| { | |
| "best_metric": 0.929534375667572, | |
| "best_model_checkpoint": "./vit-base-beans-demo-v5/checkpoint-100", | |
| "epoch": 4.0, | |
| "eval_steps": 100, | |
| "global_step": 496, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.9071108102798462, | |
| "learning_rate": 0.00019596774193548388, | |
| "loss": 0.7806, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 2.2380499839782715, | |
| "learning_rate": 0.00019193548387096775, | |
| "loss": 0.7214, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 1.4890930652618408, | |
| "learning_rate": 0.00018790322580645164, | |
| "loss": 0.6215, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 3.2323720455169678, | |
| "learning_rate": 0.00018387096774193548, | |
| "loss": 0.6378, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 2.838930606842041, | |
| "learning_rate": 0.00017983870967741935, | |
| "loss": 0.7502, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 3.2034356594085693, | |
| "learning_rate": 0.00017580645161290325, | |
| "loss": 0.5904, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 3.1891825199127197, | |
| "learning_rate": 0.00017177419354838711, | |
| "loss": 0.5718, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 2.0921356678009033, | |
| "learning_rate": 0.00016774193548387098, | |
| "loss": 0.3783, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 2.864804983139038, | |
| "learning_rate": 0.00016370967741935485, | |
| "loss": 0.6002, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 3.1752126216888428, | |
| "learning_rate": 0.00015967741935483872, | |
| "loss": 0.6438, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "eval_accuracy": 0.634703196347032, | |
| "eval_loss": 0.929534375667572, | |
| "eval_runtime": 7.2962, | |
| "eval_samples_per_second": 30.016, | |
| "eval_steps_per_second": 3.838, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 2.728193521499634, | |
| "learning_rate": 0.0001556451612903226, | |
| "loss": 0.5441, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 2.140393018722534, | |
| "learning_rate": 0.00015161290322580646, | |
| "loss": 0.4403, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 0.6765386462211609, | |
| "learning_rate": 0.00014758064516129032, | |
| "loss": 0.3251, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 0.9497590661048889, | |
| "learning_rate": 0.00014354838709677422, | |
| "loss": 0.2046, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 4.010074615478516, | |
| "learning_rate": 0.00013991935483870967, | |
| "loss": 0.3276, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 3.7631189823150635, | |
| "learning_rate": 0.00013588709677419357, | |
| "loss": 0.2937, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.5803029537200928, | |
| "learning_rate": 0.00013185483870967743, | |
| "loss": 0.1906, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 5.088043212890625, | |
| "learning_rate": 0.0001278225806451613, | |
| "loss": 0.2207, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 2.3816022872924805, | |
| "learning_rate": 0.00012379032258064514, | |
| "loss": 0.1919, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 5.558553218841553, | |
| "learning_rate": 0.00011975806451612903, | |
| "loss": 0.3105, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "eval_accuracy": 0.6575342465753424, | |
| "eval_loss": 0.9350173473358154, | |
| "eval_runtime": 7.7793, | |
| "eval_samples_per_second": 28.152, | |
| "eval_steps_per_second": 3.599, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 3.439823865890503, | |
| "learning_rate": 0.00011572580645161291, | |
| "loss": 0.3714, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 2.6023850440979004, | |
| "learning_rate": 0.00011169354838709678, | |
| "loss": 0.2869, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 3.2238519191741943, | |
| "learning_rate": 0.00010766129032258066, | |
| "loss": 0.4462, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 1.0531260967254639, | |
| "learning_rate": 0.00010362903225806453, | |
| "loss": 0.3634, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 0.5729889869689941, | |
| "learning_rate": 9.95967741935484e-05, | |
| "loss": 0.2624, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 0.1924820989370346, | |
| "learning_rate": 9.556451612903226e-05, | |
| "loss": 0.0999, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 0.39775505661964417, | |
| "learning_rate": 9.153225806451613e-05, | |
| "loss": 0.0938, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 0.22179947793483734, | |
| "learning_rate": 8.75e-05, | |
| "loss": 0.1017, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 1.6249357461929321, | |
| "learning_rate": 8.346774193548388e-05, | |
| "loss": 0.1745, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 0.34801536798477173, | |
| "learning_rate": 7.943548387096774e-05, | |
| "loss": 0.0634, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_accuracy": 0.6894977168949772, | |
| "eval_loss": 1.0781886577606201, | |
| "eval_runtime": 7.6715, | |
| "eval_samples_per_second": 28.547, | |
| "eval_steps_per_second": 3.65, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 2.6541597843170166, | |
| "learning_rate": 7.540322580645162e-05, | |
| "loss": 0.0772, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 0.1635380983352661, | |
| "learning_rate": 7.137096774193549e-05, | |
| "loss": 0.1042, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 1.126976490020752, | |
| "learning_rate": 6.733870967741935e-05, | |
| "loss": 0.1643, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 0.2140628844499588, | |
| "learning_rate": 6.330645161290322e-05, | |
| "loss": 0.0479, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 0.14856065809726715, | |
| "learning_rate": 5.9274193548387104e-05, | |
| "loss": 0.0606, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 1.9021470546722412, | |
| "learning_rate": 5.5241935483870966e-05, | |
| "loss": 0.0576, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 0.488421767950058, | |
| "learning_rate": 5.120967741935484e-05, | |
| "loss": 0.1573, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "grad_norm": 0.27475953102111816, | |
| "learning_rate": 4.7177419354838716e-05, | |
| "loss": 0.0264, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "grad_norm": 0.08814023435115814, | |
| "learning_rate": 4.3145161290322584e-05, | |
| "loss": 0.0197, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "grad_norm": 0.10707065463066101, | |
| "learning_rate": 3.911290322580645e-05, | |
| "loss": 0.0257, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "eval_accuracy": 0.6986301369863014, | |
| "eval_loss": 1.06435227394104, | |
| "eval_runtime": 7.0971, | |
| "eval_samples_per_second": 30.858, | |
| "eval_steps_per_second": 3.945, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "grad_norm": 0.06996390968561172, | |
| "learning_rate": 3.508064516129033e-05, | |
| "loss": 0.0192, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "grad_norm": 1.358115315437317, | |
| "learning_rate": 3.1048387096774195e-05, | |
| "loss": 0.0431, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "grad_norm": 0.4962191581726074, | |
| "learning_rate": 2.7016129032258064e-05, | |
| "loss": 0.0573, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "grad_norm": 0.08283121138811111, | |
| "learning_rate": 2.2983870967741935e-05, | |
| "loss": 0.0216, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "grad_norm": 0.06285007297992706, | |
| "learning_rate": 1.8951612903225807e-05, | |
| "loss": 0.0169, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "grad_norm": 0.10198648273944855, | |
| "learning_rate": 1.4919354838709679e-05, | |
| "loss": 0.0188, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "grad_norm": 1.5539321899414062, | |
| "learning_rate": 1.0887096774193549e-05, | |
| "loss": 0.0227, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "grad_norm": 0.06271003931760788, | |
| "learning_rate": 6.854838709677419e-06, | |
| "loss": 0.0212, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "grad_norm": 0.1244824230670929, | |
| "learning_rate": 2.82258064516129e-06, | |
| "loss": 0.0183, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 496, | |
| "total_flos": 6.10974224738132e+17, | |
| "train_loss": 0.25425288126233125, | |
| "train_runtime": 387.3536, | |
| "train_samples_per_second": 20.353, | |
| "train_steps_per_second": 1.28 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 496, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "total_flos": 6.10974224738132e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |