| { | |
| "best_metric": 0.9032258064516129, | |
| "best_model_checkpoint": "BEiT-TO-DA\\checkpoint-116", | |
| "epoch": 38.62068965517241, | |
| "eval_steps": 500, | |
| "global_step": 560, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.910714285714286e-05, | |
| "loss": 1.5028, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_accuracy": 0.14516129032258066, | |
| "eval_loss": 1.3862112760543823, | |
| "eval_runtime": 2.2297, | |
| "eval_samples_per_second": 27.806, | |
| "eval_steps_per_second": 1.794, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.8214285714285716e-05, | |
| "loss": 1.3358, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8064516129032258, | |
| "eval_loss": 1.0830727815628052, | |
| "eval_runtime": 2.045, | |
| "eval_samples_per_second": 30.318, | |
| "eval_steps_per_second": 1.956, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.732142857142857e-05, | |
| "loss": 1.1346, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.642857142857143e-05, | |
| "loss": 0.8919, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.8097184300422668, | |
| "eval_runtime": 1.9686, | |
| "eval_samples_per_second": 31.495, | |
| "eval_steps_per_second": 2.032, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 4.5535714285714286e-05, | |
| "loss": 0.7328, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7741935483870968, | |
| "eval_loss": 0.6545600891113281, | |
| "eval_runtime": 2.0425, | |
| "eval_samples_per_second": 30.355, | |
| "eval_steps_per_second": 1.958, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 4.464285714285715e-05, | |
| "loss": 0.6079, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.4881, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.4860498309135437, | |
| "eval_runtime": 1.9434, | |
| "eval_samples_per_second": 31.902, | |
| "eval_steps_per_second": 2.058, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 4.2857142857142856e-05, | |
| "loss": 0.4367, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.47702911496162415, | |
| "eval_runtime": 2.1423, | |
| "eval_samples_per_second": 28.94, | |
| "eval_steps_per_second": 1.867, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "learning_rate": 4.196428571428572e-05, | |
| "loss": 0.3971, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 4.107142857142857e-05, | |
| "loss": 0.3343, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.39124295115470886, | |
| "eval_runtime": 2.085, | |
| "eval_samples_per_second": 29.736, | |
| "eval_steps_per_second": 1.918, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 4.017857142857143e-05, | |
| "loss": 0.2794, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9032258064516129, | |
| "eval_loss": 0.3226192891597748, | |
| "eval_runtime": 2.0791, | |
| "eval_samples_per_second": 29.821, | |
| "eval_steps_per_second": 1.924, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 3.928571428571429e-05, | |
| "loss": 0.2586, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 3.839285714285715e-05, | |
| "loss": 0.2424, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "eval_accuracy": 0.7903225806451613, | |
| "eval_loss": 0.6426439881324768, | |
| "eval_runtime": 2.129, | |
| "eval_samples_per_second": 29.122, | |
| "eval_steps_per_second": 1.879, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.2875, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.4603542685508728, | |
| "eval_runtime": 2.0375, | |
| "eval_samples_per_second": 30.429, | |
| "eval_steps_per_second": 1.963, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 3.6607142857142853e-05, | |
| "loss": 0.226, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.3026295006275177, | |
| "eval_runtime": 1.9845, | |
| "eval_samples_per_second": 31.243, | |
| "eval_steps_per_second": 2.016, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 3.571428571428572e-05, | |
| "loss": 0.2307, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 3.4821428571428574e-05, | |
| "loss": 0.1819, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.3874911963939667, | |
| "eval_runtime": 2.167, | |
| "eval_samples_per_second": 28.611, | |
| "eval_steps_per_second": 1.846, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 3.392857142857143e-05, | |
| "loss": 0.2354, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "eval_accuracy": 0.9032258064516129, | |
| "eval_loss": 0.3412918746471405, | |
| "eval_runtime": 1.943, | |
| "eval_samples_per_second": 31.91, | |
| "eval_steps_per_second": 2.059, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 3.303571428571429e-05, | |
| "loss": 0.1651, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 3.2142857142857144e-05, | |
| "loss": 0.2264, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.8870967741935484, | |
| "eval_loss": 0.3948277235031128, | |
| "eval_runtime": 2.0535, | |
| "eval_samples_per_second": 30.193, | |
| "eval_steps_per_second": 1.948, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.1652, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.36500856280326843, | |
| "eval_runtime": 1.978, | |
| "eval_samples_per_second": 31.345, | |
| "eval_steps_per_second": 2.022, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 3.0357142857142857e-05, | |
| "loss": 0.1601, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 15.86, | |
| "learning_rate": 2.9464285714285718e-05, | |
| "loss": 0.1449, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.8870967741935484, | |
| "eval_loss": 0.3610541820526123, | |
| "eval_runtime": 2.1646, | |
| "eval_samples_per_second": 28.643, | |
| "eval_steps_per_second": 1.848, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 16.55, | |
| "learning_rate": 2.857142857142857e-05, | |
| "loss": 0.0993, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 16.97, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.4573518931865692, | |
| "eval_runtime": 1.9815, | |
| "eval_samples_per_second": 31.29, | |
| "eval_steps_per_second": 2.019, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 17.24, | |
| "learning_rate": 2.767857142857143e-05, | |
| "loss": 0.1642, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "learning_rate": 2.6785714285714288e-05, | |
| "loss": 0.1566, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.8870967741935484, | |
| "eval_loss": 0.3923678696155548, | |
| "eval_runtime": 2.043, | |
| "eval_samples_per_second": 30.348, | |
| "eval_steps_per_second": 1.958, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 18.62, | |
| "learning_rate": 2.5892857142857148e-05, | |
| "loss": 0.1399, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.48281946778297424, | |
| "eval_runtime": 2.0105, | |
| "eval_samples_per_second": 30.838, | |
| "eval_steps_per_second": 1.99, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 19.31, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.1659, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 2.4107142857142858e-05, | |
| "loss": 0.1025, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.5376753807067871, | |
| "eval_runtime": 2.1678, | |
| "eval_samples_per_second": 28.601, | |
| "eval_steps_per_second": 1.845, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 20.69, | |
| "learning_rate": 2.3214285714285715e-05, | |
| "loss": 0.0855, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 20.97, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.4958372414112091, | |
| "eval_runtime": 2.035, | |
| "eval_samples_per_second": 30.467, | |
| "eval_steps_per_second": 1.966, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 21.38, | |
| "learning_rate": 2.2321428571428575e-05, | |
| "loss": 0.1419, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.8387096774193549, | |
| "eval_loss": 0.6155916452407837, | |
| "eval_runtime": 2.099, | |
| "eval_samples_per_second": 29.538, | |
| "eval_steps_per_second": 1.906, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 22.07, | |
| "learning_rate": 2.1428571428571428e-05, | |
| "loss": 0.1266, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 22.76, | |
| "learning_rate": 2.0535714285714285e-05, | |
| "loss": 0.117, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 22.97, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.4914974272251129, | |
| "eval_runtime": 2.0475, | |
| "eval_samples_per_second": 30.281, | |
| "eval_steps_per_second": 1.954, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 23.45, | |
| "learning_rate": 1.9642857142857145e-05, | |
| "loss": 0.0905, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.5896703600883484, | |
| "eval_runtime": 2.111, | |
| "eval_samples_per_second": 29.37, | |
| "eval_steps_per_second": 1.895, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 24.14, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.1049, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 24.83, | |
| "learning_rate": 1.785714285714286e-05, | |
| "loss": 0.1199, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 24.97, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.4870842695236206, | |
| "eval_runtime": 2.0015, | |
| "eval_samples_per_second": 30.977, | |
| "eval_steps_per_second": 1.999, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 25.52, | |
| "learning_rate": 1.6964285714285715e-05, | |
| "loss": 0.1246, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.8548387096774194, | |
| "eval_loss": 0.4824499189853668, | |
| "eval_runtime": 2.0932, | |
| "eval_samples_per_second": 29.619, | |
| "eval_steps_per_second": 1.911, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 26.21, | |
| "learning_rate": 1.6071428571428572e-05, | |
| "loss": 0.1051, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 26.9, | |
| "learning_rate": 1.5178571428571429e-05, | |
| "loss": 0.0967, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 26.97, | |
| "eval_accuracy": 0.8064516129032258, | |
| "eval_loss": 0.7483570575714111, | |
| "eval_runtime": 2.0485, | |
| "eval_samples_per_second": 30.266, | |
| "eval_steps_per_second": 1.953, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "learning_rate": 1.4285714285714285e-05, | |
| "loss": 0.1025, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.8387096774193549, | |
| "eval_loss": 0.6973522901535034, | |
| "eval_runtime": 2.0675, | |
| "eval_samples_per_second": 29.988, | |
| "eval_steps_per_second": 1.935, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 28.28, | |
| "learning_rate": 1.3392857142857144e-05, | |
| "loss": 0.0955, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 28.97, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.1112, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 28.97, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.6391315460205078, | |
| "eval_runtime": 2.0205, | |
| "eval_samples_per_second": 30.686, | |
| "eval_steps_per_second": 1.98, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 29.66, | |
| "learning_rate": 1.1607142857142857e-05, | |
| "loss": 0.0715, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.6585091948509216, | |
| "eval_runtime": 2.0845, | |
| "eval_samples_per_second": 29.744, | |
| "eval_steps_per_second": 1.919, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 30.34, | |
| "learning_rate": 1.0714285714285714e-05, | |
| "loss": 0.085, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 30.97, | |
| "eval_accuracy": 0.8064516129032258, | |
| "eval_loss": 0.7087302207946777, | |
| "eval_runtime": 2.3831, | |
| "eval_samples_per_second": 26.017, | |
| "eval_steps_per_second": 1.678, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 31.03, | |
| "learning_rate": 9.821428571428573e-06, | |
| "loss": 0.0707, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 31.72, | |
| "learning_rate": 8.92857142857143e-06, | |
| "loss": 0.1032, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.8387096774193549, | |
| "eval_loss": 0.6093763113021851, | |
| "eval_runtime": 2.041, | |
| "eval_samples_per_second": 30.377, | |
| "eval_steps_per_second": 1.96, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 32.41, | |
| "learning_rate": 8.035714285714286e-06, | |
| "loss": 0.0836, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 32.97, | |
| "eval_accuracy": 0.8064516129032258, | |
| "eval_loss": 0.5578293204307556, | |
| "eval_runtime": 2.0097, | |
| "eval_samples_per_second": 30.85, | |
| "eval_steps_per_second": 1.99, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 33.1, | |
| "learning_rate": 7.142857142857143e-06, | |
| "loss": 0.1029, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 33.79, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.0716, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.5497013330459595, | |
| "eval_runtime": 2.0935, | |
| "eval_samples_per_second": 29.615, | |
| "eval_steps_per_second": 1.911, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 34.48, | |
| "learning_rate": 5.357142857142857e-06, | |
| "loss": 0.069, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 34.97, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.5093409419059753, | |
| "eval_runtime": 2.0105, | |
| "eval_samples_per_second": 30.839, | |
| "eval_steps_per_second": 1.99, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 35.17, | |
| "learning_rate": 4.464285714285715e-06, | |
| "loss": 0.0719, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 35.86, | |
| "learning_rate": 3.5714285714285714e-06, | |
| "loss": 0.0577, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.8709677419354839, | |
| "eval_loss": 0.5188572406768799, | |
| "eval_runtime": 2.0901, | |
| "eval_samples_per_second": 29.663, | |
| "eval_steps_per_second": 1.914, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 36.55, | |
| "learning_rate": 2.6785714285714285e-06, | |
| "loss": 0.0882, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 36.97, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.6530526280403137, | |
| "eval_runtime": 2.0785, | |
| "eval_samples_per_second": 29.829, | |
| "eval_steps_per_second": 1.924, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 37.24, | |
| "learning_rate": 1.7857142857142857e-06, | |
| "loss": 0.0673, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 37.93, | |
| "learning_rate": 8.928571428571428e-07, | |
| "loss": 0.0563, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.6660893559455872, | |
| "eval_runtime": 2.139, | |
| "eval_samples_per_second": 28.986, | |
| "eval_steps_per_second": 1.87, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 38.62, | |
| "learning_rate": 0.0, | |
| "loss": 0.0841, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 38.62, | |
| "eval_accuracy": 0.8225806451612904, | |
| "eval_loss": 0.6615698933601379, | |
| "eval_runtime": 1.969, | |
| "eval_samples_per_second": 31.489, | |
| "eval_steps_per_second": 2.032, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 38.62, | |
| "step": 560, | |
| "total_flos": 2.776175636329267e+18, | |
| "train_loss": 0.24997490516730717, | |
| "train_runtime": 1716.8626, | |
| "train_samples_per_second": 21.621, | |
| "train_steps_per_second": 0.326 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 560, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 40, | |
| "save_steps": 500, | |
| "total_flos": 2.776175636329267e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |