| { | |
| "best_metric": 0.21217399835586548, | |
| "best_model_checkpoint": "autotrain-swinv2-base-patch4-window8-256/checkpoint-11696", | |
| "epoch": 8.0, | |
| "eval_steps": 500, | |
| "global_step": 11696, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01709986320109439, | |
| "grad_norm": 10.912551879882812, | |
| "learning_rate": 8.207934336525308e-07, | |
| "loss": 2.9335, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03419972640218878, | |
| "grad_norm": 11.582744598388672, | |
| "learning_rate": 1.6757865937072504e-06, | |
| "loss": 2.8112, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05129958960328317, | |
| "grad_norm": 11.078932762145996, | |
| "learning_rate": 2.5307797537619698e-06, | |
| "loss": 2.6726, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06839945280437756, | |
| "grad_norm": 12.702400207519531, | |
| "learning_rate": 3.38577291381669e-06, | |
| "loss": 2.4547, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08549931600547196, | |
| "grad_norm": 16.10662078857422, | |
| "learning_rate": 4.240766073871409e-06, | |
| "loss": 2.0783, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.10259917920656635, | |
| "grad_norm": 15.413719177246094, | |
| "learning_rate": 5.0615595075239396e-06, | |
| "loss": 1.8944, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11969904240766074, | |
| "grad_norm": 14.965916633605957, | |
| "learning_rate": 5.91655266757866e-06, | |
| "loss": 1.6313, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13679890560875513, | |
| "grad_norm": 15.34847640991211, | |
| "learning_rate": 6.77154582763338e-06, | |
| "loss": 1.5534, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1538987688098495, | |
| "grad_norm": 14.655660629272461, | |
| "learning_rate": 7.592339261285911e-06, | |
| "loss": 1.4456, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17099863201094392, | |
| "grad_norm": 18.560701370239258, | |
| "learning_rate": 8.44733242134063e-06, | |
| "loss": 1.3092, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1880984952120383, | |
| "grad_norm": 29.902738571166992, | |
| "learning_rate": 9.302325581395349e-06, | |
| "loss": 1.2648, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2051983584131327, | |
| "grad_norm": 27.373172760009766, | |
| "learning_rate": 1.0157318741450068e-05, | |
| "loss": 1.5046, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22229822161422708, | |
| "grad_norm": 50.6275749206543, | |
| "learning_rate": 1.1012311901504789e-05, | |
| "loss": 1.3096, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2393980848153215, | |
| "grad_norm": 16.131093978881836, | |
| "learning_rate": 1.1867305061559508e-05, | |
| "loss": 1.1689, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.25649794801641584, | |
| "grad_norm": 30.61119270324707, | |
| "learning_rate": 1.2722298221614229e-05, | |
| "loss": 1.1555, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.27359781121751026, | |
| "grad_norm": 12.447290420532227, | |
| "learning_rate": 1.3577291381668946e-05, | |
| "loss": 0.9556, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.29069767441860467, | |
| "grad_norm": 7.643873691558838, | |
| "learning_rate": 1.4432284541723667e-05, | |
| "loss": 1.0063, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.307797537619699, | |
| "grad_norm": 21.5053768157959, | |
| "learning_rate": 1.5287277701778386e-05, | |
| "loss": 1.0971, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.32489740082079344, | |
| "grad_norm": 26.181028366088867, | |
| "learning_rate": 1.6142270861833107e-05, | |
| "loss": 0.8747, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.34199726402188785, | |
| "grad_norm": Infinity, | |
| "learning_rate": 1.6963064295485636e-05, | |
| "loss": 0.899, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3590971272229822, | |
| "grad_norm": 13.22681713104248, | |
| "learning_rate": 1.7818057455540357e-05, | |
| "loss": 0.844, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.3761969904240766, | |
| "grad_norm": 20.26528549194336, | |
| "learning_rate": 1.8673050615595075e-05, | |
| "loss": 0.9843, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.393296853625171, | |
| "grad_norm": 13.295209884643555, | |
| "learning_rate": 1.9528043775649796e-05, | |
| "loss": 0.8477, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4103967168262654, | |
| "grad_norm": 17.751012802124023, | |
| "learning_rate": 2.0383036935704516e-05, | |
| "loss": 0.9479, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4274965800273598, | |
| "grad_norm": 7.726760387420654, | |
| "learning_rate": 2.1238030095759234e-05, | |
| "loss": 0.8116, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.44459644322845415, | |
| "grad_norm": 14.156341552734375, | |
| "learning_rate": 2.2093023255813955e-05, | |
| "loss": 0.8521, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.46169630642954856, | |
| "grad_norm": 39.080848693847656, | |
| "learning_rate": 2.2948016415868672e-05, | |
| "loss": 0.9364, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.478796169630643, | |
| "grad_norm": 18.685993194580078, | |
| "learning_rate": 2.3803009575923393e-05, | |
| "loss": 0.6069, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.49589603283173733, | |
| "grad_norm": 22.621519088745117, | |
| "learning_rate": 2.4658002735978114e-05, | |
| "loss": 0.6538, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5129958960328317, | |
| "grad_norm": 31.690587997436523, | |
| "learning_rate": 2.5512995896032832e-05, | |
| "loss": 0.7562, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5300957592339262, | |
| "grad_norm": 7.775012016296387, | |
| "learning_rate": 2.6367989056087556e-05, | |
| "loss": 0.692, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5471956224350205, | |
| "grad_norm": 24.86013412475586, | |
| "learning_rate": 2.7222982216142274e-05, | |
| "loss": 0.7072, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5642954856361149, | |
| "grad_norm": 6.716433525085449, | |
| "learning_rate": 2.807797537619699e-05, | |
| "loss": 0.6732, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5813953488372093, | |
| "grad_norm": 14.250188827514648, | |
| "learning_rate": 2.893296853625171e-05, | |
| "loss": 0.8366, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5984952120383037, | |
| "grad_norm": 17.7914981842041, | |
| "learning_rate": 2.9787961696306433e-05, | |
| "loss": 0.7953, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.615595075239398, | |
| "grad_norm": 13.2632417678833, | |
| "learning_rate": 3.064295485636115e-05, | |
| "loss": 0.7905, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6326949384404925, | |
| "grad_norm": 34.731689453125, | |
| "learning_rate": 3.149794801641587e-05, | |
| "loss": 0.741, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6497948016415869, | |
| "grad_norm": 23.923887252807617, | |
| "learning_rate": 3.235294117647059e-05, | |
| "loss": 0.7366, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6668946648426812, | |
| "grad_norm": 12.461081504821777, | |
| "learning_rate": 3.3207934336525306e-05, | |
| "loss": 0.8837, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.6839945280437757, | |
| "grad_norm": 14.00641918182373, | |
| "learning_rate": 3.406292749658003e-05, | |
| "loss": 0.6269, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.70109439124487, | |
| "grad_norm": 23.20526123046875, | |
| "learning_rate": 3.491792065663475e-05, | |
| "loss": 0.8368, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7181942544459644, | |
| "grad_norm": 29.71381187438965, | |
| "learning_rate": 3.577291381668947e-05, | |
| "loss": 0.5834, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7352941176470589, | |
| "grad_norm": 13.521023750305176, | |
| "learning_rate": 3.662790697674418e-05, | |
| "loss": 0.682, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.7523939808481532, | |
| "grad_norm": 18.163925170898438, | |
| "learning_rate": 3.748290013679891e-05, | |
| "loss": 0.7037, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7694938440492476, | |
| "grad_norm": 4.578676223754883, | |
| "learning_rate": 3.8337893296853625e-05, | |
| "loss": 0.6831, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.786593707250342, | |
| "grad_norm": 37.317405700683594, | |
| "learning_rate": 3.9192886456908346e-05, | |
| "loss": 0.6257, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8036935704514364, | |
| "grad_norm": 19.161466598510742, | |
| "learning_rate": 4.004787961696307e-05, | |
| "loss": 0.6827, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8207934336525308, | |
| "grad_norm": 12.673524856567383, | |
| "learning_rate": 4.090287277701779e-05, | |
| "loss": 0.7644, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8378932968536251, | |
| "grad_norm": 19.195293426513672, | |
| "learning_rate": 4.17578659370725e-05, | |
| "loss": 0.8797, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.8549931600547196, | |
| "grad_norm": 12.568032264709473, | |
| "learning_rate": 4.261285909712722e-05, | |
| "loss": 0.8057, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.872093023255814, | |
| "grad_norm": 6.0890984535217285, | |
| "learning_rate": 4.3467852257181944e-05, | |
| "loss": 0.7327, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.8891928864569083, | |
| "grad_norm": 17.399168014526367, | |
| "learning_rate": 4.4322845417236665e-05, | |
| "loss": 0.6994, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9062927496580028, | |
| "grad_norm": 22.242332458496094, | |
| "learning_rate": 4.517783857729138e-05, | |
| "loss": 0.7407, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9233926128590971, | |
| "grad_norm": 34.255836486816406, | |
| "learning_rate": 4.6032831737346106e-05, | |
| "loss": 0.7177, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9404924760601915, | |
| "grad_norm": 32.492916107177734, | |
| "learning_rate": 4.688782489740082e-05, | |
| "loss": 0.6823, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.957592339261286, | |
| "grad_norm": 8.38550853729248, | |
| "learning_rate": 4.774281805745554e-05, | |
| "loss": 0.6598, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9746922024623803, | |
| "grad_norm": 14.937362670898438, | |
| "learning_rate": 4.859781121751026e-05, | |
| "loss": 0.6052, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.9917920656634747, | |
| "grad_norm": 12.976387023925781, | |
| "learning_rate": 4.945280437756498e-05, | |
| "loss": 0.6612, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7494025264595425, | |
| "eval_f1_macro": 0.4161312340366561, | |
| "eval_f1_micro": 0.7494025264595425, | |
| "eval_f1_weighted": 0.7356756793477198, | |
| "eval_loss": 0.6534971594810486, | |
| "eval_precision_macro": 0.5527584385412905, | |
| "eval_precision_micro": 0.7494025264595425, | |
| "eval_precision_weighted": 0.8030313567462574, | |
| "eval_recall_macro": 0.42879797708152634, | |
| "eval_recall_micro": 0.7494025264595425, | |
| "eval_recall_weighted": 0.7494025264595425, | |
| "eval_runtime": 25.4169, | |
| "eval_samples_per_second": 115.238, | |
| "eval_steps_per_second": 7.239, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 1.008891928864569, | |
| "grad_norm": 9.59922981262207, | |
| "learning_rate": 4.996580027359781e-05, | |
| "loss": 0.5404, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.0259917920656634, | |
| "grad_norm": 8.981080055236816, | |
| "learning_rate": 4.9870801033591734e-05, | |
| "loss": 0.6632, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.043091655266758, | |
| "grad_norm": 19.810388565063477, | |
| "learning_rate": 4.977580179358565e-05, | |
| "loss": 0.7257, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.0601915184678523, | |
| "grad_norm": 24.40424346923828, | |
| "learning_rate": 4.9680802553579575e-05, | |
| "loss": 0.4992, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.0772913816689467, | |
| "grad_norm": 17.77897071838379, | |
| "learning_rate": 4.958580331357349e-05, | |
| "loss": 0.7102, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.094391244870041, | |
| "grad_norm": 9.230584144592285, | |
| "learning_rate": 4.9490804073567415e-05, | |
| "loss": 0.5452, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.1114911080711354, | |
| "grad_norm": 9.320358276367188, | |
| "learning_rate": 4.939580483356133e-05, | |
| "loss": 0.6861, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.1285909712722297, | |
| "grad_norm": 19.349388122558594, | |
| "learning_rate": 4.9300805593555256e-05, | |
| "loss": 0.6399, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.1456908344733243, | |
| "grad_norm": 8.893238067626953, | |
| "learning_rate": 4.920580635354917e-05, | |
| "loss": 0.541, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.1627906976744187, | |
| "grad_norm": 7.8030219078063965, | |
| "learning_rate": 4.9110807113543096e-05, | |
| "loss": 0.6742, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.179890560875513, | |
| "grad_norm": 30.97956657409668, | |
| "learning_rate": 4.901580787353701e-05, | |
| "loss": 0.5536, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.1969904240766074, | |
| "grad_norm": 28.708515167236328, | |
| "learning_rate": 4.892080863353094e-05, | |
| "loss": 0.5912, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.2140902872777017, | |
| "grad_norm": 8.513506889343262, | |
| "learning_rate": 4.8825809393524854e-05, | |
| "loss": 0.6994, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.231190150478796, | |
| "grad_norm": 3.460026741027832, | |
| "learning_rate": 4.873081015351878e-05, | |
| "loss": 0.5413, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.2482900136798905, | |
| "grad_norm": 11.066655158996582, | |
| "learning_rate": 4.8635810913512694e-05, | |
| "loss": 0.5687, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.265389876880985, | |
| "grad_norm": 11.196250915527344, | |
| "learning_rate": 4.854081167350661e-05, | |
| "loss": 0.6434, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.2824897400820794, | |
| "grad_norm": 6.403533935546875, | |
| "learning_rate": 4.8445812433500535e-05, | |
| "loss": 0.6681, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.2995896032831737, | |
| "grad_norm": 0.6531491875648499, | |
| "learning_rate": 4.835081319349445e-05, | |
| "loss": 0.6018, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.316689466484268, | |
| "grad_norm": 17.756439208984375, | |
| "learning_rate": 4.8255813953488375e-05, | |
| "loss": 0.6509, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.3337893296853625, | |
| "grad_norm": 6.791726589202881, | |
| "learning_rate": 4.816081471348229e-05, | |
| "loss": 0.7089, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.350889192886457, | |
| "grad_norm": 11.32545280456543, | |
| "learning_rate": 4.8065815473476216e-05, | |
| "loss": 0.6242, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.3679890560875512, | |
| "grad_norm": 23.671077728271484, | |
| "learning_rate": 4.797081623347013e-05, | |
| "loss": 0.7434, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.3850889192886457, | |
| "grad_norm": 2.791210174560547, | |
| "learning_rate": 4.7875816993464056e-05, | |
| "loss": 0.5504, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.40218878248974, | |
| "grad_norm": 8.67063045501709, | |
| "learning_rate": 4.778081775345797e-05, | |
| "loss": 0.756, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.4192886456908345, | |
| "grad_norm": 3.611431360244751, | |
| "learning_rate": 4.76858185134519e-05, | |
| "loss": 0.6977, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.4363885088919288, | |
| "grad_norm": 6.853910446166992, | |
| "learning_rate": 4.7590819273445814e-05, | |
| "loss": 0.4457, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.4534883720930232, | |
| "grad_norm": 5.6227264404296875, | |
| "learning_rate": 4.749582003343974e-05, | |
| "loss": 0.5169, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.4705882352941178, | |
| "grad_norm": 10.246155738830566, | |
| "learning_rate": 4.7400820793433654e-05, | |
| "loss": 0.7258, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.487688098495212, | |
| "grad_norm": 10.109542846679688, | |
| "learning_rate": 4.730582155342758e-05, | |
| "loss": 0.5037, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.5047879616963065, | |
| "grad_norm": 29.261133193969727, | |
| "learning_rate": 4.7210822313421495e-05, | |
| "loss": 0.5408, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.5218878248974008, | |
| "grad_norm": 22.082735061645508, | |
| "learning_rate": 4.711582307341542e-05, | |
| "loss": 0.6156, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.5389876880984952, | |
| "grad_norm": 4.202814102172852, | |
| "learning_rate": 4.7020823833409335e-05, | |
| "loss": 0.3967, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.5560875512995898, | |
| "grad_norm": 19.054508209228516, | |
| "learning_rate": 4.692582459340326e-05, | |
| "loss": 0.6536, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.573187414500684, | |
| "grad_norm": 10.150508880615234, | |
| "learning_rate": 4.6830825353397176e-05, | |
| "loss": 0.6868, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.5902872777017785, | |
| "grad_norm": 9.790059089660645, | |
| "learning_rate": 4.673582611339109e-05, | |
| "loss": 0.5342, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.6073871409028728, | |
| "grad_norm": 20.408973693847656, | |
| "learning_rate": 4.664082687338501e-05, | |
| "loss": 0.6302, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.6244870041039672, | |
| "grad_norm": 3.9431777000427246, | |
| "learning_rate": 4.6545827633378933e-05, | |
| "loss": 0.5242, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.6415868673050615, | |
| "grad_norm": 17.057270050048828, | |
| "learning_rate": 4.645082839337285e-05, | |
| "loss": 0.5606, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.658686730506156, | |
| "grad_norm": 15.276410102844238, | |
| "learning_rate": 4.6355829153366774e-05, | |
| "loss": 0.6188, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.6757865937072505, | |
| "grad_norm": 4.771947383880615, | |
| "learning_rate": 4.626082991336069e-05, | |
| "loss": 0.4203, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.6928864569083446, | |
| "grad_norm": 11.805036544799805, | |
| "learning_rate": 4.6165830673354615e-05, | |
| "loss": 0.6164, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.7099863201094392, | |
| "grad_norm": 15.42662239074707, | |
| "learning_rate": 4.607463140294878e-05, | |
| "loss": 0.561, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.7270861833105335, | |
| "grad_norm": 27.376279830932617, | |
| "learning_rate": 4.59796321629427e-05, | |
| "loss": 0.5232, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.744186046511628, | |
| "grad_norm": 14.586313247680664, | |
| "learning_rate": 4.5888432892536865e-05, | |
| "loss": 0.5013, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.7612859097127223, | |
| "grad_norm": 4.355943202972412, | |
| "learning_rate": 4.579343365253078e-05, | |
| "loss": 0.623, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.7783857729138166, | |
| "grad_norm": 14.60937213897705, | |
| "learning_rate": 4.5698434412524706e-05, | |
| "loss": 0.5106, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.7954856361149112, | |
| "grad_norm": 11.220321655273438, | |
| "learning_rate": 4.560343517251862e-05, | |
| "loss": 0.4515, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.8125854993160053, | |
| "grad_norm": 10.763425827026367, | |
| "learning_rate": 4.5508435932512546e-05, | |
| "loss": 0.5121, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.8296853625171, | |
| "grad_norm": 14.884260177612305, | |
| "learning_rate": 4.541343669250646e-05, | |
| "loss": 0.5982, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.8467852257181943, | |
| "grad_norm": 10.640288352966309, | |
| "learning_rate": 4.531843745250039e-05, | |
| "loss": 0.4412, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.8638850889192886, | |
| "grad_norm": 4.36832332611084, | |
| "learning_rate": 4.5223438212494304e-05, | |
| "loss": 0.57, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.8809849521203832, | |
| "grad_norm": 1.8027430772781372, | |
| "learning_rate": 4.512843897248822e-05, | |
| "loss": 0.5363, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.8980848153214773, | |
| "grad_norm": 3.5775504112243652, | |
| "learning_rate": 4.5033439732482144e-05, | |
| "loss": 0.433, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.915184678522572, | |
| "grad_norm": 6.087740421295166, | |
| "learning_rate": 4.493844049247606e-05, | |
| "loss": 0.505, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.9322845417236663, | |
| "grad_norm": 10.784486770629883, | |
| "learning_rate": 4.484344125246998e-05, | |
| "loss": 0.54, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.9493844049247606, | |
| "grad_norm": 16.345617294311523, | |
| "learning_rate": 4.47484420124639e-05, | |
| "loss": 0.5492, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.966484268125855, | |
| "grad_norm": 8.161157608032227, | |
| "learning_rate": 4.465344277245782e-05, | |
| "loss": 0.547, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.9835841313269493, | |
| "grad_norm": 0.8972734808921814, | |
| "learning_rate": 4.455844353245174e-05, | |
| "loss": 0.5136, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8866507340389211, | |
| "eval_f1_macro": 0.6620993762486447, | |
| "eval_f1_micro": 0.8866507340389211, | |
| "eval_f1_weighted": 0.8825688205984242, | |
| "eval_loss": 0.3399985134601593, | |
| "eval_precision_macro": 0.7711949393609613, | |
| "eval_precision_micro": 0.8866507340389211, | |
| "eval_precision_weighted": 0.8986612163367862, | |
| "eval_recall_macro": 0.637716282980808, | |
| "eval_recall_micro": 0.8866507340389211, | |
| "eval_recall_weighted": 0.8866507340389211, | |
| "eval_runtime": 25.8144, | |
| "eval_samples_per_second": 113.464, | |
| "eval_steps_per_second": 7.128, | |
| "step": 2924 | |
| }, | |
| { | |
| "epoch": 2.000683994528044, | |
| "grad_norm": 13.184772491455078, | |
| "learning_rate": 4.446344429244566e-05, | |
| "loss": 0.5396, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.017783857729138, | |
| "grad_norm": 25.030324935913086, | |
| "learning_rate": 4.436844505243958e-05, | |
| "loss": 0.4869, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.0348837209302326, | |
| "grad_norm": 4.769693851470947, | |
| "learning_rate": 4.42734458124335e-05, | |
| "loss": 0.6058, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.0519835841313268, | |
| "grad_norm": 7.207031726837158, | |
| "learning_rate": 4.417844657242742e-05, | |
| "loss": 0.4164, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.0690834473324213, | |
| "grad_norm": 8.032303810119629, | |
| "learning_rate": 4.408344733242134e-05, | |
| "loss": 0.6109, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.086183310533516, | |
| "grad_norm": 9.863656044006348, | |
| "learning_rate": 4.3988448092415264e-05, | |
| "loss": 0.5524, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.10328317373461, | |
| "grad_norm": 12.57875919342041, | |
| "learning_rate": 4.389344885240918e-05, | |
| "loss": 0.5071, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.1203830369357046, | |
| "grad_norm": 14.003454208374023, | |
| "learning_rate": 4.3798449612403104e-05, | |
| "loss": 0.5256, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.1374829001367988, | |
| "grad_norm": 18.200809478759766, | |
| "learning_rate": 4.370345037239703e-05, | |
| "loss": 0.5496, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.1545827633378933, | |
| "grad_norm": 4.887876033782959, | |
| "learning_rate": 4.3608451132390945e-05, | |
| "loss": 0.4347, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.1716826265389875, | |
| "grad_norm": 19.693723678588867, | |
| "learning_rate": 4.351345189238487e-05, | |
| "loss": 0.5804, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.188782489740082, | |
| "grad_norm": 4.4787726402282715, | |
| "learning_rate": 4.3418452652378786e-05, | |
| "loss": 0.7788, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.2058823529411766, | |
| "grad_norm": 8.199685096740723, | |
| "learning_rate": 4.33234534123727e-05, | |
| "loss": 0.5392, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.2229822161422708, | |
| "grad_norm": 3.600613594055176, | |
| "learning_rate": 4.322845417236662e-05, | |
| "loss": 0.4586, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.2400820793433653, | |
| "grad_norm": 1.0569565296173096, | |
| "learning_rate": 4.313345493236054e-05, | |
| "loss": 0.4908, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.2571819425444595, | |
| "grad_norm": 9.293862342834473, | |
| "learning_rate": 4.303845569235446e-05, | |
| "loss": 0.4662, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.274281805745554, | |
| "grad_norm": 20.558799743652344, | |
| "learning_rate": 4.2943456452348384e-05, | |
| "loss": 0.4093, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.2913816689466486, | |
| "grad_norm": 9.981971740722656, | |
| "learning_rate": 4.28484572123423e-05, | |
| "loss": 0.5925, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.3084815321477428, | |
| "grad_norm": 9.498967170715332, | |
| "learning_rate": 4.2753457972336224e-05, | |
| "loss": 0.398, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.3255813953488373, | |
| "grad_norm": 15.670341491699219, | |
| "learning_rate": 4.265845873233014e-05, | |
| "loss": 0.5653, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.3426812585499315, | |
| "grad_norm": 2.6220016479492188, | |
| "learning_rate": 4.2563459492324065e-05, | |
| "loss": 0.2821, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.359781121751026, | |
| "grad_norm": 4.382536888122559, | |
| "learning_rate": 4.246846025231798e-05, | |
| "loss": 0.5604, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.37688098495212, | |
| "grad_norm": 29.831298828125, | |
| "learning_rate": 4.2373461012311905e-05, | |
| "loss": 0.6324, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.3939808481532148, | |
| "grad_norm": 11.329654693603516, | |
| "learning_rate": 4.227846177230582e-05, | |
| "loss": 0.4776, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.4110807113543093, | |
| "grad_norm": 14.24655818939209, | |
| "learning_rate": 4.2183462532299746e-05, | |
| "loss": 0.3838, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.4281805745554035, | |
| "grad_norm": 10.425446510314941, | |
| "learning_rate": 4.208846329229366e-05, | |
| "loss": 0.4854, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.445280437756498, | |
| "grad_norm": 12.805408477783203, | |
| "learning_rate": 4.1993464052287586e-05, | |
| "loss": 0.4318, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.462380300957592, | |
| "grad_norm": 2.7314155101776123, | |
| "learning_rate": 4.18984648122815e-05, | |
| "loss": 0.4299, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.4794801641586868, | |
| "grad_norm": 18.400419235229492, | |
| "learning_rate": 4.180346557227543e-05, | |
| "loss": 0.51, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.496580027359781, | |
| "grad_norm": 7.5878214836120605, | |
| "learning_rate": 4.1708466332269344e-05, | |
| "loss": 0.4057, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.5136798905608755, | |
| "grad_norm": 6.642499923706055, | |
| "learning_rate": 4.161346709226327e-05, | |
| "loss": 0.5965, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.53077975376197, | |
| "grad_norm": 3.847749948501587, | |
| "learning_rate": 4.1518467852257184e-05, | |
| "loss": 0.5182, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.547879616963064, | |
| "grad_norm": 11.216913223266602, | |
| "learning_rate": 4.14234686122511e-05, | |
| "loss": 0.4942, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.5649794801641588, | |
| "grad_norm": 18.338762283325195, | |
| "learning_rate": 4.1328469372245025e-05, | |
| "loss": 0.5502, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.582079343365253, | |
| "grad_norm": 2.1774024963378906, | |
| "learning_rate": 4.123347013223894e-05, | |
| "loss": 0.557, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.5991792065663475, | |
| "grad_norm": 11.719930648803711, | |
| "learning_rate": 4.1138470892232865e-05, | |
| "loss": 0.5172, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.616279069767442, | |
| "grad_norm": 10.061673164367676, | |
| "learning_rate": 4.104347165222678e-05, | |
| "loss": 0.4343, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.633378932968536, | |
| "grad_norm": 8.708327293395996, | |
| "learning_rate": 4.0948472412220706e-05, | |
| "loss": 0.4235, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.650478796169631, | |
| "grad_norm": 13.297274589538574, | |
| "learning_rate": 4.085347317221462e-05, | |
| "loss": 0.4919, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.667578659370725, | |
| "grad_norm": 1.7467930316925049, | |
| "learning_rate": 4.0758473932208547e-05, | |
| "loss": 0.5269, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.6846785225718195, | |
| "grad_norm": 3.1984822750091553, | |
| "learning_rate": 4.066347469220246e-05, | |
| "loss": 0.5333, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.701778385772914, | |
| "grad_norm": 4.631045341491699, | |
| "learning_rate": 4.056847545219639e-05, | |
| "loss": 0.5227, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.718878248974008, | |
| "grad_norm": 34.19083786010742, | |
| "learning_rate": 4.0473476212190304e-05, | |
| "loss": 0.5034, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.7359781121751023, | |
| "grad_norm": 11.861445426940918, | |
| "learning_rate": 4.037847697218423e-05, | |
| "loss": 0.4182, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.753077975376197, | |
| "grad_norm": 2.626359224319458, | |
| "learning_rate": 4.0283477732178144e-05, | |
| "loss": 0.4524, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.7701778385772915, | |
| "grad_norm": 3.4258358478546143, | |
| "learning_rate": 4.018847849217207e-05, | |
| "loss": 0.6167, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.7872777017783856, | |
| "grad_norm": 8.724355697631836, | |
| "learning_rate": 4.0093479252165985e-05, | |
| "loss": 0.4887, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.80437756497948, | |
| "grad_norm": 4.958431720733643, | |
| "learning_rate": 3.999848001215991e-05, | |
| "loss": 0.5661, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.8214774281805743, | |
| "grad_norm": 22.518157958984375, | |
| "learning_rate": 3.9903480772153826e-05, | |
| "loss": 0.4875, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.838577291381669, | |
| "grad_norm": 2.556485176086426, | |
| "learning_rate": 3.980848153214774e-05, | |
| "loss": 0.4512, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.8556771545827635, | |
| "grad_norm": 22.274921417236328, | |
| "learning_rate": 3.971348229214166e-05, | |
| "loss": 0.6026, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.8727770177838576, | |
| "grad_norm": 11.015833854675293, | |
| "learning_rate": 3.961848305213558e-05, | |
| "loss": 0.4365, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.889876880984952, | |
| "grad_norm": 5.808104515075684, | |
| "learning_rate": 3.95234838121295e-05, | |
| "loss": 0.4042, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.9069767441860463, | |
| "grad_norm": 7.572930335998535, | |
| "learning_rate": 3.9428484572123424e-05, | |
| "loss": 0.4788, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.924076607387141, | |
| "grad_norm": 10.827695846557617, | |
| "learning_rate": 3.933348533211734e-05, | |
| "loss": 0.4966, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.9411764705882355, | |
| "grad_norm": 7.229436874389648, | |
| "learning_rate": 3.9238486092111264e-05, | |
| "loss": 0.4786, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.9582763337893296, | |
| "grad_norm": 1.738945484161377, | |
| "learning_rate": 3.914348685210518e-05, | |
| "loss": 0.3888, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.975376196990424, | |
| "grad_norm": 7.930450439453125, | |
| "learning_rate": 3.9048487612099105e-05, | |
| "loss": 0.5177, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.9924760601915183, | |
| "grad_norm": 3.204556465148926, | |
| "learning_rate": 3.895348837209303e-05, | |
| "loss": 0.5221, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8921133492659611, | |
| "eval_f1_macro": 0.7248099411197927, | |
| "eval_f1_micro": 0.8921133492659611, | |
| "eval_f1_weighted": 0.8900566376282183, | |
| "eval_loss": 0.30895766615867615, | |
| "eval_precision_macro": 0.8455393630579494, | |
| "eval_precision_micro": 0.8921133492659611, | |
| "eval_precision_weighted": 0.9014059400435374, | |
| "eval_recall_macro": 0.6934276077306252, | |
| "eval_recall_micro": 0.8921133492659611, | |
| "eval_recall_weighted": 0.8921133492659611, | |
| "eval_runtime": 25.8185, | |
| "eval_samples_per_second": 113.446, | |
| "eval_steps_per_second": 7.127, | |
| "step": 4386 | |
| }, | |
| { | |
| "epoch": 3.009575923392613, | |
| "grad_norm": 4.8183465003967285, | |
| "learning_rate": 3.8858489132086945e-05, | |
| "loss": 0.5111, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.026675786593707, | |
| "grad_norm": 11.976899147033691, | |
| "learning_rate": 3.876348989208087e-05, | |
| "loss": 0.5971, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.0437756497948016, | |
| "grad_norm": 7.26112699508667, | |
| "learning_rate": 3.8668490652074786e-05, | |
| "loss": 0.5008, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.060875512995896, | |
| "grad_norm": 16.217424392700195, | |
| "learning_rate": 3.857349141206871e-05, | |
| "loss": 0.542, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.0779753761969904, | |
| "grad_norm": 4.9412922859191895, | |
| "learning_rate": 3.8478492172062626e-05, | |
| "loss": 0.4046, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.095075239398085, | |
| "grad_norm": 14.565752983093262, | |
| "learning_rate": 3.838349293205655e-05, | |
| "loss": 0.2833, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.112175102599179, | |
| "grad_norm": 12.100059509277344, | |
| "learning_rate": 3.828849369205047e-05, | |
| "loss": 0.6031, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.1292749658002736, | |
| "grad_norm": 7.3751325607299805, | |
| "learning_rate": 3.8193494452044384e-05, | |
| "loss": 0.3429, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.146374829001368, | |
| "grad_norm": 1.8577054738998413, | |
| "learning_rate": 3.809849521203831e-05, | |
| "loss": 0.4626, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.1634746922024624, | |
| "grad_norm": 21.437606811523438, | |
| "learning_rate": 3.8003495972032224e-05, | |
| "loss": 0.5272, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.180574555403557, | |
| "grad_norm": 0.7049314379692078, | |
| "learning_rate": 3.790849673202614e-05, | |
| "loss": 0.4756, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.197674418604651, | |
| "grad_norm": 26.833208084106445, | |
| "learning_rate": 3.7813497492020065e-05, | |
| "loss": 0.4575, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.2147742818057456, | |
| "grad_norm": 3.7546021938323975, | |
| "learning_rate": 3.771849825201398e-05, | |
| "loss": 0.3827, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.23187414500684, | |
| "grad_norm": 0.08676711469888687, | |
| "learning_rate": 3.7623499012007905e-05, | |
| "loss": 0.5096, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.2489740082079344, | |
| "grad_norm": 6.28751802444458, | |
| "learning_rate": 3.752849977200182e-05, | |
| "loss": 0.5496, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.266073871409029, | |
| "grad_norm": 4.246407985687256, | |
| "learning_rate": 3.7433500531995746e-05, | |
| "loss": 0.4028, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.283173734610123, | |
| "grad_norm": 7.93133020401001, | |
| "learning_rate": 3.733850129198966e-05, | |
| "loss": 0.55, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.3002735978112177, | |
| "grad_norm": 4.031394004821777, | |
| "learning_rate": 3.7243502051983587e-05, | |
| "loss": 0.6104, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.317373461012312, | |
| "grad_norm": 6.97119140625, | |
| "learning_rate": 3.7148502811977503e-05, | |
| "loss": 0.3561, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.3344733242134064, | |
| "grad_norm": 4.752348899841309, | |
| "learning_rate": 3.705350357197143e-05, | |
| "loss": 0.4267, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.3515731874145005, | |
| "grad_norm": 5.945026397705078, | |
| "learning_rate": 3.6958504331965344e-05, | |
| "loss": 0.5058, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.368673050615595, | |
| "grad_norm": 3.1954736709594727, | |
| "learning_rate": 3.686350509195927e-05, | |
| "loss": 0.4159, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.3857729138166897, | |
| "grad_norm": 20.407272338867188, | |
| "learning_rate": 3.676850585195319e-05, | |
| "loss": 0.5054, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.402872777017784, | |
| "grad_norm": 9.6301851272583, | |
| "learning_rate": 3.667350661194711e-05, | |
| "loss": 0.4432, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.4199726402188784, | |
| "grad_norm": 7.080709934234619, | |
| "learning_rate": 3.657850737194103e-05, | |
| "loss": 0.3967, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.4370725034199725, | |
| "grad_norm": 6.897956848144531, | |
| "learning_rate": 3.648350813193495e-05, | |
| "loss": 0.3478, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 3.454172366621067, | |
| "grad_norm": 2.1270782947540283, | |
| "learning_rate": 3.6388508891928866e-05, | |
| "loss": 0.4657, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 3.471272229822161, | |
| "grad_norm": 8.693300247192383, | |
| "learning_rate": 3.629350965192278e-05, | |
| "loss": 0.3595, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 3.488372093023256, | |
| "grad_norm": 5.767752170562744, | |
| "learning_rate": 3.6198510411916706e-05, | |
| "loss": 0.6431, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.5054719562243504, | |
| "grad_norm": 2.4627044200897217, | |
| "learning_rate": 3.610351117191062e-05, | |
| "loss": 0.4672, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 3.5225718194254445, | |
| "grad_norm": 11.726508140563965, | |
| "learning_rate": 3.600851193190455e-05, | |
| "loss": 0.5105, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 3.539671682626539, | |
| "grad_norm": 12.632149696350098, | |
| "learning_rate": 3.5913512691898464e-05, | |
| "loss": 0.43, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 3.556771545827633, | |
| "grad_norm": 12.43307113647461, | |
| "learning_rate": 3.581851345189239e-05, | |
| "loss": 0.5033, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.573871409028728, | |
| "grad_norm": 5.589097499847412, | |
| "learning_rate": 3.5723514211886304e-05, | |
| "loss": 0.5718, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 3.5909712722298224, | |
| "grad_norm": 7.462125778198242, | |
| "learning_rate": 3.562851497188023e-05, | |
| "loss": 0.5025, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.6080711354309165, | |
| "grad_norm": 2.099492073059082, | |
| "learning_rate": 3.5533515731874145e-05, | |
| "loss": 0.3758, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 3.625170998632011, | |
| "grad_norm": 6.787677764892578, | |
| "learning_rate": 3.543851649186807e-05, | |
| "loss": 0.3685, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.6422708618331052, | |
| "grad_norm": 11.592167854309082, | |
| "learning_rate": 3.5343517251861985e-05, | |
| "loss": 0.4173, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 3.6593707250342, | |
| "grad_norm": 5.881494522094727, | |
| "learning_rate": 3.524851801185591e-05, | |
| "loss": 0.3484, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.6764705882352944, | |
| "grad_norm": 5.682451248168945, | |
| "learning_rate": 3.5153518771849826e-05, | |
| "loss": 0.4324, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 3.6935704514363885, | |
| "grad_norm": 0.20855078101158142, | |
| "learning_rate": 3.505851953184375e-05, | |
| "loss": 0.4147, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.7106703146374826, | |
| "grad_norm": 8.257678031921387, | |
| "learning_rate": 3.4963520291837666e-05, | |
| "loss": 0.494, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 3.7277701778385772, | |
| "grad_norm": 9.025782585144043, | |
| "learning_rate": 3.486852105183159e-05, | |
| "loss": 0.4538, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.744870041039672, | |
| "grad_norm": 8.485616683959961, | |
| "learning_rate": 3.477352181182551e-05, | |
| "loss": 0.4405, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 3.761969904240766, | |
| "grad_norm": 7.031386375427246, | |
| "learning_rate": 3.467852257181943e-05, | |
| "loss": 0.5225, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.7790697674418605, | |
| "grad_norm": 6.483025550842285, | |
| "learning_rate": 3.458352333181335e-05, | |
| "loss": 0.5224, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 3.7961696306429547, | |
| "grad_norm": 13.42219066619873, | |
| "learning_rate": 3.4488524091807264e-05, | |
| "loss": 0.4022, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.8132694938440492, | |
| "grad_norm": 11.677425384521484, | |
| "learning_rate": 3.439352485180119e-05, | |
| "loss": 0.4221, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 3.830369357045144, | |
| "grad_norm": 83.29227447509766, | |
| "learning_rate": 3.4298525611795105e-05, | |
| "loss": 0.4575, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.847469220246238, | |
| "grad_norm": 13.023981094360352, | |
| "learning_rate": 3.420352637178903e-05, | |
| "loss": 0.4159, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 3.8645690834473325, | |
| "grad_norm": 3.78824520111084, | |
| "learning_rate": 3.4108527131782945e-05, | |
| "loss": 0.3825, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.8816689466484267, | |
| "grad_norm": 23.089439392089844, | |
| "learning_rate": 3.401352789177687e-05, | |
| "loss": 0.4796, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 3.8987688098495212, | |
| "grad_norm": 8.51207160949707, | |
| "learning_rate": 3.3918528651770786e-05, | |
| "loss": 0.389, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.915868673050616, | |
| "grad_norm": 25.023662567138672, | |
| "learning_rate": 3.382352941176471e-05, | |
| "loss": 0.4779, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 3.93296853625171, | |
| "grad_norm": 10.165434837341309, | |
| "learning_rate": 3.3728530171758627e-05, | |
| "loss": 0.4599, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.9500683994528045, | |
| "grad_norm": 18.18007469177246, | |
| "learning_rate": 3.363353093175255e-05, | |
| "loss": 0.4497, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 3.9671682626538987, | |
| "grad_norm": 4.1317315101623535, | |
| "learning_rate": 3.353853169174647e-05, | |
| "loss": 0.3715, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.9842681258549932, | |
| "grad_norm": 10.207343101501465, | |
| "learning_rate": 3.344353245174039e-05, | |
| "loss": 0.3945, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.899283031751451, | |
| "eval_f1_macro": 0.7167734728855646, | |
| "eval_f1_micro": 0.899283031751451, | |
| "eval_f1_weighted": 0.8949067494674995, | |
| "eval_loss": 0.2918279469013214, | |
| "eval_precision_macro": 0.8481396171573297, | |
| "eval_precision_micro": 0.899283031751451, | |
| "eval_precision_weighted": 0.9104794727753962, | |
| "eval_recall_macro": 0.677056197036938, | |
| "eval_recall_micro": 0.899283031751451, | |
| "eval_recall_weighted": 0.899283031751451, | |
| "eval_runtime": 26.2772, | |
| "eval_samples_per_second": 111.465, | |
| "eval_steps_per_second": 7.002, | |
| "step": 5848 | |
| }, | |
| { | |
| "epoch": 4.001367989056088, | |
| "grad_norm": 9.241109848022461, | |
| "learning_rate": 3.334853321173431e-05, | |
| "loss": 0.3062, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 4.0184678522571815, | |
| "grad_norm": 2.8735108375549316, | |
| "learning_rate": 3.325353397172823e-05, | |
| "loss": 0.3876, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 4.035567715458276, | |
| "grad_norm": 7.347777843475342, | |
| "learning_rate": 3.315853473172215e-05, | |
| "loss": 0.5103, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.052667578659371, | |
| "grad_norm": 17.826900482177734, | |
| "learning_rate": 3.306353549171607e-05, | |
| "loss": 0.4089, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 4.069767441860465, | |
| "grad_norm": 4.9833292961120605, | |
| "learning_rate": 3.296853625170999e-05, | |
| "loss": 0.4656, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 4.08686730506156, | |
| "grad_norm": 9.229751586914062, | |
| "learning_rate": 3.2873537011703906e-05, | |
| "loss": 0.3591, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 4.1039671682626535, | |
| "grad_norm": 0.23064318299293518, | |
| "learning_rate": 3.277853777169782e-05, | |
| "loss": 0.4878, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.121067031463748, | |
| "grad_norm": 23.045246124267578, | |
| "learning_rate": 3.2683538531691746e-05, | |
| "loss": 0.3487, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 4.138166894664843, | |
| "grad_norm": 8.428695678710938, | |
| "learning_rate": 3.258853929168566e-05, | |
| "loss": 0.3685, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 4.155266757865937, | |
| "grad_norm": 23.741397857666016, | |
| "learning_rate": 3.249354005167959e-05, | |
| "loss": 0.4878, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 4.172366621067032, | |
| "grad_norm": 4.824987888336182, | |
| "learning_rate": 3.2398540811673504e-05, | |
| "loss": 0.3667, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.1894664842681255, | |
| "grad_norm": 5.439081192016602, | |
| "learning_rate": 3.230354157166743e-05, | |
| "loss": 0.449, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 4.20656634746922, | |
| "grad_norm": 4.361891269683838, | |
| "learning_rate": 3.2208542331661344e-05, | |
| "loss": 0.3391, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 4.223666210670315, | |
| "grad_norm": 68.6259994506836, | |
| "learning_rate": 3.211354309165527e-05, | |
| "loss": 0.4126, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 4.240766073871409, | |
| "grad_norm": 5.1873860359191895, | |
| "learning_rate": 3.201854385164919e-05, | |
| "loss": 0.5427, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.257865937072504, | |
| "grad_norm": 4.0612359046936035, | |
| "learning_rate": 3.192354461164311e-05, | |
| "loss": 0.45, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 4.2749658002735975, | |
| "grad_norm": 7.9056010246276855, | |
| "learning_rate": 3.182854537163703e-05, | |
| "loss": 0.3444, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 4.292065663474692, | |
| "grad_norm": 13.213556289672852, | |
| "learning_rate": 3.173354613163095e-05, | |
| "loss": 0.3836, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 4.309165526675787, | |
| "grad_norm": 2.609199047088623, | |
| "learning_rate": 3.163854689162487e-05, | |
| "loss": 0.3189, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.326265389876881, | |
| "grad_norm": 5.747865676879883, | |
| "learning_rate": 3.154354765161879e-05, | |
| "loss": 0.4637, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 4.343365253077975, | |
| "grad_norm": 11.459451675415039, | |
| "learning_rate": 3.144854841161271e-05, | |
| "loss": 0.4867, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 4.3604651162790695, | |
| "grad_norm": 38.839962005615234, | |
| "learning_rate": 3.135354917160663e-05, | |
| "loss": 0.3632, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 4.377564979480164, | |
| "grad_norm": 16.965190887451172, | |
| "learning_rate": 3.125854993160055e-05, | |
| "loss": 0.4038, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.394664842681259, | |
| "grad_norm": 9.547988891601562, | |
| "learning_rate": 3.116355069159447e-05, | |
| "loss": 0.3246, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 4.411764705882353, | |
| "grad_norm": 29.236087799072266, | |
| "learning_rate": 3.106855145158839e-05, | |
| "loss": 0.4722, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 4.428864569083447, | |
| "grad_norm": 9.634954452514648, | |
| "learning_rate": 3.0973552211582304e-05, | |
| "loss": 0.4658, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 4.4459644322845415, | |
| "grad_norm": 14.585807800292969, | |
| "learning_rate": 3.087855297157623e-05, | |
| "loss": 0.3389, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.463064295485636, | |
| "grad_norm": 9.657065391540527, | |
| "learning_rate": 3.0783553731570145e-05, | |
| "loss": 0.3755, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 4.480164158686731, | |
| "grad_norm": 10.945516586303711, | |
| "learning_rate": 3.068855449156407e-05, | |
| "loss": 0.4174, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 4.497264021887825, | |
| "grad_norm": 7.025084018707275, | |
| "learning_rate": 3.0593555251557986e-05, | |
| "loss": 0.5521, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 4.514363885088919, | |
| "grad_norm": 4.908510684967041, | |
| "learning_rate": 3.049855601155191e-05, | |
| "loss": 0.3591, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 4.5314637482900135, | |
| "grad_norm": 3.1344048976898193, | |
| "learning_rate": 3.0403556771545826e-05, | |
| "loss": 0.3548, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 4.548563611491108, | |
| "grad_norm": 4.515110015869141, | |
| "learning_rate": 3.030855753153975e-05, | |
| "loss": 0.4855, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 4.565663474692203, | |
| "grad_norm": 7.656958103179932, | |
| "learning_rate": 3.0213558291533667e-05, | |
| "loss": 0.4507, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 4.582763337893297, | |
| "grad_norm": 16.05572509765625, | |
| "learning_rate": 3.011855905152759e-05, | |
| "loss": 0.481, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 4.599863201094391, | |
| "grad_norm": 3.04164457321167, | |
| "learning_rate": 3.0023559811521507e-05, | |
| "loss": 0.3844, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 4.6169630642954855, | |
| "grad_norm": 11.163993835449219, | |
| "learning_rate": 2.992856057151543e-05, | |
| "loss": 0.4604, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 4.63406292749658, | |
| "grad_norm": 2.3913285732269287, | |
| "learning_rate": 2.983356133150935e-05, | |
| "loss": 0.4375, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 4.651162790697675, | |
| "grad_norm": 12.693867683410645, | |
| "learning_rate": 2.9738562091503268e-05, | |
| "loss": 0.3717, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 4.668262653898768, | |
| "grad_norm": 5.616913795471191, | |
| "learning_rate": 2.964356285149719e-05, | |
| "loss": 0.2993, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 4.685362517099863, | |
| "grad_norm": 10.53511905670166, | |
| "learning_rate": 2.954856361149111e-05, | |
| "loss": 0.4742, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 4.7024623803009575, | |
| "grad_norm": 4.093250751495361, | |
| "learning_rate": 2.9453564371485032e-05, | |
| "loss": 0.3256, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 4.719562243502052, | |
| "grad_norm": 12.562928199768066, | |
| "learning_rate": 2.935856513147895e-05, | |
| "loss": 0.3647, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 4.736662106703147, | |
| "grad_norm": 28.220108032226562, | |
| "learning_rate": 2.9263565891472873e-05, | |
| "loss": 0.4505, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 4.75376196990424, | |
| "grad_norm": 11.514904975891113, | |
| "learning_rate": 2.916856665146679e-05, | |
| "loss": 0.3662, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 4.770861833105335, | |
| "grad_norm": 3.687920093536377, | |
| "learning_rate": 2.907356741146071e-05, | |
| "loss": 0.3526, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 4.7879616963064295, | |
| "grad_norm": 7.425134181976318, | |
| "learning_rate": 2.897856817145463e-05, | |
| "loss": 0.3395, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.805061559507524, | |
| "grad_norm": 2.660003185272217, | |
| "learning_rate": 2.888356893144855e-05, | |
| "loss": 0.4889, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 4.822161422708619, | |
| "grad_norm": 7.089037895202637, | |
| "learning_rate": 2.8788569691442467e-05, | |
| "loss": 0.4015, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 4.839261285909712, | |
| "grad_norm": 11.357674598693848, | |
| "learning_rate": 2.869357045143639e-05, | |
| "loss": 0.4199, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 4.856361149110807, | |
| "grad_norm": 18.09085464477539, | |
| "learning_rate": 2.8598571211430308e-05, | |
| "loss": 0.4895, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 4.8734610123119015, | |
| "grad_norm": 10.047117233276367, | |
| "learning_rate": 2.850357197142423e-05, | |
| "loss": 0.4722, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 4.890560875512996, | |
| "grad_norm": 21.57419776916504, | |
| "learning_rate": 2.840857273141815e-05, | |
| "loss": 0.4878, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 4.907660738714091, | |
| "grad_norm": 7.3810577392578125, | |
| "learning_rate": 2.8313573491412072e-05, | |
| "loss": 0.4413, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 4.924760601915184, | |
| "grad_norm": 1.9230318069458008, | |
| "learning_rate": 2.821857425140599e-05, | |
| "loss": 0.3885, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 4.941860465116279, | |
| "grad_norm": 6.324495792388916, | |
| "learning_rate": 2.812357501139991e-05, | |
| "loss": 0.4513, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 4.9589603283173735, | |
| "grad_norm": 6.727700710296631, | |
| "learning_rate": 2.802857577139383e-05, | |
| "loss": 0.4085, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 4.976060191518468, | |
| "grad_norm": 7.250006675720215, | |
| "learning_rate": 2.793357653138775e-05, | |
| "loss": 0.3343, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 4.993160054719562, | |
| "grad_norm": 2.118319034576416, | |
| "learning_rate": 2.7838577291381667e-05, | |
| "loss": 0.4999, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8828951860703311, | |
| "eval_f1_macro": 0.7464038256987945, | |
| "eval_f1_micro": 0.8828951860703311, | |
| "eval_f1_weighted": 0.8814097485811152, | |
| "eval_loss": 0.3324004113674164, | |
| "eval_precision_macro": 0.8694718296335959, | |
| "eval_precision_micro": 0.8828951860703311, | |
| "eval_precision_weighted": 0.9022243801832591, | |
| "eval_recall_macro": 0.7003746302122184, | |
| "eval_recall_micro": 0.8828951860703311, | |
| "eval_recall_weighted": 0.8828951860703311, | |
| "eval_runtime": 25.9346, | |
| "eval_samples_per_second": 112.938, | |
| "eval_steps_per_second": 7.095, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 5.010259917920656, | |
| "grad_norm": 0.7182475328445435, | |
| "learning_rate": 2.774357805137559e-05, | |
| "loss": 0.4188, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 5.027359781121751, | |
| "grad_norm": 1.1450774669647217, | |
| "learning_rate": 2.7648578811369507e-05, | |
| "loss": 0.394, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 5.0444596443228455, | |
| "grad_norm": 5.432520866394043, | |
| "learning_rate": 2.755357957136343e-05, | |
| "loss": 0.4667, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 5.06155950752394, | |
| "grad_norm": 9.457201957702637, | |
| "learning_rate": 2.7458580331357355e-05, | |
| "loss": 0.3702, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 5.078659370725034, | |
| "grad_norm": 4.041426181793213, | |
| "learning_rate": 2.736358109135127e-05, | |
| "loss": 0.4232, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 5.095759233926128, | |
| "grad_norm": 5.445071220397949, | |
| "learning_rate": 2.7268581851345192e-05, | |
| "loss": 0.5922, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 5.112859097127223, | |
| "grad_norm": 21.097936630249023, | |
| "learning_rate": 2.717358261133911e-05, | |
| "loss": 0.3139, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 5.1299589603283176, | |
| "grad_norm": 7.353487968444824, | |
| "learning_rate": 2.7078583371333032e-05, | |
| "loss": 0.365, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 5.147058823529412, | |
| "grad_norm": 0.25362271070480347, | |
| "learning_rate": 2.698358413132695e-05, | |
| "loss": 0.4082, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 5.164158686730506, | |
| "grad_norm": 13.88097858428955, | |
| "learning_rate": 2.6888584891320873e-05, | |
| "loss": 0.3095, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 5.1812585499316, | |
| "grad_norm": 0.10476674884557724, | |
| "learning_rate": 2.679358565131479e-05, | |
| "loss": 0.4199, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 5.198358413132695, | |
| "grad_norm": 15.92738151550293, | |
| "learning_rate": 2.6698586411308713e-05, | |
| "loss": 0.3436, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 5.2154582763337896, | |
| "grad_norm": 3.1535122394561768, | |
| "learning_rate": 2.660358717130263e-05, | |
| "loss": 0.3798, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 5.232558139534884, | |
| "grad_norm": 20.45937728881836, | |
| "learning_rate": 2.6508587931296554e-05, | |
| "loss": 0.4386, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 5.249658002735978, | |
| "grad_norm": 11.347773551940918, | |
| "learning_rate": 2.641358869129047e-05, | |
| "loss": 0.2972, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 5.266757865937072, | |
| "grad_norm": 2.8144240379333496, | |
| "learning_rate": 2.631858945128439e-05, | |
| "loss": 0.3987, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 5.283857729138167, | |
| "grad_norm": 5.629673480987549, | |
| "learning_rate": 2.6223590211278308e-05, | |
| "loss": 0.3263, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 5.300957592339262, | |
| "grad_norm": 2.6866259574890137, | |
| "learning_rate": 2.6128590971272232e-05, | |
| "loss": 0.3661, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 5.318057455540355, | |
| "grad_norm": 14.74667739868164, | |
| "learning_rate": 2.603359173126615e-05, | |
| "loss": 0.3379, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 5.33515731874145, | |
| "grad_norm": 3.7586936950683594, | |
| "learning_rate": 2.5938592491260072e-05, | |
| "loss": 0.3503, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 5.352257181942544, | |
| "grad_norm": 4.488011360168457, | |
| "learning_rate": 2.584359325125399e-05, | |
| "loss": 0.4632, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 5.369357045143639, | |
| "grad_norm": 2.083674430847168, | |
| "learning_rate": 2.5748594011247913e-05, | |
| "loss": 0.3565, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 5.386456908344734, | |
| "grad_norm": 7.482815265655518, | |
| "learning_rate": 2.565359477124183e-05, | |
| "loss": 0.3507, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 5.403556771545827, | |
| "grad_norm": 15.02985954284668, | |
| "learning_rate": 2.5558595531235753e-05, | |
| "loss": 0.3787, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 5.420656634746922, | |
| "grad_norm": 8.766378402709961, | |
| "learning_rate": 2.546359629122967e-05, | |
| "loss": 0.4566, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 5.437756497948016, | |
| "grad_norm": 13.31812858581543, | |
| "learning_rate": 2.536859705122359e-05, | |
| "loss": 0.4214, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 5.454856361149111, | |
| "grad_norm": 3.4309494495391846, | |
| "learning_rate": 2.5273597811217514e-05, | |
| "loss": 0.4198, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 5.471956224350206, | |
| "grad_norm": 13.40803337097168, | |
| "learning_rate": 2.517859857121143e-05, | |
| "loss": 0.3288, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 5.489056087551299, | |
| "grad_norm": 2.447211265563965, | |
| "learning_rate": 2.5083599331205355e-05, | |
| "loss": 0.3237, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 5.506155950752394, | |
| "grad_norm": 3.199465036392212, | |
| "learning_rate": 2.498860009119927e-05, | |
| "loss": 0.455, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 5.523255813953488, | |
| "grad_norm": 20.3509521484375, | |
| "learning_rate": 2.4893600851193192e-05, | |
| "loss": 0.5065, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 5.540355677154583, | |
| "grad_norm": 51.09640884399414, | |
| "learning_rate": 2.4798601611187112e-05, | |
| "loss": 0.3341, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 5.557455540355678, | |
| "grad_norm": 2.58319091796875, | |
| "learning_rate": 2.4703602371181032e-05, | |
| "loss": 0.3664, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 5.574555403556771, | |
| "grad_norm": 8.091569900512695, | |
| "learning_rate": 2.460860313117495e-05, | |
| "loss": 0.3487, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 5.591655266757866, | |
| "grad_norm": 3.2208356857299805, | |
| "learning_rate": 2.451360389116887e-05, | |
| "loss": 0.4357, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 5.60875512995896, | |
| "grad_norm": 7.119288444519043, | |
| "learning_rate": 2.441860465116279e-05, | |
| "loss": 0.4647, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 5.625854993160055, | |
| "grad_norm": 2.9200692176818848, | |
| "learning_rate": 2.432360541115671e-05, | |
| "loss": 0.4414, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 5.642954856361149, | |
| "grad_norm": 17.426734924316406, | |
| "learning_rate": 2.422860617115063e-05, | |
| "loss": 0.3779, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 5.660054719562243, | |
| "grad_norm": 6.002254009246826, | |
| "learning_rate": 2.413360693114455e-05, | |
| "loss": 0.4117, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 5.677154582763338, | |
| "grad_norm": 10.85733413696289, | |
| "learning_rate": 2.4038607691138474e-05, | |
| "loss": 0.3918, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 5.694254445964432, | |
| "grad_norm": 22.411169052124023, | |
| "learning_rate": 2.3943608451132395e-05, | |
| "loss": 0.3103, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 5.711354309165527, | |
| "grad_norm": 14.168360710144043, | |
| "learning_rate": 2.384860921112631e-05, | |
| "loss": 0.3941, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 5.728454172366621, | |
| "grad_norm": 3.775557279586792, | |
| "learning_rate": 2.3753609971120232e-05, | |
| "loss": 0.4347, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 5.745554035567715, | |
| "grad_norm": 2.539484977722168, | |
| "learning_rate": 2.3658610731114152e-05, | |
| "loss": 0.273, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 5.76265389876881, | |
| "grad_norm": 2.8777148723602295, | |
| "learning_rate": 2.3563611491108072e-05, | |
| "loss": 0.3594, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 5.779753761969904, | |
| "grad_norm": 3.811925172805786, | |
| "learning_rate": 2.3468612251101993e-05, | |
| "loss": 0.3306, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 5.796853625170999, | |
| "grad_norm": 10.664482116699219, | |
| "learning_rate": 2.3373613011095913e-05, | |
| "loss": 0.3625, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 5.813953488372093, | |
| "grad_norm": 1.653448462486267, | |
| "learning_rate": 2.3278613771089833e-05, | |
| "loss": 0.298, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 5.831053351573187, | |
| "grad_norm": 3.4559919834136963, | |
| "learning_rate": 2.3183614531083753e-05, | |
| "loss": 0.355, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 5.848153214774282, | |
| "grad_norm": 4.02964973449707, | |
| "learning_rate": 2.3088615291077674e-05, | |
| "loss": 0.4186, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 5.865253077975376, | |
| "grad_norm": 7.618357181549072, | |
| "learning_rate": 2.2993616051071594e-05, | |
| "loss": 0.5639, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 5.882352941176471, | |
| "grad_norm": 4.395501136779785, | |
| "learning_rate": 2.289861681106551e-05, | |
| "loss": 0.3863, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 5.899452804377565, | |
| "grad_norm": 4.189037322998047, | |
| "learning_rate": 2.280361757105943e-05, | |
| "loss": 0.3274, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 5.916552667578659, | |
| "grad_norm": 1.825620412826538, | |
| "learning_rate": 2.270861833105335e-05, | |
| "loss": 0.4074, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 5.933652530779754, | |
| "grad_norm": 2.863705635070801, | |
| "learning_rate": 2.2613619091047272e-05, | |
| "loss": 0.3973, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 5.950752393980848, | |
| "grad_norm": 3.3748676776885986, | |
| "learning_rate": 2.2518619851041192e-05, | |
| "loss": 0.3318, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 5.967852257181942, | |
| "grad_norm": 18.595125198364258, | |
| "learning_rate": 2.2423620611035112e-05, | |
| "loss": 0.355, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 5.984952120383037, | |
| "grad_norm": 8.303060531616211, | |
| "learning_rate": 2.2328621371029033e-05, | |
| "loss": 0.3325, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9235233868214408, | |
| "eval_f1_macro": 0.8015916762918697, | |
| "eval_f1_micro": 0.9235233868214408, | |
| "eval_f1_weighted": 0.922485221004273, | |
| "eval_loss": 0.24165508151054382, | |
| "eval_precision_macro": 0.8743011626241778, | |
| "eval_precision_micro": 0.9235233868214408, | |
| "eval_precision_weighted": 0.9276221151549562, | |
| "eval_recall_macro": 0.7664921252239268, | |
| "eval_recall_micro": 0.9235233868214408, | |
| "eval_recall_weighted": 0.9235233868214408, | |
| "eval_runtime": 25.9565, | |
| "eval_samples_per_second": 112.842, | |
| "eval_steps_per_second": 7.089, | |
| "step": 8772 | |
| }, | |
| { | |
| "epoch": 6.002051983584131, | |
| "grad_norm": 11.892972946166992, | |
| "learning_rate": 2.2233622131022953e-05, | |
| "loss": 0.3357, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 6.019151846785226, | |
| "grad_norm": 7.985339164733887, | |
| "learning_rate": 2.2138622891016873e-05, | |
| "loss": 0.4488, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 6.03625170998632, | |
| "grad_norm": 6.227213382720947, | |
| "learning_rate": 2.2043623651010793e-05, | |
| "loss": 0.3902, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 6.053351573187414, | |
| "grad_norm": 19.529422760009766, | |
| "learning_rate": 2.194862441100471e-05, | |
| "loss": 0.4248, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 6.070451436388509, | |
| "grad_norm": 19.405685424804688, | |
| "learning_rate": 2.185362517099863e-05, | |
| "loss": 0.2371, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 6.087551299589603, | |
| "grad_norm": 31.696195602416992, | |
| "learning_rate": 2.1758625930992554e-05, | |
| "loss": 0.511, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 6.104651162790698, | |
| "grad_norm": 23.88582992553711, | |
| "learning_rate": 2.1663626690986475e-05, | |
| "loss": 0.2996, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 6.121751025991792, | |
| "grad_norm": 7.8858561515808105, | |
| "learning_rate": 2.1568627450980395e-05, | |
| "loss": 0.2554, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 6.138850889192886, | |
| "grad_norm": 2.5783660411834717, | |
| "learning_rate": 2.1473628210974315e-05, | |
| "loss": 0.4588, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 6.155950752393981, | |
| "grad_norm": 2.6995346546173096, | |
| "learning_rate": 2.1378628970968235e-05, | |
| "loss": 0.3833, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 6.173050615595075, | |
| "grad_norm": 6.289373874664307, | |
| "learning_rate": 2.1283629730962156e-05, | |
| "loss": 0.4185, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 6.19015047879617, | |
| "grad_norm": 6.533155918121338, | |
| "learning_rate": 2.1188630490956073e-05, | |
| "loss": 0.2384, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 6.207250341997264, | |
| "grad_norm": 17.285852432250977, | |
| "learning_rate": 2.1093631250949993e-05, | |
| "loss": 0.2801, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 6.224350205198358, | |
| "grad_norm": 9.399192810058594, | |
| "learning_rate": 2.0998632010943913e-05, | |
| "loss": 0.5708, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 6.241450068399453, | |
| "grad_norm": 12.317584991455078, | |
| "learning_rate": 2.0903632770937833e-05, | |
| "loss": 0.4415, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 6.258549931600547, | |
| "grad_norm": 7.904637336730957, | |
| "learning_rate": 2.0808633530931754e-05, | |
| "loss": 0.3887, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 6.275649794801642, | |
| "grad_norm": 8.931986808776855, | |
| "learning_rate": 2.0713634290925674e-05, | |
| "loss": 0.4239, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 6.292749658002736, | |
| "grad_norm": 7.259034633636475, | |
| "learning_rate": 2.0618635050919594e-05, | |
| "loss": 0.3405, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 6.30984952120383, | |
| "grad_norm": 2.570568084716797, | |
| "learning_rate": 2.0523635810913514e-05, | |
| "loss": 0.3136, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 6.326949384404925, | |
| "grad_norm": 16.150344848632812, | |
| "learning_rate": 2.0428636570907435e-05, | |
| "loss": 0.4056, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 6.344049247606019, | |
| "grad_norm": 15.508628845214844, | |
| "learning_rate": 2.0333637330901355e-05, | |
| "loss": 0.4423, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 6.361149110807114, | |
| "grad_norm": 3.140324354171753, | |
| "learning_rate": 2.0238638090895272e-05, | |
| "loss": 0.3972, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 6.378248974008208, | |
| "grad_norm": 2.326601505279541, | |
| "learning_rate": 2.0143638850889192e-05, | |
| "loss": 0.3066, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 6.395348837209302, | |
| "grad_norm": 2.924823760986328, | |
| "learning_rate": 2.0048639610883112e-05, | |
| "loss": 0.3117, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 6.412448700410397, | |
| "grad_norm": 13.392170906066895, | |
| "learning_rate": 1.9953640370877033e-05, | |
| "loss": 0.2947, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 6.429548563611491, | |
| "grad_norm": 2.736306667327881, | |
| "learning_rate": 1.9858641130870953e-05, | |
| "loss": 0.2432, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 6.446648426812586, | |
| "grad_norm": 2.8124167919158936, | |
| "learning_rate": 1.9763641890864873e-05, | |
| "loss": 0.4835, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 6.46374829001368, | |
| "grad_norm": 8.63110637664795, | |
| "learning_rate": 1.9668642650858794e-05, | |
| "loss": 0.3444, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 6.480848153214774, | |
| "grad_norm": 12.762578010559082, | |
| "learning_rate": 1.9573643410852714e-05, | |
| "loss": 0.2832, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 6.497948016415869, | |
| "grad_norm": 4.230663776397705, | |
| "learning_rate": 1.9478644170846634e-05, | |
| "loss": 0.3595, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 6.515047879616963, | |
| "grad_norm": 2.9571053981781006, | |
| "learning_rate": 1.9383644930840554e-05, | |
| "loss": 0.3388, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 6.532147742818058, | |
| "grad_norm": 3.3192389011383057, | |
| "learning_rate": 1.9288645690834475e-05, | |
| "loss": 0.4937, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 6.549247606019152, | |
| "grad_norm": 4.746058940887451, | |
| "learning_rate": 1.9193646450828395e-05, | |
| "loss": 0.4209, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 6.566347469220246, | |
| "grad_norm": 1.2305563688278198, | |
| "learning_rate": 1.9098647210822315e-05, | |
| "loss": 0.361, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 6.583447332421341, | |
| "grad_norm": 0.1572732776403427, | |
| "learning_rate": 1.9003647970816235e-05, | |
| "loss": 0.3816, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 6.600547195622435, | |
| "grad_norm": 8.715572357177734, | |
| "learning_rate": 1.8908648730810156e-05, | |
| "loss": 0.3678, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 6.617647058823529, | |
| "grad_norm": 4.759888172149658, | |
| "learning_rate": 1.8813649490804076e-05, | |
| "loss": 0.3702, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 6.634746922024624, | |
| "grad_norm": 3.071633815765381, | |
| "learning_rate": 1.8718650250797996e-05, | |
| "loss": 0.3771, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 6.651846785225718, | |
| "grad_norm": 13.45046615600586, | |
| "learning_rate": 1.8623651010791917e-05, | |
| "loss": 0.5068, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 6.668946648426813, | |
| "grad_norm": 15.248006820678711, | |
| "learning_rate": 1.8528651770785833e-05, | |
| "loss": 0.3403, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 6.686046511627907, | |
| "grad_norm": 2.1745738983154297, | |
| "learning_rate": 1.8433652530779754e-05, | |
| "loss": 0.3092, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 6.703146374829001, | |
| "grad_norm": 10.927750587463379, | |
| "learning_rate": 1.8338653290773674e-05, | |
| "loss": 0.3776, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 6.720246238030096, | |
| "grad_norm": 24.628570556640625, | |
| "learning_rate": 1.8243654050767594e-05, | |
| "loss": 0.4011, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 6.73734610123119, | |
| "grad_norm": 1.9271844625473022, | |
| "learning_rate": 1.8148654810761515e-05, | |
| "loss": 0.3472, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 6.754445964432285, | |
| "grad_norm": 11.969741821289062, | |
| "learning_rate": 1.8053655570755435e-05, | |
| "loss": 0.3599, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 6.771545827633379, | |
| "grad_norm": 10.276418685913086, | |
| "learning_rate": 1.7958656330749355e-05, | |
| "loss": 0.3054, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 6.788645690834473, | |
| "grad_norm": 1.4558168649673462, | |
| "learning_rate": 1.7863657090743275e-05, | |
| "loss": 0.363, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 6.805745554035568, | |
| "grad_norm": 6.96836519241333, | |
| "learning_rate": 1.7768657850737196e-05, | |
| "loss": 0.4066, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 6.822845417236662, | |
| "grad_norm": 11.515777587890625, | |
| "learning_rate": 1.7673658610731113e-05, | |
| "loss": 0.3551, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 6.839945280437757, | |
| "grad_norm": 13.478057861328125, | |
| "learning_rate": 1.7578659370725033e-05, | |
| "loss": 0.3799, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 6.857045143638851, | |
| "grad_norm": 11.620177268981934, | |
| "learning_rate": 1.7483660130718953e-05, | |
| "loss": 0.3375, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 6.874145006839945, | |
| "grad_norm": 3.876800537109375, | |
| "learning_rate": 1.7388660890712873e-05, | |
| "loss": 0.5015, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 6.89124487004104, | |
| "grad_norm": 1.9693479537963867, | |
| "learning_rate": 1.7293661650706794e-05, | |
| "loss": 0.2689, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 6.908344733242134, | |
| "grad_norm": 14.633630752563477, | |
| "learning_rate": 1.7198662410700717e-05, | |
| "loss": 0.3271, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 6.925444596443229, | |
| "grad_norm": 4.947651386260986, | |
| "learning_rate": 1.7103663170694638e-05, | |
| "loss": 0.434, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 6.942544459644322, | |
| "grad_norm": 2.0154216289520264, | |
| "learning_rate": 1.7008663930688558e-05, | |
| "loss": 0.4467, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 6.959644322845417, | |
| "grad_norm": 14.244186401367188, | |
| "learning_rate": 1.6913664690682475e-05, | |
| "loss": 0.3106, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 6.976744186046512, | |
| "grad_norm": 4.09621000289917, | |
| "learning_rate": 1.6818665450676395e-05, | |
| "loss": 0.3063, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 6.993844049247606, | |
| "grad_norm": 2.0784831047058105, | |
| "learning_rate": 1.6723666210670315e-05, | |
| "loss": 0.3515, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9238648002731308, | |
| "eval_f1_macro": 0.8426039748613977, | |
| "eval_f1_micro": 0.9238648002731308, | |
| "eval_f1_weighted": 0.9239964103657616, | |
| "eval_loss": 0.22421783208847046, | |
| "eval_precision_macro": 0.9018755264534066, | |
| "eval_precision_micro": 0.9238648002731308, | |
| "eval_precision_weighted": 0.9296443693001667, | |
| "eval_recall_macro": 0.8036953584993286, | |
| "eval_recall_micro": 0.9238648002731308, | |
| "eval_recall_weighted": 0.9238648002731308, | |
| "eval_runtime": 25.5062, | |
| "eval_samples_per_second": 114.835, | |
| "eval_steps_per_second": 7.214, | |
| "step": 10234 | |
| }, | |
| { | |
| "epoch": 7.010943912448701, | |
| "grad_norm": 4.232228755950928, | |
| "learning_rate": 1.6628666970664236e-05, | |
| "loss": 0.4255, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 7.028043775649794, | |
| "grad_norm": 12.369464874267578, | |
| "learning_rate": 1.6533667730658156e-05, | |
| "loss": 0.3828, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 7.045143638850889, | |
| "grad_norm": 8.454262733459473, | |
| "learning_rate": 1.6438668490652076e-05, | |
| "loss": 0.287, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 7.062243502051984, | |
| "grad_norm": 2.064643144607544, | |
| "learning_rate": 1.6343669250645996e-05, | |
| "loss": 0.3688, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 7.079343365253078, | |
| "grad_norm": 22.918745040893555, | |
| "learning_rate": 1.6248670010639917e-05, | |
| "loss": 0.3487, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 7.096443228454173, | |
| "grad_norm": 3.0556788444519043, | |
| "learning_rate": 1.6153670770633837e-05, | |
| "loss": 0.2822, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 7.113543091655266, | |
| "grad_norm": 11.955765724182129, | |
| "learning_rate": 1.6058671530627757e-05, | |
| "loss": 0.4273, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 7.130642954856361, | |
| "grad_norm": 13.998736381530762, | |
| "learning_rate": 1.5963672290621674e-05, | |
| "loss": 0.3034, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 7.147742818057456, | |
| "grad_norm": 3.8617422580718994, | |
| "learning_rate": 1.5868673050615594e-05, | |
| "loss": 0.3679, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 7.16484268125855, | |
| "grad_norm": 1.1045358180999756, | |
| "learning_rate": 1.5773673810609515e-05, | |
| "loss": 0.4234, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 7.181942544459645, | |
| "grad_norm": 5.257449626922607, | |
| "learning_rate": 1.5678674570603435e-05, | |
| "loss": 0.3665, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 7.199042407660738, | |
| "grad_norm": 7.530872821807861, | |
| "learning_rate": 1.5583675330597355e-05, | |
| "loss": 0.4602, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 7.216142270861833, | |
| "grad_norm": 0.4516567885875702, | |
| "learning_rate": 1.5488676090591275e-05, | |
| "loss": 0.3864, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 7.233242134062928, | |
| "grad_norm": 22.563528060913086, | |
| "learning_rate": 1.5393676850585196e-05, | |
| "loss": 0.3435, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 7.250341997264022, | |
| "grad_norm": 13.331972122192383, | |
| "learning_rate": 1.5298677610579116e-05, | |
| "loss": 0.2879, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 7.267441860465116, | |
| "grad_norm": 9.387733459472656, | |
| "learning_rate": 1.5203678370573035e-05, | |
| "loss": 0.3333, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 7.2845417236662104, | |
| "grad_norm": 7.556676864624023, | |
| "learning_rate": 1.5108679130566955e-05, | |
| "loss": 0.4042, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 7.301641586867305, | |
| "grad_norm": 6.794082164764404, | |
| "learning_rate": 1.5013679890560875e-05, | |
| "loss": 0.4062, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 7.3187414500684, | |
| "grad_norm": 2.1838858127593994, | |
| "learning_rate": 1.4918680650554795e-05, | |
| "loss": 0.3238, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 7.335841313269494, | |
| "grad_norm": 10.469732284545898, | |
| "learning_rate": 1.4823681410548717e-05, | |
| "loss": 0.4204, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 7.352941176470588, | |
| "grad_norm": 2.8655450344085693, | |
| "learning_rate": 1.4728682170542638e-05, | |
| "loss": 0.3095, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 7.3700410396716824, | |
| "grad_norm": 7.8528337478637695, | |
| "learning_rate": 1.4633682930536558e-05, | |
| "loss": 0.2917, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 7.387140902872777, | |
| "grad_norm": 1.5195367336273193, | |
| "learning_rate": 1.4538683690530477e-05, | |
| "loss": 0.2933, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 7.404240766073872, | |
| "grad_norm": 12.097100257873535, | |
| "learning_rate": 1.4443684450524397e-05, | |
| "loss": 0.4429, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 7.421340629274966, | |
| "grad_norm": 16.263513565063477, | |
| "learning_rate": 1.4348685210518317e-05, | |
| "loss": 0.3024, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 7.43844049247606, | |
| "grad_norm": 1.629339337348938, | |
| "learning_rate": 1.4253685970512237e-05, | |
| "loss": 0.2379, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 7.4555403556771545, | |
| "grad_norm": 0.13442982733249664, | |
| "learning_rate": 1.4158686730506158e-05, | |
| "loss": 0.3297, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 7.472640218878249, | |
| "grad_norm": 22.14508628845215, | |
| "learning_rate": 1.4063687490500076e-05, | |
| "loss": 0.3236, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 7.489740082079344, | |
| "grad_norm": 0.142426535487175, | |
| "learning_rate": 1.3968688250493997e-05, | |
| "loss": 0.3527, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 7.506839945280438, | |
| "grad_norm": 0.38898319005966187, | |
| "learning_rate": 1.3873689010487917e-05, | |
| "loss": 0.3389, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 7.523939808481532, | |
| "grad_norm": 0.7467881441116333, | |
| "learning_rate": 1.3778689770481837e-05, | |
| "loss": 0.3385, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 7.5410396716826265, | |
| "grad_norm": 12.254922866821289, | |
| "learning_rate": 1.3683690530475757e-05, | |
| "loss": 0.4477, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 7.558139534883721, | |
| "grad_norm": 8.779296875, | |
| "learning_rate": 1.3588691290469676e-05, | |
| "loss": 0.3468, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 7.575239398084816, | |
| "grad_norm": 39.5240478515625, | |
| "learning_rate": 1.3493692050463596e-05, | |
| "loss": 0.3673, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 7.592339261285909, | |
| "grad_norm": 11.708680152893066, | |
| "learning_rate": 1.3398692810457516e-05, | |
| "loss": 0.3424, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 7.609439124487004, | |
| "grad_norm": 4.1340131759643555, | |
| "learning_rate": 1.3303693570451437e-05, | |
| "loss": 0.3234, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 7.6265389876880985, | |
| "grad_norm": 34.217979431152344, | |
| "learning_rate": 1.3208694330445357e-05, | |
| "loss": 0.3146, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 7.643638850889193, | |
| "grad_norm": 2.088963508605957, | |
| "learning_rate": 1.3113695090439276e-05, | |
| "loss": 0.3924, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 7.660738714090288, | |
| "grad_norm": 14.35055160522461, | |
| "learning_rate": 1.3018695850433196e-05, | |
| "loss": 0.3071, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 7.677838577291381, | |
| "grad_norm": 7.640285491943359, | |
| "learning_rate": 1.2923696610427116e-05, | |
| "loss": 0.3105, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 7.694938440492476, | |
| "grad_norm": 10.794476509094238, | |
| "learning_rate": 1.283249734002128e-05, | |
| "loss": 0.3436, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 7.7120383036935705, | |
| "grad_norm": 21.634641647338867, | |
| "learning_rate": 1.27374981000152e-05, | |
| "loss": 0.2365, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 7.729138166894665, | |
| "grad_norm": 1.7207705974578857, | |
| "learning_rate": 1.2642498860009118e-05, | |
| "loss": 0.3978, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 7.74623803009576, | |
| "grad_norm": 2.8081226348876953, | |
| "learning_rate": 1.2547499620003039e-05, | |
| "loss": 0.2167, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 7.763337893296853, | |
| "grad_norm": 3.3401083946228027, | |
| "learning_rate": 1.245250037999696e-05, | |
| "loss": 0.3313, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 7.780437756497948, | |
| "grad_norm": 24.61179542541504, | |
| "learning_rate": 1.2357501139990881e-05, | |
| "loss": 0.4088, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 7.7975376196990425, | |
| "grad_norm": 3.2282018661499023, | |
| "learning_rate": 1.22625018999848e-05, | |
| "loss": 0.3965, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 7.814637482900137, | |
| "grad_norm": 5.257138252258301, | |
| "learning_rate": 1.216750265997872e-05, | |
| "loss": 0.275, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 7.831737346101232, | |
| "grad_norm": 1.5772024393081665, | |
| "learning_rate": 1.207250341997264e-05, | |
| "loss": 0.1914, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 7.848837209302325, | |
| "grad_norm": 7.5829033851623535, | |
| "learning_rate": 1.197750417996656e-05, | |
| "loss": 0.5063, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 7.86593707250342, | |
| "grad_norm": 3.2422189712524414, | |
| "learning_rate": 1.188250493996048e-05, | |
| "loss": 0.3235, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 7.8830369357045145, | |
| "grad_norm": 2.9335296154022217, | |
| "learning_rate": 1.1787505699954401e-05, | |
| "loss": 0.2887, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 7.900136798905609, | |
| "grad_norm": 0.015060103498399258, | |
| "learning_rate": 1.1692506459948321e-05, | |
| "loss": 0.4049, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 7.917236662106703, | |
| "grad_norm": 9.77502155303955, | |
| "learning_rate": 1.1597507219942242e-05, | |
| "loss": 0.2858, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 7.934336525307797, | |
| "grad_norm": 1.4075678586959839, | |
| "learning_rate": 1.1502507979936162e-05, | |
| "loss": 0.3153, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 7.951436388508892, | |
| "grad_norm": 12.93342399597168, | |
| "learning_rate": 1.140750873993008e-05, | |
| "loss": 0.3199, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 7.9685362517099865, | |
| "grad_norm": 19.707088470458984, | |
| "learning_rate": 1.1312509499924e-05, | |
| "loss": 0.5197, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 7.985636114911081, | |
| "grad_norm": 2.147141695022583, | |
| "learning_rate": 1.1217510259917921e-05, | |
| "loss": 0.338, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9320587231136906, | |
| "eval_f1_macro": 0.8801923417646881, | |
| "eval_f1_micro": 0.9320587231136906, | |
| "eval_f1_weighted": 0.9322151264732859, | |
| "eval_loss": 0.21217399835586548, | |
| "eval_precision_macro": 0.9267115227700036, | |
| "eval_precision_micro": 0.9320587231136906, | |
| "eval_precision_weighted": 0.9357267323781668, | |
| "eval_recall_macro": 0.8522160392320227, | |
| "eval_recall_micro": 0.9320587231136906, | |
| "eval_recall_weighted": 0.9320587231136906, | |
| "eval_runtime": 25.9574, | |
| "eval_samples_per_second": 112.839, | |
| "eval_steps_per_second": 7.089, | |
| "step": 11696 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 14620, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 9.590567882317627e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |