| { | |
| "best_metric": 1.1938023567199707, | |
| "best_model_checkpoint": "autotrain-v2v7o-9tu3d/checkpoint-1340", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 1340, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.018656716417910446, | |
| "grad_norm": 3.044177532196045, | |
| "learning_rate": 4.6641791044776116e-07, | |
| "loss": 3.5829, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03731343283582089, | |
| "grad_norm": 3.1452090740203857, | |
| "learning_rate": 9.328358208955223e-07, | |
| "loss": 3.5707, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.055970149253731345, | |
| "grad_norm": 3.5051560401916504, | |
| "learning_rate": 1.3992537313432837e-06, | |
| "loss": 3.5587, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07462686567164178, | |
| "grad_norm": 3.155848741531372, | |
| "learning_rate": 1.8656716417910446e-06, | |
| "loss": 3.5653, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09328358208955224, | |
| "grad_norm": 3.025758743286133, | |
| "learning_rate": 2.3320895522388064e-06, | |
| "loss": 3.5639, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11194029850746269, | |
| "grad_norm": 2.966825485229492, | |
| "learning_rate": 2.7985074626865674e-06, | |
| "loss": 3.5187, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13059701492537312, | |
| "grad_norm": 2.870944023132324, | |
| "learning_rate": 3.2649253731343283e-06, | |
| "loss": 3.5382, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14925373134328357, | |
| "grad_norm": 3.2777228355407715, | |
| "learning_rate": 3.7313432835820893e-06, | |
| "loss": 3.5311, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16791044776119404, | |
| "grad_norm": 2.8916268348693848, | |
| "learning_rate": 4.1977611940298515e-06, | |
| "loss": 3.5348, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1865671641791045, | |
| "grad_norm": 2.667680263519287, | |
| "learning_rate": 4.664179104477613e-06, | |
| "loss": 3.5209, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.20522388059701493, | |
| "grad_norm": 2.6136128902435303, | |
| "learning_rate": 5.130597014925373e-06, | |
| "loss": 3.5293, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22388059701492538, | |
| "grad_norm": 3.2311267852783203, | |
| "learning_rate": 5.597014925373135e-06, | |
| "loss": 3.5536, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24253731343283583, | |
| "grad_norm": 2.513153076171875, | |
| "learning_rate": 6.063432835820896e-06, | |
| "loss": 3.5286, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26119402985074625, | |
| "grad_norm": 2.53420090675354, | |
| "learning_rate": 6.529850746268657e-06, | |
| "loss": 3.5305, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2798507462686567, | |
| "grad_norm": 2.7235593795776367, | |
| "learning_rate": 6.996268656716418e-06, | |
| "loss": 3.5244, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 2.621687889099121, | |
| "learning_rate": 7.4626865671641785e-06, | |
| "loss": 3.4981, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.31716417910447764, | |
| "grad_norm": 2.5116870403289795, | |
| "learning_rate": 7.92910447761194e-06, | |
| "loss": 3.5208, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3358208955223881, | |
| "grad_norm": 2.6995580196380615, | |
| "learning_rate": 8.395522388059703e-06, | |
| "loss": 3.5069, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.35447761194029853, | |
| "grad_norm": 2.7225232124328613, | |
| "learning_rate": 8.861940298507463e-06, | |
| "loss": 3.517, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.373134328358209, | |
| "grad_norm": 2.9143073558807373, | |
| "learning_rate": 9.328358208955226e-06, | |
| "loss": 3.4939, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3917910447761194, | |
| "grad_norm": 2.6115267276763916, | |
| "learning_rate": 9.794776119402986e-06, | |
| "loss": 3.4971, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.41044776119402987, | |
| "grad_norm": 2.9028003215789795, | |
| "learning_rate": 1.0261194029850747e-05, | |
| "loss": 3.5239, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4291044776119403, | |
| "grad_norm": 2.7557551860809326, | |
| "learning_rate": 1.0727611940298509e-05, | |
| "loss": 3.4567, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.44776119402985076, | |
| "grad_norm": 3.2114148139953613, | |
| "learning_rate": 1.119402985074627e-05, | |
| "loss": 3.5006, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.4664179104477612, | |
| "grad_norm": 3.3313052654266357, | |
| "learning_rate": 1.166044776119403e-05, | |
| "loss": 3.4342, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.48507462686567165, | |
| "grad_norm": 3.8037962913513184, | |
| "learning_rate": 1.2126865671641792e-05, | |
| "loss": 3.5062, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.503731343283582, | |
| "grad_norm": 3.6135289669036865, | |
| "learning_rate": 1.2593283582089551e-05, | |
| "loss": 3.4769, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5223880597014925, | |
| "grad_norm": 3.9982452392578125, | |
| "learning_rate": 1.3059701492537313e-05, | |
| "loss": 3.5184, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5410447761194029, | |
| "grad_norm": 4.104573726654053, | |
| "learning_rate": 1.3526119402985074e-05, | |
| "loss": 3.4843, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5597014925373134, | |
| "grad_norm": 4.863430976867676, | |
| "learning_rate": 1.3992537313432836e-05, | |
| "loss": 3.4106, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5783582089552238, | |
| "grad_norm": 4.079490661621094, | |
| "learning_rate": 1.4458955223880596e-05, | |
| "loss": 3.4743, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 3.870924949645996, | |
| "learning_rate": 1.4925373134328357e-05, | |
| "loss": 3.3934, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6156716417910447, | |
| "grad_norm": 3.8417205810546875, | |
| "learning_rate": 1.539179104477612e-05, | |
| "loss": 3.4258, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6343283582089553, | |
| "grad_norm": 4.541757583618164, | |
| "learning_rate": 1.585820895522388e-05, | |
| "loss": 3.4659, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6529850746268657, | |
| "grad_norm": 4.8343424797058105, | |
| "learning_rate": 1.6324626865671644e-05, | |
| "loss": 3.4103, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6716417910447762, | |
| "grad_norm": 5.048968315124512, | |
| "learning_rate": 1.6791044776119406e-05, | |
| "loss": 3.459, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6902985074626866, | |
| "grad_norm": 4.90827751159668, | |
| "learning_rate": 1.7257462686567165e-05, | |
| "loss": 3.4152, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7089552238805971, | |
| "grad_norm": 4.624119758605957, | |
| "learning_rate": 1.7723880597014927e-05, | |
| "loss": 3.4129, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7276119402985075, | |
| "grad_norm": 4.981058120727539, | |
| "learning_rate": 1.819029850746269e-05, | |
| "loss": 3.4204, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.746268656716418, | |
| "grad_norm": 4.767482280731201, | |
| "learning_rate": 1.865671641791045e-05, | |
| "loss": 3.3169, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.7649253731343284, | |
| "grad_norm": 4.601747512817383, | |
| "learning_rate": 1.912313432835821e-05, | |
| "loss": 3.278, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.7835820895522388, | |
| "grad_norm": 6.704329013824463, | |
| "learning_rate": 1.9589552238805972e-05, | |
| "loss": 3.3873, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8022388059701493, | |
| "grad_norm": 5.003252983093262, | |
| "learning_rate": 2.0055970149253735e-05, | |
| "loss": 3.3281, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8208955223880597, | |
| "grad_norm": 5.126171588897705, | |
| "learning_rate": 2.0522388059701493e-05, | |
| "loss": 3.2048, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8395522388059702, | |
| "grad_norm": 5.796511173248291, | |
| "learning_rate": 2.0988805970149256e-05, | |
| "loss": 3.2639, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.8582089552238806, | |
| "grad_norm": 5.77567720413208, | |
| "learning_rate": 2.1455223880597018e-05, | |
| "loss": 3.2442, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.8768656716417911, | |
| "grad_norm": 5.419305324554443, | |
| "learning_rate": 2.1921641791044777e-05, | |
| "loss": 3.123, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.8955223880597015, | |
| "grad_norm": 5.75105619430542, | |
| "learning_rate": 2.238805970149254e-05, | |
| "loss": 3.0977, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.914179104477612, | |
| "grad_norm": 8.254322052001953, | |
| "learning_rate": 2.28544776119403e-05, | |
| "loss": 3.086, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.9328358208955224, | |
| "grad_norm": 6.343637943267822, | |
| "learning_rate": 2.332089552238806e-05, | |
| "loss": 3.3336, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.9514925373134329, | |
| "grad_norm": 5.593357563018799, | |
| "learning_rate": 2.3787313432835822e-05, | |
| "loss": 3.2817, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.9701492537313433, | |
| "grad_norm": 8.842999458312988, | |
| "learning_rate": 2.4253731343283584e-05, | |
| "loss": 3.1676, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.9888059701492538, | |
| "grad_norm": 8.3274507522583, | |
| "learning_rate": 2.4720149253731347e-05, | |
| "loss": 3.1462, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.155, | |
| "eval_f1_macro": 0.09100175187131708, | |
| "eval_f1_micro": 0.155, | |
| "eval_f1_weighted": 0.08955112279025322, | |
| "eval_loss": 3.2174994945526123, | |
| "eval_precision_macro": 0.07134410305142012, | |
| "eval_precision_micro": 0.155, | |
| "eval_precision_weighted": 0.07162192482314433, | |
| "eval_recall_macro": 0.16519607843137257, | |
| "eval_recall_micro": 0.155, | |
| "eval_recall_weighted": 0.155, | |
| "eval_runtime": 16.5066, | |
| "eval_samples_per_second": 12.116, | |
| "eval_steps_per_second": 0.788, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.007462686567164, | |
| "grad_norm": 6.375778675079346, | |
| "learning_rate": 2.5186567164179102e-05, | |
| "loss": 3.0937, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.0261194029850746, | |
| "grad_norm": 9.297365188598633, | |
| "learning_rate": 2.5652985074626868e-05, | |
| "loss": 3.0123, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.044776119402985, | |
| "grad_norm": 6.8317646980285645, | |
| "learning_rate": 2.6119402985074626e-05, | |
| "loss": 3.0876, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.0634328358208955, | |
| "grad_norm": 6.641528606414795, | |
| "learning_rate": 2.658582089552239e-05, | |
| "loss": 2.9694, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.0820895522388059, | |
| "grad_norm": 6.846834659576416, | |
| "learning_rate": 2.7052238805970147e-05, | |
| "loss": 3.0272, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.1007462686567164, | |
| "grad_norm": 6.250601291656494, | |
| "learning_rate": 2.7518656716417913e-05, | |
| "loss": 2.9646, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.1194029850746268, | |
| "grad_norm": 6.11965274810791, | |
| "learning_rate": 2.7985074626865672e-05, | |
| "loss": 2.8305, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.1380597014925373, | |
| "grad_norm": 7.502304553985596, | |
| "learning_rate": 2.8451492537313434e-05, | |
| "loss": 2.7931, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.1567164179104479, | |
| "grad_norm": 6.00736665725708, | |
| "learning_rate": 2.8917910447761193e-05, | |
| "loss": 2.8108, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.1753731343283582, | |
| "grad_norm": 7.524813175201416, | |
| "learning_rate": 2.9384328358208955e-05, | |
| "loss": 2.8187, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 7.996919631958008, | |
| "learning_rate": 2.9850746268656714e-05, | |
| "loss": 2.8544, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.212686567164179, | |
| "grad_norm": 7.103128910064697, | |
| "learning_rate": 3.031716417910448e-05, | |
| "loss": 2.8545, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.2313432835820897, | |
| "grad_norm": 5.705358505249023, | |
| "learning_rate": 3.078358208955224e-05, | |
| "loss": 2.5292, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 6.714156150817871, | |
| "learning_rate": 3.125e-05, | |
| "loss": 2.5716, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.2686567164179103, | |
| "grad_norm": 7.388309955596924, | |
| "learning_rate": 3.171641791044776e-05, | |
| "loss": 2.8626, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.287313432835821, | |
| "grad_norm": 6.7939934730529785, | |
| "learning_rate": 3.2182835820895525e-05, | |
| "loss": 2.7215, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.3059701492537314, | |
| "grad_norm": 6.774124622344971, | |
| "learning_rate": 3.264925373134329e-05, | |
| "loss": 2.7824, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.3246268656716418, | |
| "grad_norm": 7.475742340087891, | |
| "learning_rate": 3.311567164179105e-05, | |
| "loss": 2.7855, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.3432835820895521, | |
| "grad_norm": 6.67882776260376, | |
| "learning_rate": 3.358208955223881e-05, | |
| "loss": 2.7686, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.3619402985074627, | |
| "grad_norm": 6.921255111694336, | |
| "learning_rate": 3.404850746268657e-05, | |
| "loss": 2.6471, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.3805970149253732, | |
| "grad_norm": 5.301546573638916, | |
| "learning_rate": 3.451492537313433e-05, | |
| "loss": 2.343, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.3992537313432836, | |
| "grad_norm": 8.802115440368652, | |
| "learning_rate": 3.498134328358209e-05, | |
| "loss": 2.6459, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.417910447761194, | |
| "grad_norm": 10.76266860961914, | |
| "learning_rate": 3.5447761194029854e-05, | |
| "loss": 2.245, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.4365671641791045, | |
| "grad_norm": 6.216732978820801, | |
| "learning_rate": 3.5914179104477616e-05, | |
| "loss": 2.1281, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.455223880597015, | |
| "grad_norm": 7.1129231452941895, | |
| "learning_rate": 3.638059701492538e-05, | |
| "loss": 2.6988, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.4738805970149254, | |
| "grad_norm": 6.50852108001709, | |
| "learning_rate": 3.6847014925373134e-05, | |
| "loss": 2.3069, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.4925373134328357, | |
| "grad_norm": 7.611819744110107, | |
| "learning_rate": 3.73134328358209e-05, | |
| "loss": 2.3428, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.5111940298507462, | |
| "grad_norm": 8.59446907043457, | |
| "learning_rate": 3.777985074626866e-05, | |
| "loss": 2.278, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.5298507462686568, | |
| "grad_norm": 9.75235652923584, | |
| "learning_rate": 3.824626865671642e-05, | |
| "loss": 2.1313, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.5485074626865671, | |
| "grad_norm": 7.297292709350586, | |
| "learning_rate": 3.871268656716418e-05, | |
| "loss": 2.2863, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.5671641791044775, | |
| "grad_norm": 9.046923637390137, | |
| "learning_rate": 3.9179104477611945e-05, | |
| "loss": 2.5253, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.585820895522388, | |
| "grad_norm": 7.503407955169678, | |
| "learning_rate": 3.96455223880597e-05, | |
| "loss": 2.2857, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.6044776119402986, | |
| "grad_norm": 7.656423568725586, | |
| "learning_rate": 4.011194029850747e-05, | |
| "loss": 2.5456, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.623134328358209, | |
| "grad_norm": 7.632299423217773, | |
| "learning_rate": 4.0578358208955225e-05, | |
| "loss": 2.6645, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.6417910447761193, | |
| "grad_norm": 8.021852493286133, | |
| "learning_rate": 4.104477611940299e-05, | |
| "loss": 2.1135, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.6604477611940298, | |
| "grad_norm": 10.39423656463623, | |
| "learning_rate": 4.151119402985075e-05, | |
| "loss": 2.1937, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.6791044776119404, | |
| "grad_norm": 12.216695785522461, | |
| "learning_rate": 4.197761194029851e-05, | |
| "loss": 1.9825, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.6977611940298507, | |
| "grad_norm": 7.088036060333252, | |
| "learning_rate": 4.244402985074627e-05, | |
| "loss": 1.7742, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.716417910447761, | |
| "grad_norm": 7.268665313720703, | |
| "learning_rate": 4.2910447761194036e-05, | |
| "loss": 2.1474, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.7350746268656716, | |
| "grad_norm": 9.286911010742188, | |
| "learning_rate": 4.337686567164179e-05, | |
| "loss": 2.163, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.7537313432835822, | |
| "grad_norm": 10.1356782913208, | |
| "learning_rate": 4.384328358208955e-05, | |
| "loss": 2.0548, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.7723880597014925, | |
| "grad_norm": 10.506839752197266, | |
| "learning_rate": 4.4309701492537316e-05, | |
| "loss": 2.0674, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.7910447761194028, | |
| "grad_norm": 7.393703937530518, | |
| "learning_rate": 4.477611940298508e-05, | |
| "loss": 2.1359, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.8097014925373134, | |
| "grad_norm": 8.782050132751465, | |
| "learning_rate": 4.524253731343284e-05, | |
| "loss": 2.0858, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.828358208955224, | |
| "grad_norm": 8.55767822265625, | |
| "learning_rate": 4.57089552238806e-05, | |
| "loss": 2.1815, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.8470149253731343, | |
| "grad_norm": 6.715104579925537, | |
| "learning_rate": 4.617537313432836e-05, | |
| "loss": 2.1025, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.8656716417910446, | |
| "grad_norm": 10.000481605529785, | |
| "learning_rate": 4.664179104477612e-05, | |
| "loss": 1.9817, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.8843283582089554, | |
| "grad_norm": 7.614371299743652, | |
| "learning_rate": 4.710820895522388e-05, | |
| "loss": 2.2921, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.9029850746268657, | |
| "grad_norm": 7.8831658363342285, | |
| "learning_rate": 4.7574626865671644e-05, | |
| "loss": 1.6008, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.921641791044776, | |
| "grad_norm": 8.811071395874023, | |
| "learning_rate": 4.8041044776119407e-05, | |
| "loss": 2.131, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.9402985074626866, | |
| "grad_norm": 8.217741966247559, | |
| "learning_rate": 4.850746268656717e-05, | |
| "loss": 2.0112, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.9589552238805972, | |
| "grad_norm": 8.56834602355957, | |
| "learning_rate": 4.8973880597014924e-05, | |
| "loss": 1.8353, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.9776119402985075, | |
| "grad_norm": 12.720284461975098, | |
| "learning_rate": 4.944029850746269e-05, | |
| "loss": 1.5857, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.9962686567164178, | |
| "grad_norm": 10.004925727844238, | |
| "learning_rate": 4.990671641791045e-05, | |
| "loss": 2.0364, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.425, | |
| "eval_f1_macro": 0.34724867475550414, | |
| "eval_f1_micro": 0.425, | |
| "eval_f1_weighted": 0.35814368775359484, | |
| "eval_loss": 2.165677785873413, | |
| "eval_precision_macro": 0.36154654169360045, | |
| "eval_precision_micro": 0.425, | |
| "eval_precision_weighted": 0.3759306318681318, | |
| "eval_recall_macro": 0.423249299719888, | |
| "eval_recall_micro": 0.425, | |
| "eval_recall_weighted": 0.425, | |
| "eval_runtime": 22.9282, | |
| "eval_samples_per_second": 8.723, | |
| "eval_steps_per_second": 0.567, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.014925373134328, | |
| "grad_norm": 7.278439044952393, | |
| "learning_rate": 4.995854063018242e-05, | |
| "loss": 1.297, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.033582089552239, | |
| "grad_norm": 13.718267440795898, | |
| "learning_rate": 4.990671641791045e-05, | |
| "loss": 1.9014, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.0522388059701493, | |
| "grad_norm": 11.233081817626953, | |
| "learning_rate": 4.985489220563848e-05, | |
| "loss": 1.9761, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0708955223880596, | |
| "grad_norm": 11.126688003540039, | |
| "learning_rate": 4.98030679933665e-05, | |
| "loss": 1.7989, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.08955223880597, | |
| "grad_norm": 10.091024398803711, | |
| "learning_rate": 4.975124378109453e-05, | |
| "loss": 1.4646, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.1082089552238807, | |
| "grad_norm": 8.601826667785645, | |
| "learning_rate": 4.9699419568822556e-05, | |
| "loss": 1.8275, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.126865671641791, | |
| "grad_norm": 12.649139404296875, | |
| "learning_rate": 4.964759535655058e-05, | |
| "loss": 1.5706, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.1455223880597014, | |
| "grad_norm": 12.75953483581543, | |
| "learning_rate": 4.959577114427861e-05, | |
| "loss": 1.8285, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.1641791044776117, | |
| "grad_norm": 9.071000099182129, | |
| "learning_rate": 4.954394693200663e-05, | |
| "loss": 1.6905, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.1828358208955225, | |
| "grad_norm": 9.372791290283203, | |
| "learning_rate": 4.949212271973466e-05, | |
| "loss": 1.8672, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.201492537313433, | |
| "grad_norm": 9.870670318603516, | |
| "learning_rate": 4.944029850746269e-05, | |
| "loss": 1.6721, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.220149253731343, | |
| "grad_norm": 7.243556976318359, | |
| "learning_rate": 4.9388474295190717e-05, | |
| "loss": 1.6255, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.2388059701492535, | |
| "grad_norm": 5.97638463973999, | |
| "learning_rate": 4.933665008291874e-05, | |
| "loss": 1.6361, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.2574626865671643, | |
| "grad_norm": 10.749024391174316, | |
| "learning_rate": 4.928482587064677e-05, | |
| "loss": 1.8244, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.2761194029850746, | |
| "grad_norm": 5.7718186378479, | |
| "learning_rate": 4.9233001658374794e-05, | |
| "loss": 1.2545, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.294776119402985, | |
| "grad_norm": 11.37286376953125, | |
| "learning_rate": 4.9181177446102824e-05, | |
| "loss": 1.7528, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.3134328358208958, | |
| "grad_norm": 10.693103790283203, | |
| "learning_rate": 4.912935323383085e-05, | |
| "loss": 1.6237, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.332089552238806, | |
| "grad_norm": 8.764638900756836, | |
| "learning_rate": 4.907752902155888e-05, | |
| "loss": 1.6663, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.3507462686567164, | |
| "grad_norm": 6.239055156707764, | |
| "learning_rate": 4.90257048092869e-05, | |
| "loss": 1.473, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.3694029850746268, | |
| "grad_norm": 12.444618225097656, | |
| "learning_rate": 4.8973880597014924e-05, | |
| "loss": 1.6951, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.388059701492537, | |
| "grad_norm": 10.495733261108398, | |
| "learning_rate": 4.8922056384742954e-05, | |
| "loss": 1.362, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.406716417910448, | |
| "grad_norm": 6.903144836425781, | |
| "learning_rate": 4.887023217247098e-05, | |
| "loss": 1.3483, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.425373134328358, | |
| "grad_norm": 8.9871826171875, | |
| "learning_rate": 4.881840796019901e-05, | |
| "loss": 1.4329, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.4440298507462686, | |
| "grad_norm": 12.040112495422363, | |
| "learning_rate": 4.876658374792704e-05, | |
| "loss": 1.7239, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.4626865671641793, | |
| "grad_norm": 12.604753494262695, | |
| "learning_rate": 4.8714759535655055e-05, | |
| "loss": 1.0981, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.4813432835820897, | |
| "grad_norm": 8.3511323928833, | |
| "learning_rate": 4.8662935323383085e-05, | |
| "loss": 1.5905, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 6.789451599121094, | |
| "learning_rate": 4.8611111111111115e-05, | |
| "loss": 1.283, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.5186567164179103, | |
| "grad_norm": 6.545684337615967, | |
| "learning_rate": 4.855928689883914e-05, | |
| "loss": 1.3954, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.5373134328358207, | |
| "grad_norm": 11.606388092041016, | |
| "learning_rate": 4.850746268656717e-05, | |
| "loss": 1.6337, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.5559701492537314, | |
| "grad_norm": 6.834081172943115, | |
| "learning_rate": 4.845563847429519e-05, | |
| "loss": 1.0763, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.574626865671642, | |
| "grad_norm": 9.194738388061523, | |
| "learning_rate": 4.8403814262023216e-05, | |
| "loss": 1.3672, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.593283582089552, | |
| "grad_norm": 7.910974025726318, | |
| "learning_rate": 4.8351990049751246e-05, | |
| "loss": 1.2458, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.611940298507463, | |
| "grad_norm": 8.25942611694336, | |
| "learning_rate": 4.830016583747927e-05, | |
| "loss": 1.4986, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.6305970149253732, | |
| "grad_norm": 5.547841548919678, | |
| "learning_rate": 4.82483416252073e-05, | |
| "loss": 1.488, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.6492537313432836, | |
| "grad_norm": 8.00507640838623, | |
| "learning_rate": 4.819651741293533e-05, | |
| "loss": 1.5729, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.667910447761194, | |
| "grad_norm": 10.959121704101562, | |
| "learning_rate": 4.814469320066335e-05, | |
| "loss": 1.393, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.6865671641791042, | |
| "grad_norm": 12.941998481750488, | |
| "learning_rate": 4.8092868988391376e-05, | |
| "loss": 1.6623, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.705223880597015, | |
| "grad_norm": 10.971231460571289, | |
| "learning_rate": 4.8041044776119407e-05, | |
| "loss": 1.2405, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.7238805970149254, | |
| "grad_norm": 6.511988639831543, | |
| "learning_rate": 4.798922056384743e-05, | |
| "loss": 1.4053, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.7425373134328357, | |
| "grad_norm": 16.730091094970703, | |
| "learning_rate": 4.793739635157546e-05, | |
| "loss": 1.6276, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.7611940298507465, | |
| "grad_norm": 8.781760215759277, | |
| "learning_rate": 4.7885572139303484e-05, | |
| "loss": 1.4323, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.779850746268657, | |
| "grad_norm": 8.811301231384277, | |
| "learning_rate": 4.7833747927031514e-05, | |
| "loss": 1.3947, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.798507462686567, | |
| "grad_norm": 6.631250381469727, | |
| "learning_rate": 4.778192371475954e-05, | |
| "loss": 1.3312, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.8171641791044775, | |
| "grad_norm": 13.198403358459473, | |
| "learning_rate": 4.773009950248756e-05, | |
| "loss": 1.3845, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.835820895522388, | |
| "grad_norm": 13.531638145446777, | |
| "learning_rate": 4.767827529021559e-05, | |
| "loss": 1.6808, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.8544776119402986, | |
| "grad_norm": 8.421935081481934, | |
| "learning_rate": 4.762645107794362e-05, | |
| "loss": 1.3985, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.873134328358209, | |
| "grad_norm": 11.863781929016113, | |
| "learning_rate": 4.7574626865671644e-05, | |
| "loss": 1.3896, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.8917910447761193, | |
| "grad_norm": 12.868813514709473, | |
| "learning_rate": 4.7522802653399674e-05, | |
| "loss": 1.5119, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.91044776119403, | |
| "grad_norm": 9.192615509033203, | |
| "learning_rate": 4.74709784411277e-05, | |
| "loss": 1.2569, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.9291044776119404, | |
| "grad_norm": 6.196342945098877, | |
| "learning_rate": 4.741915422885572e-05, | |
| "loss": 1.0811, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.9477611940298507, | |
| "grad_norm": 13.614799499511719, | |
| "learning_rate": 4.736733001658375e-05, | |
| "loss": 1.298, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.966417910447761, | |
| "grad_norm": 8.10826301574707, | |
| "learning_rate": 4.7315505804311775e-05, | |
| "loss": 1.152, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "grad_norm": 15.173196792602539, | |
| "learning_rate": 4.7263681592039805e-05, | |
| "loss": 1.5199, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.545, | |
| "eval_f1_macro": 0.4674843775735156, | |
| "eval_f1_micro": 0.545, | |
| "eval_f1_weighted": 0.4795885366054804, | |
| "eval_loss": 1.6110626459121704, | |
| "eval_precision_macro": 0.48627026568203036, | |
| "eval_precision_micro": 0.545, | |
| "eval_precision_weighted": 0.5079666305916306, | |
| "eval_recall_macro": 0.5418767507002801, | |
| "eval_recall_micro": 0.545, | |
| "eval_recall_weighted": 0.545, | |
| "eval_runtime": 16.2687, | |
| "eval_samples_per_second": 12.294, | |
| "eval_steps_per_second": 0.799, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 3.003731343283582, | |
| "grad_norm": 6.815779209136963, | |
| "learning_rate": 4.7211857379767835e-05, | |
| "loss": 0.8911, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.0223880597014925, | |
| "grad_norm": 8.526626586914062, | |
| "learning_rate": 4.716003316749585e-05, | |
| "loss": 1.2486, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.041044776119403, | |
| "grad_norm": 13.612601280212402, | |
| "learning_rate": 4.710820895522388e-05, | |
| "loss": 1.0458, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.0597014925373136, | |
| "grad_norm": 11.703157424926758, | |
| "learning_rate": 4.705638474295191e-05, | |
| "loss": 1.0918, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.078358208955224, | |
| "grad_norm": 12.684816360473633, | |
| "learning_rate": 4.7004560530679936e-05, | |
| "loss": 1.3883, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.0970149253731343, | |
| "grad_norm": 6.267882823944092, | |
| "learning_rate": 4.6952736318407966e-05, | |
| "loss": 0.951, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.1156716417910446, | |
| "grad_norm": 6.348490238189697, | |
| "learning_rate": 4.690091210613599e-05, | |
| "loss": 1.1257, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.1343283582089554, | |
| "grad_norm": 15.914472579956055, | |
| "learning_rate": 4.684908789386401e-05, | |
| "loss": 1.2884, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.1529850746268657, | |
| "grad_norm": 8.568007469177246, | |
| "learning_rate": 4.679726368159204e-05, | |
| "loss": 1.1261, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.171641791044776, | |
| "grad_norm": 12.23885440826416, | |
| "learning_rate": 4.6745439469320066e-05, | |
| "loss": 1.0198, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.1902985074626864, | |
| "grad_norm": 18.287586212158203, | |
| "learning_rate": 4.6693615257048096e-05, | |
| "loss": 0.9699, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.208955223880597, | |
| "grad_norm": 6.615917682647705, | |
| "learning_rate": 4.664179104477612e-05, | |
| "loss": 1.0768, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.2276119402985075, | |
| "grad_norm": 4.780862808227539, | |
| "learning_rate": 4.658996683250415e-05, | |
| "loss": 0.8213, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.246268656716418, | |
| "grad_norm": 12.68826675415039, | |
| "learning_rate": 4.6538142620232173e-05, | |
| "loss": 1.1507, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.264925373134328, | |
| "grad_norm": 6.729078769683838, | |
| "learning_rate": 4.64863184079602e-05, | |
| "loss": 1.0932, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.283582089552239, | |
| "grad_norm": 7.569777488708496, | |
| "learning_rate": 4.643449419568823e-05, | |
| "loss": 1.0546, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.3022388059701493, | |
| "grad_norm": 11.058488845825195, | |
| "learning_rate": 4.638266998341626e-05, | |
| "loss": 1.1409, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.3208955223880596, | |
| "grad_norm": 3.7507197856903076, | |
| "learning_rate": 4.633084577114428e-05, | |
| "loss": 0.8494, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.33955223880597, | |
| "grad_norm": 7.260571479797363, | |
| "learning_rate": 4.627902155887231e-05, | |
| "loss": 0.9837, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.3582089552238807, | |
| "grad_norm": 10.02383041381836, | |
| "learning_rate": 4.6227197346600334e-05, | |
| "loss": 1.1109, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.376865671641791, | |
| "grad_norm": 13.254716873168945, | |
| "learning_rate": 4.617537313432836e-05, | |
| "loss": 1.1479, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.3955223880597014, | |
| "grad_norm": 10.84252643585205, | |
| "learning_rate": 4.612354892205639e-05, | |
| "loss": 1.5116, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.4141791044776117, | |
| "grad_norm": 9.479592323303223, | |
| "learning_rate": 4.607172470978441e-05, | |
| "loss": 1.3419, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.4328358208955225, | |
| "grad_norm": 11.758341789245605, | |
| "learning_rate": 4.601990049751244e-05, | |
| "loss": 1.0607, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.451492537313433, | |
| "grad_norm": 12.16956901550293, | |
| "learning_rate": 4.596807628524047e-05, | |
| "loss": 1.2986, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.470149253731343, | |
| "grad_norm": 16.24034309387207, | |
| "learning_rate": 4.591625207296849e-05, | |
| "loss": 0.9493, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.4888059701492535, | |
| "grad_norm": 5.806056499481201, | |
| "learning_rate": 4.586442786069652e-05, | |
| "loss": 0.854, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.5074626865671643, | |
| "grad_norm": 31.129661560058594, | |
| "learning_rate": 4.581260364842455e-05, | |
| "loss": 0.9267, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.5261194029850746, | |
| "grad_norm": 10.927783012390137, | |
| "learning_rate": 4.576077943615257e-05, | |
| "loss": 0.9969, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.544776119402985, | |
| "grad_norm": 19.682039260864258, | |
| "learning_rate": 4.57089552238806e-05, | |
| "loss": 0.9038, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.5634328358208958, | |
| "grad_norm": 12.5592679977417, | |
| "learning_rate": 4.5657131011608626e-05, | |
| "loss": 1.1389, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.582089552238806, | |
| "grad_norm": 10.561615943908691, | |
| "learning_rate": 4.560530679933665e-05, | |
| "loss": 0.8216, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.6007462686567164, | |
| "grad_norm": 6.979230880737305, | |
| "learning_rate": 4.555348258706468e-05, | |
| "loss": 0.968, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.6194029850746268, | |
| "grad_norm": 8.147384643554688, | |
| "learning_rate": 4.55016583747927e-05, | |
| "loss": 1.0937, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.638059701492537, | |
| "grad_norm": 14.14340591430664, | |
| "learning_rate": 4.544983416252073e-05, | |
| "loss": 1.1813, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.656716417910448, | |
| "grad_norm": 10.942938804626465, | |
| "learning_rate": 4.539800995024876e-05, | |
| "loss": 0.9468, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.675373134328358, | |
| "grad_norm": 7.3539838790893555, | |
| "learning_rate": 4.5346185737976786e-05, | |
| "loss": 0.7761, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.6940298507462686, | |
| "grad_norm": 5.7556843757629395, | |
| "learning_rate": 4.529436152570481e-05, | |
| "loss": 0.848, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.7126865671641793, | |
| "grad_norm": 7.527775287628174, | |
| "learning_rate": 4.524253731343284e-05, | |
| "loss": 0.9969, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 3.7313432835820897, | |
| "grad_norm": 3.0273733139038086, | |
| "learning_rate": 4.5190713101160863e-05, | |
| "loss": 0.6834, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 4.489946365356445, | |
| "learning_rate": 4.5138888888888894e-05, | |
| "loss": 0.6737, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 3.7686567164179103, | |
| "grad_norm": 14.911623001098633, | |
| "learning_rate": 4.508706467661692e-05, | |
| "loss": 0.7317, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.7873134328358207, | |
| "grad_norm": 18.18294906616211, | |
| "learning_rate": 4.503524046434495e-05, | |
| "loss": 1.1278, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 3.8059701492537314, | |
| "grad_norm": 19.4583797454834, | |
| "learning_rate": 4.498341625207297e-05, | |
| "loss": 1.29, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.824626865671642, | |
| "grad_norm": 5.924612522125244, | |
| "learning_rate": 4.4931592039800994e-05, | |
| "loss": 0.6149, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.843283582089552, | |
| "grad_norm": 9.307779312133789, | |
| "learning_rate": 4.4879767827529024e-05, | |
| "loss": 0.7793, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.861940298507463, | |
| "grad_norm": 4.041577339172363, | |
| "learning_rate": 4.482794361525705e-05, | |
| "loss": 0.5555, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 3.8805970149253732, | |
| "grad_norm": 12.132710456848145, | |
| "learning_rate": 4.477611940298508e-05, | |
| "loss": 1.5825, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.8992537313432836, | |
| "grad_norm": 11.172629356384277, | |
| "learning_rate": 4.472429519071311e-05, | |
| "loss": 1.2408, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 3.917910447761194, | |
| "grad_norm": 9.551191329956055, | |
| "learning_rate": 4.4672470978441125e-05, | |
| "loss": 1.2186, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.9365671641791042, | |
| "grad_norm": 6.64059591293335, | |
| "learning_rate": 4.4620646766169155e-05, | |
| "loss": 0.9661, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 3.955223880597015, | |
| "grad_norm": 11.107884407043457, | |
| "learning_rate": 4.4568822553897185e-05, | |
| "loss": 0.8848, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.9738805970149254, | |
| "grad_norm": 14.161197662353516, | |
| "learning_rate": 4.451699834162521e-05, | |
| "loss": 1.0005, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 3.9925373134328357, | |
| "grad_norm": 6.025814533233643, | |
| "learning_rate": 4.446517412935324e-05, | |
| "loss": 0.7212, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.685, | |
| "eval_f1_macro": 0.6471554397159933, | |
| "eval_f1_micro": 0.685, | |
| "eval_f1_weighted": 0.6601559323029912, | |
| "eval_loss": 1.3381353616714478, | |
| "eval_precision_macro": 0.6763538748832867, | |
| "eval_precision_micro": 0.685, | |
| "eval_precision_weighted": 0.6935892857142857, | |
| "eval_recall_macro": 0.6763305322128851, | |
| "eval_recall_micro": 0.685, | |
| "eval_recall_weighted": 0.685, | |
| "eval_runtime": 18.3933, | |
| "eval_samples_per_second": 10.873, | |
| "eval_steps_per_second": 0.707, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 4.0111940298507465, | |
| "grad_norm": 2.7835495471954346, | |
| "learning_rate": 4.441334991708126e-05, | |
| "loss": 0.6966, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.029850746268656, | |
| "grad_norm": 10.108399391174316, | |
| "learning_rate": 4.4361525704809285e-05, | |
| "loss": 1.1173, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.048507462686567, | |
| "grad_norm": 4.910768032073975, | |
| "learning_rate": 4.4309701492537316e-05, | |
| "loss": 0.6386, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.067164179104478, | |
| "grad_norm": 15.222392082214355, | |
| "learning_rate": 4.425787728026534e-05, | |
| "loss": 0.8312, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.085820895522388, | |
| "grad_norm": 5.983860969543457, | |
| "learning_rate": 4.420605306799337e-05, | |
| "loss": 0.5752, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.104477611940299, | |
| "grad_norm": 14.089235305786133, | |
| "learning_rate": 4.41542288557214e-05, | |
| "loss": 0.8314, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.123134328358209, | |
| "grad_norm": 6.404083251953125, | |
| "learning_rate": 4.410240464344942e-05, | |
| "loss": 0.9376, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.141791044776119, | |
| "grad_norm": 7.631033897399902, | |
| "learning_rate": 4.4050580431177446e-05, | |
| "loss": 0.8144, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.16044776119403, | |
| "grad_norm": 7.922863483428955, | |
| "learning_rate": 4.3998756218905476e-05, | |
| "loss": 0.5124, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.17910447761194, | |
| "grad_norm": 5.675821304321289, | |
| "learning_rate": 4.39469320066335e-05, | |
| "loss": 0.778, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.197761194029851, | |
| "grad_norm": 10.380525588989258, | |
| "learning_rate": 4.389510779436153e-05, | |
| "loss": 0.824, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.2164179104477615, | |
| "grad_norm": 10.827855110168457, | |
| "learning_rate": 4.384328358208955e-05, | |
| "loss": 0.7693, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.235074626865671, | |
| "grad_norm": 8.679563522338867, | |
| "learning_rate": 4.3791459369817584e-05, | |
| "loss": 0.7922, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.253731343283582, | |
| "grad_norm": 6.679693222045898, | |
| "learning_rate": 4.373963515754561e-05, | |
| "loss": 0.629, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.272388059701493, | |
| "grad_norm": 3.585435390472412, | |
| "learning_rate": 4.368781094527363e-05, | |
| "loss": 0.4874, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.291044776119403, | |
| "grad_norm": 16.56496238708496, | |
| "learning_rate": 4.363598673300166e-05, | |
| "loss": 0.9753, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.309701492537314, | |
| "grad_norm": 7.082263469696045, | |
| "learning_rate": 4.358416252072969e-05, | |
| "loss": 0.7295, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.3283582089552235, | |
| "grad_norm": 14.251839637756348, | |
| "learning_rate": 4.3532338308457714e-05, | |
| "loss": 0.4749, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.347014925373134, | |
| "grad_norm": 10.183996200561523, | |
| "learning_rate": 4.3480514096185744e-05, | |
| "loss": 0.9388, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.365671641791045, | |
| "grad_norm": 11.86069393157959, | |
| "learning_rate": 4.342868988391377e-05, | |
| "loss": 0.8015, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.384328358208955, | |
| "grad_norm": 8.476336479187012, | |
| "learning_rate": 4.337686567164179e-05, | |
| "loss": 0.8224, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.402985074626866, | |
| "grad_norm": 14.603717803955078, | |
| "learning_rate": 4.332504145936982e-05, | |
| "loss": 0.5892, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.4216417910447765, | |
| "grad_norm": 4.239728927612305, | |
| "learning_rate": 4.3273217247097845e-05, | |
| "loss": 0.5871, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.440298507462686, | |
| "grad_norm": 7.121862411499023, | |
| "learning_rate": 4.3221393034825875e-05, | |
| "loss": 0.6393, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.458955223880597, | |
| "grad_norm": 7.324426651000977, | |
| "learning_rate": 4.3169568822553905e-05, | |
| "loss": 0.6936, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.477611940298507, | |
| "grad_norm": 10.668092727661133, | |
| "learning_rate": 4.311774461028192e-05, | |
| "loss": 0.7822, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.496268656716418, | |
| "grad_norm": 8.423562049865723, | |
| "learning_rate": 4.306592039800995e-05, | |
| "loss": 0.6684, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.514925373134329, | |
| "grad_norm": 14.151291847229004, | |
| "learning_rate": 4.301409618573798e-05, | |
| "loss": 0.82, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.5335820895522385, | |
| "grad_norm": 8.986945152282715, | |
| "learning_rate": 4.2962271973466006e-05, | |
| "loss": 0.7146, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.552238805970149, | |
| "grad_norm": 15.039894104003906, | |
| "learning_rate": 4.2910447761194036e-05, | |
| "loss": 0.9461, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.57089552238806, | |
| "grad_norm": 14.114082336425781, | |
| "learning_rate": 4.285862354892206e-05, | |
| "loss": 0.7352, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.58955223880597, | |
| "grad_norm": 9.63955307006836, | |
| "learning_rate": 4.280679933665008e-05, | |
| "loss": 0.6686, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.608208955223881, | |
| "grad_norm": 9.240038871765137, | |
| "learning_rate": 4.275497512437811e-05, | |
| "loss": 0.966, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 4.6268656716417915, | |
| "grad_norm": 30.38275718688965, | |
| "learning_rate": 4.2703150912106136e-05, | |
| "loss": 0.6015, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 4.645522388059701, | |
| "grad_norm": 3.4407896995544434, | |
| "learning_rate": 4.2651326699834166e-05, | |
| "loss": 0.5758, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 4.664179104477612, | |
| "grad_norm": 2.6312341690063477, | |
| "learning_rate": 4.259950248756219e-05, | |
| "loss": 0.7366, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 4.682835820895522, | |
| "grad_norm": 5.770228862762451, | |
| "learning_rate": 4.254767827529022e-05, | |
| "loss": 0.5223, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 4.701492537313433, | |
| "grad_norm": 9.451111793518066, | |
| "learning_rate": 4.249585406301824e-05, | |
| "loss": 0.6489, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 4.720149253731344, | |
| "grad_norm": 9.007664680480957, | |
| "learning_rate": 4.244402985074627e-05, | |
| "loss": 0.6277, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 4.7388059701492535, | |
| "grad_norm": 11.020060539245605, | |
| "learning_rate": 4.23922056384743e-05, | |
| "loss": 0.5666, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 4.757462686567164, | |
| "grad_norm": 4.820965766906738, | |
| "learning_rate": 4.234038142620233e-05, | |
| "loss": 0.6379, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 4.776119402985074, | |
| "grad_norm": 7.916781902313232, | |
| "learning_rate": 4.228855721393035e-05, | |
| "loss": 0.7684, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 4.794776119402985, | |
| "grad_norm": 9.316835403442383, | |
| "learning_rate": 4.223673300165838e-05, | |
| "loss": 0.8076, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 4.813432835820896, | |
| "grad_norm": 11.205623626708984, | |
| "learning_rate": 4.2184908789386404e-05, | |
| "loss": 0.7485, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 4.832089552238806, | |
| "grad_norm": 9.189976692199707, | |
| "learning_rate": 4.213308457711443e-05, | |
| "loss": 0.8462, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 4.850746268656716, | |
| "grad_norm": 10.27412223815918, | |
| "learning_rate": 4.208126036484246e-05, | |
| "loss": 0.6273, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 4.869402985074627, | |
| "grad_norm": 9.638339042663574, | |
| "learning_rate": 4.202943615257048e-05, | |
| "loss": 0.7762, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 4.888059701492537, | |
| "grad_norm": 5.456606388092041, | |
| "learning_rate": 4.197761194029851e-05, | |
| "loss": 0.6509, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 4.906716417910448, | |
| "grad_norm": 8.314762115478516, | |
| "learning_rate": 4.192578772802654e-05, | |
| "loss": 0.644, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 4.925373134328359, | |
| "grad_norm": 5.1142072677612305, | |
| "learning_rate": 4.187396351575456e-05, | |
| "loss": 0.7036, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 4.9440298507462686, | |
| "grad_norm": 6.1798577308654785, | |
| "learning_rate": 4.182213930348259e-05, | |
| "loss": 0.4501, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 4.962686567164179, | |
| "grad_norm": 12.538804054260254, | |
| "learning_rate": 4.177031509121062e-05, | |
| "loss": 0.9835, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 4.981343283582089, | |
| "grad_norm": 11.114508628845215, | |
| "learning_rate": 4.171849087893864e-05, | |
| "loss": 0.4535, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 24.881195068359375, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 0.5743, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.69, | |
| "eval_f1_macro": 0.6552111287405404, | |
| "eval_f1_micro": 0.69, | |
| "eval_f1_weighted": 0.6724256854256855, | |
| "eval_loss": 1.1938023567199707, | |
| "eval_precision_macro": 0.687453314659197, | |
| "eval_precision_micro": 0.69, | |
| "eval_precision_weighted": 0.7043214285714287, | |
| "eval_recall_macro": 0.6736694677871148, | |
| "eval_recall_micro": 0.69, | |
| "eval_recall_weighted": 0.69, | |
| "eval_runtime": 20.8959, | |
| "eval_samples_per_second": 9.571, | |
| "eval_steps_per_second": 0.622, | |
| "step": 1340 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 5360, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 354552510105600.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |