| { | |
| "best_metric": 0.9049533605575562, | |
| "best_model_checkpoint": "./vit-large-brain-xray/checkpoint-300", | |
| "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": 0.6702606678009033, | |
| "learning_rate": 0.00019722222222222225, | |
| "loss": 1.3662, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 1.5160739421844482, | |
| "learning_rate": 0.00019444444444444446, | |
| "loss": 1.2341, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 3.668222665786743, | |
| "learning_rate": 0.00019166666666666667, | |
| "loss": 0.9384, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 2.245746612548828, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.6538, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 2.4727349281311035, | |
| "learning_rate": 0.00018611111111111112, | |
| "loss": 0.4997, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 3.7319023609161377, | |
| "learning_rate": 0.00018333333333333334, | |
| "loss": 0.3898, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3888888888888889, | |
| "grad_norm": 0.6027088165283203, | |
| "learning_rate": 0.00018055555555555557, | |
| "loss": 0.3304, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.37244492769241333, | |
| "learning_rate": 0.00017777777777777779, | |
| "loss": 0.2543, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.6335736513137817, | |
| "learning_rate": 0.000175, | |
| "loss": 0.3538, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 1.6272715330123901, | |
| "learning_rate": 0.00017222222222222224, | |
| "loss": 0.352, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "eval_accuracy": 0.6294416243654822, | |
| "eval_loss": 1.2266901731491089, | |
| "eval_runtime": 8.2003, | |
| "eval_samples_per_second": 48.047, | |
| "eval_steps_per_second": 6.097, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6111111111111112, | |
| "grad_norm": 1.5688170194625854, | |
| "learning_rate": 0.00016944444444444445, | |
| "loss": 0.3773, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.8265367150306702, | |
| "learning_rate": 0.0001666666666666667, | |
| "loss": 0.226, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.7222222222222222, | |
| "grad_norm": 2.6476309299468994, | |
| "learning_rate": 0.0001638888888888889, | |
| "loss": 0.3515, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.7777777777777778, | |
| "grad_norm": 0.8978700637817383, | |
| "learning_rate": 0.0001611111111111111, | |
| "loss": 0.2367, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 1.1820647716522217, | |
| "learning_rate": 0.00015833333333333332, | |
| "loss": 0.3112, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 2.701751708984375, | |
| "learning_rate": 0.00015555555555555556, | |
| "loss": 0.2741, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9444444444444444, | |
| "grad_norm": 1.249694585800171, | |
| "learning_rate": 0.00015277777777777777, | |
| "loss": 0.2529, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.23323917388916016, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 0.2239, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.0555555555555556, | |
| "grad_norm": 0.4952305853366852, | |
| "learning_rate": 0.00014722222222222223, | |
| "loss": 0.1749, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 0.2073395550251007, | |
| "learning_rate": 0.00014444444444444444, | |
| "loss": 0.1612, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "eval_accuracy": 0.7538071065989848, | |
| "eval_loss": 1.0894657373428345, | |
| "eval_runtime": 7.609, | |
| "eval_samples_per_second": 51.781, | |
| "eval_steps_per_second": 6.571, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.1666666666666667, | |
| "grad_norm": 0.16568297147750854, | |
| "learning_rate": 0.00014166666666666668, | |
| "loss": 0.0718, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.2222222222222223, | |
| "grad_norm": 0.0984596461057663, | |
| "learning_rate": 0.0001388888888888889, | |
| "loss": 0.1395, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.2777777777777777, | |
| "grad_norm": 0.6732985973358154, | |
| "learning_rate": 0.00013611111111111113, | |
| "loss": 0.1421, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 2.5877740383148193, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 0.1118, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 3.4756956100463867, | |
| "learning_rate": 0.00013055555555555555, | |
| "loss": 0.1562, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.4444444444444444, | |
| "grad_norm": 0.09811172634363174, | |
| "learning_rate": 0.00012777777777777776, | |
| "loss": 0.1463, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 3.1999194622039795, | |
| "learning_rate": 0.000125, | |
| "loss": 0.1179, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 0.45012134313583374, | |
| "learning_rate": 0.00012222222222222224, | |
| "loss": 0.0929, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.6111111111111112, | |
| "grad_norm": 1.862654209136963, | |
| "learning_rate": 0.00011944444444444445, | |
| "loss": 0.186, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 1.39608895778656, | |
| "learning_rate": 0.00011666666666666668, | |
| "loss": 0.0473, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "eval_accuracy": 0.7741116751269036, | |
| "eval_loss": 0.9049533605575562, | |
| "eval_runtime": 6.1753, | |
| "eval_samples_per_second": 63.803, | |
| "eval_steps_per_second": 8.097, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.7222222222222223, | |
| "grad_norm": 0.0962180569767952, | |
| "learning_rate": 0.00011388888888888889, | |
| "loss": 0.0858, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.08585009723901749, | |
| "learning_rate": 0.00011111111111111112, | |
| "loss": 0.0519, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.8333333333333335, | |
| "grad_norm": 0.06303343176841736, | |
| "learning_rate": 0.00010833333333333333, | |
| "loss": 0.0221, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.8888888888888888, | |
| "grad_norm": 5.68204402923584, | |
| "learning_rate": 0.00010555555555555557, | |
| "loss": 0.0589, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 5.385427474975586, | |
| "learning_rate": 0.00010277777777777778, | |
| "loss": 0.1881, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 3.201244831085205, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0758, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.0555555555555554, | |
| "grad_norm": 1.2587229013442993, | |
| "learning_rate": 9.722222222222223e-05, | |
| "loss": 0.0461, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.111111111111111, | |
| "grad_norm": 0.07729563117027283, | |
| "learning_rate": 9.444444444444444e-05, | |
| "loss": 0.0155, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.1666666666666665, | |
| "grad_norm": 5.077848434448242, | |
| "learning_rate": 9.166666666666667e-05, | |
| "loss": 0.0724, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.05171338841319084, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 0.0525, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "eval_accuracy": 0.7690355329949239, | |
| "eval_loss": 1.0663037300109863, | |
| "eval_runtime": 6.6314, | |
| "eval_samples_per_second": 59.415, | |
| "eval_steps_per_second": 7.54, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.2777777777777777, | |
| "grad_norm": 0.056168586015701294, | |
| "learning_rate": 8.611111111111112e-05, | |
| "loss": 0.1214, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 0.2696777582168579, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 0.0148, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.388888888888889, | |
| "grad_norm": 0.045138537883758545, | |
| "learning_rate": 8.055555555555556e-05, | |
| "loss": 0.0175, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.03756405785679817, | |
| "learning_rate": 7.777777777777778e-05, | |
| "loss": 0.0264, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.17634020745754242, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0101, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.5555555555555554, | |
| "grad_norm": 0.037890926003456116, | |
| "learning_rate": 7.222222222222222e-05, | |
| "loss": 0.0484, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.611111111111111, | |
| "grad_norm": 4.751524448394775, | |
| "learning_rate": 6.944444444444444e-05, | |
| "loss": 0.0525, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.15853020548820496, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.0719, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.7222222222222223, | |
| "grad_norm": 0.039081115275621414, | |
| "learning_rate": 6.388888888888888e-05, | |
| "loss": 0.0085, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.4480770230293274, | |
| "learning_rate": 6.111111111111112e-05, | |
| "loss": 0.0123, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "eval_accuracy": 0.7461928934010152, | |
| "eval_loss": 1.2449774742126465, | |
| "eval_runtime": 5.9167, | |
| "eval_samples_per_second": 66.591, | |
| "eval_steps_per_second": 8.451, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.8333333333333335, | |
| "grad_norm": 7.5741801261901855, | |
| "learning_rate": 5.833333333333334e-05, | |
| "loss": 0.0278, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.888888888888889, | |
| "grad_norm": 0.1201184019446373, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 0.0091, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.9444444444444446, | |
| "grad_norm": 0.032710809260606766, | |
| "learning_rate": 5.2777777777777784e-05, | |
| "loss": 0.0077, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.03236711025238037, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0557, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 3.0555555555555554, | |
| "grad_norm": 0.034722838550806046, | |
| "learning_rate": 4.722222222222222e-05, | |
| "loss": 0.0191, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.111111111111111, | |
| "grad_norm": 4.018179416656494, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 0.0176, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 3.1666666666666665, | |
| "grad_norm": 0.5732712745666504, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 0.0087, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 3.2222222222222223, | |
| "grad_norm": 0.027404414489865303, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 0.0079, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 3.2777777777777777, | |
| "grad_norm": 0.02965979278087616, | |
| "learning_rate": 3.611111111111111e-05, | |
| "loss": 0.0068, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 0.026871565729379654, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 0.0066, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "eval_accuracy": 0.7817258883248731, | |
| "eval_loss": 1.1282514333724976, | |
| "eval_runtime": 6.7045, | |
| "eval_samples_per_second": 58.767, | |
| "eval_steps_per_second": 7.458, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.388888888888889, | |
| "grad_norm": 0.03278065472841263, | |
| "learning_rate": 3.055555555555556e-05, | |
| "loss": 0.0086, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 3.4444444444444446, | |
| "grad_norm": 0.07111264020204544, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 0.0295, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "grad_norm": 0.028257286176085472, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.0065, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 3.5555555555555554, | |
| "grad_norm": 0.02719848044216633, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 0.0418, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.611111111111111, | |
| "grad_norm": 0.026137089356780052, | |
| "learning_rate": 1.9444444444444445e-05, | |
| "loss": 0.0073, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.6666666666666665, | |
| "grad_norm": 0.030431417748332024, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.0101, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.7222222222222223, | |
| "grad_norm": 0.025364473462104797, | |
| "learning_rate": 1.388888888888889e-05, | |
| "loss": 0.0062, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.7777777777777777, | |
| "grad_norm": 0.02630157209932804, | |
| "learning_rate": 1.1111111111111112e-05, | |
| "loss": 0.0076, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.8333333333333335, | |
| "grad_norm": 0.025917503982782364, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0062, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "grad_norm": 6.676637649536133, | |
| "learning_rate": 5.555555555555556e-06, | |
| "loss": 0.0126, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "eval_accuracy": 0.7842639593908629, | |
| "eval_loss": 1.1716859340667725, | |
| "eval_runtime": 6.745, | |
| "eval_samples_per_second": 58.414, | |
| "eval_steps_per_second": 7.413, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.9444444444444446, | |
| "grad_norm": 0.025717712938785553, | |
| "learning_rate": 2.777777777777778e-06, | |
| "loss": 0.008, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.07326529920101166, | |
| "learning_rate": 0.0, | |
| "loss": 0.0065, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 720, | |
| "total_flos": 3.16768696086528e+18, | |
| "train_loss": 0.16034429804939362, | |
| "train_runtime": 1070.1524, | |
| "train_samples_per_second": 10.727, | |
| "train_steps_per_second": 0.673 | |
| } | |
| ], | |
| "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 | |
| } | |