| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.017883569023870095, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.8e-05, | |
| "loss": 2.2791, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.6e-05, | |
| "loss": 2.27, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.4e-05, | |
| "loss": 2.2689, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 2.2549, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9e-05, | |
| "loss": 2.2822, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 2.2564, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.6e-05, | |
| "loss": 2.2551, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.4e-05, | |
| "loss": 2.2471, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.2e-05, | |
| "loss": 2.2515, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8e-05, | |
| "loss": 2.2825, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 2.2584, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.6e-05, | |
| "loss": 2.2232, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.4e-05, | |
| "loss": 2.2463, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.2e-05, | |
| "loss": 2.2775, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7e-05, | |
| "loss": 2.2533, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 2.2596, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.6e-05, | |
| "loss": 2.2307, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 2.2671, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.2e-05, | |
| "loss": 2.2191, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-05, | |
| "loss": 2.2374, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.8e-05, | |
| "loss": 2.2755, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 2.249, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 2.2641, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 2.2875, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 2.2802, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "eval_accuracy": 0.5345466022376205, | |
| "eval_loss": 2.2552647590637207, | |
| "eval_runtime": 2008.2531, | |
| "eval_samples_per_second": 76.764, | |
| "eval_steps_per_second": 1.2, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.2486, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 2.2124, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 2.2253, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.2e-05, | |
| "loss": 2.2211, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2316, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.8e-05, | |
| "loss": 2.2463, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-05, | |
| "loss": 2.252, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 2.267, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.2416, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3e-05, | |
| "loss": 2.2452, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.242, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 2.2572, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.2346, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 2.2022, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 2.2317, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.2348, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.2453, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 2.2165, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.2548, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2867, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.2381, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6e-06, | |
| "loss": 2.242, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.267, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.2344, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0, | |
| "loss": 2.2277, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "eval_accuracy": 0.5368494657483851, | |
| "eval_loss": 2.2406654357910156, | |
| "eval_runtime": 1837.3548, | |
| "eval_samples_per_second": 83.904, | |
| "eval_steps_per_second": 1.311, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "step": 1000, | |
| "total_flos": 1.7866929733632e+17, | |
| "train_loss": 2.2496585121154786, | |
| "train_runtime": 5217.4701, | |
| "train_samples_per_second": 12.266, | |
| "train_steps_per_second": 0.192 | |
| } | |
| ], | |
| "max_steps": 1000, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.7866929733632e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |