Team Ai
Modelpublic

souvik-biswas/codegemma-webdev-lora

sourceHugging Faceotherupdated 17d agoView on Hugging Face
0likes48downloads
trainer_state.json636 linesDownload Raw Back to root
1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 3.0,6  "eval_steps": 200,7  "global_step": 804,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.037418147801683815,14      "grad_norm": 0.4778975248336792,15      "learning_rate": 4.390243902439025e-05,16      "loss": 0.6015380859375,17      "step": 1018    },19    {20      "epoch": 0.07483629560336763,21      "grad_norm": 0.6195058822631836,22      "learning_rate": 9.26829268292683e-05,23      "loss": 0.3497171401977539,24      "step": 2025    },26    {27      "epoch": 0.11225444340505145,28      "grad_norm": 0.39988505840301514,29      "learning_rate": 0.00014146341463414634,30      "loss": 0.07948170900344849,31      "step": 3032    },33    {34      "epoch": 0.14967259120673526,35      "grad_norm": 0.21401679515838623,36      "learning_rate": 0.0001902439024390244,37      "loss": 0.03847042918205261,38      "step": 4039    },40    {41      "epoch": 0.18709073900841908,42      "grad_norm": 0.09664341062307358,43      "learning_rate": 0.0001999457547918641,44      "loss": 0.032358613610267636,45      "step": 5046    },47    {48      "epoch": 0.2245088868101029,49      "grad_norm": 0.10169447958469391,50      "learning_rate": 0.0001997254844924806,51      "loss": 0.03248982727527618,52      "step": 6053    },54    {55      "epoch": 0.26192703461178674,56      "grad_norm": 0.10140790790319443,57      "learning_rate": 0.00019933617186395917,58      "loss": 0.027928534150123595,59      "step": 7060    },61    {62      "epoch": 0.2993451824134705,63      "grad_norm": 0.07614590972661972,64      "learning_rate": 0.00019877847682100294,65      "loss": 0.03329395055770874,66      "step": 8067    },68    {69      "epoch": 0.33676333021515437,70      "grad_norm": 0.06965985894203186,71      "learning_rate": 0.00019805334469939528,72      "loss": 0.029634883999824523,73      "step": 9074    },75    {76      "epoch": 0.37418147801683815,77      "grad_norm": 0.11948850005865097,78      "learning_rate": 0.0001971620046535835,79      "loss": 0.029285627603530883,80      "step": 10081    },82    {83      "epoch": 0.411599625818522,84      "grad_norm": 0.06373175233602524,85      "learning_rate": 0.0001961059675731678,86      "loss": 0.02732127606868744,87      "step": 11088    },89    {90      "epoch": 0.4490177736202058,91      "grad_norm": 0.04860406368970871,92      "learning_rate": 0.00019488702352182786,93      "loss": 0.027704697847366334,94      "step": 12095    },96    {97      "epoch": 0.4864359214218896,98      "grad_norm": 0.12966613471508026,99      "learning_rate": 0.0001935072387030279,100      "loss": 0.0346372127532959,101      "step": 130102    },103    {104      "epoch": 0.5238540692235735,105      "grad_norm": 0.07111155986785889,106      "learning_rate": 0.00019196895195764362,107      "loss": 0.029640913009643555,108      "step": 140109    },110    {111      "epoch": 0.5612722170252572,112      "grad_norm": 0.07331525534391403,113      "learning_rate": 0.00019027477079944788,114      "loss": 0.027310377359390257,115      "step": 150116    },117    {118      "epoch": 0.598690364826941,119      "grad_norm": 0.06717830151319504,120      "learning_rate": 0.00018842756699517538,121      "loss": 0.02696542739868164,122      "step": 160123    },124    {125      "epoch": 0.6361085126286249,126      "grad_norm": 0.05457225814461708,127      "learning_rate": 0.0001864304716966581,128      "loss": 0.02828747034072876,129      "step": 170130    },131    {132      "epoch": 0.6735266604303087,133      "grad_norm": 0.051271021366119385,134      "learning_rate": 0.00018428687013328336,135      "loss": 0.025368821620941163,136      "step": 180137    },138    {139      "epoch": 0.7109448082319925,140      "grad_norm": 0.030938835814595222,141      "learning_rate": 0.00018200039587377097,142      "loss": 0.02721564769744873,143      "step": 190144    },145    {146      "epoch": 0.7483629560336763,147      "grad_norm": 0.030669471248984337,148      "learning_rate": 0.00017957492466699605,149      "loss": 0.02553218901157379,150      "step": 200151    },152    {153      "epoch": 0.7483629560336763,154      "eval_loss": 0.028854047879576683,155      "eval_runtime": 112.1375,156      "eval_samples_per_second": 4.236,157      "eval_steps_per_second": 1.061,158      "step": 200159    },160    {161      "epoch": 0.7857811038353602,162      "grad_norm": 0.055315256118774414,163      "learning_rate": 0.00017701456787229804,164      "loss": 0.027117228507995604,165      "step": 210166    },167    {168      "epoch": 0.823199251637044,169      "grad_norm": 0.06075793504714966,170      "learning_rate": 0.000174323665490412,171      "loss": 0.026607033610343934,172      "step": 220173    },174    {175      "epoch": 0.8606173994387278,176      "grad_norm": 0.03559086099267006,177      "learning_rate": 0.00017150677880683514,178      "loss": 0.026579588651657104,179      "step": 230180    },181    {182      "epoch": 0.8980355472404116,183      "grad_norm": 0.056787408888339996,184      "learning_rate": 0.00016856868266009874,185      "loss": 0.0274478018283844,186      "step": 240187    },188    {189      "epoch": 0.9354536950420954,190      "grad_norm": 0.02651643380522728,191      "learning_rate": 0.0001655143573480515,192      "loss": 0.027723655104637146,193      "step": 250194    },195    {196      "epoch": 0.9728718428437793,197      "grad_norm": 0.04855380207300186,198      "learning_rate": 0.00016234898018587337,199      "loss": 0.030566543340682983,200      "step": 260201    },202    {203      "epoch": 1.0074836295603367,204      "grad_norm": 0.038310177624225616,205      "learning_rate": 0.00015907791673013016,206      "loss": 0.03060322403907776,207      "step": 270208    },209    {210      "epoch": 1.0449017773620206,211      "grad_norm": 0.049252696335315704,212      "learning_rate": 0.00015570671168374438,213      "loss": 0.025193819403648378,214      "step": 280215    },216    {217      "epoch": 1.0823199251637043,218      "grad_norm": 0.058025866746902466,219      "learning_rate": 0.0001522410794972995,220      "loss": 0.02864006459712982,221      "step": 290222    },223    {224      "epoch": 1.1197380729653883,225      "grad_norm": 0.02883829176425934,226      "learning_rate": 0.00014868689468260874,227      "loss": 0.026764112710952758,228      "step": 300229    },230    {231      "epoch": 1.157156220767072,232      "grad_norm": 0.028890302404761314,233      "learning_rate": 0.00014505018185496802,234      "loss": 0.02434994578361511,235      "step": 310236    },237    {238      "epoch": 1.1945743685687558,239      "grad_norm": 0.020377911627292633,240      "learning_rate": 0.00014133710552097174,241      "loss": 0.02561296820640564,242      "step": 320243    },244    {245      "epoch": 1.2319925163704397,246      "grad_norm": 0.02488919533789158,247      "learning_rate": 0.0001375539596292025,248      "loss": 0.025670444965362547,249      "step": 330250    },251    {252      "epoch": 1.2694106641721234,253      "grad_norm": 0.031621985137462616,254      "learning_rate": 0.0001337071569015063,255      "loss": 0.025845205783843993,256      "step": 340257    },258    {259      "epoch": 1.3068288119738072,260      "grad_norm": 0.034460149705410004,261      "learning_rate": 0.00012980321796293836,262      "loss": 0.02731255292892456,263      "step": 350264    },265    {266      "epoch": 1.3442469597754911,267      "grad_norm": 0.07521438598632812,268      "learning_rate": 0.00012584876028880453,269      "loss": 0.027375590801239014,270      "step": 360271    },272    {273      "epoch": 1.3816651075771749,274      "grad_norm": 0.044437043368816376,275      "learning_rate": 0.00012185048698753403,276      "loss": 0.026981157064437867,277      "step": 370278    },279    {280      "epoch": 1.4190832553788588,281      "grad_norm": 0.02358797751367092,282      "learning_rate": 0.00011781517543839768,283      "loss": 0.026310837268829344,284      "step": 380285    },286    {287      "epoch": 1.4565014031805426,288      "grad_norm": 0.05172913521528244,289      "learning_rate": 0.00011374966580333147,290      "loss": 0.026896259188652037,291      "step": 390292    },293    {294      "epoch": 1.4939195509822265,295      "grad_norm": 0.027119116857647896,296      "learning_rate": 0.00010966084943233817,297      "loss": 0.027932611107826234,298      "step": 400299    },300    {301      "epoch": 1.4939195509822265,302      "eval_loss": 0.028618767857551575,303      "eval_runtime": 111.9869,304      "eval_samples_per_second": 4.242,305      "eval_steps_per_second": 1.063,306      "step": 400307    },308    {309      "epoch": 1.5313376987839102,310      "grad_norm": 0.027601879090070724,311      "learning_rate": 0.00010555565718212197,312      "loss": 0.02582102417945862,313      "step": 410314    },315    {316      "epoch": 1.568755846585594,317      "grad_norm": 0.025182051584124565,318      "learning_rate": 0.00010144104766775572,319      "loss": 0.029713955521583558,320      "step": 420321    },322    {323      "epoch": 1.606173994387278,324      "grad_norm": 0.033296652138233185,325      "learning_rate": 9.732399546729578e-05,326      "loss": 0.028361794352531434,327      "step": 430328    },329    {330      "epoch": 1.6435921421889617,331      "grad_norm": 0.016636556014418602,332      "learning_rate": 9.321147929933847e-05,333      "loss": 0.026169759035110474,334      "step": 440335    },336    {337      "epoch": 1.6810102899906454,338      "grad_norm": 0.03441890701651573,339      "learning_rate": 8.911047019355764e-05,340      "loss": 0.03007696270942688,341      "step": 450342    },343    {344      "epoch": 1.7184284377923293,345      "grad_norm": 0.024368125945329666,346      "learning_rate": 8.502791967427618e-05,347      "loss": 0.02840135395526886,348      "step": 460349    },350    {351      "epoch": 1.755846585594013,352      "grad_norm": 0.030341170728206635,353      "learning_rate": 8.097074797710006e-05,354      "loss": 0.027544933557510375,355      "step": 470356    },357    {358      "epoch": 1.7932647333956968,359      "grad_norm": 0.03189330920577049,360      "learning_rate": 7.69458323185892e-05,361      "loss": 0.026382505893707275,362      "step": 480363    },364    {365      "epoch": 1.8306828811973808,366      "grad_norm": 0.030166875571012497,367      "learning_rate": 7.295999523884921e-05,368      "loss": 0.026857292652130126,369      "step": 490370    },371    {372      "epoch": 1.8681010289990645,373      "grad_norm": 0.022643962875008583,374      "learning_rate": 6.901999303680358e-05,375      "loss": 0.024822740256786345,376      "step": 500377    },378    {379      "epoch": 1.9055191768007482,380      "grad_norm": 0.025209374725818634,381      "learning_rate": 6.513250431775004e-05,382      "loss": 0.02585785388946533,383      "step": 510384    },385    {386      "epoch": 1.9429373246024322,387      "grad_norm": 0.02024596370756626,388      "learning_rate": 6.13041186726131e-05,389      "loss": 0.023086994886398315,390      "step": 520391    },392    {393      "epoch": 1.9803554724041161,394      "grad_norm": 0.03264998644590378,395      "learning_rate": 5.754132550808344e-05,396      "loss": 0.027441507577896117,397      "step": 530398    },399    {400      "epoch": 2.0149672591206733,401      "grad_norm": 0.02290951833128929,402      "learning_rate": 5.385050304657669e-05,403      "loss": 0.027830862998962404,404      "step": 540405    },406    {407      "epoch": 2.0523854069223573,408      "grad_norm": 0.04214494302868843,409      "learning_rate": 5.0237907514658176e-05,410      "loss": 0.028206586837768555,411      "step": 550412    },413    {414      "epoch": 2.0898035547240412,415      "grad_norm": 0.017041301354765892,416      "learning_rate": 4.6709662538260267e-05,417      "loss": 0.026084256172180176,418      "step": 560419    },420    {421      "epoch": 2.127221702525725,422      "grad_norm": 0.023693662136793137,423      "learning_rate": 4.327174876266743e-05,424      "loss": 0.02829332947731018,425      "step": 570426    },427    {428      "epoch": 2.1646398503274087,429      "grad_norm": 0.0256296806037426,430      "learning_rate": 3.992999371486508e-05,431      "loss": 0.026536005735397338,432      "step": 580433    },434    {435      "epoch": 2.2020579981290926,436      "grad_norm": 0.029245780780911446,437      "learning_rate": 3.6690061925434814e-05,438      "loss": 0.02760576903820038,439      "step": 590440    },441    {442      "epoch": 2.2394761459307766,443      "grad_norm": 0.03691583499312401,444      "learning_rate": 3.355744532674211e-05,445      "loss": 0.02619142234325409,446      "step": 600447    },448    {449      "epoch": 2.2394761459307766,450      "eval_loss": 0.028461672365665436,451      "eval_runtime": 111.6932,452      "eval_samples_per_second": 4.253,453      "eval_steps_per_second": 1.065,454      "step": 600455    },456    {457      "epoch": 2.27689429373246,458      "grad_norm": 0.025541063398122787,459      "learning_rate": 3.0537453943690074e-05,460      "loss": 0.025616389513015748,461      "step": 610462    },463    {464      "epoch": 2.314312441534144,465      "grad_norm": 0.02622550167143345,466      "learning_rate": 2.763520689282122e-05,467      "loss": 0.023774966597557068,468      "step": 620469    },470    {471      "epoch": 2.351730589335828,472      "grad_norm": 0.032154783606529236,473      "learning_rate": 2.485562370502279e-05,474      "loss": 0.025150638818740845,475      "step": 630476    },477    {478      "epoch": 2.3891487371375115,479      "grad_norm": 0.022133855149149895,480      "learning_rate": 2.220341598654565e-05,481      "loss": 0.023556211590766908,482      "step": 640483    },484    {485      "epoch": 2.4265668849391955,486      "grad_norm": 0.020089825615286827,487      "learning_rate": 1.9683079432470775e-05,488      "loss": 0.02604680061340332,489      "step": 650490    },491    {492      "epoch": 2.4639850327408794,493      "grad_norm": 0.026050923392176628,494      "learning_rate": 1.729888620616228e-05,495      "loss": 0.024851146340370178,496      "step": 660497    },498    {499      "epoch": 2.501403180542563,500      "grad_norm": 0.02465742453932762,501      "learning_rate": 1.5054877697623304e-05,502      "loss": 0.026841408014297484,503      "step": 670504    },505    {506      "epoch": 2.538821328344247,507      "grad_norm": 0.024667708203196526,508      "learning_rate": 1.2954857673030807e-05,509      "loss": 0.026710045337677003,510      "step": 680511    },512    {513      "epoch": 2.576239476145931,514      "grad_norm": 0.026438958942890167,515      "learning_rate": 1.1002385827060602e-05,516      "loss": 0.02445058524608612,517      "step": 690518    },519    {520      "epoch": 2.6136576239476144,521      "grad_norm": 0.02051541581749916,522      "learning_rate": 9.200771748932513e-06,523      "loss": 0.02866950035095215,524      "step": 700525    },526    {527      "epoch": 2.6510757717492983,528      "grad_norm": 0.03139111399650574,529      "learning_rate": 7.553069312403349e-06,530      "loss": 0.02889391779899597,531      "step": 710532    },533    {534      "epoch": 2.6884939195509823,535      "grad_norm": 0.024994639679789543,536      "learning_rate": 6.062071499217081e-06,537      "loss": 0.02809330224990845,538      "step": 720539    },540    {541      "epoch": 2.725912067352666,542      "grad_norm": 0.024395441636443138,543      "learning_rate": 4.73030566478706e-06,544      "loss": 0.02776895761489868,545      "step": 730546    },547    {548      "epoch": 2.7633302151543497,549      "grad_norm": 0.022323913872241974,550      "learning_rate": 3.5600292541351e-06,551      "loss": 0.02733958661556244,552      "step": 740553    },554    {555      "epoch": 2.8007483629560337,556      "grad_norm": 0.023678358644247055,557      "learning_rate": 2.5532259753494827e-06,558      "loss": 0.026967012882232667,559      "step": 750560    },561    {562      "epoch": 2.8381665107577176,563      "grad_norm": 0.0324757881462574,564      "learning_rate": 1.7116024370477878e-06,565      "loss": 0.024685764312744142,566      "step": 760567    },568    {569      "epoch": 2.875584658559401,570      "grad_norm": 0.031328026205301285,571      "learning_rate": 1.036585255544764e-06,572      "loss": 0.023696568608283997,573      "step": 770574    },575    {576      "epoch": 2.913002806361085,577      "grad_norm": 0.02811577543616295,578      "learning_rate": 5.293186366284176e-07,579      "loss": 0.02752375602722168,580      "step": 780581    },582    {583      "epoch": 2.950420954162769,584      "grad_norm": 0.025879858061671257,585      "learning_rate": 1.9066243604367594e-07,586      "loss": 0.030453172326087952,587      "step": 790588    },589    {590      "epoch": 2.987839101964453,591      "grad_norm": 0.023125233128666878,592      "learning_rate": 2.1190701971052218e-08,593      "loss": 0.02600964605808258,594      "step": 800595    },596    {597      "epoch": 2.987839101964453,598      "eval_loss": 0.0283315721899271,599      "eval_runtime": 111.8308,600      "eval_samples_per_second": 4.247,601      "eval_steps_per_second": 1.064,602      "step": 800603    },604    {605      "epoch": 3.0,606      "step": 804,607      "total_flos": 1.1187677673658778e+17,608      "train_loss": 0.039224508037762854,609      "train_runtime": 8954.9146,610      "train_samples_per_second": 1.432,611      "train_steps_per_second": 0.09612    }613  ],614  "logging_steps": 10,615  "max_steps": 804,616  "num_input_tokens_seen": 0,617  "num_train_epochs": 3,618  "save_steps": 200,619  "stateful_callbacks": {620    "TrainerControl": {621      "args": {622        "should_epoch_stop": false,623        "should_evaluate": false,624        "should_log": false,625        "should_save": true,626        "should_training_stop": true627      },628      "attributes": {}629    }630  },631  "total_flos": 1.1187677673658778e+17,632  "train_batch_size": 2,633  "trial_name": null,634  "trial_params": null635}636