| { | |
| "best_metric": 0.907917320728302, | |
| "best_model_checkpoint": "./vit-base-brain-xray/checkpoint-100", | |
| "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": 5.699078559875488, | |
| "learning_rate": 0.00019722222222222225, | |
| "loss": 1.2918, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 3.1500043869018555, | |
| "learning_rate": 0.00019444444444444446, | |
| "loss": 0.8867, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 3.1283438205718994, | |
| "learning_rate": 0.00019166666666666667, | |
| "loss": 0.5707, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 2.686929225921631, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.3967, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 1.111426830291748, | |
| "learning_rate": 0.00018611111111111112, | |
| "loss": 0.3759, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 2.6397788524627686, | |
| "learning_rate": 0.00018333333333333334, | |
| "loss": 0.2927, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3888888888888889, | |
| "grad_norm": 0.6170921325683594, | |
| "learning_rate": 0.00018055555555555557, | |
| "loss": 0.2122, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.33072641491889954, | |
| "learning_rate": 0.00017777777777777779, | |
| "loss": 0.2795, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.4631454944610596, | |
| "learning_rate": 0.000175, | |
| "loss": 0.3315, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 7.331964492797852, | |
| "learning_rate": 0.00017222222222222224, | |
| "loss": 0.2478, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "eval_accuracy": 0.6903553299492385, | |
| "eval_loss": 0.907917320728302, | |
| "eval_runtime": 5.7847, | |
| "eval_samples_per_second": 68.111, | |
| "eval_steps_per_second": 8.644, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6111111111111112, | |
| "grad_norm": 3.3308780193328857, | |
| "learning_rate": 0.00016944444444444445, | |
| "loss": 0.4174, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 3.7708303928375244, | |
| "learning_rate": 0.0001666666666666667, | |
| "loss": 0.2337, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7222222222222222, | |
| "grad_norm": 0.6579460501670837, | |
| "learning_rate": 0.0001638888888888889, | |
| "loss": 0.271, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.7777777777777778, | |
| "grad_norm": 2.6412248611450195, | |
| "learning_rate": 0.0001611111111111111, | |
| "loss": 0.2476, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 4.0377197265625, | |
| "learning_rate": 0.00015833333333333332, | |
| "loss": 0.1787, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 5.922617435455322, | |
| "learning_rate": 0.00015555555555555556, | |
| "loss": 0.1848, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9444444444444444, | |
| "grad_norm": 4.513651371002197, | |
| "learning_rate": 0.00015277777777777777, | |
| "loss": 0.2581, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.23064616322517395, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 0.1437, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0555555555555556, | |
| "grad_norm": 2.2603862285614014, | |
| "learning_rate": 0.00014722222222222223, | |
| "loss": 0.1052, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 4.710208892822266, | |
| "learning_rate": 0.00014444444444444444, | |
| "loss": 0.1499, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "eval_accuracy": 0.7182741116751269, | |
| "eval_loss": 1.1542534828186035, | |
| "eval_runtime": 5.8503, | |
| "eval_samples_per_second": 67.347, | |
| "eval_steps_per_second": 8.547, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1666666666666667, | |
| "grad_norm": 0.15027037262916565, | |
| "learning_rate": 0.00014166666666666668, | |
| "loss": 0.0747, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.2222222222222223, | |
| "grad_norm": 0.3188658356666565, | |
| "learning_rate": 0.0001388888888888889, | |
| "loss": 0.1555, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.2777777777777777, | |
| "grad_norm": 3.695791482925415, | |
| "learning_rate": 0.00013611111111111113, | |
| "loss": 0.1876, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 2.009079694747925, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 0.1054, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 0.07869268208742142, | |
| "learning_rate": 0.00013055555555555555, | |
| "loss": 0.0381, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.4444444444444444, | |
| "grad_norm": 0.9898849725723267, | |
| "learning_rate": 0.00012777777777777776, | |
| "loss": 0.1144, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.509087085723877, | |
| "learning_rate": 0.000125, | |
| "loss": 0.1002, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 0.08736218512058258, | |
| "learning_rate": 0.00012222222222222224, | |
| "loss": 0.1103, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.6111111111111112, | |
| "grad_norm": 0.24889719486236572, | |
| "learning_rate": 0.00011944444444444445, | |
| "loss": 0.0987, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 2.0384163856506348, | |
| "learning_rate": 0.00011666666666666668, | |
| "loss": 0.0872, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "eval_accuracy": 0.7614213197969543, | |
| "eval_loss": 1.146877646446228, | |
| "eval_runtime": 6.3643, | |
| "eval_samples_per_second": 61.908, | |
| "eval_steps_per_second": 7.856, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.7222222222222223, | |
| "grad_norm": 0.47840073704719543, | |
| "learning_rate": 0.00011388888888888889, | |
| "loss": 0.0468, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.06918664276599884, | |
| "learning_rate": 0.00011111111111111112, | |
| "loss": 0.0191, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.8333333333333335, | |
| "grad_norm": 0.7651189565658569, | |
| "learning_rate": 0.00010833333333333333, | |
| "loss": 0.1305, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.8888888888888888, | |
| "grad_norm": 1.8910560607910156, | |
| "learning_rate": 0.00010555555555555557, | |
| "loss": 0.0355, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 0.07424870133399963, | |
| "learning_rate": 0.00010277777777777778, | |
| "loss": 0.0567, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.05715026333928108, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0991, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.0555555555555554, | |
| "grad_norm": 2.061647415161133, | |
| "learning_rate": 9.722222222222223e-05, | |
| "loss": 0.0603, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.111111111111111, | |
| "grad_norm": 0.04806053638458252, | |
| "learning_rate": 9.444444444444444e-05, | |
| "loss": 0.0325, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.1666666666666665, | |
| "grad_norm": 0.05517473816871643, | |
| "learning_rate": 9.166666666666667e-05, | |
| "loss": 0.0301, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.8571930527687073, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 0.0118, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "eval_accuracy": 0.7258883248730964, | |
| "eval_loss": 1.2360663414001465, | |
| "eval_runtime": 6.5101, | |
| "eval_samples_per_second": 60.522, | |
| "eval_steps_per_second": 7.68, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.2777777777777777, | |
| "grad_norm": 0.040004875510931015, | |
| "learning_rate": 8.611111111111112e-05, | |
| "loss": 0.0118, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 7.572581768035889, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 0.0314, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.388888888888889, | |
| "grad_norm": 0.037391383200883865, | |
| "learning_rate": 8.055555555555556e-05, | |
| "loss": 0.0088, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.034802764654159546, | |
| "learning_rate": 7.777777777777778e-05, | |
| "loss": 0.0147, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.06982697546482086, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0085, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.5555555555555554, | |
| "grad_norm": 0.03507501631975174, | |
| "learning_rate": 7.222222222222222e-05, | |
| "loss": 0.0087, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.611111111111111, | |
| "grad_norm": 4.626717567443848, | |
| "learning_rate": 6.944444444444444e-05, | |
| "loss": 0.0216, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.0349123477935791, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.05, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.7222222222222223, | |
| "grad_norm": 0.03614797443151474, | |
| "learning_rate": 6.388888888888888e-05, | |
| "loss": 0.0085, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.06703199446201324, | |
| "learning_rate": 6.111111111111112e-05, | |
| "loss": 0.0077, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "eval_accuracy": 0.766497461928934, | |
| "eval_loss": 1.2023228406906128, | |
| "eval_runtime": 5.9054, | |
| "eval_samples_per_second": 66.719, | |
| "eval_steps_per_second": 8.467, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.8333333333333335, | |
| "grad_norm": 0.03837073966860771, | |
| "learning_rate": 5.833333333333334e-05, | |
| "loss": 0.007, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.888888888888889, | |
| "grad_norm": 0.031188365072011948, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 0.0072, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.9444444444444446, | |
| "grad_norm": 0.15879875421524048, | |
| "learning_rate": 5.305555555555556e-05, | |
| "loss": 0.0145, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.051313698291778564, | |
| "learning_rate": 5.027777777777778e-05, | |
| "loss": 0.0067, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 3.0555555555555554, | |
| "grad_norm": 0.027085309848189354, | |
| "learning_rate": 4.75e-05, | |
| "loss": 0.0063, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.111111111111111, | |
| "grad_norm": 0.02693328447639942, | |
| "learning_rate": 4.472222222222223e-05, | |
| "loss": 0.0066, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 3.1666666666666665, | |
| "grad_norm": 0.027860965579748154, | |
| "learning_rate": 4.194444444444445e-05, | |
| "loss": 0.006, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 3.2222222222222223, | |
| "grad_norm": 0.025451989844441414, | |
| "learning_rate": 3.9166666666666665e-05, | |
| "loss": 0.006, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 3.2777777777777777, | |
| "grad_norm": 0.02524595521390438, | |
| "learning_rate": 3.638888888888889e-05, | |
| "loss": 0.0059, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 0.02702181413769722, | |
| "learning_rate": 3.3611111111111116e-05, | |
| "loss": 0.0057, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "eval_accuracy": 0.7639593908629442, | |
| "eval_loss": 1.2470098733901978, | |
| "eval_runtime": 5.9411, | |
| "eval_samples_per_second": 66.318, | |
| "eval_steps_per_second": 8.416, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.388888888888889, | |
| "grad_norm": 0.027771031484007835, | |
| "learning_rate": 3.0833333333333335e-05, | |
| "loss": 0.0057, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 3.4444444444444446, | |
| "grad_norm": 0.02504831738770008, | |
| "learning_rate": 2.8055555555555557e-05, | |
| "loss": 0.0313, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "grad_norm": 0.027147900313138962, | |
| "learning_rate": 2.527777777777778e-05, | |
| "loss": 0.0056, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 3.5555555555555554, | |
| "grad_norm": 0.026741115376353264, | |
| "learning_rate": 2.25e-05, | |
| "loss": 0.0059, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.611111111111111, | |
| "grad_norm": 0.05946130305528641, | |
| "learning_rate": 1.9722222222222224e-05, | |
| "loss": 0.0055, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.6666666666666665, | |
| "grad_norm": 0.025428051128983498, | |
| "learning_rate": 1.6944444444444446e-05, | |
| "loss": 0.0054, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.7222222222222223, | |
| "grad_norm": 0.023423248901963234, | |
| "learning_rate": 1.4166666666666668e-05, | |
| "loss": 0.0053, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.7777777777777777, | |
| "grad_norm": 0.023936165496706963, | |
| "learning_rate": 1.138888888888889e-05, | |
| "loss": 0.0057, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.8333333333333335, | |
| "grad_norm": 0.023328222334384918, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0052, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "grad_norm": 0.025306882336735725, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0053, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "eval_accuracy": 0.7766497461928934, | |
| "eval_loss": 1.209647297859192, | |
| "eval_runtime": 7.4912, | |
| "eval_samples_per_second": 52.595, | |
| "eval_steps_per_second": 6.675, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.9444444444444446, | |
| "grad_norm": 0.023267222568392754, | |
| "learning_rate": 3.0555555555555556e-06, | |
| "loss": 0.0057, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.03729303553700447, | |
| "learning_rate": 2.777777777777778e-07, | |
| "loss": 0.0057, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 720, | |
| "total_flos": 8.896239873264845e+17, | |
| "train_loss": 0.12508326765770714, | |
| "train_runtime": 729.602, | |
| "train_samples_per_second": 15.735, | |
| "train_steps_per_second": 0.987 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 720, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "total_flos": 8.896239873264845e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |