854127870886,
"grad_norm": 0.14350184798240662,
"learning_rate": 2.350392809802464e-06,
"loss": 0.006315089464187622,
"step": 92200
},
{
"epoch": 1.7894553072625698,
"grad_norm": 0.7373741865158081,
"learning_rate": 2.3396161348377574e-06,
"loss": 0.01009372591972351,
"step": 92250
},
{
"epoch": 1.7904252017380509,
"grad_norm": 0.000679600634612143,
"learning_rate": 2.328839459873051e-06,
"loss": 0.010176174640655518,
"step": 92300
},
{
"epoch": 1.7913950962135319,
"grad_norm": 0.09259176254272461,
"learning_rate": 2.3180627849083446e-06,
"loss": 0.008696593642234803,
"step": 92350
},
{
"epoch": 1.7923649906890131,
"grad_norm": 0.0009009715868160129,
"learning_rate": 2.307286109943638e-06,
"loss": 0.012147856950759888,
"step": 92400
},
{
"epoch": 1.7933348851644941,
"grad_norm": 0.3308854401111603,
"learning_rate": 2.2965094349789317e-06,
"loss": 0.008714711666107178,
"step": 92450
},
{
"epoch": 1.7943047796399751,
"grad_norm": 0.1769242137670517,
"learning_rate": 2.2857327600142253e-06,
"loss": 0.007047352194786072,
"step": 92500
},
{
"epoch": 1.7952746741154564,
"grad_norm": 0.09431857615709305,
"learning_rate": 2.274956085049519e-06,
"loss": 0.005283564925193786,
"step": 92550
},
{
"epoch": 1.7962445685909372,
"grad_norm": 0.32506683468818665,
"learning_rate": 2.264179410084813e-06,
"loss": 0.010197395086288452,
"step": 92600
},
{
"epoch": 1.7972144630664184,
"grad_norm": 0.013492737896740437,
"learning_rate": 2.253402735120106e-06,
"loss": 0.01551008939743042,
"step": 92650
},
{
"epoch": 1.7981843575418994,
"grad_norm": 0.10829909145832062,
"learning_rate": 2.2426260601553996e-06,
"loss": 0.008766326308250427,
"step": 92700
},
{
"epoch": 1.7991542520173804,
"grad_norm": 0.005105915013700724,
"learning_rate": 2.2318493851906932e-06,
"loss": 0.00998417854309082,
"step": 92750
},
{
"epoch": 1.8001241464928617,
"grad_norm": 0.000455458095530048,
"learning_rate": 2.2210727102259872e-06,
"loss": 0.006275529861450195,
"step": 92800
},
{
"epoch": 1.8010940409683427,
"grad_norm": 0.03558963164687157,
"learning_rate": 2.210296035261281e-06,
"loss": 0.009625229835510254,
"step": 92850
},
{
"epoch": 1.8020639354438237,
"grad_norm": 0.08278805017471313,
"learning_rate": 2.1995193602965744e-06,
"loss": 0.01499701976776123,
"step": 92900
},
{
"epoch": 1.803033829919305,
"grad_norm": 0.25602424144744873,
"learning_rate": 2.188742685331868e-06,
"loss": 0.008170786499977111,
"step": 92950
},
{
"epoch": 1.8040037243947857,
"grad_norm": 0.15193909406661987,
"learning_rate": 2.1779660103671616e-06,
"loss": 0.00834970772266388,
"step": 93000
},
{
"epoch": 1.804973618870267,
"grad_norm": 6.9982476234436035,
"learning_rate": 2.167189335402455e-06,
"loss": 0.010644700527191162,
"step": 93050
},
{
"epoch": 1.805943513345748,
"grad_norm": 0.46758341789245605,
"learning_rate": 2.1564126604377487e-06,
"loss": 0.0072982382774353025,
"step": 93100
},
{
"epoch": 1.806913407821229,
"grad_norm": 1.786726713180542,
"learning_rate": 2.1456359854730423e-06,
"loss": 0.012052642107009888,
"step": 93150
},
{
"epoch": 1.8078833022967102,
"grad_norm": 0.01810626871883869,
"learning_rate": 2.134859310508336e-06,
"loss": 0.005456081032752991,
"step": 93200
},
{
"epoch": 1.808853196772191,
"grad_norm": 0.3778686225414276,
"learning_rate": 2.1240826355436295e-06,
"loss": 0.010128356218338012,
"step": 93250
},
{
"epoch": 1.8098230912476723,
"grad_norm": 0.04733676090836525,
"learning_rate": 2.1133059605789235e-06,
"loss": 0.008798363804817199,
"step": 93300
},
{
"epoch": 1.8107929857231533,
"grad_norm": 0.34491416811943054,
"learning_rate": 2.1025292856142166e-06,
"loss": 0.008397238254547119,
"step": 93350
},
{
"epoch": 1.8117628801986343,
"grad_norm": 0.0004525998665485531,
"learning_rate": 2.0917526106495102e-06,
"loss": 0.0062387073040008545,
"step": 93400
},
{
"epoch": 1.8127327746741155,
"grad_norm": 0.042098741978406906,
"learning_rate": 2.080975935684804e-06,
"loss": 0.0034190592169761657,
"step": 93450
},
{
"epoch": 1.8137026691495965,
"grad_norm": 0.11118258535861969,
"learning_rate": 2.070199260720098e-06,
"loss": 0.006020020842552185,
"step": 93500
},
{
"epoch": 1.8146725636250776,
"grad_norm": 0.04114688187837601,
"learning_rate": 2.0594225857553914e-06,
"loss": 0.011320677995681762,
"step": 93550
},
{
"epoch": 1.8156424581005588,
"grad_norm": 0.3252732455730438,
"learning_rate": 2.0486459107906846e-06,
"loss": 0.008717839121818543,
"step": 93600
},
{
"epoch": 1.8166123525760396,
"grad_norm": 0.0947057381272316,
"learning_rate": 2.037869235825978e-06,
"loss": 0.009167813062667847,
"step": 93650
},
{
"epoch": 1.8175822470515208,
"grad_norm": 0.16303987801074982,
"learning_rate": 2.027092560861272e-06,
"loss": 0.01040002703666687,
"step": 93700
},
{
"epoch": 1.8185521415270018,
"grad_norm": 0.027486305683851242,
"learning_rate": 2.0163158858965657e-06,
"loss": 0.013106870651245116,
"step": 93750
},
{
"epoch": 1.8195220360024829,
"grad_norm": 0.2683391273021698,
"learning_rate": 2.0055392109318593e-06,
"loss": 0.008514492511749268,
"step": 93800
},
{
"epoch": 1.820491930477964,
"grad_norm": 0.00037964186049066484,
"learning_rate": 1.994762535967153e-06,
"loss": 0.00887359380722046,
"step": 93850
},
{
"epoch": 1.821461824953445,
"grad_norm": 0.006205215584486723,
"learning_rate": 1.9839858610024465e-06,
"loss": 0.008083742260932923,
"step": 93900
},
{
"epoch": 1.8224317194289261,
"grad_norm": 0.17823991179466248,
"learning_rate": 1.97320918603774e-06,
"loss": 0.00835272490978241,
"step": 93950
},
{
"epoch": 1.8234016139044074,
"grad_norm": 1.6592928171157837,
"learning_rate": 1.9624325110730336e-06,
"loss": 0.010166020393371581,
"step": 94000
},
{
"epoch": 1.8243715083798882,
"grad_norm": 0.0015608917456120253,
"learning_rate": 1.9516558361083272e-06,
"loss": 0.0065430748462677,
"step": 94050
},
{
"epoch": 1.8253414028553694,
"grad_norm": 0.2366517335176468,
"learning_rate": 1.940879161143621e-06,
"loss": 0.008816433548927307,
"step": 94100
},
{
"epoch": 1.8263112973308504,
"grad_norm": 0.9170594215393066,
"learning_rate": 1.9301024861789144e-06,
"loss": 0.011960110664367675,
"step": 94150
},
{
"epoch": 1.8272811918063314,
"grad_norm": 0.10817422717809677,
"learning_rate": 1.9193258112142084e-06,
"loss": 0.009775087833404542,
"step": 94200
},
{
"epoch": 1.8282510862818127,
"grad_norm": 0.0007111942977644503,
"learning_rate": 1.908549136249502e-06,
"loss": 0.011418261528015138,
"step": 94250
},
{
"epoch": 1.8292209807572934,
"grad_norm": 0.011562325060367584,
"learning_rate": 1.8977724612847954e-06,
"loss": 0.011568971872329713,
"step": 94300
},
{
"epoch": 1.8301908752327747,
"grad_norm": 0.6854749321937561,
"learning_rate": 1.886995786320089e-06,
"loss": 0.009997607469558715,
"step": 94350
},
{
"epoch": 1.8311607697082557,
"grad_norm": 1.0283762216567993,
"learning_rate": 1.8762191113553827e-06,
"loss": 0.013303570747375489,
"step": 94400
},
{
"epoch": 1.8321306641837367,
"grad_norm": 0.0014454243937507272,
"learning_rate": 1.8654424363906761e-06,
"loss": 0.006436108350753784,
"step": 94450
},
{
"epoch": 1.833100558659218,
"grad_norm": 2.3428142070770264,
"learning_rate": 1.8546657614259697e-06,
"loss": 0.006638326644897461,
"step": 94500
},
{
"epoch": 1.834070453134699,
"grad_norm": 0.0006349983741529286,
"learning_rate": 1.8438890864612635e-06,
"loss": 0.00946279227733612,
"step": 94550
},
{
"epoch": 1.83504034761018,
"grad_norm": 0.05324838310480118,
"learning_rate": 1.833112411496557e-06,
"loss": 0.008550931811332702,
"step": 94600
},
{
"epoch": 1.8360102420856612,
"grad_norm": 0.20900948345661163,
"learning_rate": 1.8223357365318507e-06,
"loss": 0.010544004440307618,
"step": 94650
},
{
"epoch": 1.836980136561142,
"grad_norm": 0.0015481601003557444,
"learning_rate": 1.811559061567144e-06,
"loss": 0.0073264044523239135,
"step": 94700
},
{
"epoch": 1.8379500310366232,
"grad_norm": 0.024757418781518936,
"learning_rate": 1.800782386602438e-06,
"loss": 0.00889215886592865,
"step": 94750
},
{
"epoch": 1.8389199255121043,
"grad_norm": 0.07515113055706024,
"learning_rate": 1.7900057116377314e-06,
"loss": 0.009042514562606812,
"step": 94800
},
{
"epoch": 1.8398898199875853,
"grad_norm": 0.1305868774652481,
"learning_rate": 1.779229036673025e-06,
"loss": 0.007908307313919068,
"step": 94850
},
{
"epoch": 1.8408597144630665,
"grad_norm": 0.6719141006469727,
"learning_rate": 1.7684523617083186e-06,
"loss": 0.007264495491981506,
"step": 94900
},
{
"epoch": 1.8418296089385475,
"grad_norm": 0.48758819699287415,
"learning_rate": 1.7576756867436124e-06,
"loss": 0.005666710138320923,
"step": 94950
},
{
"epoch": 1.8427995034140285,
"grad_norm": 0.10828927904367447,
"learning_rate": 1.746899011778906e-06,
"loss": 0.011984080076217651,
"step": 95000
},
{
"epoch": 1.8437693978895098,
"grad_norm": 0.10528331995010376,
"learning_rate": 1.7361223368141993e-06,
"loss": 0.010076775550842285,
"step": 95050
},
{
"epoch": 1.8447392923649906,
"grad_norm": 0.484756201505661,
"learning_rate": 1.7253456618494931e-06,
"loss": 0.0060277307033538815,
"step": 95100
},
{
"epoch": 1.8457091868404718,
"grad_norm": 0.07647203654050827,
"learning_rate": 1.7145689868847867e-06,
"loss": 0.011150398254394532,
"step": 95150
},
{
"epoch": 1.8466790813159528,
"grad_norm": 0.0008887226576916873,
"learning_rate": 1.7037923119200803e-06,
"loss": 0.003633313179016113,
"step": 95200
},
{
"epoch": 1.8476489757914338,
"grad_norm": 0.08629953116178513,
"learning_rate": 1.6930156369553739e-06,
"loss": 0.008564449548721313,
"step": 95250
},
{
"epoch": 1.848618870266915,
"grad_norm": 0.3029954731464386,
"learning_rate": 1.6822389619906677e-06,
"loss": 0.00506857693195343,
"step": 95300
},
{
"epoch": 1.8495887647423959,
"grad_norm": 0.16498436033725739,
"learning_rate": 1.6714622870259612e-06,
"loss": 0.011419548988342285,
"step": 95350
},
{
"epoch": 1.850558659217877,
"grad_norm": 0.15406468510627747,
"learning_rate": 1.6606856120612546e-06,
"loss": 0.011674436330795289,
"step": 95400
},
{
"epoch": 1.8515285536933581,
"grad_norm": 0.03525263071060181,
"learning_rate": 1.6499089370965484e-06,
"loss": 0.005115324854850769,
"step": 95450
},
{
"epoch": 1.8524984481688391,
"grad_norm": 0.050088539719581604,
"learning_rate": 1.639132262131842e-06,
"loss": 0.007713648080825805,
"step": 95500
},
{
"epoch": 1.8534683426443204,
"grad_norm": 0.3537728190422058,
"learning_rate": 1.6283555871671356e-06,
"loss": 0.00649250328540802,
"step": 95550
},
{
"epoch": 1.8544382371198014,
"grad_norm": 0.08216680586338043,
"learning_rate": 1.6175789122024292e-06,
"loss": 0.008591524958610534,
"step": 95600
},
{
"epoch": 1.8554081315952824,
"grad_norm": 0.004408149980008602,
"learning_rate": 1.606802237237723e-06,
"loss": 0.007959678173065185,
"step": 95650
},
{
"epoch": 1.8563780260707636,
"grad_norm": 0.03372584655880928,
"learning_rate": 1.5960255622730163e-06,
"loss": 0.007181422710418701,
"step": 95700
},
{
"epoch": 1.8573479205462444,
"grad_norm": 0.7478105425834656,
"learning_rate": 1.58524888730831e-06,
"loss": 0.0070823818445205685,
"step": 95750
},
{
"epoch": 1.8583178150217257,
"grad_norm": 0.04272932559251785,
"learning_rate": 1.5744722123436037e-06,
"loss": 0.008029258847236632,
"step": 95800
},
{
"epoch": 1.8592877094972067,
"grad_norm": 0.14882275462150574,
"learning_rate": 1.5636955373788973e-06,
"loss": 0.012594330310821533,
"step": 95850
},
{
"epoch": 1.8602576039726877,
"grad_norm": 0.1323213130235672,
"learning_rate": 1.5529188624141909e-06,
"loss": 0.011510368585586548,
"step": 95900
},
{
"epoch": 1.861227498448169,
"grad_norm": 0.2851249873638153,
"learning_rate": 1.5421421874494844e-06,
"loss": 0.007271660566329956,
"step": 95950
},
{
"epoch": 1.86219739292365,
"grad_norm": 2.1523916721343994,
"learning_rate": 1.5313655124847782e-06,
"loss": 0.007877467274665833,
"step": 96000
},
{
"epoch": 1.863167287399131,
"grad_norm": 0.08266094326972961,
"learning_rate": 1.5205888375200716e-06,
"loss": 0.01112757921218872,
"step": 96050
},
{
"epoch": 1.8641371818746122,
"grad_norm": 1.0002660751342773,
"learning_rate": 1.5098121625553652e-06,
"loss": 0.008937347531318664,
"step": 96100
},
{
"epoch": 1.865107076350093,
"grad_norm": 0.05023910477757454,
"learning_rate": 1.499035487590659e-06,
"loss": 0.012742213010787963,
"step": 96150
},
{
"epoch": 1.8660769708255742,
"grad_norm": 0.018416233360767365,
"learning_rate": 1.4882588126259526e-06,
"loss": 0.005195416808128357,
"step": 96200
},
{
"epoch": 1.8670468653010552,
"grad_norm": 0.0056945099495351315,
"learning_rate": 1.4774821376612462e-06,
"loss": 0.006476567387580872,
"step": 96250
},
{
"epoch": 1.8680167597765363,
"grad_norm": 0.0035495914053171873,
"learning_rate": 1.4667054626965395e-06,
"loss": 0.010579663515090942,
"step": 96300
},
{
"epoch": 1.8689866542520175,
"grad_norm": 0.17893879115581512,
"learning_rate": 1.4559287877318335e-06,
"loss": 0.007443230748176575,
"step": 96350
},
{
"epoch": 1.8699565487274985,
"grad_norm": 0.012545437552034855,
"learning_rate": 1.445152112767127e-06,
"loss": 0.009089780449867248,
"step": 96400
},
{
"epoch": 1.8709264432029795,
"grad_norm": 0.0454813688993454,
"learning_rate": 1.4343754378024205e-06,
"loss": 0.007944419980049133,
"step": 96450
},
{
"epoch": 1.8718963376784605,
"grad_norm": 0.18183587491512299,
"learning_rate": 1.423598762837714e-06,
"loss": 0.008951717615127563,
"step": 96500
},
{
"epoch": 1.8728662321539415,
"grad_norm": 0.027300484478473663,
"learning_rate": 1.4128220878730079e-06,
"loss": 0.005890120267868042,
"step": 96550
},
{
"epoch": 1.8738361266294228,
"grad_norm": 0.09700526297092438,
"learning_rate": 1.4020454129083015e-06,
"loss": 0.008143134117126465,
"step": 96600
},
{
"epoch": 1.8748060211049038,
"grad_norm": 0.009609153494238853,
"learning_rate": 1.3912687379435948e-06,
"loss": 0.008795841336250306,
"step": 96650
},
{
"epoch": 1.8757759155803848,
"grad_norm": 0.028468972072005272,
"learning_rate": 1.3804920629788886e-06,
"loss": 0.006313281655311584,
"step": 96700
},
{
"epoch": 1.876745810055866,
"grad_norm": 0.4592001736164093,
"learning_rate": 1.3697153880141822e-06,
"loss": 0.005694099068641663,
"step": 96750
},
{
"epoch": 1.8777157045313468,
"grad_norm": 0.6127103567123413,
"learning_rate": 1.3589387130494758e-06,
"loss": 0.009227088689804076,
"step": 96800
},
{
"epoch": 1.878685599006828,
"grad_norm": 0.1674927920103073,
"learning_rate": 1.3481620380847694e-06,
"loss": 0.007014630436897278,
"step": 96850
},
{
"epoch": 1.879655493482309,
"grad_norm": 0.001068423269316554,
"learning_rate": 1.3373853631200632e-06,
"loss": 0.004451268017292022,
"step": 96900
},
{
"epoch": 1.88062538795779,
"grad_norm": 0.06357524544000626,
"learning_rate": 1.3266086881553567e-06,
"loss": 0.005129663348197937,
"step": 96950
},
{
"epoch": 1.8815952824332713,
"grad_norm": 0.06591501086950302,
"learning_rate": 1.3158320131906501e-06,
"loss": 0.007831496596336364,
"step": 97000
},
{
"epoch": 1.8825651769087524,
"grad_norm": 0.0360637903213501,
"learning_rate": 1.305055338225944e-06,
"loss": 0.006132184267044067,
"step": 97050
},
{
"epoch": 1.8835350713842334,
"grad_norm": 0.01282025221735239,
"learning_rate": 1.2942786632612375e-06,
"loss": 0.012564661502838135,
"step": 97100
},
{
"epoch": 1.8845049658597146,
"grad_norm": 0.07650037109851837,
"learning_rate": 1.283501988296531e-06,
"loss": 0.005789422392845154,
"step": 97150
},
{
"epoch": 1.8854748603351954,
"grad_norm": 0.17856217920780182,
"learning_rate": 1.2727253133318247e-06,
"loss": 0.009976030588150024,
"step": 97200
},
{
"epoch": 1.8864447548106766,
"grad_norm": 0.01678593084216118,
"learning_rate": 1.2619486383671185e-06,
"loss": 0.006699115633964538,
"step": 97250
},
{
"epoch": 1.8874146492861577,
"grad_norm": 0.1250247359275818,
"learning_rate": 1.2511719634024118e-06,
"loss": 0.012197649478912354,
"step": 97300
},
{
"epoch": 1.8883845437616387,
"grad_norm": 0.026601802557706833,
"learning_rate": 1.2403952884377056e-06,
"loss": 0.007774943709373474,
"step": 97350
},
{
"epoch": 1.88935443823712,
"grad_norm": 0.0064043947495520115,
"learning_rate": 1.2296186134729992e-06,
"loss": 0.006855222582817078,
"step": 97400
},
{
"epoch": 1.890324332712601,
"grad_norm": 0.04244533181190491,
"learning_rate": 1.2188419385082928e-06,
"loss": 0.010524344444274903,
"step": 97450
},
{
"epoch": 1.891294227188082,
"grad_norm": 0.000445271551143378,
"learning_rate": 1.2080652635435864e-06,
"loss": 0.011611403226852416,
"step": 97500
},
{
"epoch": 1.8922641216635632,
"grad_norm": 0.3114107847213745,
"learning_rate": 1.19728858857888e-06,
"loss": 0.010610769987106323,
"step": 97550
},
{
"epoch": 1.893234016139044,
"grad_norm": 0.7320522665977478,
"learning_rate": 1.1865119136141735e-06,
"loss": 0.012082910537719727,
"step": 97600
},
{
"epoch": 1.8942039106145252,
"grad_norm": 3.072538137435913,
"learning_rate": 1.1757352386494671e-06,
"loss": 0.012612748146057128,
"step": 97650
},
{
"epoch": 1.8951738050900062,
"grad_norm": 0.0223554614931345,
"learning_rate": 1.1649585636847607e-06,
"loss": 0.0050801122188568115,
"step": 97700
},
{
"epoch": 1.8961436995654872,
"grad_norm": 0.04654264822602272,
"learning_rate": 1.1541818887200545e-06,
"loss": 0.011660633087158203,
"step": 97750
},
{
"epoch": 1.8971135940409685,
"grad_norm": 0.323238730430603,
"learning_rate": 1.143405213755348e-06,
"loss": 0.005464251041412353,
"step": 97800
},
{
"epoch": 1.8980834885164493,
"grad_norm": 0.3786829113960266,
"learning_rate": 1.1326285387906417e-06,
"loss": 0.009254413843154907,
"step": 97850
},
{
"epoch": 1.8990533829919305,
"grad_norm": 0.0881204605102539,
"learning_rate": 1.1218518638259352e-06,
"loss": 0.006912820339202881,
"step": 97900
},
{
"epoch": 1.9000232774674115,
"grad_norm": 0.011120930314064026,
"learning_rate": 1.1110751888612288e-06,
"loss": 0.009433020949363708,
"step": 97950
},
{
"epoch": 1.9009931719428925,
"grad_norm": 2.018315553665161,
"learning_rate": 1.1002985138965224e-06,
"loss": 0.005587102174758911,
"step": 98000
},
{
"epoch": 1.9019630664183738,
"grad_norm": 0.12855657935142517,
"learning_rate": 1.089521838931816e-06,
"loss": 0.008500252962112426,
"step": 98050
},
{
"epoch": 1.9029329608938548,
"grad_norm": 0.07897940278053284,
"learning_rate": 1.0787451639671096e-06,
"loss": 0.010591692924499511,
"step": 98100
},
{
"epoch": 1.9039028553693358,
"grad_norm": 0.11750440299510956,
"learning_rate": 1.0679684890024034e-06,
"loss": 0.01251029372215271,
"step": 98150
},
{
"epoch": 1.904872749844817,
"grad_norm": 0.020481739193201065,
"learning_rate": 1.057191814037697e-06,
"loss": 0.007386518716812134,
"step": 98200
},
{
"epoch": 1.9058426443202978,
"grad_norm": 0.0157831571996212,
"learning_rate": 1.0464151390729905e-06,
"loss": 0.007652405500411987,
"step": 98250
},
{
"epoch": 1.906812538795779,
"grad_norm": 1.2562167644500732,
"learning_rate": 1.0356384641082841e-06,
"loss": 0.009157900810241698,
"step": 98300
},
{
"epoch": 1.90778243327126,
"grad_norm": 0.2843869626522064,
"learning_rate": 1.0248617891435777e-06,
"loss": 0.008741449117660522,
"step": 98350
},
{
"epoch": 1.908752327746741,
"grad_norm": 0.8916972875595093,
"learning_rate": 1.0140851141788713e-06,
"loss": 0.010915225744247437,
"step": 98400
},
{
"epoch": 1.9097222222222223,
"grad_norm": 0.3004150390625,
"learning_rate": 1.0033084392141649e-06,
"loss": 0.010254523754119872,
"step": 98450
},
{
"epoch": 1.9106921166977033,
"grad_norm": 0.04915625974535942,
"learning_rate": 9.925317642494585e-07,
"loss": 0.0065569406747818,
"step": 98500
},
{
"epoch": 1.9116620111731844,
"grad_norm": 0.12359398603439331,
"learning_rate": 9.817550892847523e-07,
"loss": 0.013229272365570068,
"step": 98550
},
{
"epoch": 1.9126319056486656,
"grad_norm": 0.3123663663864136,
"learning_rate": 9.709784143200458e-07,
"loss": 0.007916935682296754,
"step": 98600
},
{
"epoch": 1.9136018001241464,
"grad_norm": 0.014977834187448025,
"learning_rate": 9.602017393553394e-07,
"loss": 0.011487702131271363,
"step": 98650
},
{
"epoch": 1.9145716945996276,
"grad_norm": 0.056082405149936676,
"learning_rate": 9.49425064390633e-07,
"loss": 0.011379657983779908,
"step": 98700
},
{
"epoch": 1.9155415890751086,
"grad_norm": 2.501650094985962,
"learning_rate": 9.386483894259266e-07,
"loss": 0.011750187873840332,
"step": 98750
},
{
"epoch": 1.9165114835505896,
"grad_norm": 0.020055390894412994,
"learning_rate": 9.278717144612203e-07,
"loss": 0.006895939707756042,
"step": 98800
},
{
"epoch": 1.9174813780260709,
"grad_norm": 0.1815965622663498,
"learning_rate": 9.170950394965137e-07,
"loss": 0.007907973527908325,
"step": 98850
},
{
"epoch": 1.9184512725015517,
"grad_norm": 1.2448716163635254,
"learning_rate": 9.063183645318074e-07,
"loss": 0.008194063901901244,
"step": 98900
},
{
"epoch": 1.919421166977033,
"grad_norm": 0.056585829704999924,
"learning_rate": 8.955416895671011e-07,
"loss": 0.006504110693931579,
"step": 98950
},
{
"epoch": 1.920391061452514,
"grad_norm": 0.11757718771696091,
"learning_rate": 8.847650146023946e-07,
"loss": 0.007295999526977539,
"step": 99000
},
{
"epoch": 1.921360955927995,
"grad_norm": 0.24114497005939484,
"learning_rate": 8.739883396376883e-07,
"loss": 0.009650174379348755,
"step": 99050
},
{
"epoch": 1.9223308504034762,
"grad_norm": 0.33005982637405396,
"learning_rate": 8.632116646729819e-07,
"loss": 0.006440539360046387,
"step": 99100
},
{
"epoch": 1.9233007448789572,
"grad_norm": 0.08689321577548981,
"learning_rate": 8.524349897082755e-07,
"loss": 0.011413204669952392,
"step": 99150
},
{
"epoch": 1.9242706393544382,
"grad_norm": 0.05016558989882469,
"learning_rate": 8.41658314743569e-07,
"loss": 0.009782095551490783,
"step": 99200
},
{
"epoch": 1.9252405338299194,
"grad_norm": 0.1457662582397461,
"learning_rate": 8.308816397788627e-07,
"loss": 0.008605839610099792,
"step": 99250
},
{
"epoch": 1.9262104283054002,
"grad_norm": 0.6099115014076233,
"learning_rate": 8.201049648141562e-07,
"loss": 0.005791289806365967,
"step": 99300
},
{
"epoch": 1.9271803227808815,
"grad_norm": 0.001070057158358395,
"learning_rate": 8.093282898494499e-07,
"loss": 0.009243248105049133,
"step": 99350
},
{
"epoch": 1.9281502172563625,
"grad_norm": 0.10314682126045227,
"learning_rate": 7.985516148847436e-07,
"loss": 0.0075647300481796265,
"step": 99400
},
{
"epoch": 1.9291201117318435,
"grad_norm": 0.6658992171287537,
"learning_rate": 7.877749399200372e-07,
"loss": 0.0070918196439743045,
"step": 99450
},
{
"epoch": 1.9300900062073247,
"grad_norm": 0.015978632494807243,
"learning_rate": 7.769982649553308e-07,
"loss": 0.0076174306869506835,
"step": 99500
},
{
"epoch": 1.9310599006828058,
"grad_norm": 0.01855318248271942,
"learning_rate": 7.662215899906243e-07,
"loss": 0.008746723532676698,
"step": 99550
},
{
"epoch": 1.9320297951582868,
"grad_norm": 0.10632064193487167,
"learning_rate": 7.55444915025918e-07,
"loss": 0.005162266492843628,
"step": 99600
},
{
"epoch": 1.932999689633768,
"grad_norm": 0.0408286415040493,
"learning_rate": 7.446682400612115e-07,
"loss": 0.009361888766288757,
"step": 99650
},
{
"epoch": 1.9339695841092488,
"grad_norm": 0.7107354402542114,
"learning_rate": 7.338915650965052e-07,
"loss": 0.00840396523475647,
"step": 99700
},
{
"epoch": 1.93493947858473,
"grad_norm": 0.330807626247406,
"learning_rate": 7.231148901317988e-07,
"loss": 0.01021736979484558,
"step": 99750
},
{
"epoch": 1.935909373060211,
"grad_norm": 0.08130278438329697,
"learning_rate": 7.123382151670924e-07,
"loss": 0.011657531261444093,
"step": 99800
},
{
"epoch": 1.936879267535692,
"grad_norm": 0.0005203223554417491,
"learning_rate": 7.01561540202386e-07,
"loss": 0.009198451638221741,
"step": 99850
},
{
"epoch": 1.9378491620111733,
"grad_norm": 0.6889069676399231,
"learning_rate": 6.907848652376796e-07,
"loss": 0.010637754201889038,
"step": 99900
},
{
"epoch": 1.938819056486654,
"grad_norm": 0.30668315291404724,
"learning_rate": 6.800081902729733e-07,
"loss": 0.011177635192871094,
"step": 99950
},
{
"epoch": 1.9397889509621353,
"grad_norm": 0.13162565231323242,
"learning_rate": 6.692315153082668e-07,
"loss": 0.00863505244255066,
"step": 100000
},
{
"epoch": 1.9407588454376163,
"grad_norm": 0.0011848299764096737,
"learning_rate": 6.584548403435605e-07,
"loss": 0.005853920578956604,
"step": 100050
},
{
"epoch": 1.9417287399130974,
"grad_norm": 0.34323039650917053,
"learning_rate": 6.47678165378854e-07,
"loss": 0.008631235361099244,
"step": 100100
},
{
"epoch": 1.9426986343885786,
"grad_norm": 0.2779937982559204,
"learning_rate": 6.369014904141477e-07,
"loss": 0.006678025126457214,
"step": 100150
},
{
"epoch": 1.9436685288640596,
"grad_norm": 0.002062960295006633,
"learning_rate": 6.261248154494413e-07,
"loss": 0.00972768485546112,
"step": 100200
},
{
"epoch": 1.9446384233395406,
"grad_norm": 0.0016356809064745903,
"learning_rate": 6.153481404847349e-07,
"loss": 0.007599831819534302,
"step": 100250
},
{
"epoch": 1.9456083178150219,
"grad_norm": 0.005748541094362736,
"learning_rate": 6.045714655200285e-07,
"loss": 0.009428836703300476,
"step": 100300
},
{
"epoch": 1.9465782122905027,
"grad_norm": 0.1627378761768341,
"learning_rate": 5.937947905553221e-07,
"loss": 0.008739418983459472,
"step": 100350
},
{
"epoch": 1.9475481067659839,
"grad_norm": 0.025353165343403816,
"learning_rate": 5.830181155906158e-07,
"loss": 0.01084277629852295,
"step": 100400
},
{
"epoch": 1.948518001241465,
"grad_norm": 0.056178461760282516,
"learning_rate": 5.722414406259094e-07,
"loss": 0.0063758170604705814,
"step": 100450
},
{
"epoch": 1.949487895716946,
"grad_norm": 0.003460692474618554,
"learning_rate": 5.614647656612029e-07,
"loss": 0.006974753141403198,
"step": 100500
},
{
"epoch": 1.9504577901924272,
"grad_norm": 0.04695584252476692,
"learning_rate": 5.506880906964965e-07,
"loss": 0.006639321446418762,
"step": 100550
},
{
"epoch": 1.9514276846679082,
"grad_norm": 0.01810281164944172,
"learning_rate": 5.399114157317901e-07,
"loss": 0.007115241289138794,
"step": 100600
},
{
"epoch": 1.9523975791433892,
"grad_norm": 0.31708040833473206,
"learning_rate": 5.291347407670838e-07,
"loss": 0.008110795021057129,
"step": 100650
},
{
"epoch": 1.9533674736188704,
"grad_norm": 0.0004253904044162482,
"learning_rate": 5.183580658023774e-07,
"loss": 0.008554291725158692,
"step": 100700
},
{
"epoch": 1.9543373680943512,
"grad_norm": 0.08235584944486618,
"learning_rate": 5.07581390837671e-07,
"loss": 0.00849874496459961,
"step": 100750
},
{
"epoch": 1.9553072625698324,
"grad_norm": 0.020426906645298004,
"learning_rate": 4.968047158729647e-07,
"loss": 0.00592732846736908,
"step": 100800
},
{
"epoch": 1.9562771570453135,
"grad_norm": 0.06692051887512207,
"learning_rate": 4.860280409082582e-07,
"loss": 0.010624276399612427,
"step": 100850
},
{
"epoch": 1.9572470515207945,
"grad_norm": 0.0674823522567749,
"learning_rate": 4.752513659435518e-07,
"loss": 0.008916394710540771,
"step": 100900
},
{
"epoch": 1.9582169459962757,
"grad_norm": 0.004914544988423586,
"learning_rate": 4.644746909788454e-07,
"loss": 0.009018712043762207,
"step": 100950
},
{
"epoch": 1.9591868404717565,
"grad_norm": 0.43417787551879883,
"learning_rate": 4.5369801601413904e-07,
"loss": 0.007368384003639221,
"step": 101000
},
{
"epoch": 1.9601567349472377,
"grad_norm": 0.0006635240279138088,
"learning_rate": 4.429213410494326e-07,
"loss": 0.012534016370773315,
"step": 101050
},
{
"epoch": 1.9611266294227188,
"grad_norm": 0.004774922039359808,
"learning_rate": 4.321446660847262e-07,
"loss": 0.008402658700942993,
"step": 101100
},
{
"epoch": 1.9620965238981998,
"grad_norm": 0.36077457666397095,
"learning_rate": 4.2136799112001984e-07,
"loss": 0.007970858812332154,
"step": 101150
},
{
"epoch": 1.963066418373681,
"grad_norm": 0.003512367606163025,
"learning_rate": 4.105913161553135e-07,
"loss": 0.006853591799736023,
"step": 101200
},
{
"epoch": 1.964036312849162,
"grad_norm": 0.2746148407459259,
"learning_rate": 3.998146411906071e-07,
"loss": 0.009138476252555847,
"step": 101250
},
{
"epoch": 1.965006207324643,
"grad_norm": 0.025977415964007378,
"learning_rate": 3.890379662259007e-07,
"loss": 0.012794718742370606,
"step": 101300
},
{
"epoch": 1.9659761018001243,
"grad_norm": 0.002292331773787737,
"learning_rate": 3.782612912611943e-07,
"loss": 0.009966141581535339,
"step": 101350
},
{
"epoch": 1.966945996275605,
"grad_norm": 0.015616307966411114,
"learning_rate": 3.674846162964879e-07,
"loss": 0.009807459712028503,
"step": 101400
},
{
"epoch": 1.9679158907510863,
"grad_norm": 0.049884483218193054,
"learning_rate": 3.567079413317815e-07,
"loss": 0.011683801412582398,
"step": 101450
},
{
"epoch": 1.9688857852265673,
"grad_norm": 0.042611073702573776,
"learning_rate": 3.459312663670751e-07,
"loss": 0.010766392946243286,
"step": 101500
},
{
"epoch": 1.9698556797020483,
"grad_norm": 0.29214608669281006,
"learning_rate": 3.351545914023687e-07,
"loss": 0.00970311403274536,
"step": 101550
},
{
"epoch": 1.9708255741775296,
"grad_norm": 0.09411260485649109,
"learning_rate": 3.2437791643766235e-07,
"loss": 0.01058212399482727,
"step": 101600
},
{
"epoch": 1.9717954686530106,
"grad_norm": 0.0006795229273848236,
"learning_rate": 3.13601241472956e-07,
"loss": 0.00846306324005127,
"step": 101650
},
{
"epoch": 1.9727653631284916,
"grad_norm": 0.37969887256622314,
"learning_rate": 3.028245665082496e-07,
"loss": 0.010985256433486938,
"step": 101700
},
{
"epoch": 1.9737352576039728,
"grad_norm": 0.009496955201029778,
"learning_rate": 2.9204789154354316e-07,
"loss": 0.012427825927734375,
"step": 101750
},
{
"epoch": 1.9747051520794536,
"grad_norm": 0.011450120247900486,
"learning_rate": 2.812712165788368e-07,
"loss": 0.00921747922897339,
"step": 101800
},
{
"epoch": 1.9756750465549349,
"grad_norm": 0.0739368200302124,
"learning_rate": 2.704945416141304e-07,
"loss": 0.01020667552947998,
"step": 101850
},
{
"epoch": 1.9766449410304159,
"grad_norm": 0.02177155390381813,
"learning_rate": 2.59717866649424e-07,
"loss": 0.008021920323371887,
"step": 101900
},
{
"epoch": 1.977614835505897,
"grad_norm": 0.04332567751407623,
"learning_rate": 2.489411916847176e-07,
"loss": 0.0072528403997421265,
"step": 101950
},
{
"epoch": 1.9785847299813781,
"grad_norm": 0.08101877570152283,
"learning_rate": 2.3816451672001123e-07,
"loss": 0.010537716150283814,
"step": 102000
},
{
"epoch": 1.9795546244568591,
"grad_norm": 1.1570383310317993,
"learning_rate": 2.2738784175530481e-07,
"loss": 0.004843094348907471,
"step": 102050
},
{
"epoch": 1.9805245189323402,
"grad_norm": 0.05882470682263374,
"learning_rate": 2.1661116679059845e-07,
"loss": 0.008976127505302429,
"step": 102100
},
{
"epoch": 1.9814944134078212,
"grad_norm": 0.05905785411596298,
"learning_rate": 2.0583449182589206e-07,
"loss": 0.00710660994052887,
"step": 102150
},
{
"epoch": 1.9824643078833022,
"grad_norm": 5.326664447784424,
"learning_rate": 1.9505781686118567e-07,
"loss": 0.00559435248374939,
"step": 102200
},
{
"epoch": 1.9834342023587834,
"grad_norm": 0.5267305970191956,
"learning_rate": 1.8428114189647925e-07,
"loss": 0.006998690366744995,
"step": 102250
},
{
"epoch": 1.9844040968342644,
"grad_norm": 2.7762629985809326,
"learning_rate": 1.735044669317729e-07,
"loss": 0.011264873743057251,
"step": 102300
},
{
"epoch": 1.9853739913097455,
"grad_norm": 0.010550913400948048,
"learning_rate": 1.627277919670665e-07,
"loss": 0.008160226345062256,
"step": 102350
},
{
"epoch": 1.9863438857852267,
"grad_norm": 0.004286179319024086,
"learning_rate": 1.519511170023601e-07,
"loss": 0.0065465766191482545,
"step": 102400
},
{
"epoch": 1.9873137802607075,
"grad_norm": 0.22992965579032898,
"learning_rate": 1.4117444203765372e-07,
"loss": 0.009190858602523803,
"step": 102450
},
{
"epoch": 1.9882836747361887,
"grad_norm": 0.03422815352678299,
"learning_rate": 1.3039776707294733e-07,
"loss": 0.007535686492919922,
"step": 102500
},
{
"epoch": 1.9892535692116697,
"grad_norm": 0.44593170285224915,
"learning_rate": 1.1962109210824094e-07,
"loss": 0.008048560619354248,
"step": 102550
},
{
"epoch": 1.9902234636871508,
"grad_norm": 1.0705064535140991,
"learning_rate": 1.0884441714353453e-07,
"loss": 0.0067255586385726925,
"step": 102600
},
{
"epoch": 1.991193358162632,
"grad_norm": 0.004993344657123089,
"learning_rate": 9.806774217882816e-08,
"loss": 0.005870773196220398,
"step": 102650
},
{
"epoch": 1.992163252638113,
"grad_norm": 0.17308974266052246,
"learning_rate": 8.729106721412177e-08,
"loss": 0.006675005555152893,
"step": 102700
},
{
"epoch": 1.993133147113594,
"grad_norm": 0.7259581089019775,
"learning_rate": 7.651439224941537e-08,
"loss": 0.00922674298286438,
"step": 102750
},
{
"epoch": 1.9941030415890753,
"grad_norm": 0.04203763231635094,
"learning_rate": 6.573771728470898e-08,
"loss": 0.014431644678115845,
"step": 102800
},
{
"epoch": 1.995072936064556,
"grad_norm": 0.0592789426445961,
"learning_rate": 5.496104232000259e-08,
"loss": 0.008624550700187684,
"step": 102850
},
{
"epoch": 1.9960428305400373,
"grad_norm": 0.00026709120720624924,
"learning_rate": 4.41843673552962e-08,
"loss": 0.010920953750610352,
"step": 102900
},
{
"epoch": 1.9970127250155183,
"grad_norm": 0.4239858388900757,
"learning_rate": 3.340769239058981e-08,
"loss": 0.009018361568450928,
"step": 102950
},
{
"epoch": 1.9979826194909993,
"grad_norm": 0.01616194099187851,
"learning_rate": 2.2631017425883423e-08,
"loss": 0.007892364263534546,
"step": 103000
},
{
"epoch": 1.9989525139664805,
"grad_norm": 0.0008136804681271315,
"learning_rate": 1.1854342461177029e-08,
"loss": 0.009902952313423157,
"step": 103050
},
{
"epoch": 1.9999224084419616,
"grad_norm": 0.010138635523617268,
"learning_rate": 1.077667496470639e-09,
"loss": 0.006877103447914123,
"step": 103100
},
{
"epoch": 2.0,
"eval_ce-val_accuracy": 0.9968681824263258,
"eval_ce-val_accuracy_threshold": 0.20284655690193176,
"eval_ce-val_average_precision": 0.9996460692260924,
"eval_ce-val_f1": 0.9917682773667099,
"eval_ce-val_f1_threshold": 0.19957852363586426,
"eval_ce-val_precision": 0.9875653889031816,
"eval_ce-val_recall": 0.9960070920963487,
"eval_loss": 0.008623547852039337,
"eval_runtime": 312.9714,
"eval_samples_per_second": 1170.206,
"eval_steps_per_second": 18.286,
"step": 103104
}
],
"logging_steps": 50,
"max_steps": 103104,
"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": 0.0,
"train_batch_size": 64,
"trial_name": null,
"trial_params": null
}
{
"best_global_step": 103104,
"best_metric": 0.9996460692260924,
"best_model_checkpoint": "../ce_model_100k/checkpoint-103104",
"epoch": 2.0,
"eval_steps": 500,
"global_step": 103104,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0009698944754810676,
"grad_norm": 27.08783531188965,
"learning_rate": 9.504412763068568e-08,
"loss": 1.3570481872558593,
"step": 50
},
{
"epoch": 0.0019397889509621352,
"grad_norm": 20.9023494720459,
"learning_rate": 1.92027931335467e-07,
"loss": 1.1735592651367188,
"step": 100
},
{
"epoch": 0.002909683426443203,
"grad_norm": 8.505711555480957,
"learning_rate": 2.890117350402483e-07,
"loss": 0.8990924072265625,
"step": 150
},
{
"epoch": 0.0038795779019242705,
"grad_norm": 4.790644645690918,
"learning_rate": 3.859955387450296e-07,
"loss": 0.6421822357177734,
"step": 200
},
{
"epoch": 0.004849472377405338,
"grad_norm": 5.505950927734375,
"learning_rate": 4.829793424498109e-07,
"loss": 0.45252552032470705,
"step": 250
},
{
"epoch": 0.005819366852886406,
"grad_norm": 6.148208141326904,
"learning_rate": 5.799631461545922e-07,
"loss": 0.3296080780029297,
"step": 300
},
{
"epoch": 0.006789261328367473,
"grad_norm": 2.746929168701172,
"learning_rate": 6.769469498593735e-07,
"loss": 0.21170696258544922,
"step": 350
},
{
"epoch": 0.007759155803848541,
"grad_norm": 3.268200397491455,
"learning_rate": 7.739307535641548e-07,
"loss": 0.2188690185546875,
"step": 400
},
{
"epoch": 0.008729050279329608,
"grad_norm": 4.391243934631348,
"learning_rate": 8.709145572689362e-07,
"loss": 0.21369564056396484,
"step": 450
},
{
"epoch": 0.009698944754810677,
"grad_norm": 3.1883127689361572,
"learning_rate": 9.678983609737174e-07,
"loss": 0.17295127868652344,
"step": 500
},
{
"epoch": 0.010668839230291743,
"grad_norm": 3.346428394317627,
"learning_rate": 1.0648821646784986e-06,
"loss": 0.16976539611816407,
"step": 550
},
{
"epoch": 0.011638733705772812,
"grad_norm": 2.9807257652282715,
"learning_rate": 1.1618659683832801e-06,
"loss": 0.16226112365722656,
"step": 600
},
{
"epoch": 0.01260862818125388,
"grad_norm": 2.836063861846924,
"learning_rate": 1.2588497720880614e-06,
"loss": 0.1293662452697754,