Team Ai
Modelpublic

DCAgent/a1-stack_csharp

sourceHugging Faceotherupdated 7mo agoView on Hugging Face
0likes16downloads
trainer_state.json10465 linesDownload Raw Back to root
1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 7.0,6  "eval_steps": 500,7  "global_step": 4739,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.007385524372230428,14      "grad_norm": 19.97872360716285,15      "learning_rate": 3.3755274261603377e-07,16      "loss": 0.624,17      "loss_nan_ranks": 0,18      "loss_rank_avg": 0.7236873507499695,19      "step": 5,20      "valid_targets_mean": 4191.2,21      "valid_targets_min": 56022    },23    {24      "epoch": 0.014771048744460856,25      "grad_norm": 16.297750328748872,26      "learning_rate": 7.59493670886076e-07,27      "loss": 0.6447,28      "loss_nan_ranks": 0,29      "loss_rank_avg": 0.6396403312683105,30      "step": 10,31      "valid_targets_mean": 3906.9,32      "valid_targets_min": 61733    },34    {35      "epoch": 0.022156573116691284,36      "grad_norm": 13.56762186389351,37      "learning_rate": 1.1814345991561182e-06,38      "loss": 0.6682,39      "loss_nan_ranks": 0,40      "loss_rank_avg": 0.5922929048538208,41      "step": 15,42      "valid_targets_mean": 5468.2,43      "valid_targets_min": 57344    },45    {46      "epoch": 0.029542097488921712,47      "grad_norm": 11.653889962740626,48      "learning_rate": 1.6033755274261605e-06,49      "loss": 0.5585,50      "loss_nan_ranks": 0,51      "loss_rank_avg": 0.5627623796463013,52      "step": 20,53      "valid_targets_mean": 4871.1,54      "valid_targets_min": 58155    },56    {57      "epoch": 0.03692762186115214,58      "grad_norm": 8.00562953617597,59      "learning_rate": 2.0253164556962026e-06,60      "loss": 0.5008,61      "loss_nan_ranks": 0,62      "loss_rank_avg": 0.5002298355102539,63      "step": 25,64      "valid_targets_mean": 4081.2,65      "valid_targets_min": 40066    },67    {68      "epoch": 0.04431314623338257,69      "grad_norm": 4.397412308851758,70      "learning_rate": 2.447257383966245e-06,71      "loss": 0.4932,72      "loss_nan_ranks": 0,73      "loss_rank_avg": 0.4381677806377411,74      "step": 30,75      "valid_targets_mean": 5501.4,76      "valid_targets_min": 63877    },78    {79      "epoch": 0.051698670605613,80      "grad_norm": 2.3819720550400656,81      "learning_rate": 2.8691983122362873e-06,82      "loss": 0.4548,83      "loss_nan_ranks": 0,84      "loss_rank_avg": 0.4095567464828491,85      "step": 35,86      "valid_targets_mean": 6018.8,87      "valid_targets_min": 47988    },89    {90      "epoch": 0.059084194977843424,91      "grad_norm": 1.5356961908394742,92      "learning_rate": 3.2911392405063294e-06,93      "loss": 0.4411,94      "loss_nan_ranks": 0,95      "loss_rank_avg": 0.3926240801811218,96      "step": 40,97      "valid_targets_mean": 4400.6,98      "valid_targets_min": 51799    },100    {101      "epoch": 0.06646971935007386,102      "grad_norm": 1.0870293491792131,103      "learning_rate": 3.713080168776372e-06,104      "loss": 0.4433,105      "loss_nan_ranks": 0,106      "loss_rank_avg": 0.417349636554718,107      "step": 45,108      "valid_targets_mean": 5836.1,109      "valid_targets_min": 762110    },111    {112      "epoch": 0.07385524372230429,113      "grad_norm": 0.943385047160246,114      "learning_rate": 4.135021097046414e-06,115      "loss": 0.4154,116      "loss_nan_ranks": 0,117      "loss_rank_avg": 0.41143539547920227,118      "step": 50,119      "valid_targets_mean": 5084.6,120      "valid_targets_min": 639121    },122    {123      "epoch": 0.08124076809453472,124      "grad_norm": 1.011998448398478,125      "learning_rate": 4.556962025316456e-06,126      "loss": 0.4151,127      "loss_nan_ranks": 0,128      "loss_rank_avg": 0.39654070138931274,129      "step": 55,130      "valid_targets_mean": 5247.8,131      "valid_targets_min": 681132    },133    {134      "epoch": 0.08862629246676514,135      "grad_norm": 0.686487046735958,136      "learning_rate": 4.978902953586498e-06,137      "loss": 0.3792,138      "loss_nan_ranks": 0,139      "loss_rank_avg": 0.33487510681152344,140      "step": 60,141      "valid_targets_mean": 5221.6,142      "valid_targets_min": 490143    },144    {145      "epoch": 0.09601181683899557,146      "grad_norm": 0.571553738713959,147      "learning_rate": 5.40084388185654e-06,148      "loss": 0.3743,149      "loss_nan_ranks": 0,150      "loss_rank_avg": 0.3575626313686371,151      "step": 65,152      "valid_targets_mean": 6041.3,153      "valid_targets_min": 618154    },155    {156      "epoch": 0.103397341211226,157      "grad_norm": 0.49365785429432335,158      "learning_rate": 5.8227848101265824e-06,159      "loss": 0.3463,160      "loss_nan_ranks": 0,161      "loss_rank_avg": 0.30460304021835327,162      "step": 70,163      "valid_targets_mean": 5298.4,164      "valid_targets_min": 535165    },166    {167      "epoch": 0.11078286558345643,168      "grad_norm": 0.5050364830057085,169      "learning_rate": 6.244725738396625e-06,170      "loss": 0.3399,171      "loss_nan_ranks": 0,172      "loss_rank_avg": 0.3081740140914917,173      "step": 75,174      "valid_targets_mean": 5255.7,175      "valid_targets_min": 652176    },177    {178      "epoch": 0.11816838995568685,179      "grad_norm": 0.5226990318037864,180      "learning_rate": 6.666666666666667e-06,181      "loss": 0.3904,182      "loss_nan_ranks": 0,183      "loss_rank_avg": 0.3145563006401062,184      "step": 80,185      "valid_targets_mean": 4663.9,186      "valid_targets_min": 392187    },188    {189      "epoch": 0.1255539143279173,190      "grad_norm": 0.4881100768231626,191      "learning_rate": 7.08860759493671e-06,192      "loss": 0.3726,193      "loss_nan_ranks": 0,194      "loss_rank_avg": 0.30497312545776367,195      "step": 85,196      "valid_targets_mean": 5632.1,197      "valid_targets_min": 719198    },199    {200      "epoch": 0.1329394387001477,201      "grad_norm": 0.5237726651638548,202      "learning_rate": 7.510548523206752e-06,203      "loss": 0.3544,204      "loss_nan_ranks": 0,205      "loss_rank_avg": 0.3278190791606903,206      "step": 90,207      "valid_targets_mean": 6220.0,208      "valid_targets_min": 751209    },210    {211      "epoch": 0.14032496307237813,212      "grad_norm": 0.4650280516019618,213      "learning_rate": 7.932489451476793e-06,214      "loss": 0.3182,215      "loss_nan_ranks": 0,216      "loss_rank_avg": 0.28953981399536133,217      "step": 95,218      "valid_targets_mean": 5031.3,219      "valid_targets_min": 642220    },221    {222      "epoch": 0.14771048744460857,223      "grad_norm": 0.46371951812182943,224      "learning_rate": 8.354430379746837e-06,225      "loss": 0.3099,226      "loss_nan_ranks": 0,227      "loss_rank_avg": 0.33909645676612854,228      "step": 100,229      "valid_targets_mean": 6196.9,230      "valid_targets_min": 545231    },232    {233      "epoch": 0.155096011816839,234      "grad_norm": 0.4568538569230339,235      "learning_rate": 8.776371308016879e-06,236      "loss": 0.3471,237      "loss_nan_ranks": 0,238      "loss_rank_avg": 0.2957189679145813,239      "step": 105,240      "valid_targets_mean": 6063.0,241      "valid_targets_min": 584242    },243    {244      "epoch": 0.16248153618906944,245      "grad_norm": 0.4578520750020651,246      "learning_rate": 9.198312236286921e-06,247      "loss": 0.3276,248      "loss_nan_ranks": 0,249      "loss_rank_avg": 0.29967236518859863,250      "step": 110,251      "valid_targets_mean": 6171.0,252      "valid_targets_min": 540253    },254    {255      "epoch": 0.16986706056129985,256      "grad_norm": 0.4955996123561873,257      "learning_rate": 9.620253164556963e-06,258      "loss": 0.3176,259      "loss_nan_ranks": 0,260      "loss_rank_avg": 0.2830328941345215,261      "step": 115,262      "valid_targets_mean": 4840.9,263      "valid_targets_min": 454264    },265    {266      "epoch": 0.17725258493353027,267      "grad_norm": 0.5471230185393541,268      "learning_rate": 1.0042194092827005e-05,269      "loss": 0.3219,270      "loss_nan_ranks": 0,271      "loss_rank_avg": 0.36016958951950073,272      "step": 120,273      "valid_targets_mean": 4376.1,274      "valid_targets_min": 605275    },276    {277      "epoch": 0.18463810930576072,278      "grad_norm": 0.5075473057243887,279      "learning_rate": 1.0464135021097049e-05,280      "loss": 0.3135,281      "loss_nan_ranks": 0,282      "loss_rank_avg": 0.314586877822876,283      "step": 125,284      "valid_targets_mean": 4974.8,285      "valid_targets_min": 615286    },287    {288      "epoch": 0.19202363367799113,289      "grad_norm": 0.4679315871943788,290      "learning_rate": 1.088607594936709e-05,291      "loss": 0.2943,292      "loss_nan_ranks": 0,293      "loss_rank_avg": 0.27506357431411743,294      "step": 130,295      "valid_targets_mean": 5003.4,296      "valid_targets_min": 309297    },298    {299      "epoch": 0.19940915805022155,300      "grad_norm": 0.5203002455514378,301      "learning_rate": 1.1308016877637132e-05,302      "loss": 0.3111,303      "loss_nan_ranks": 0,304      "loss_rank_avg": 0.3530064821243286,305      "step": 135,306      "valid_targets_mean": 4744.3,307      "valid_targets_min": 380308    },309    {310      "epoch": 0.206794682422452,311      "grad_norm": 0.508532245402902,312      "learning_rate": 1.1729957805907175e-05,313      "loss": 0.2799,314      "loss_nan_ranks": 0,315      "loss_rank_avg": 0.26319918036460876,316      "step": 140,317      "valid_targets_mean": 4969.9,318      "valid_targets_min": 298319    },320    {321      "epoch": 0.21418020679468242,322      "grad_norm": 0.4977333684095323,323      "learning_rate": 1.2151898734177216e-05,324      "loss": 0.2864,325      "loss_nan_ranks": 0,326      "loss_rank_avg": 0.281037300825119,327      "step": 145,328      "valid_targets_mean": 4680.9,329      "valid_targets_min": 553330    },331    {332      "epoch": 0.22156573116691286,333      "grad_norm": 0.5700488921766971,334      "learning_rate": 1.2573839662447258e-05,335      "loss": 0.3138,336      "loss_nan_ranks": 0,337      "loss_rank_avg": 0.3317258059978485,338      "step": 150,339      "valid_targets_mean": 4780.8,340      "valid_targets_min": 613341    },342    {343      "epoch": 0.22895125553914328,344      "grad_norm": 0.5049210471064856,345      "learning_rate": 1.29957805907173e-05,346      "loss": 0.2939,347      "loss_nan_ranks": 0,348      "loss_rank_avg": 0.31507956981658936,349      "step": 155,350      "valid_targets_mean": 4951.3,351      "valid_targets_min": 648352    },353    {354      "epoch": 0.2363367799113737,355      "grad_norm": 0.48733685753987915,356      "learning_rate": 1.3417721518987344e-05,357      "loss": 0.3144,358      "loss_nan_ranks": 0,359      "loss_rank_avg": 0.3532077670097351,360      "step": 160,361      "valid_targets_mean": 5704.9,362      "valid_targets_min": 615363    },364    {365      "epoch": 0.24372230428360414,366      "grad_norm": 0.478477178779921,367      "learning_rate": 1.3839662447257384e-05,368      "loss": 0.2889,369      "loss_nan_ranks": 0,370      "loss_rank_avg": 0.25617825984954834,371      "step": 165,372      "valid_targets_mean": 5201.0,373      "valid_targets_min": 614374    },375    {376      "epoch": 0.2511078286558346,377      "grad_norm": 0.6185796972171613,378      "learning_rate": 1.4261603375527426e-05,379      "loss": 0.2979,380      "loss_nan_ranks": 0,381      "loss_rank_avg": 0.3018920421600342,382      "step": 170,383      "valid_targets_mean": 4082.6,384      "valid_targets_min": 595385    },386    {387      "epoch": 0.258493353028065,388      "grad_norm": 0.5575122691401694,389      "learning_rate": 1.468354430379747e-05,390      "loss": 0.2849,391      "loss_nan_ranks": 0,392      "loss_rank_avg": 0.2911016345024109,393      "step": 175,394      "valid_targets_mean": 4299.9,395      "valid_targets_min": 662396    },397    {398      "epoch": 0.2658788774002954,399      "grad_norm": 0.6276828580042122,400      "learning_rate": 1.5105485232067512e-05,401      "loss": 0.2649,402      "loss_nan_ranks": 0,403      "loss_rank_avg": 0.25065547227859497,404      "step": 180,405      "valid_targets_mean": 4217.5,406      "valid_targets_min": 388407    },408    {409      "epoch": 0.27326440177252587,410      "grad_norm": 0.4696050024959157,411      "learning_rate": 1.5527426160337554e-05,412      "loss": 0.2938,413      "loss_nan_ranks": 0,414      "loss_rank_avg": 0.2493332028388977,415      "step": 185,416      "valid_targets_mean": 5711.5,417      "valid_targets_min": 679418    },419    {420      "epoch": 0.28064992614475626,421      "grad_norm": 0.5623419824234117,422      "learning_rate": 1.5949367088607598e-05,423      "loss": 0.2589,424      "loss_nan_ranks": 0,425      "loss_rank_avg": 0.27796900272369385,426      "step": 190,427      "valid_targets_mean": 4589.1,428      "valid_targets_min": 592429    },430    {431      "epoch": 0.2880354505169867,432      "grad_norm": 0.45269545432048147,433      "learning_rate": 1.637130801687764e-05,434      "loss": 0.243,435      "loss_nan_ranks": 0,436      "loss_rank_avg": 0.23348630964756012,437      "step": 195,438      "valid_targets_mean": 6195.4,439      "valid_targets_min": 1406440    },441    {442      "epoch": 0.29542097488921715,443      "grad_norm": 0.5800039153836231,444      "learning_rate": 1.679324894514768e-05,445      "loss": 0.2751,446      "loss_nan_ranks": 0,447      "loss_rank_avg": 0.3133583664894104,448      "step": 200,449      "valid_targets_mean": 4379.1,450      "valid_targets_min": 577451    },452    {453      "epoch": 0.30280649926144754,454      "grad_norm": 0.4475247352688734,455      "learning_rate": 1.7215189873417723e-05,456      "loss": 0.2688,457      "loss_nan_ranks": 0,458      "loss_rank_avg": 0.2346859574317932,459      "step": 205,460      "valid_targets_mean": 5674.9,461      "valid_targets_min": 516462    },463    {464      "epoch": 0.310192023633678,465      "grad_norm": 0.6043039588064852,466      "learning_rate": 1.7637130801687767e-05,467      "loss": 0.284,468      "loss_nan_ranks": 0,469      "loss_rank_avg": 0.2985689640045166,470      "step": 210,471      "valid_targets_mean": 3230.7,472      "valid_targets_min": 451473    },474    {475      "epoch": 0.3175775480059084,476      "grad_norm": 0.46608512023497484,477      "learning_rate": 1.8059071729957807e-05,478      "loss": 0.2621,479      "loss_nan_ranks": 0,480      "loss_rank_avg": 0.2667817175388336,481      "step": 215,482      "valid_targets_mean": 5615.9,483      "valid_targets_min": 608484    },485    {486      "epoch": 0.3249630723781389,487      "grad_norm": 0.6393799102975143,488      "learning_rate": 1.848101265822785e-05,489      "loss": 0.2857,490      "loss_nan_ranks": 0,491      "loss_rank_avg": 0.3485134243965149,492      "step": 220,493      "valid_targets_mean": 5232.9,494      "valid_targets_min": 551495    },496    {497      "epoch": 0.33234859675036926,498      "grad_norm": 0.4985956967264833,499      "learning_rate": 1.890295358649789e-05,500      "loss": 0.2504,501      "loss_nan_ranks": 0,502      "loss_rank_avg": 0.22727638483047485,503      "step": 225,504      "valid_targets_mean": 5186.1,505      "valid_targets_min": 557506    },507    {508      "epoch": 0.3397341211225997,509      "grad_norm": 0.698867408214604,510      "learning_rate": 1.9324894514767935e-05,511      "loss": 0.277,512      "loss_nan_ranks": 0,513      "loss_rank_avg": 0.28445443511009216,514      "step": 230,515      "valid_targets_mean": 4395.3,516      "valid_targets_min": 613517    },518    {519      "epoch": 0.34711964549483015,520      "grad_norm": 0.4699161647524185,521      "learning_rate": 1.974683544303798e-05,522      "loss": 0.2863,523      "loss_nan_ranks": 0,524      "loss_rank_avg": 0.27419084310531616,525      "step": 235,526      "valid_targets_mean": 6050.6,527      "valid_targets_min": 697528    },529    {530      "epoch": 0.35450516986706054,531      "grad_norm": 0.49712308816298567,532      "learning_rate": 2.0168776371308016e-05,533      "loss": 0.2654,534      "loss_nan_ranks": 0,535      "loss_rank_avg": 0.2411026805639267,536      "step": 240,537      "valid_targets_mean": 6215.0,538      "valid_targets_min": 375539    },540    {541      "epoch": 0.361890694239291,542      "grad_norm": 0.4755998361435196,543      "learning_rate": 2.0590717299578063e-05,544      "loss": 0.2812,545      "loss_nan_ranks": 0,546      "loss_rank_avg": 0.31807953119277954,547      "step": 245,548      "valid_targets_mean": 5977.3,549      "valid_targets_min": 222550    },551    {552      "epoch": 0.36927621861152143,553      "grad_norm": 0.5522737703807714,554      "learning_rate": 2.1012658227848103e-05,555      "loss": 0.2728,556      "loss_nan_ranks": 0,557      "loss_rank_avg": 0.2637110650539398,558      "step": 250,559      "valid_targets_mean": 4707.1,560      "valid_targets_min": 424561    },562    {563      "epoch": 0.3766617429837518,564      "grad_norm": 0.5752807812899275,565      "learning_rate": 2.1434599156118144e-05,566      "loss": 0.2675,567      "loss_nan_ranks": 0,568      "loss_rank_avg": 0.252408891916275,569      "step": 255,570      "valid_targets_mean": 4646.9,571      "valid_targets_min": 620572    },573    {574      "epoch": 0.38404726735598227,575      "grad_norm": 0.4839304938690229,576      "learning_rate": 2.1856540084388188e-05,577      "loss": 0.2899,578      "loss_nan_ranks": 0,579      "loss_rank_avg": 0.311644971370697,580      "step": 260,581      "valid_targets_mean": 6014.4,582      "valid_targets_min": 320583    },584    {585      "epoch": 0.3914327917282127,586      "grad_norm": 0.3679428184390633,587      "learning_rate": 2.2278481012658228e-05,588      "loss": 0.2603,589      "loss_nan_ranks": 0,590      "loss_rank_avg": 0.1715935468673706,591      "step": 265,592      "valid_targets_mean": 7149.1,593      "valid_targets_min": 708594    },595    {596      "epoch": 0.3988183161004431,597      "grad_norm": 0.4388882366281187,598      "learning_rate": 2.270042194092827e-05,599      "loss": 0.2662,600      "loss_nan_ranks": 0,601      "loss_rank_avg": 0.21887677907943726,602      "step": 270,603      "valid_targets_mean": 5259.7,604      "valid_targets_min": 462605    },606    {607      "epoch": 0.40620384047267355,608      "grad_norm": 0.45156140818813745,609      "learning_rate": 2.3122362869198316e-05,610      "loss": 0.243,611      "loss_nan_ranks": 0,612      "loss_rank_avg": 0.1979508399963379,613      "step": 275,614      "valid_targets_mean": 5245.2,615      "valid_targets_min": 370616    },617    {618      "epoch": 0.413589364844904,619      "grad_norm": 0.5635183886045622,620      "learning_rate": 2.3544303797468356e-05,621      "loss": 0.2644,622      "loss_nan_ranks": 0,623      "loss_rank_avg": 0.2971973419189453,624      "step": 280,625      "valid_targets_mean": 4294.6,626      "valid_targets_min": 250627    },628    {629      "epoch": 0.42097488921713444,630      "grad_norm": 0.6003559911338966,631      "learning_rate": 2.39662447257384e-05,632      "loss": 0.2701,633      "loss_nan_ranks": 0,634      "loss_rank_avg": 0.27018484473228455,635      "step": 285,636      "valid_targets_mean": 4037.1,637      "valid_targets_min": 530638    },639    {640      "epoch": 0.42836041358936483,641      "grad_norm": 0.519146451009721,642      "learning_rate": 2.438818565400844e-05,643      "loss": 0.2749,644      "loss_nan_ranks": 0,645      "loss_rank_avg": 0.25247296690940857,646      "step": 290,647      "valid_targets_mean": 5198.4,648      "valid_targets_min": 757649    },650    {651      "epoch": 0.4357459379615953,652      "grad_norm": 0.5295365016140835,653      "learning_rate": 2.481012658227848e-05,654      "loss": 0.2734,655      "loss_nan_ranks": 0,656      "loss_rank_avg": 0.30645301938056946,657      "step": 295,658      "valid_targets_mean": 5075.9,659      "valid_targets_min": 586660    },661    {662      "epoch": 0.4431314623338257,663      "grad_norm": 0.5064359925137679,664      "learning_rate": 2.5232067510548524e-05,665      "loss": 0.2615,666      "loss_nan_ranks": 0,667      "loss_rank_avg": 0.27146783471107483,668      "step": 300,669      "valid_targets_mean": 5670.0,670      "valid_targets_min": 510671    },672    {673      "epoch": 0.4505169867060561,674      "grad_norm": 0.4807884411055291,675      "learning_rate": 2.5654008438818568e-05,676      "loss": 0.2637,677      "loss_nan_ranks": 0,678      "loss_rank_avg": 0.245832160115242,679      "step": 305,680      "valid_targets_mean": 4988.6,681      "valid_targets_min": 758682    },683    {684      "epoch": 0.45790251107828656,685      "grad_norm": 0.4874621524760727,686      "learning_rate": 2.6075949367088612e-05,687      "loss": 0.2355,688      "loss_nan_ranks": 0,689      "loss_rank_avg": 0.22424304485321045,690      "step": 310,691      "valid_targets_mean": 4865.2,692      "valid_targets_min": 690693    },694    {695      "epoch": 0.465288035450517,696      "grad_norm": 0.5241781469793773,697      "learning_rate": 2.6497890295358652e-05,698      "loss": 0.2674,699      "loss_nan_ranks": 0,700      "loss_rank_avg": 0.2541046142578125,701      "step": 315,702      "valid_targets_mean": 4655.8,703      "valid_targets_min": 303704    },705    {706      "epoch": 0.4726735598227474,707      "grad_norm": 0.6693714356739207,708      "learning_rate": 2.6919831223628693e-05,709      "loss": 0.2598,710      "loss_nan_ranks": 0,711      "loss_rank_avg": 0.3150583803653717,712      "step": 320,713      "valid_targets_mean": 3789.6,714      "valid_targets_min": 323715    },716    {717      "epoch": 0.48005908419497784,718      "grad_norm": 0.4049583166705286,719      "learning_rate": 2.7341772151898737e-05,720      "loss": 0.2634,721      "loss_nan_ranks": 0,722      "loss_rank_avg": 0.19988644123077393,723      "step": 325,724      "valid_targets_mean": 6250.1,725      "valid_targets_min": 565726    },727    {728      "epoch": 0.4874446085672083,729      "grad_norm": 0.5683986043691779,730      "learning_rate": 2.7763713080168777e-05,731      "loss": 0.2955,732      "loss_nan_ranks": 0,733      "loss_rank_avg": 0.30885380506515503,734      "step": 330,735      "valid_targets_mean": 4539.6,736      "valid_targets_min": 511737    },738    {739      "epoch": 0.4948301329394387,740      "grad_norm": 0.4293798580209713,741      "learning_rate": 2.8185654008438824e-05,742      "loss": 0.2476,743      "loss_nan_ranks": 0,744      "loss_rank_avg": 0.2373587191104889,745      "step": 335,746      "valid_targets_mean": 6893.3,747      "valid_targets_min": 540748    },749    {750      "epoch": 0.5022156573116692,751      "grad_norm": 0.5051385898465901,752      "learning_rate": 2.8607594936708865e-05,753      "loss": 0.2575,754      "loss_nan_ranks": 0,755      "loss_rank_avg": 0.2349412590265274,756      "step": 340,757      "valid_targets_mean": 5513.8,758      "valid_targets_min": 348759    },760    {761      "epoch": 0.5096011816838996,762      "grad_norm": 0.453650649705282,763      "learning_rate": 2.9029535864978905e-05,764      "loss": 0.2487,765      "loss_nan_ranks": 0,766      "loss_rank_avg": 0.24615350365638733,767      "step": 345,768      "valid_targets_mean": 5636.8,769      "valid_targets_min": 632770    },771    {772      "epoch": 0.51698670605613,773      "grad_norm": 0.5026094522177535,774      "learning_rate": 2.945147679324895e-05,775      "loss": 0.2865,776      "loss_nan_ranks": 0,777      "loss_rank_avg": 0.25473105907440186,778      "step": 350,779      "valid_targets_mean": 4648.2,780      "valid_targets_min": 618781    },782    {783      "epoch": 0.5243722304283605,784      "grad_norm": 0.5473929641026576,785      "learning_rate": 2.987341772151899e-05,786      "loss": 0.2454,787      "loss_nan_ranks": 0,788      "loss_rank_avg": 0.25512436032295227,789      "step": 355,790      "valid_targets_mean": 4888.8,791      "valid_targets_min": 455792    },793    {794      "epoch": 0.5317577548005908,795      "grad_norm": 0.6818407765763758,796      "learning_rate": 3.029535864978903e-05,797      "loss": 0.251,798      "loss_nan_ranks": 0,799      "loss_rank_avg": 0.26583975553512573,800      "step": 360,801      "valid_targets_mean": 3901.6,802      "valid_targets_min": 393803    },804    {805      "epoch": 0.5391432791728212,806      "grad_norm": 0.43264528877861325,807      "learning_rate": 3.0717299578059074e-05,808      "loss": 0.248,809      "loss_nan_ranks": 0,810      "loss_rank_avg": 0.2336602509021759,811      "step": 365,812      "valid_targets_mean": 6759.1,813      "valid_targets_min": 654814    },815    {816      "epoch": 0.5465288035450517,817      "grad_norm": 0.4786744793085355,818      "learning_rate": 3.113924050632912e-05,819      "loss": 0.2568,820      "loss_nan_ranks": 0,821      "loss_rank_avg": 0.23913979530334473,822      "step": 370,823      "valid_targets_mean": 5128.6,824      "valid_targets_min": 436825    },826    {827      "epoch": 0.5539143279172821,828      "grad_norm": 0.6098930815067188,829      "learning_rate": 3.156118143459916e-05,830      "loss": 0.2664,831      "loss_nan_ranks": 0,832      "loss_rank_avg": 0.27685385942459106,833      "step": 375,834      "valid_targets_mean": 4722.4,835      "valid_targets_min": 547836    },837    {838      "epoch": 0.5612998522895125,839      "grad_norm": 0.4884326875060275,840      "learning_rate": 3.19831223628692e-05,841      "loss": 0.2529,842      "loss_nan_ranks": 0,843      "loss_rank_avg": 0.269244909286499,844      "step": 380,845      "valid_targets_mean": 5746.3,846      "valid_targets_min": 649847    },848    {849      "epoch": 0.568685376661743,850      "grad_norm": 0.6966159287505779,851      "learning_rate": 3.240506329113924e-05,852      "loss": 0.2452,853      "loss_nan_ranks": 0,854      "loss_rank_avg": 0.25672098994255066,855      "step": 385,856      "valid_targets_mean": 5211.0,857      "valid_targets_min": 451858    },859    {860      "epoch": 0.5760709010339734,861      "grad_norm": 0.4826433803477916,862      "learning_rate": 3.282700421940928e-05,863      "loss": 0.2588,864      "loss_nan_ranks": 0,865      "loss_rank_avg": 0.21904008090496063,866      "step": 390,867      "valid_targets_mean": 4840.0,868      "valid_targets_min": 614869    },870    {871      "epoch": 0.5834564254062038,872      "grad_norm": 0.5161328067549831,873      "learning_rate": 3.324894514767932e-05,874      "loss": 0.2401,875      "loss_nan_ranks": 0,876      "loss_rank_avg": 0.21546295285224915,877      "step": 395,878      "valid_targets_mean": 5000.8,879      "valid_targets_min": 474880    },881    {882      "epoch": 0.5908419497784343,883      "grad_norm": 0.441903839632028,884      "learning_rate": 3.367088607594937e-05,885      "loss": 0.2326,886      "loss_nan_ranks": 0,887      "loss_rank_avg": 0.2418462634086609,888      "step": 400,889      "valid_targets_mean": 6752.2,890      "valid_targets_min": 526891    },892    {893      "epoch": 0.5982274741506647,894      "grad_norm": 0.4263460620059315,895      "learning_rate": 3.409282700421941e-05,896      "loss": 0.2489,897      "loss_nan_ranks": 0,898      "loss_rank_avg": 0.2564152479171753,899      "step": 405,900      "valid_targets_mean": 5806.5,901      "valid_targets_min": 581902    },903    {904      "epoch": 0.6056129985228951,905      "grad_norm": 0.4114391570850889,906      "learning_rate": 3.451476793248946e-05,907      "loss": 0.2561,908      "loss_nan_ranks": 0,909      "loss_rank_avg": 0.19688798487186432,910      "step": 410,911      "valid_targets_mean": 5193.4,912      "valid_targets_min": 765913    },914    {915      "epoch": 0.6129985228951256,916      "grad_norm": 0.560472208109208,917      "learning_rate": 3.49367088607595e-05,918      "loss": 0.3089,919      "loss_nan_ranks": 0,920      "loss_rank_avg": 0.3748968243598938,921      "step": 415,922      "valid_targets_mean": 6106.8,923      "valid_targets_min": 673924    },925    {926      "epoch": 0.620384047267356,927      "grad_norm": 0.4389773616307953,928      "learning_rate": 3.535864978902954e-05,929      "loss": 0.2474,930      "loss_nan_ranks": 0,931      "loss_rank_avg": 0.25939327478408813,932      "step": 420,933      "valid_targets_mean": 5600.4,934      "valid_targets_min": 587935    },936    {937      "epoch": 0.6277695716395865,938      "grad_norm": 0.49005945851489136,939      "learning_rate": 3.578059071729958e-05,940      "loss": 0.2352,941      "loss_nan_ranks": 0,942      "loss_rank_avg": 0.2415715903043747,943      "step": 425,944      "valid_targets_mean": 5480.6,945      "valid_targets_min": 557946    },947    {948      "epoch": 0.6351550960118169,949      "grad_norm": 0.6120232650144795,950      "learning_rate": 3.6202531645569626e-05,951      "loss": 0.2384,952      "loss_nan_ranks": 0,953      "loss_rank_avg": 0.27258408069610596,954      "step": 430,955      "valid_targets_mean": 4020.8,956      "valid_targets_min": 575957    },958    {959      "epoch": 0.6425406203840472,960      "grad_norm": 0.5517761299010163,961      "learning_rate": 3.6624472573839666e-05,962      "loss": 0.248,963      "loss_nan_ranks": 0,964      "loss_rank_avg": 0.2571716904640198,965      "step": 435,966      "valid_targets_mean": 3986.9,967      "valid_targets_min": 570968    },969    {970      "epoch": 0.6499261447562777,971      "grad_norm": 0.4519259500134533,972      "learning_rate": 3.704641350210971e-05,973      "loss": 0.2584,974      "loss_nan_ranks": 0,975      "loss_rank_avg": 0.23544706404209137,976      "step": 440,977      "valid_targets_mean": 5446.1,978      "valid_targets_min": 798979    },980    {981      "epoch": 0.6573116691285081,982      "grad_norm": 0.4627225505498187,983      "learning_rate": 3.746835443037975e-05,984      "loss": 0.2572,985      "loss_nan_ranks": 0,986      "loss_rank_avg": 0.22768300771713257,987      "step": 445,988      "valid_targets_mean": 5305.9,989      "valid_targets_min": 612990    },991    {992      "epoch": 0.6646971935007385,993      "grad_norm": 0.5459862764397345,994      "learning_rate": 3.7890295358649794e-05,995      "loss": 0.2509,996      "loss_nan_ranks": 0,997      "loss_rank_avg": 0.2680595815181732,998      "step": 450,999      "valid_targets_mean": 3753.4,1000      "valid_targets_min": 6371001    },1002    {1003      "epoch": 0.672082717872969,1004      "grad_norm": 0.5037347511570361,1005      "learning_rate": 3.8312236286919835e-05,1006      "loss": 0.2492,1007      "loss_nan_ranks": 0,1008      "loss_rank_avg": 0.2467479407787323,1009      "step": 455,1010      "valid_targets_mean": 5501.6,1011      "valid_targets_min": 3101012    },1013    {1014      "epoch": 0.6794682422451994,1015      "grad_norm": 0.4327775289602796,1016      "learning_rate": 3.873417721518988e-05,1017      "loss": 0.2367,1018      "loss_nan_ranks": 0,1019      "loss_rank_avg": 0.19664692878723145,1020      "step": 460,1021      "valid_targets_mean": 5931.2,1022      "valid_targets_min": 9131023    },1024    {1025      "epoch": 0.6868537666174298,1026      "grad_norm": 0.4409637060996679,1027      "learning_rate": 3.915611814345992e-05,1028      "loss": 0.2239,1029      "loss_nan_ranks": 0,1030      "loss_rank_avg": 0.21506735682487488,1031      "step": 465,1032      "valid_targets_mean": 6059.1,1033      "valid_targets_min": 6471034    },1035    {1036      "epoch": 0.6942392909896603,1037      "grad_norm": 0.5504576504647792,1038      "learning_rate": 3.957805907172996e-05,1039      "loss": 0.2497,1040      "loss_nan_ranks": 0,1041      "loss_rank_avg": 0.2300184667110443,1042      "step": 470,1043      "valid_targets_mean": 3839.0,1044      "valid_targets_min": 5571045    },1046    {1047      "epoch": 0.7016248153618907,1048      "grad_norm": 0.429274562141855,1049      "learning_rate": 4e-05,1050      "loss": 0.2322,1051      "loss_nan_ranks": 0,1052      "loss_rank_avg": 0.2237507700920105,1053      "step": 475,1054      "valid_targets_mean": 6357.5,1055      "valid_targets_min": 7421056    },1057    {1058      "epoch": 0.7090103397341211,1059      "grad_norm": 0.7536006693457217,1060      "learning_rate": 3.9999864355811366e-05,1061      "loss": 0.2377,1062      "loss_nan_ranks": 0,1063      "loss_rank_avg": 0.22905516624450684,1064      "step": 480,1065      "valid_targets_mean": 4381.5,1066      "valid_targets_min": 7291067    },1068    {1069      "epoch": 0.7163958641063516,1070      "grad_norm": 0.43953758941215776,1071      "learning_rate": 3.99994574250854e-05,1072      "loss": 0.2316,1073      "loss_nan_ranks": 0,1074      "loss_rank_avg": 0.20200951397418976,1075      "step": 485,1076      "valid_targets_mean": 5644.1,1077      "valid_targets_min": 6831078    },1079    {1080      "epoch": 0.723781388478582,1081      "grad_norm": 0.4723664561759756,1082      "learning_rate": 3.999877921334187e-05,1083      "loss": 0.2475,1084      "loss_nan_ranks": 0,1085      "loss_rank_avg": 0.26116621494293213,1086      "step": 490,1087      "valid_targets_mean": 5649.9,1088      "valid_targets_min": 5981089    },1090    {1091      "epoch": 0.7311669128508124,1092      "grad_norm": 0.5372942073144541,1093      "learning_rate": 3.999782972978035e-05,1094      "loss": 0.2515,1095      "loss_nan_ranks": 0,1096      "loss_rank_avg": 0.2597278356552124,1097      "step": 495,1098      "valid_targets_mean": 4376.8,1099      "valid_targets_min": 7111100    },1101    {1102      "epoch": 0.7385524372230429,1103      "grad_norm": 0.4923885109877066,1104      "learning_rate": 3.9996608987280005e-05,1105      "loss": 0.2537,1106      "loss_nan_ranks": 0,1107      "loss_rank_avg": 0.3310204744338989,1108      "step": 500,1109      "valid_targets_mean": 5373.9,1110      "valid_targets_min": 3621111    },1112    {1113      "epoch": 0.7459379615952733,1114      "grad_norm": 0.4360411302968856,1115      "learning_rate": 3.999511700239951e-05,1116      "loss": 0.2356,1117      "loss_nan_ranks": 0,1118      "loss_rank_avg": 0.23595163226127625,1119      "step": 505,1120      "valid_targets_mean": 5867.7,1121      "valid_targets_min": 7471122    },1123    {1124      "epoch": 0.7533234859675036,1125      "grad_norm": 0.4125019822494474,1126      "learning_rate": 3.999335379537677e-05,1127      "loss": 0.2542,1128      "loss_nan_ranks": 0,1129      "loss_rank_avg": 0.22982177138328552,1130      "step": 510,1131      "valid_targets_mean": 5860.2,1132      "valid_targets_min": 2841133    },1134    {1135      "epoch": 0.7607090103397341,1136      "grad_norm": 0.4816111850196653,1137      "learning_rate": 3.999131939012866e-05,1138      "loss": 0.2421,1139      "loss_nan_ranks": 0,1140      "loss_rank_avg": 0.23170827329158783,1141      "step": 515,1142      "valid_targets_mean": 5939.9,1143      "valid_targets_min": 5731144    },1145    {1146      "epoch": 0.7680945347119645,1147      "grad_norm": 0.5607421617686298,1148      "learning_rate": 3.998901381425072e-05,1149      "loss": 0.2555,1150      "loss_nan_ranks": 0,1151      "loss_rank_avg": 0.27282047271728516,1152      "step": 520,1153      "valid_targets_mean": 4010.6,1154      "valid_targets_min": 3081155    },1156    {1157      "epoch": 0.7754800590841949,1158      "grad_norm": 0.4899434940882494,1159      "learning_rate": 3.998643709901673e-05,1160      "loss": 0.2255,1161      "loss_nan_ranks": 0,1162      "loss_rank_avg": 0.2315795123577118,1163      "step": 525,1164      "valid_targets_mean": 4446.6,1165      "valid_targets_min": 7551166    },1167    {1168      "epoch": 0.7828655834564254,1169      "grad_norm": 0.4551383882144779,1170      "learning_rate": 3.998358927937835e-05,1171      "loss": 0.2489,1172      "loss_nan_ranks": 0,1173      "loss_rank_avg": 0.21953409910202026,1174      "step": 530,1175      "valid_targets_mean": 5531.2,1176      "valid_targets_min": 5381177    },1178    {1179      "epoch": 0.7902511078286558,1180      "grad_norm": 0.4743224847774692,1181      "learning_rate": 3.9980470393964584e-05,1182      "loss": 0.2604,1183      "loss_nan_ranks": 0,1184      "loss_rank_avg": 0.23994377255439758,1185      "step": 535,1186      "valid_targets_mean": 4992.5,1187      "valid_targets_min": 4991188    },1189    {1190      "epoch": 0.7976366322008862,1191      "grad_norm": 0.42125006599133685,1192      "learning_rate": 3.997708048508131e-05,1193      "loss": 0.2232,1194      "loss_nan_ranks": 0,1195      "loss_rank_avg": 0.19587251543998718,1196      "step": 540,1197      "valid_targets_mean": 5289.6,1198      "valid_targets_min": 6671199    },1200    {

Showing the first 1,200 of 10465 lines. Download the file for the rest.