Team Ai
Modelpublic

codeaidbackUp/CouplingSmells-Detection-Adpater

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes11downloads
trainer_state.json1384 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": 100,7  "global_step": 1728,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.017361111111111112,14      "grad_norm": 2.007795810699463,15      "learning_rate": 5.202312138728324e-06,16      "loss": 1.463,17      "step": 1018    },19    {20      "epoch": 0.034722222222222224,21      "grad_norm": 1.1297450065612793,22      "learning_rate": 1.0982658959537573e-05,23      "loss": 1.2884,24      "step": 2025    },26    {27      "epoch": 0.052083333333333336,28      "grad_norm": 0.774868369102478,29      "learning_rate": 1.676300578034682e-05,30      "loss": 1.0796,31      "step": 3032    },33    {34      "epoch": 0.06944444444444445,35      "grad_norm": 0.5096573233604431,36      "learning_rate": 2.254335260115607e-05,37      "loss": 0.8443,38      "step": 4039    },40    {41      "epoch": 0.08680555555555555,42      "grad_norm": 0.6606957316398621,43      "learning_rate": 2.832369942196532e-05,44      "loss": 0.756,45      "step": 5046    },47    {48      "epoch": 0.10416666666666667,49      "grad_norm": 0.8858688473701477,50      "learning_rate": 3.410404624277457e-05,51      "loss": 0.6759,52      "step": 6053    },54    {55      "epoch": 0.12152777777777778,56      "grad_norm": 1.686498999595642,57      "learning_rate": 3.988439306358382e-05,58      "loss": 0.636,59      "step": 7060    },61    {62      "epoch": 0.1388888888888889,63      "grad_norm": 0.681108295917511,64      "learning_rate": 4.566473988439307e-05,65      "loss": 0.5915,66      "step": 8067    },68    {69      "epoch": 0.15625,70      "grad_norm": 0.657633900642395,71      "learning_rate": 5.1445086705202317e-05,72      "loss": 0.6144,73      "step": 9074    },75    {76      "epoch": 0.1736111111111111,77      "grad_norm": 0.9662413001060486,78      "learning_rate": 5.722543352601156e-05,79      "loss": 0.656,80      "step": 10081    },82    {83      "epoch": 0.1736111111111111,84      "eval_loss": 0.5766947269439697,85      "eval_runtime": 513.1213,86      "eval_samples_per_second": 2.247,87      "eval_steps_per_second": 2.247,88      "step": 10089    },90    {91      "epoch": 0.1909722222222222,92      "grad_norm": 0.6163507699966431,93      "learning_rate": 6.300578034682081e-05,94      "loss": 0.5819,95      "step": 11096    },97    {98      "epoch": 0.20833333333333334,99      "grad_norm": 0.7107399106025696,100      "learning_rate": 6.878612716763007e-05,101      "loss": 0.5718,102      "step": 120103    },104    {105      "epoch": 0.22569444444444445,106      "grad_norm": 0.6259875893592834,107      "learning_rate": 7.456647398843931e-05,108      "loss": 0.658,109      "step": 130110    },111    {112      "epoch": 0.24305555555555555,113      "grad_norm": 0.5605979561805725,114      "learning_rate": 8.034682080924855e-05,115      "loss": 0.549,116      "step": 140117    },118    {119      "epoch": 0.2604166666666667,120      "grad_norm": 0.5297986268997192,121      "learning_rate": 8.612716763005781e-05,122      "loss": 0.5671,123      "step": 150124    },125    {126      "epoch": 0.2777777777777778,127      "grad_norm": 0.5555981397628784,128      "learning_rate": 9.190751445086706e-05,129      "loss": 0.5569,130      "step": 160131    },132    {133      "epoch": 0.2951388888888889,134      "grad_norm": 0.5598844885826111,135      "learning_rate": 9.768786127167631e-05,136      "loss": 0.5862,137      "step": 170138    },139    {140      "epoch": 0.3125,141      "grad_norm": 0.5559885501861572,142      "learning_rate": 9.999632653293776e-05,143      "loss": 0.5738,144      "step": 180145    },146    {147      "epoch": 0.3298611111111111,148      "grad_norm": 0.4869542717933655,149      "learning_rate": 9.99738795222662e-05,150      "loss": 0.5789,151      "step": 190152    },153    {154      "epoch": 0.3472222222222222,155      "grad_norm": 0.5302107334136963,156      "learning_rate": 9.993103546669093e-05,157      "loss": 0.5695,158      "step": 200159    },160    {161      "epoch": 0.3472222222222222,162      "eval_loss": 0.5406240224838257,163      "eval_runtime": 515.9751,164      "eval_samples_per_second": 2.235,165      "eval_steps_per_second": 2.235,166      "step": 200167    },168    {169      "epoch": 0.3645833333333333,170      "grad_norm": 0.4969506561756134,171      "learning_rate": 9.98678118531898e-05,172      "loss": 0.5943,173      "step": 210174    },175    {176      "epoch": 0.3819444444444444,177      "grad_norm": 0.5448970794677734,178      "learning_rate": 9.97842344867413e-05,179      "loss": 0.5719,180      "step": 220181    },182    {183      "epoch": 0.3993055555555556,184      "grad_norm": 0.5558710098266602,185      "learning_rate": 9.968033747979195e-05,186      "loss": 0.5303,187      "step": 230188    },189    {190      "epoch": 0.4166666666666667,191      "grad_norm": 0.5268853902816772,192      "learning_rate": 9.955616323833343e-05,193      "loss": 0.5646,194      "step": 240195    },196    {197      "epoch": 0.4340277777777778,198      "grad_norm": 0.5985867381095886,199      "learning_rate": 9.941176244459418e-05,200      "loss": 0.5558,201      "step": 250202    },203    {204      "epoch": 0.4513888888888889,205      "grad_norm": 0.37411603331565857,206      "learning_rate": 9.924719403635333e-05,207      "loss": 0.5372,208      "step": 260209    },210    {211      "epoch": 0.46875,212      "grad_norm": 0.4203139841556549,213      "learning_rate": 9.906252518288501e-05,214      "loss": 0.5492,215      "step": 270216    },217    {218      "epoch": 0.4861111111111111,219      "grad_norm": 0.5267091393470764,220      "learning_rate": 9.885783125754291e-05,221      "loss": 0.5233,222      "step": 280223    },224    {225      "epoch": 0.5034722222222222,226      "grad_norm": 0.5201379656791687,227      "learning_rate": 9.863319580699635e-05,228      "loss": 0.5909,229      "step": 290230    },231    {232      "epoch": 0.5208333333333334,233      "grad_norm": 0.419812947511673,234      "learning_rate": 9.838871051713033e-05,235      "loss": 0.5849,236      "step": 300237    },238    {239      "epoch": 0.5208333333333334,240      "eval_loss": 0.4985196888446808,241      "eval_runtime": 506.4652,242      "eval_samples_per_second": 2.277,243      "eval_steps_per_second": 2.277,244      "step": 300245    },246    {247      "epoch": 0.5381944444444444,248      "grad_norm": 0.5255069136619568,249      "learning_rate": 9.81244751756236e-05,250      "loss": 0.5894,251      "step": 310252    },253    {254      "epoch": 0.5555555555555556,255      "grad_norm": 0.5103104114532471,256      "learning_rate": 9.784059763121993e-05,257      "loss": 0.5187,258      "step": 320259    },260    {261      "epoch": 0.5729166666666666,262      "grad_norm": 0.6287491321563721,263      "learning_rate": 9.753719374970908e-05,264      "loss": 0.5181,265      "step": 330266    },267    {268      "epoch": 0.5902777777777778,269      "grad_norm": 0.8683649897575378,270      "learning_rate": 9.721438736663585e-05,271      "loss": 0.5413,272      "step": 340273    },274    {275      "epoch": 0.6076388888888888,276      "grad_norm": 0.7942582964897156,277      "learning_rate": 9.687231023675597e-05,278      "loss": 0.4949,279      "step": 350280    },281    {282      "epoch": 0.625,283      "grad_norm": 0.476675808429718,284      "learning_rate": 9.651110198025997e-05,285      "loss": 0.5139,286      "step": 360287    },288    {289      "epoch": 0.6423611111111112,290      "grad_norm": 0.42043331265449524,291      "learning_rate": 9.613091002578649e-05,292      "loss": 0.5163,293      "step": 370294    },295    {296      "epoch": 0.6597222222222222,297      "grad_norm": 0.4923442602157593,298      "learning_rate": 9.573188955024882e-05,299      "loss": 0.5453,300      "step": 380301    },302    {303      "epoch": 0.6770833333333334,304      "grad_norm": 0.4754851758480072,305      "learning_rate": 9.531420341549865e-05,306      "loss": 0.5674,307      "step": 390308    },309    {310      "epoch": 0.6944444444444444,311      "grad_norm": 0.49470463395118713,312      "learning_rate": 9.48780221018535e-05,313      "loss": 0.527,314      "step": 400315    },316    {317      "epoch": 0.6944444444444444,318      "eval_loss": 0.48567065596580505,319      "eval_runtime": 512.591,320      "eval_samples_per_second": 2.249,321      "eval_steps_per_second": 2.249,322      "step": 400323    },324    {325      "epoch": 0.7118055555555556,326      "grad_norm": 0.5066745281219482,327      "learning_rate": 9.442352363851428e-05,328      "loss": 0.555,329      "step": 410330    },331    {332      "epoch": 0.7291666666666666,333      "grad_norm": 0.3770899772644043,334      "learning_rate": 9.395089353090213e-05,335      "loss": 0.4918,336      "step": 420337    },338    {339      "epoch": 0.7465277777777778,340      "grad_norm": 0.497141569852829,341      "learning_rate": 9.346032468494353e-05,342      "loss": 0.5761,343      "step": 430344    },345    {346      "epoch": 0.7638888888888888,347      "grad_norm": 0.3677353858947754,348      "learning_rate": 9.295201732833487e-05,349      "loss": 0.5363,350      "step": 440351    },352    {353      "epoch": 0.78125,354      "grad_norm": 0.5019080638885498,355      "learning_rate": 9.242617892881889e-05,356      "loss": 0.5385,357      "step": 450358    },359    {360      "epoch": 0.7986111111111112,361      "grad_norm": 0.4239552617073059,362      "learning_rate": 9.188302410950565e-05,363      "loss": 0.5026,364      "step": 460365    },366    {367      "epoch": 0.8159722222222222,368      "grad_norm": 0.4331854283809662,369      "learning_rate": 9.13227745612734e-05,370      "loss": 0.5322,371      "step": 470372    },373    {374      "epoch": 0.8333333333333334,375      "grad_norm": 0.4505394399166107,376      "learning_rate": 9.074565895228442e-05,377      "loss": 0.5145,378      "step": 480379    },380    {381      "epoch": 0.8506944444444444,382      "grad_norm": 0.4377307891845703,383      "learning_rate": 9.015191283465319e-05,384      "loss": 0.5339,385      "step": 490386    },387    {388      "epoch": 0.8680555555555556,389      "grad_norm": 0.5486445426940918,390      "learning_rate": 8.954177854830487e-05,391      "loss": 0.4782,392      "step": 500393    },394    {395      "epoch": 0.8680555555555556,396      "eval_loss": 0.4704386591911316,397      "eval_runtime": 507.1518,398      "eval_samples_per_second": 2.273,399      "eval_steps_per_second": 2.273,400      "step": 500401    },402    {403      "epoch": 0.8854166666666666,404      "grad_norm": 0.505261242389679,405      "learning_rate": 8.891550512206322e-05,406      "loss": 0.51,407      "step": 510408    },409    {410      "epoch": 0.9027777777777778,411      "grad_norm": 0.4194414019584656,412      "learning_rate": 8.827334817200845e-05,413      "loss": 0.5049,414      "step": 520415    },416    {417      "epoch": 0.9201388888888888,418      "grad_norm": 0.5332881212234497,419      "learning_rate": 8.76155697971465e-05,420      "loss": 0.5303,421      "step": 530422    },423    {424      "epoch": 0.9375,425      "grad_norm": 0.4372180700302124,426      "learning_rate": 8.694243847243224e-05,427      "loss": 0.5237,428      "step": 540429    },430    {431      "epoch": 0.9548611111111112,432      "grad_norm": 0.5241920948028564,433      "learning_rate": 8.625422893919027e-05,434      "loss": 0.5211,435      "step": 550436    },437    {438      "epoch": 0.9722222222222222,439      "grad_norm": 0.4739208221435547,440      "learning_rate": 8.555122209297817e-05,441      "loss": 0.4999,442      "step": 560443    },444    {445      "epoch": 0.9895833333333334,446      "grad_norm": 0.4937109649181366,447      "learning_rate": 8.483370486893783e-05,448      "loss": 0.5348,449      "step": 570450    },451    {452      "epoch": 1.0069444444444444,453      "grad_norm": 0.40740323066711426,454      "learning_rate": 8.410197012468156e-05,455      "loss": 0.487,456      "step": 580457    },458    {459      "epoch": 1.0243055555555556,460      "grad_norm": 0.4971655607223511,461      "learning_rate": 8.335631652076125e-05,462      "loss": 0.4467,463      "step": 590464    },465    {466      "epoch": 1.0416666666666667,467      "grad_norm": 0.5765105485916138,468      "learning_rate": 8.259704839876871e-05,469      "loss": 0.4357,470      "step": 600471    },472    {473      "epoch": 1.0416666666666667,474      "eval_loss": 0.4711858034133911,475      "eval_runtime": 502.4975,476      "eval_samples_per_second": 2.295,477      "eval_steps_per_second": 2.295,478      "step": 600479    },480    {481      "epoch": 1.0590277777777777,482      "grad_norm": 0.6608824133872986,483      "learning_rate": 8.182447565711737e-05,484      "loss": 0.4109,485      "step": 610486    },487    {488      "epoch": 1.0763888888888888,489      "grad_norm": 0.7043329477310181,490      "learning_rate": 8.103891362455602e-05,491      "loss": 0.425,492      "step": 620493    },494    {495      "epoch": 1.09375,496      "grad_norm": 0.6336780786514282,497      "learning_rate": 8.024068293146612e-05,498      "loss": 0.4188,499      "step": 630500    },501    {502      "epoch": 1.1111111111111112,503      "grad_norm": 0.6143578886985779,504      "learning_rate": 7.943010937899511e-05,505      "loss": 0.4329,506      "step": 640507    },508    {509      "epoch": 1.1284722222222223,510      "grad_norm": 0.4573313891887665,511      "learning_rate": 7.860752380607936e-05,512      "loss": 0.4484,513      "step": 650514    },515    {516      "epoch": 1.1458333333333333,517      "grad_norm": 0.49324455857276917,518      "learning_rate": 7.777326195441088e-05,519      "loss": 0.4482,520      "step": 660521    },522    {523      "epoch": 1.1631944444444444,524      "grad_norm": 0.5778242349624634,525      "learning_rate": 7.692766433140302e-05,526      "loss": 0.4013,527      "step": 670528    },529    {530      "epoch": 1.1805555555555556,531      "grad_norm": 0.515394926071167,532      "learning_rate": 7.607107607121086e-05,533      "loss": 0.4149,534      "step": 680535    },536    {537      "epoch": 1.1979166666666667,538      "grad_norm": 0.5518147349357605,539      "learning_rate": 7.520384679386344e-05,540      "loss": 0.4334,541      "step": 690542    },543    {544      "epoch": 1.2152777777777777,545      "grad_norm": 0.6633225679397583,546      "learning_rate": 7.432633046256472e-05,547      "loss": 0.4033,548      "step": 700549    },550    {551      "epoch": 1.2152777777777777,552      "eval_loss": 0.4653947055339813,553      "eval_runtime": 508.2251,554      "eval_samples_per_second": 2.269,555      "eval_steps_per_second": 2.269,556      "step": 700557    },558    {559      "epoch": 1.2326388888888888,560      "grad_norm": 0.48088809847831726,561      "learning_rate": 7.343888523922219e-05,562      "loss": 0.4328,563      "step": 710564    },565    {566      "epoch": 1.25,567      "grad_norm": 0.6473700404167175,568      "learning_rate": 7.25418733382615e-05,569      "loss": 0.4019,570      "step": 720571    },572    {573      "epoch": 1.2673611111111112,574      "grad_norm": 0.6023727059364319,575      "learning_rate": 7.163566087878726e-05,576      "loss": 0.4238,577      "step": 730578    },579    {580      "epoch": 1.2847222222222223,581      "grad_norm": 0.5473999977111816,582      "learning_rate": 7.072061773514991e-05,583      "loss": 0.4204,584      "step": 740585    },586    {587      "epoch": 1.3020833333333333,588      "grad_norm": 0.5065605640411377,589      "learning_rate": 6.979711738598004e-05,590      "loss": 0.4175,591      "step": 750592    },593    {594      "epoch": 1.3194444444444444,595      "grad_norm": 0.48466363549232483,596      "learning_rate": 6.886553676175155e-05,597      "loss": 0.4559,598      "step": 760599    },600    {601      "epoch": 1.3368055555555556,602      "grad_norm": 0.3937870264053345,603      "learning_rate": 6.792625609093593e-05,604      "loss": 0.4052,605      "step": 770606    },607    {608      "epoch": 1.3541666666666667,609      "grad_norm": 0.5539367198944092,610      "learning_rate": 6.697965874481038e-05,611      "loss": 0.4268,612      "step": 780613    },614    {615      "epoch": 1.3715277777777777,616      "grad_norm": 0.5883862376213074,617      "learning_rate": 6.602613108098327e-05,618      "loss": 0.3801,619      "step": 790620    },621    {622      "epoch": 1.3888888888888888,623      "grad_norm": 0.5152338743209839,624      "learning_rate": 6.506606228570077e-05,625      "loss": 0.4438,626      "step": 800627    },628    {629      "epoch": 1.3888888888888888,630      "eval_loss": 0.4593656063079834,631      "eval_runtime": 507.6748,632      "eval_samples_per_second": 2.271,633      "eval_steps_per_second": 2.271,634      "step": 800635    },636    {637      "epoch": 1.40625,638      "grad_norm": 0.5437167286872864,639      "learning_rate": 6.409984421499877e-05,640      "loss": 0.4497,641      "step": 810642    },643    {644      "epoch": 1.4236111111111112,645      "grad_norm": 0.560701847076416,646      "learning_rate": 6.312787123476523e-05,647      "loss": 0.4215,648      "step": 820649    },650    {651      "epoch": 1.4409722222222223,652      "grad_norm": 0.4475458860397339,653      "learning_rate": 6.215054005977829e-05,654      "loss": 0.417,655      "step": 830656    },657    {658      "epoch": 1.4583333333333333,659      "grad_norm": 0.6264685988426208,660      "learning_rate": 6.11682495917852e-05,661      "loss": 0.3694,662      "step": 840663    },664    {665      "epoch": 1.4756944444444444,666      "grad_norm": 0.4710831940174103,667      "learning_rate": 6.0181400756689185e-05,668      "loss": 0.4168,669      "step": 850670    },671    {672      "epoch": 1.4930555555555556,673      "grad_norm": 0.4411803185939789,674      "learning_rate": 5.919039634090979e-05,675      "loss": 0.4061,676      "step": 860677    },678    {679      "epoch": 1.5104166666666665,680      "grad_norm": 0.5613175630569458,681      "learning_rate": 5.819564082698392e-05,682      "loss": 0.4011,683      "step": 870684    },685    {686      "epoch": 1.5277777777777777,687      "grad_norm": 0.5883941054344177,688      "learning_rate": 5.7197540228474634e-05,689      "loss": 0.4046,690      "step": 880691    },692    {693      "epoch": 1.5451388888888888,694      "grad_norm": 0.7705323100090027,695      "learning_rate": 5.6196501924255083e-05,696      "loss": 0.4447,697      "step": 890698    },699    {700      "epoch": 1.5625,701      "grad_norm": 0.4412948489189148,702      "learning_rate": 5.5192934492235116e-05,703      "loss": 0.4297,704      "step": 900705    },706    {707      "epoch": 1.5625,708      "eval_loss": 0.45315322279930115,709      "eval_runtime": 509.3304,710      "eval_samples_per_second": 2.264,711      "eval_steps_per_second": 2.264,712      "step": 900713    },714    {715      "epoch": 1.5798611111111112,716      "grad_norm": 0.4683547914028168,717      "learning_rate": 5.4187247542598495e-05,718      "loss": 0.4094,719      "step": 910720    },721    {722      "epoch": 1.5972222222222223,723      "grad_norm": 0.5470554232597351,724      "learning_rate": 5.317985155061882e-05,725      "loss": 0.413,726      "step": 920727    },728    {729      "epoch": 1.6145833333333335,730      "grad_norm": 0.476554274559021,731      "learning_rate": 5.217115768912242e-05,732      "loss": 0.3978,733      "step": 930734    },735    {736      "epoch": 1.6319444444444444,737      "grad_norm": 0.4833614230155945,738      "learning_rate": 5.116157766066637e-05,739      "loss": 0.4377,740      "step": 940741    },742    {743      "epoch": 1.6493055555555556,744      "grad_norm": 0.5570272207260132,745      "learning_rate": 5.015152352950045e-05,746      "loss": 0.4299,747      "step": 950748    },749    {750      "epoch": 1.6666666666666665,751      "grad_norm": 0.37459680438041687,752      "learning_rate": 4.914140755338144e-05,753      "loss": 0.4285,754      "step": 960755    },756    {757      "epoch": 1.6840277777777777,758      "grad_norm": 0.462679922580719,759      "learning_rate": 4.813164201530838e-05,760      "loss": 0.4038,761      "step": 970762    },763    {764      "epoch": 1.7013888888888888,765      "grad_norm": 0.5221256017684937,766      "learning_rate": 4.712263905524754e-05,767      "loss": 0.4016,768      "step": 980769    },770    {771      "epoch": 1.71875,772      "grad_norm": 0.6251868009567261,773      "learning_rate": 4.611481050191587e-05,774      "loss": 0.4188,775      "step": 990776    },777    {778      "epoch": 1.7361111111111112,779      "grad_norm": 0.5684912204742432,780      "learning_rate": 4.5108567704691405e-05,781      "loss": 0.4012,782      "step": 1000783    },784    {785      "epoch": 1.7361111111111112,786      "eval_loss": 0.4480222463607788,787      "eval_runtime": 505.0708,788      "eval_samples_per_second": 2.283,789      "eval_steps_per_second": 2.283,790      "step": 1000791    },792    {793      "epoch": 1.7534722222222223,794      "grad_norm": 0.44916579127311707,795      "learning_rate": 4.4104321365719216e-05,796      "loss": 0.3071,797      "step": 1010798    },799    {800      "epoch": 1.7708333333333335,801      "grad_norm": 0.4596291482448578,802      "learning_rate": 4.310248137228176e-05,803      "loss": 0.302,804      "step": 1020805    },806    {807      "epoch": 1.7881944444444444,808      "grad_norm": 0.44650307297706604,809      "learning_rate": 4.210345662950161e-05,810      "loss": 0.2602,811      "step": 1030812    },813    {814      "epoch": 1.8055555555555556,815      "grad_norm": 0.4998283088207245,816      "learning_rate": 4.11076548934452e-05,817      "loss": 0.3086,818      "step": 1040819    },820    {821      "epoch": 1.8229166666666665,822      "grad_norm": 0.4060577154159546,823      "learning_rate": 4.011548260469552e-05,824      "loss": 0.2801,825      "step": 1050826    },827    {828      "epoch": 1.8402777777777777,829      "grad_norm": 0.8512741327285767,830      "learning_rate": 3.9127344722461615e-05,831      "loss": 0.3004,832      "step": 1060833    },834    {835      "epoch": 1.8576388888888888,836      "grad_norm": 0.6027821898460388,837      "learning_rate": 3.8143644559292927e-05,838      "loss": 0.2837,839      "step": 1070840    },841    {842      "epoch": 1.875,843      "grad_norm": 0.5069570541381836,844      "learning_rate": 3.716478361646555e-05,845      "loss": 0.2604,846      "step": 1080847    },848    {849      "epoch": 1.8923611111111112,850      "grad_norm": 0.56328284740448,851      "learning_rate": 3.6191161420107864e-05,852      "loss": 0.304,853      "step": 1090854    },855    {856      "epoch": 1.9097222222222223,857      "grad_norm": 0.5753824710845947,858      "learning_rate": 3.5223175358132256e-05,859      "loss": 0.2866,860      "step": 1100861    },862    {863      "epoch": 1.9097222222222223,864      "eval_loss": 0.48055773973464966,865      "eval_runtime": 491.0229,866      "eval_samples_per_second": 2.348,867      "eval_steps_per_second": 2.348,868      "step": 1100869    },870    {871      "epoch": 1.9270833333333335,872      "grad_norm": 0.6144798398017883,873      "learning_rate": 3.426122051803969e-05,874      "loss": 0.2847,875      "step": 1110876    },877    {878      "epoch": 1.9444444444444444,879      "grad_norm": 0.7388680577278137,880      "learning_rate": 3.330568952566302e-05,881      "loss": 0.278,882      "step": 1120883    },884    {885      "epoch": 1.9618055555555556,886      "grad_norm": 0.7205446362495422,887      "learning_rate": 3.235697238491516e-05,888      "loss": 0.2883,889      "step": 1130890    },891    {892      "epoch": 1.9791666666666665,893      "grad_norm": 0.6098833680152893,894      "learning_rate": 3.141545631860746e-05,895      "loss": 0.2924,896      "step": 1140897    },898    {899      "epoch": 1.9965277777777777,900      "grad_norm": 0.5834055542945862,901      "learning_rate": 3.0481525610402978e-05,902      "loss": 0.3073,903      "step": 1150904    },905    {906      "epoch": 2.013888888888889,907      "grad_norm": 0.6533185839653015,908      "learning_rate": 2.9555561447969683e-05,909      "loss": 0.304,910      "step": 1160911    },912    {913      "epoch": 2.03125,914      "grad_norm": 0.6724961996078491,915      "learning_rate": 2.8637941767396954e-05,916      "loss": 0.2801,917      "step": 1170918    },919    {920      "epoch": 2.048611111111111,921      "grad_norm": 0.771094024181366,922      "learning_rate": 2.7729041098939585e-05,923      "loss": 0.287,924      "step": 1180925    },926    {927      "epoch": 2.0659722222222223,928      "grad_norm": 0.6162275075912476,929      "learning_rate": 2.682923041415163e-05,930      "loss": 0.3276,931      "step": 1190932    },933    {934      "epoch": 2.0833333333333335,935      "grad_norm": 0.7060827016830444,936      "learning_rate": 2.5938876974472802e-05,937      "loss": 0.2718,938      "step": 1200939    },940    {941      "epoch": 2.0833333333333335,942      "eval_loss": 0.47958508133888245,943      "eval_runtime": 491.4469,944      "eval_samples_per_second": 2.346,945      "eval_steps_per_second": 2.346,946      "step": 1200947    },948    {949      "epoch": 2.1006944444444446,950      "grad_norm": 0.7561212182044983,951      "learning_rate": 2.5058344181329273e-05,952      "loss": 0.2929,953      "step": 1210954    },955    {956      "epoch": 2.1180555555555554,957      "grad_norm": 0.5527466535568237,958      "learning_rate": 2.418799142780982e-05,959      "loss": 0.3054,960      "step": 1220961    },962    {963      "epoch": 2.1354166666666665,964      "grad_norm": 0.6940686702728271,965      "learning_rate": 2.332817395197813e-05,966      "loss": 0.3087,967      "step": 1230968    },969    {970      "epoch": 2.1527777777777777,971      "grad_norm": 0.7222234010696411,972      "learning_rate": 2.2479242691880887e-05,973      "loss": 0.286,974      "step": 1240975    },976    {977      "epoch": 2.170138888888889,978      "grad_norm": 0.6128625273704529,979      "learning_rate": 2.164154414231091e-05,980      "loss": 0.2902,981      "step": 1250982    },983    {984      "epoch": 2.1875,985      "grad_norm": 0.672198474407196,986      "learning_rate": 2.0815420213383936e-05,987      "loss": 0.2631,988      "step": 1260989    },990    {991      "epoch": 2.204861111111111,992      "grad_norm": 0.5412243008613586,993      "learning_rate": 2.0001208090986557e-05,994      "loss": 0.3163,995      "step": 1270996    },997    {998      "epoch": 2.2222222222222223,999      "grad_norm": 0.8546826839447021,1000      "learning_rate": 1.9199240099152498e-05,1001      "loss": 0.3257,1002      "step": 12801003    },1004    {1005      "epoch": 2.2395833333333335,1006      "grad_norm": 0.7093095183372498,1007      "learning_rate": 1.8409843564423056e-05,1008      "loss": 0.2952,1009      "step": 12901010    },1011    {1012      "epoch": 2.2569444444444446,1013      "grad_norm": 0.6084073781967163,1014      "learning_rate": 1.7633340682247558e-05,1015      "loss": 0.3169,1016      "step": 13001017    },1018    {1019      "epoch": 2.2569444444444446,1020      "eval_loss": 0.468641072511673,1021      "eval_runtime": 486.238,1022      "eval_samples_per_second": 2.371,1023      "eval_steps_per_second": 2.371,1024      "step": 13001025    },1026    {1027      "epoch": 2.2743055555555554,1028      "grad_norm": 0.7556958198547363,1029      "learning_rate": 1.6870048385477848e-05,1030      "loss": 0.3121,1031      "step": 13101032    },1033    {1034      "epoch": 2.2916666666666665,1035      "grad_norm": 0.6000959873199463,1036      "learning_rate": 1.6120278215010886e-05,1037      "loss": 0.2928,1038      "step": 13201039    },1040    {1041      "epoch": 2.3090277777777777,1042      "grad_norm": 0.6133815050125122,1043      "learning_rate": 1.5384336192632104e-05,1044      "loss": 0.2578,1045      "step": 13301046    },1047    {1048      "epoch": 2.326388888888889,1049      "grad_norm": 0.7813517451286316,1050      "learning_rate": 1.4662522696111225e-05,1051      "loss": 0.2856,1052      "step": 13401053    },1054    {1055      "epoch": 2.34375,1056      "grad_norm": 0.738869309425354,1057      "learning_rate": 1.395513233660199e-05,1058      "loss": 0.2829,1059      "step": 13501060    },1061    {1062      "epoch": 2.361111111111111,1063      "grad_norm": 0.7364549040794373,1064      "learning_rate": 1.3262453838395205e-05,1065      "loss": 0.3188,1066      "step": 13601067    },1068    {1069      "epoch": 2.3784722222222223,1070      "grad_norm": 0.5563517212867737,1071      "learning_rate": 1.2584769921074962e-05,1072      "loss": 0.295,1073      "step": 13701074    },1075    {1076      "epoch": 2.3958333333333335,1077      "grad_norm": 0.6138696074485779,1078      "learning_rate": 1.1922357184125327e-05,1079      "loss": 0.3112,1080      "step": 13801081    },1082    {1083      "epoch": 2.4131944444444446,1084      "grad_norm": 0.6921333074569702,1085      "learning_rate": 1.1275485994035107e-05,1086      "loss": 0.2837,1087      "step": 13901088    },1089    {1090      "epoch": 2.4305555555555554,1091      "grad_norm": 0.6158069968223572,1092      "learning_rate": 1.064442037394679e-05,1093      "loss": 0.276,1094      "step": 14001095    },1096    {1097      "epoch": 2.4305555555555554,1098      "eval_loss": 0.46923181414604187,1099      "eval_runtime": 491.6516,1100      "eval_samples_per_second": 2.345,1101      "eval_steps_per_second": 2.345,1102      "step": 14001103    },1104    {1105      "epoch": 2.4479166666666665,1106      "grad_norm": 0.5451982021331787,1107      "learning_rate": 1.0029417895894211e-05,1108      "loss": 0.305,1109      "step": 14101110    },1111    {1112      "epoch": 2.4652777777777777,1113      "grad_norm": 0.5262075662612915,1114      "learning_rate": 9.430729575673658e-06,1115      "loss": 0.263,1116      "step": 14201117    },1118    {1119      "epoch": 2.482638888888889,1120      "grad_norm": 0.5978283286094666,1121      "learning_rate": 8.848599770390653e-06,1122      "loss": 0.2709,1123      "step": 14301124    },1125    {1126      "epoch": 2.5,1127      "grad_norm": 0.7257421612739563,1128      "learning_rate": 8.283266078724571e-06,1129      "loss": 0.2834,1130      "step": 14401131    },1132    {1133      "epoch": 2.517361111111111,1134      "grad_norm": 0.682624101638794,1135      "learning_rate": 7.734959243951783e-06,1136      "loss": 0.2785,1137      "step": 14501138    },1139    {1140      "epoch": 2.5347222222222223,1141      "grad_norm": 0.5846174955368042,1142      "learning_rate": 7.203903059766848e-06,1143      "loss": 0.2671,1144      "step": 14601145    },1146    {1147      "epoch": 2.5520833333333335,1148      "grad_norm": 0.7986878752708435,1149      "learning_rate": 6.690314278940207e-06,1150      "loss": 0.2646,1151      "step": 14701152    },1153    {1154      "epoch": 2.5694444444444446,1155      "grad_norm": 0.6198819875717163,1156      "learning_rate": 6.194402524849585e-06,1157      "loss": 0.2905,1158      "step": 14801159    },1160    {1161      "epoch": 2.5868055555555554,1162      "grad_norm": 0.5760586261749268,1163      "learning_rate": 5.716370205921434e-06,1164      "loss": 0.2948,1165      "step": 14901166    },1167    {1168      "epoch": 2.6041666666666665,1169      "grad_norm": 0.707608163356781,1170      "learning_rate": 5.256412433017005e-06,1171      "loss": 0.2616,1172      "step": 15001173    },1174    {1175      "epoch": 2.6041666666666665,1176      "eval_loss": 0.4693165123462677,1177      "eval_runtime": 489.2257,1178      "eval_samples_per_second": 2.357,1179      "eval_steps_per_second": 2.357,1180      "step": 15001181    },1182    {1183      "epoch": 2.6215277777777777,1184      "grad_norm": 0.43746083974838257,1185      "learning_rate": 4.814716939797109e-06,1186      "loss": 0.2411,1187      "step": 15101188    },1189    {1190      "epoch": 2.638888888888889,1191      "grad_norm": 0.6086592674255371,1192      "learning_rate": 4.391464006097845e-06,1193      "loss": 0.2909,1194      "step": 15201195    },1196    {1197      "epoch": 2.65625,1198      "grad_norm": 0.6783342957496643,1199      "learning_rate": 3.986826384348608e-06,1200      "loss": 0.2828,

Showing the first 1,200 of 1384 lines. Download the file for the rest.