|
{ |
|
"best_metric": 1.0935020446777344, |
|
"best_model_checkpoint": "./vit-large-brain-xray/checkpoint-200", |
|
"epoch": 4.0, |
|
"eval_steps": 100, |
|
"global_step": 720, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.05555555555555555, |
|
"grad_norm": 1.083911657333374, |
|
"learning_rate": 0.00019722222222222225, |
|
"loss": 1.3206, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.1111111111111111, |
|
"grad_norm": 1.5310566425323486, |
|
"learning_rate": 0.00019444444444444446, |
|
"loss": 0.9361, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.16666666666666666, |
|
"grad_norm": 3.7486021518707275, |
|
"learning_rate": 0.00019166666666666667, |
|
"loss": 0.5999, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.2222222222222222, |
|
"grad_norm": 3.448930025100708, |
|
"learning_rate": 0.00018888888888888888, |
|
"loss": 0.5138, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.2777777777777778, |
|
"grad_norm": 1.6228857040405273, |
|
"learning_rate": 0.00018611111111111112, |
|
"loss": 0.4566, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.3333333333333333, |
|
"grad_norm": 2.215758800506592, |
|
"learning_rate": 0.00018333333333333334, |
|
"loss": 0.4068, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.3888888888888889, |
|
"grad_norm": 1.5400978326797485, |
|
"learning_rate": 0.00018055555555555557, |
|
"loss": 0.3305, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.4444444444444444, |
|
"grad_norm": 1.4063529968261719, |
|
"learning_rate": 0.00017777777777777779, |
|
"loss": 0.4062, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"grad_norm": 0.7729310393333435, |
|
"learning_rate": 0.00017527777777777778, |
|
"loss": 0.2615, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.5555555555555556, |
|
"grad_norm": 6.676679611206055, |
|
"learning_rate": 0.00017250000000000002, |
|
"loss": 0.2144, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.5555555555555556, |
|
"eval_accuracy": 0.6269035532994924, |
|
"eval_loss": 1.2678815126419067, |
|
"eval_runtime": 7.3748, |
|
"eval_samples_per_second": 53.425, |
|
"eval_steps_per_second": 6.78, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.6111111111111112, |
|
"grad_norm": 1.6695863008499146, |
|
"learning_rate": 0.00016972222222222223, |
|
"loss": 0.3772, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.6666666666666666, |
|
"grad_norm": 0.457354336977005, |
|
"learning_rate": 0.00016694444444444447, |
|
"loss": 0.261, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.7222222222222222, |
|
"grad_norm": 6.663379192352295, |
|
"learning_rate": 0.00016416666666666668, |
|
"loss": 0.2832, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.7777777777777778, |
|
"grad_norm": 0.6463492512702942, |
|
"learning_rate": 0.0001613888888888889, |
|
"loss": 0.1613, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.8333333333333334, |
|
"grad_norm": 5.534128189086914, |
|
"learning_rate": 0.0001586111111111111, |
|
"loss": 0.452, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.8888888888888888, |
|
"grad_norm": 1.487303614616394, |
|
"learning_rate": 0.00015583333333333334, |
|
"loss": 0.3549, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.9444444444444444, |
|
"grad_norm": 8.732577323913574, |
|
"learning_rate": 0.00015305555555555556, |
|
"loss": 0.26, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 0.3070247173309326, |
|
"learning_rate": 0.0001502777777777778, |
|
"loss": 0.1787, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 1.0555555555555556, |
|
"grad_norm": 0.15718808770179749, |
|
"learning_rate": 0.0001475, |
|
"loss": 0.156, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 1.1111111111111112, |
|
"grad_norm": 1.4413491487503052, |
|
"learning_rate": 0.00014472222222222222, |
|
"loss": 0.1091, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.1111111111111112, |
|
"eval_accuracy": 0.7081218274111675, |
|
"eval_loss": 1.0935020446777344, |
|
"eval_runtime": 6.8075, |
|
"eval_samples_per_second": 57.877, |
|
"eval_steps_per_second": 7.345, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.1666666666666667, |
|
"grad_norm": 1.1086454391479492, |
|
"learning_rate": 0.00014194444444444446, |
|
"loss": 0.1243, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 1.2222222222222223, |
|
"grad_norm": 0.09483372420072556, |
|
"learning_rate": 0.00013916666666666667, |
|
"loss": 0.1712, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 1.2777777777777777, |
|
"grad_norm": 2.7763671875, |
|
"learning_rate": 0.0001363888888888889, |
|
"loss": 0.2156, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 1.3333333333333333, |
|
"grad_norm": 1.6603256464004517, |
|
"learning_rate": 0.00013361111111111112, |
|
"loss": 0.1186, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 1.3888888888888888, |
|
"grad_norm": 5.325634479522705, |
|
"learning_rate": 0.00013083333333333333, |
|
"loss": 0.1065, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 1.4444444444444444, |
|
"grad_norm": 0.07357333600521088, |
|
"learning_rate": 0.00012805555555555555, |
|
"loss": 0.1192, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"grad_norm": 3.8914055824279785, |
|
"learning_rate": 0.00012527777777777778, |
|
"loss": 0.0622, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 1.5555555555555556, |
|
"grad_norm": 0.37973469495773315, |
|
"learning_rate": 0.00012250000000000002, |
|
"loss": 0.2217, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 1.6111111111111112, |
|
"grad_norm": 0.08939097821712494, |
|
"learning_rate": 0.00011972222222222222, |
|
"loss": 0.1868, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 1.6666666666666665, |
|
"grad_norm": 0.29244643449783325, |
|
"learning_rate": 0.00011694444444444446, |
|
"loss": 0.1078, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.6666666666666665, |
|
"eval_accuracy": 0.7588832487309645, |
|
"eval_loss": 1.1237390041351318, |
|
"eval_runtime": 6.6691, |
|
"eval_samples_per_second": 59.078, |
|
"eval_steps_per_second": 7.497, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.7222222222222223, |
|
"grad_norm": 2.0514800548553467, |
|
"learning_rate": 0.00011416666666666667, |
|
"loss": 0.0995, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 1.7777777777777777, |
|
"grad_norm": 0.3739979863166809, |
|
"learning_rate": 0.0001113888888888889, |
|
"loss": 0.0752, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 1.8333333333333335, |
|
"grad_norm": 0.07262101769447327, |
|
"learning_rate": 0.00010861111111111111, |
|
"loss": 0.1004, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 1.8888888888888888, |
|
"grad_norm": 1.6550071239471436, |
|
"learning_rate": 0.00010583333333333334, |
|
"loss": 0.0593, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 1.9444444444444444, |
|
"grad_norm": 1.8396694660186768, |
|
"learning_rate": 0.00010305555555555555, |
|
"loss": 0.1425, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 0.061273444443941116, |
|
"learning_rate": 0.00010027777777777779, |
|
"loss": 0.0944, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 2.0555555555555554, |
|
"grad_norm": 1.0921915769577026, |
|
"learning_rate": 9.75e-05, |
|
"loss": 0.118, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 2.111111111111111, |
|
"grad_norm": 0.16066594421863556, |
|
"learning_rate": 9.472222222222222e-05, |
|
"loss": 0.0479, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 2.1666666666666665, |
|
"grad_norm": 0.06149543076753616, |
|
"learning_rate": 9.194444444444445e-05, |
|
"loss": 0.0161, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 2.2222222222222223, |
|
"grad_norm": 0.10219839215278625, |
|
"learning_rate": 8.916666666666667e-05, |
|
"loss": 0.016, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.2222222222222223, |
|
"eval_accuracy": 0.7563451776649747, |
|
"eval_loss": 1.235643982887268, |
|
"eval_runtime": 6.0001, |
|
"eval_samples_per_second": 65.666, |
|
"eval_steps_per_second": 8.333, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.2777777777777777, |
|
"grad_norm": 0.36931440234184265, |
|
"learning_rate": 8.63888888888889e-05, |
|
"loss": 0.0674, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 2.3333333333333335, |
|
"grad_norm": 0.05050282925367355, |
|
"learning_rate": 8.361111111111111e-05, |
|
"loss": 0.0404, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 2.388888888888889, |
|
"grad_norm": 0.0446784570813179, |
|
"learning_rate": 8.083333333333334e-05, |
|
"loss": 0.0205, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 2.4444444444444446, |
|
"grad_norm": 0.04102100059390068, |
|
"learning_rate": 7.805555555555556e-05, |
|
"loss": 0.0207, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"grad_norm": 4.75399112701416, |
|
"learning_rate": 7.527777777777777e-05, |
|
"loss": 0.0526, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 2.5555555555555554, |
|
"grad_norm": 0.046476561576128006, |
|
"learning_rate": 7.25e-05, |
|
"loss": 0.0461, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 2.611111111111111, |
|
"grad_norm": 5.1403632164001465, |
|
"learning_rate": 6.972222222222223e-05, |
|
"loss": 0.0424, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 2.6666666666666665, |
|
"grad_norm": 0.04179733246564865, |
|
"learning_rate": 6.694444444444444e-05, |
|
"loss": 0.0831, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 2.7222222222222223, |
|
"grad_norm": 0.0394105389714241, |
|
"learning_rate": 6.416666666666668e-05, |
|
"loss": 0.0102, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 2.7777777777777777, |
|
"grad_norm": 0.039878394454717636, |
|
"learning_rate": 6.13888888888889e-05, |
|
"loss": 0.0095, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.7777777777777777, |
|
"eval_accuracy": 0.7588832487309645, |
|
"eval_loss": 1.2315524816513062, |
|
"eval_runtime": 6.6635, |
|
"eval_samples_per_second": 59.128, |
|
"eval_steps_per_second": 7.504, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 2.8333333333333335, |
|
"grad_norm": 3.2955222129821777, |
|
"learning_rate": 5.8611111111111114e-05, |
|
"loss": 0.0223, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 2.888888888888889, |
|
"grad_norm": 0.03517436608672142, |
|
"learning_rate": 5.583333333333334e-05, |
|
"loss": 0.0083, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 2.9444444444444446, |
|
"grad_norm": 0.03378378599882126, |
|
"learning_rate": 5.305555555555556e-05, |
|
"loss": 0.017, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"grad_norm": 0.029930729418992996, |
|
"learning_rate": 5.027777777777778e-05, |
|
"loss": 0.0076, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 3.0555555555555554, |
|
"grad_norm": 0.030685801059007645, |
|
"learning_rate": 4.75e-05, |
|
"loss": 0.0082, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 3.111111111111111, |
|
"grad_norm": 0.029904644936323166, |
|
"learning_rate": 4.472222222222223e-05, |
|
"loss": 0.0073, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 3.1666666666666665, |
|
"grad_norm": 0.02878139540553093, |
|
"learning_rate": 4.194444444444445e-05, |
|
"loss": 0.0069, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 3.2222222222222223, |
|
"grad_norm": 0.028465483337640762, |
|
"learning_rate": 3.9166666666666665e-05, |
|
"loss": 0.0067, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 3.2777777777777777, |
|
"grad_norm": 0.026676874607801437, |
|
"learning_rate": 3.638888888888889e-05, |
|
"loss": 0.0067, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 3.3333333333333335, |
|
"grad_norm": 0.028720058500766754, |
|
"learning_rate": 3.3611111111111116e-05, |
|
"loss": 0.0066, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.3333333333333335, |
|
"eval_accuracy": 0.7588832487309645, |
|
"eval_loss": 1.3164998292922974, |
|
"eval_runtime": 6.9479, |
|
"eval_samples_per_second": 56.707, |
|
"eval_steps_per_second": 7.196, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.388888888888889, |
|
"grad_norm": 0.030909936875104904, |
|
"learning_rate": 3.0833333333333335e-05, |
|
"loss": 0.0066, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 3.4444444444444446, |
|
"grad_norm": 0.02741164341568947, |
|
"learning_rate": 2.8055555555555557e-05, |
|
"loss": 0.0429, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 3.5, |
|
"grad_norm": 0.028826788067817688, |
|
"learning_rate": 2.527777777777778e-05, |
|
"loss": 0.0079, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 3.5555555555555554, |
|
"grad_norm": 0.02735409140586853, |
|
"learning_rate": 2.25e-05, |
|
"loss": 0.0066, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 3.611111111111111, |
|
"grad_norm": 0.10283248126506805, |
|
"learning_rate": 1.9722222222222224e-05, |
|
"loss": 0.0159, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 3.6666666666666665, |
|
"grad_norm": 0.028924111276865005, |
|
"learning_rate": 1.6944444444444446e-05, |
|
"loss": 0.0064, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 3.7222222222222223, |
|
"grad_norm": 0.025118501856923103, |
|
"learning_rate": 1.4166666666666668e-05, |
|
"loss": 0.0062, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 3.7777777777777777, |
|
"grad_norm": 0.05767374858260155, |
|
"learning_rate": 1.138888888888889e-05, |
|
"loss": 0.0065, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 3.8333333333333335, |
|
"grad_norm": 0.0323575958609581, |
|
"learning_rate": 8.611111111111112e-06, |
|
"loss": 0.0064, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 3.888888888888889, |
|
"grad_norm": 0.026371095329523087, |
|
"learning_rate": 5.833333333333334e-06, |
|
"loss": 0.0161, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 3.888888888888889, |
|
"eval_accuracy": 0.7614213197969543, |
|
"eval_loss": 1.341164469718933, |
|
"eval_runtime": 5.8873, |
|
"eval_samples_per_second": 66.924, |
|
"eval_steps_per_second": 8.493, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 3.9444444444444446, |
|
"grad_norm": 0.025476330891251564, |
|
"learning_rate": 3.0555555555555556e-06, |
|
"loss": 0.0062, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"grad_norm": 0.041727062314748764, |
|
"learning_rate": 2.777777777777778e-07, |
|
"loss": 0.0097, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"step": 720, |
|
"total_flos": 3.16768696086528e+18, |
|
"train_loss": 0.15098576029348704, |
|
"train_runtime": 924.305, |
|
"train_samples_per_second": 12.42, |
|
"train_steps_per_second": 0.779 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 720, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 4, |
|
"save_steps": 100, |
|
"total_flos": 3.16768696086528e+18, |
|
"train_batch_size": 16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|