Team Ai
Modelpublic

dzungpham/graphcodebert-code-classification

sourceHugging Facemitupdated 5mo agoView on Hugging Face
0likes
trainer_state.json604 linesDownload Raw Back to checkpoint-800
1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 0.0512,6  "eval_steps": 1000,7  "global_step": 800,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.00064,14      "grad_norm": 1.8688431978225708,15      "learning_rate": 3.840409643695328e-08,16      "loss": 0.7168,17      "step": 1018    },19    {20      "epoch": 0.00128,21      "grad_norm": 1.835353970527649,22      "learning_rate": 8.10753147002347e-08,23      "loss": 0.7179,24      "step": 2025    },26    {27      "epoch": 0.00192,28      "grad_norm": 1.2541388273239136,29      "learning_rate": 1.2374653296351612e-07,30      "loss": 0.7177,31      "step": 3032    },33    {34      "epoch": 0.00256,35      "grad_norm": 1.1381633281707764,36      "learning_rate": 1.6641775122679754e-07,37      "loss": 0.7312,38      "step": 4039    },40    {41      "epoch": 0.0032,42      "grad_norm": 1.6313824653625488,43      "learning_rate": 2.0908896949007894e-07,44      "loss": 0.7167,45      "step": 5046    },47    {48      "epoch": 0.00384,49      "grad_norm": 1.092464566230774,50      "learning_rate": 2.517601877533604e-07,51      "loss": 0.7201,52      "step": 6053    },54    {55      "epoch": 0.00448,56      "grad_norm": 1.164099097251892,57      "learning_rate": 2.944314060166418e-07,58      "loss": 0.7195,59      "step": 7060    },61    {62      "epoch": 0.00512,63      "grad_norm": 1.9283920526504517,64      "learning_rate": 3.371026242799232e-07,65      "loss": 0.7223,66      "step": 8067    },68    {69      "epoch": 0.00576,70      "grad_norm": 1.8716129064559937,71      "learning_rate": 3.7977384254320464e-07,72      "loss": 0.7151,73      "step": 9074    },75    {76      "epoch": 0.0064,77      "grad_norm": 1.5499262809753418,78      "learning_rate": 4.22445060806486e-07,79      "loss": 0.7211,80      "step": 10081    },82    {83      "epoch": 0.00704,84      "grad_norm": 1.375114917755127,85      "learning_rate": 4.651162790697675e-07,86      "loss": 0.7247,87      "step": 11088    },89    {90      "epoch": 0.00768,91      "grad_norm": 1.384915828704834,92      "learning_rate": 5.077874973330489e-07,93      "loss": 0.7156,94      "step": 12095    },96    {97      "epoch": 0.00832,98      "grad_norm": 1.6376659870147705,99      "learning_rate": 5.504587155963304e-07,100      "loss": 0.7196,101      "step": 130102    },103    {104      "epoch": 0.00896,105      "grad_norm": 1.709489345550537,106      "learning_rate": 5.931299338596117e-07,107      "loss": 0.7215,108      "step": 140109    },110    {111      "epoch": 0.0096,112      "grad_norm": 1.350512146949768,113      "learning_rate": 6.358011521228932e-07,114      "loss": 0.71,115      "step": 150116    },117    {118      "epoch": 0.01024,119      "grad_norm": 2.33050537109375,120      "learning_rate": 6.784723703861745e-07,121      "loss": 0.7191,122      "step": 160123    },124    {125      "epoch": 0.01088,126      "grad_norm": 1.0042874813079834,127      "learning_rate": 7.21143588649456e-07,128      "loss": 0.72,129      "step": 170130    },131    {132      "epoch": 0.01152,133      "grad_norm": 1.1835744380950928,134      "learning_rate": 7.638148069127374e-07,135      "loss": 0.7122,136      "step": 180137    },138    {139      "epoch": 0.01216,140      "grad_norm": 1.949506402015686,141      "learning_rate": 8.064860251760189e-07,142      "loss": 0.7091,143      "step": 190144    },145    {146      "epoch": 0.0128,147      "grad_norm": 1.139626383781433,148      "learning_rate": 8.491572434393003e-07,149      "loss": 0.7046,150      "step": 200151    },152    {153      "epoch": 0.01344,154      "grad_norm": 1.7734779119491577,155      "learning_rate": 8.918284617025817e-07,156      "loss": 0.7135,157      "step": 210158    },159    {160      "epoch": 0.01408,161      "grad_norm": 1.3442974090576172,162      "learning_rate": 9.344996799658632e-07,163      "loss": 0.7127,164      "step": 220165    },166    {167      "epoch": 0.01472,168      "grad_norm": 1.6148335933685303,169      "learning_rate": 9.771708982291445e-07,170      "loss": 0.7168,171      "step": 230172    },173    {174      "epoch": 0.01536,175      "grad_norm": 0.9220213294029236,176      "learning_rate": 1.0198421164924258e-06,177      "loss": 0.7109,178      "step": 240179    },180    {181      "epoch": 0.016,182      "grad_norm": 1.5631815195083618,183      "learning_rate": 1.0625133347557074e-06,184      "loss": 0.7149,185      "step": 250186    },187    {188      "epoch": 0.01664,189      "grad_norm": 1.5052822828292847,190      "learning_rate": 1.1051845530189888e-06,191      "loss": 0.7112,192      "step": 260193    },194    {195      "epoch": 0.01728,196      "grad_norm": 1.6714903116226196,197      "learning_rate": 1.1478557712822702e-06,198      "loss": 0.7128,199      "step": 270200    },201    {202      "epoch": 0.01792,203      "grad_norm": 1.6667262315750122,204      "learning_rate": 1.1905269895455517e-06,205      "loss": 0.7091,206      "step": 280207    },208    {209      "epoch": 0.01856,210      "grad_norm": 1.6861907243728638,211      "learning_rate": 1.233198207808833e-06,212      "loss": 0.7078,213      "step": 290214    },215    {216      "epoch": 0.0192,217      "grad_norm": 1.6259212493896484,218      "learning_rate": 1.2758694260721145e-06,219      "loss": 0.7087,220      "step": 300221    },222    {223      "epoch": 0.01984,224      "grad_norm": 1.207320213317871,225      "learning_rate": 1.318540644335396e-06,226      "loss": 0.7087,227      "step": 310228    },229    {230      "epoch": 0.02048,231      "grad_norm": 1.087124228477478,232      "learning_rate": 1.3612118625986772e-06,233      "loss": 0.7174,234      "step": 320235    },236    {237      "epoch": 0.02112,238      "grad_norm": 1.5668721199035645,239      "learning_rate": 1.4038830808619588e-06,240      "loss": 0.7066,241      "step": 330242    },243    {244      "epoch": 0.02176,245      "grad_norm": 1.5332859754562378,246      "learning_rate": 1.4465542991252401e-06,247      "loss": 0.7115,248      "step": 340249    },250    {251      "epoch": 0.0224,252      "grad_norm": 2.092994451522827,253      "learning_rate": 1.4892255173885215e-06,254      "loss": 0.7085,255      "step": 350256    },257    {258      "epoch": 0.02304,259      "grad_norm": 1.266851544380188,260      "learning_rate": 1.531896735651803e-06,261      "loss": 0.7091,262      "step": 360263    },264    {265      "epoch": 0.02368,266      "grad_norm": 1.290616512298584,267      "learning_rate": 1.5745679539150842e-06,268      "loss": 0.7103,269      "step": 370270    },271    {272      "epoch": 0.02432,273      "grad_norm": 1.2045000791549683,274      "learning_rate": 1.6172391721783658e-06,275      "loss": 0.705,276      "step": 380277    },278    {279      "epoch": 0.02496,280      "grad_norm": 1.828832983970642,281      "learning_rate": 1.6599103904416472e-06,282      "loss": 0.7038,283      "step": 390284    },285    {286      "epoch": 0.0256,287      "grad_norm": 1.7793196439743042,288      "learning_rate": 1.7025816087049288e-06,289      "loss": 0.7085,290      "step": 400291    },292    {293      "epoch": 0.02624,294      "grad_norm": 0.8949472904205322,295      "learning_rate": 1.7452528269682101e-06,296      "loss": 0.7045,297      "step": 410298    },299    {300      "epoch": 0.02688,301      "grad_norm": 1.3071945905685425,302      "learning_rate": 1.7879240452314913e-06,303      "loss": 0.702,304      "step": 420305    },306    {307      "epoch": 0.02752,308      "grad_norm": 1.7497148513793945,309      "learning_rate": 1.8305952634947729e-06,310      "loss": 0.7073,311      "step": 430312    },313    {314      "epoch": 0.02816,315      "grad_norm": 1.0856297016143799,316      "learning_rate": 1.8732664817580542e-06,317      "loss": 0.7008,318      "step": 440319    },320    {321      "epoch": 0.0288,322      "grad_norm": 1.369019627571106,323      "learning_rate": 1.915937700021336e-06,324      "loss": 0.7059,325      "step": 450326    },327    {328      "epoch": 0.02944,329      "grad_norm": 2.0961010456085205,330      "learning_rate": 1.958608918284617e-06,331      "loss": 0.6952,332      "step": 460333    },334    {335      "epoch": 0.03008,336      "grad_norm": 1.6076347827911377,337      "learning_rate": 2.0012801365478988e-06,338      "loss": 0.7074,339      "step": 470340    },341    {342      "epoch": 0.03072,343      "grad_norm": 1.3129311800003052,344      "learning_rate": 2.04395135481118e-06,345      "loss": 0.6994,346      "step": 480347    },348    {349      "epoch": 0.03136,350      "grad_norm": 1.3621476888656616,351      "learning_rate": 2.0866225730744615e-06,352      "loss": 0.696,353      "step": 490354    },355    {356      "epoch": 0.032,357      "grad_norm": 0.9786806106567383,358      "learning_rate": 2.129293791337743e-06,359      "loss": 0.7011,360      "step": 500361    },362    {363      "epoch": 0.03264,364      "grad_norm": 1.4611176252365112,365      "learning_rate": 2.1719650096010242e-06,366      "loss": 0.699,367      "step": 510368    },369    {370      "epoch": 0.03328,371      "grad_norm": 1.0675945281982422,372      "learning_rate": 2.214636227864306e-06,373      "loss": 0.6906,374      "step": 520375    },376    {377      "epoch": 0.03392,378      "grad_norm": 1.6564017534255981,379      "learning_rate": 2.257307446127587e-06,380      "loss": 0.6923,381      "step": 530382    },383    {384      "epoch": 0.03456,385      "grad_norm": 1.228119134902954,386      "learning_rate": 2.2999786643908685e-06,387      "loss": 0.6928,388      "step": 540389    },390    {391      "epoch": 0.0352,392      "grad_norm": 1.9868593215942383,393      "learning_rate": 2.34264988265415e-06,394      "loss": 0.6912,395      "step": 550396    },397    {398      "epoch": 0.03584,399      "grad_norm": 1.2531176805496216,400      "learning_rate": 2.3853211009174317e-06,401      "loss": 0.6923,402      "step": 560403    },404    {405      "epoch": 0.03648,406      "grad_norm": 1.413602352142334,407      "learning_rate": 2.427992319180713e-06,408      "loss": 0.6993,409      "step": 570410    },411    {412      "epoch": 0.03712,413      "grad_norm": 0.9655390977859497,414      "learning_rate": 2.470663537443994e-06,415      "loss": 0.7017,416      "step": 580417    },418    {419      "epoch": 0.03776,420      "grad_norm": 1.962438941001892,421      "learning_rate": 2.5133347557072756e-06,422      "loss": 0.6933,423      "step": 590424    },425    {426      "epoch": 0.0384,427      "grad_norm": 1.1099931001663208,428      "learning_rate": 2.556005973970557e-06,429      "loss": 0.6925,430      "step": 600431    },432    {433      "epoch": 0.03904,434      "grad_norm": 1.1766624450683594,435      "learning_rate": 2.5986771922338383e-06,436      "loss": 0.6894,437      "step": 610438    },439    {440      "epoch": 0.03968,441      "grad_norm": 1.574353814125061,442      "learning_rate": 2.64134841049712e-06,443      "loss": 0.6923,444      "step": 620445    },446    {447      "epoch": 0.04032,448      "grad_norm": 1.194074273109436,449      "learning_rate": 2.6840196287604015e-06,450      "loss": 0.6855,451      "step": 630452    },453    {454      "epoch": 0.04096,455      "grad_norm": 1.1750593185424805,456      "learning_rate": 2.7266908470236826e-06,457      "loss": 0.7002,458      "step": 640459    },460    {461      "epoch": 0.0416,462      "grad_norm": 1.3165347576141357,463      "learning_rate": 2.7693620652869642e-06,464      "loss": 0.6893,465      "step": 650466    },467    {468      "epoch": 0.04224,469      "grad_norm": 1.1079384088516235,470      "learning_rate": 2.812033283550246e-06,471      "loss": 0.6852,472      "step": 660473    },474    {475      "epoch": 0.04288,476      "grad_norm": 1.231327772140503,477      "learning_rate": 2.8547045018135274e-06,478      "loss": 0.695,479      "step": 670480    },481    {482      "epoch": 0.04352,483      "grad_norm": 1.966036319732666,484      "learning_rate": 2.897375720076808e-06,485      "loss": 0.6833,486      "step": 680487    },488    {489      "epoch": 0.04416,490      "grad_norm": 1.0460131168365479,491      "learning_rate": 2.9400469383400897e-06,492      "loss": 0.6802,493      "step": 690494    },495    {496      "epoch": 0.0448,497      "grad_norm": 1.2445200681686401,498      "learning_rate": 2.9827181566033713e-06,499      "loss": 0.6883,500      "step": 700501    },502    {503      "epoch": 0.04544,504      "grad_norm": 1.1759636402130127,505      "learning_rate": 3.025389374866653e-06,506      "loss": 0.682,507      "step": 710508    },509    {510      "epoch": 0.04608,511      "grad_norm": 1.2478561401367188,512      "learning_rate": 3.068060593129934e-06,513      "loss": 0.6802,514      "step": 720515    },516    {517      "epoch": 0.04672,518      "grad_norm": 0.8945108652114868,519      "learning_rate": 3.1107318113932156e-06,520      "loss": 0.6873,521      "step": 730522    },523    {524      "epoch": 0.04736,525      "grad_norm": 1.6810317039489746,526      "learning_rate": 3.153403029656497e-06,527      "loss": 0.6828,528      "step": 740529    },530    {531      "epoch": 0.048,532      "grad_norm": 2.882283926010132,533      "learning_rate": 3.1960742479197783e-06,534      "loss": 0.6797,535      "step": 750536    },537    {538      "epoch": 0.04864,539      "grad_norm": 1.575766921043396,540      "learning_rate": 3.23874546618306e-06,541      "loss": 0.6826,542      "step": 760543    },544    {545      "epoch": 0.04928,546      "grad_norm": 1.8044737577438354,547      "learning_rate": 3.2814166844463415e-06,548      "loss": 0.6837,549      "step": 770550    },551    {552      "epoch": 0.04992,553      "grad_norm": 1.4755513668060303,554      "learning_rate": 3.324087902709623e-06,555      "loss": 0.6815,556      "step": 780557    },558    {559      "epoch": 0.05056,560      "grad_norm": 2.0735654830932617,561      "learning_rate": 3.3667591209729038e-06,562      "loss": 0.6773,563      "step": 790564    },565    {566      "epoch": 0.0512,567      "grad_norm": 0.9823655486106873,568      "learning_rate": 3.4094303392361854e-06,569      "loss": 0.6689,570      "step": 800571    }572  ],573  "logging_steps": 10,574  "max_steps": 46875,575  "num_input_tokens_seen": 0,576  "num_train_epochs": 3,577  "save_steps": 200,578  "stateful_callbacks": {579    "EarlyStoppingCallback": {580      "args": {581        "early_stopping_patience": 3,582        "early_stopping_threshold": 0.0583      },584      "attributes": {585        "early_stopping_patience_counter": 0586      }587    },588    "TrainerControl": {589      "args": {590        "should_epoch_stop": false,591        "should_evaluate": false,592        "should_log": false,593        "should_save": true,594        "should_training_stop": false595      },596      "attributes": {}597    }598  },599  "total_flos": 6731943018000000.0,600  "train_batch_size": 32,601  "trial_name": null,602  "trial_params": null603}604