| { | |
| "best_metric": 2.190300464630127, | |
| "best_model_checkpoint": "./output_c/checkpoint-615", | |
| "epoch": 59.0, | |
| "global_step": 885, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 3.077622652053833, | |
| "eval_runtime": 0.2951, | |
| "eval_samples_per_second": 494.804, | |
| "eval_steps_per_second": 16.945, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 2.9188127517700195, | |
| "eval_runtime": 0.3714, | |
| "eval_samples_per_second": 393.09, | |
| "eval_steps_per_second": 13.462, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 2.8426482677459717, | |
| "eval_runtime": 0.2609, | |
| "eval_samples_per_second": 559.557, | |
| "eval_steps_per_second": 19.163, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 2.6221985816955566, | |
| "eval_runtime": 0.2726, | |
| "eval_samples_per_second": 535.638, | |
| "eval_steps_per_second": 18.344, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 2.6243655681610107, | |
| "eval_runtime": 0.2669, | |
| "eval_samples_per_second": 546.955, | |
| "eval_steps_per_second": 18.731, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 2.6885011196136475, | |
| "eval_runtime": 0.2608, | |
| "eval_samples_per_second": 559.912, | |
| "eval_steps_per_second": 19.175, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 2.4477977752685547, | |
| "eval_runtime": 0.266, | |
| "eval_samples_per_second": 548.901, | |
| "eval_steps_per_second": 18.798, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 2.588456153869629, | |
| "eval_runtime": 0.3103, | |
| "eval_samples_per_second": 470.587, | |
| "eval_steps_per_second": 16.116, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 2.4026825428009033, | |
| "eval_runtime": 0.2649, | |
| "eval_samples_per_second": 551.053, | |
| "eval_steps_per_second": 18.872, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 2.3897533416748047, | |
| "eval_runtime": 0.3189, | |
| "eval_samples_per_second": 457.83, | |
| "eval_steps_per_second": 15.679, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_loss": 2.473085641860962, | |
| "eval_runtime": 0.3415, | |
| "eval_samples_per_second": 427.488, | |
| "eval_steps_per_second": 14.64, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_loss": 2.448983907699585, | |
| "eval_runtime": 0.2684, | |
| "eval_samples_per_second": 543.977, | |
| "eval_steps_per_second": 18.629, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_loss": 2.3829126358032227, | |
| "eval_runtime": 0.2785, | |
| "eval_samples_per_second": 524.252, | |
| "eval_steps_per_second": 17.954, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_loss": 2.501011610031128, | |
| "eval_runtime": 0.2671, | |
| "eval_samples_per_second": 546.582, | |
| "eval_steps_per_second": 18.719, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_loss": 2.427177906036377, | |
| "eval_runtime": 0.2718, | |
| "eval_samples_per_second": 537.134, | |
| "eval_steps_per_second": 18.395, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_loss": 2.421048879623413, | |
| "eval_runtime": 0.2742, | |
| "eval_samples_per_second": 532.533, | |
| "eval_steps_per_second": 18.237, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_loss": 2.342533826828003, | |
| "eval_runtime": 0.2891, | |
| "eval_samples_per_second": 505.042, | |
| "eval_steps_per_second": 17.296, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_loss": 2.398080348968506, | |
| "eval_runtime": 0.2794, | |
| "eval_samples_per_second": 522.517, | |
| "eval_steps_per_second": 17.894, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_loss": 2.3011465072631836, | |
| "eval_runtime": 0.2774, | |
| "eval_samples_per_second": 526.407, | |
| "eval_steps_per_second": 18.028, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_loss": 2.5109691619873047, | |
| "eval_runtime": 0.2617, | |
| "eval_samples_per_second": 557.964, | |
| "eval_steps_per_second": 19.108, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_loss": 2.381415843963623, | |
| "eval_runtime": 0.2801, | |
| "eval_samples_per_second": 521.25, | |
| "eval_steps_per_second": 17.851, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_loss": 2.3828046321868896, | |
| "eval_runtime": 0.2743, | |
| "eval_samples_per_second": 532.299, | |
| "eval_steps_per_second": 18.229, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_loss": 2.353680372238159, | |
| "eval_runtime": 0.2779, | |
| "eval_samples_per_second": 525.432, | |
| "eval_steps_per_second": 17.994, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_loss": 2.29482364654541, | |
| "eval_runtime": 0.2832, | |
| "eval_samples_per_second": 515.574, | |
| "eval_steps_per_second": 17.657, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_loss": 2.3079590797424316, | |
| "eval_runtime": 0.3243, | |
| "eval_samples_per_second": 450.187, | |
| "eval_steps_per_second": 15.417, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_loss": 2.379464864730835, | |
| "eval_runtime": 0.2934, | |
| "eval_samples_per_second": 497.692, | |
| "eval_steps_per_second": 17.044, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_loss": 2.37467885017395, | |
| "eval_runtime": 0.2869, | |
| "eval_samples_per_second": 508.855, | |
| "eval_steps_per_second": 17.427, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_loss": 2.3926637172698975, | |
| "eval_runtime": 0.2792, | |
| "eval_samples_per_second": 523.011, | |
| "eval_steps_per_second": 17.911, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_loss": 2.2542331218719482, | |
| "eval_runtime": 0.2705, | |
| "eval_samples_per_second": 539.84, | |
| "eval_steps_per_second": 18.488, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_loss": 2.312037944793701, | |
| "eval_runtime": 0.2823, | |
| "eval_samples_per_second": 517.14, | |
| "eval_steps_per_second": 17.71, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_loss": 2.2595930099487305, | |
| "eval_runtime": 0.2709, | |
| "eval_samples_per_second": 538.869, | |
| "eval_steps_per_second": 18.454, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_loss": 2.3319013118743896, | |
| "eval_runtime": 0.2718, | |
| "eval_samples_per_second": 537.135, | |
| "eval_steps_per_second": 18.395, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_loss": 2.321133852005005, | |
| "eval_runtime": 0.3056, | |
| "eval_samples_per_second": 477.789, | |
| "eval_steps_per_second": 16.363, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_loss": 2.3662209510803223, | |
| "eval_runtime": 0.2727, | |
| "eval_samples_per_second": 535.325, | |
| "eval_steps_per_second": 18.333, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_loss": 2.3607561588287354, | |
| "eval_runtime": 0.2769, | |
| "eval_samples_per_second": 527.269, | |
| "eval_steps_per_second": 18.057, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_loss": 2.2733652591705322, | |
| "eval_runtime": 0.3144, | |
| "eval_samples_per_second": 464.449, | |
| "eval_steps_per_second": 15.906, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_loss": 2.332275390625, | |
| "eval_runtime": 0.2823, | |
| "eval_samples_per_second": 517.096, | |
| "eval_steps_per_second": 17.709, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_loss": 2.3226001262664795, | |
| "eval_runtime": 0.2722, | |
| "eval_samples_per_second": 536.297, | |
| "eval_steps_per_second": 18.366, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_loss": 2.2499899864196777, | |
| "eval_runtime": 0.2823, | |
| "eval_samples_per_second": 517.269, | |
| "eval_steps_per_second": 17.715, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_loss": 2.3148353099823, | |
| "eval_runtime": 0.2735, | |
| "eval_samples_per_second": 533.755, | |
| "eval_steps_per_second": 18.279, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 40.73, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 2.4323, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_loss": 2.190300464630127, | |
| "eval_runtime": 0.279, | |
| "eval_samples_per_second": 523.306, | |
| "eval_steps_per_second": 17.921, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_loss": 2.2688183784484863, | |
| "eval_runtime": 0.2786, | |
| "eval_samples_per_second": 524.054, | |
| "eval_steps_per_second": 17.947, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_loss": 2.3206570148468018, | |
| "eval_runtime": 0.3166, | |
| "eval_samples_per_second": 461.08, | |
| "eval_steps_per_second": 15.79, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_loss": 2.398860454559326, | |
| "eval_runtime": 0.2802, | |
| "eval_samples_per_second": 521.028, | |
| "eval_steps_per_second": 17.843, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_loss": 2.329181432723999, | |
| "eval_runtime": 0.2845, | |
| "eval_samples_per_second": 513.222, | |
| "eval_steps_per_second": 17.576, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_loss": 2.301910877227783, | |
| "eval_runtime": 0.2825, | |
| "eval_samples_per_second": 516.788, | |
| "eval_steps_per_second": 17.698, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_loss": 2.286062002182007, | |
| "eval_runtime": 0.2774, | |
| "eval_samples_per_second": 526.332, | |
| "eval_steps_per_second": 18.025, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_loss": 2.2627930641174316, | |
| "eval_runtime": 0.3491, | |
| "eval_samples_per_second": 418.174, | |
| "eval_steps_per_second": 14.321, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_loss": 2.3683576583862305, | |
| "eval_runtime": 0.2828, | |
| "eval_samples_per_second": 516.226, | |
| "eval_steps_per_second": 17.679, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_loss": 2.3841121196746826, | |
| "eval_runtime": 0.2738, | |
| "eval_samples_per_second": 533.204, | |
| "eval_steps_per_second": 18.26, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_loss": 2.3427212238311768, | |
| "eval_runtime": 0.2841, | |
| "eval_samples_per_second": 513.874, | |
| "eval_steps_per_second": 17.598, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_loss": 2.3786392211914062, | |
| "eval_runtime": 0.2882, | |
| "eval_samples_per_second": 506.555, | |
| "eval_steps_per_second": 17.348, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_loss": 2.3314857482910156, | |
| "eval_runtime": 0.3054, | |
| "eval_samples_per_second": 478.014, | |
| "eval_steps_per_second": 16.37, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_loss": 2.4228127002716064, | |
| "eval_runtime": 0.2752, | |
| "eval_samples_per_second": 530.586, | |
| "eval_steps_per_second": 18.171, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_loss": 2.2979846000671387, | |
| "eval_runtime": 0.2804, | |
| "eval_samples_per_second": 520.739, | |
| "eval_steps_per_second": 17.834, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_loss": 2.288037061691284, | |
| "eval_runtime": 0.2706, | |
| "eval_samples_per_second": 539.536, | |
| "eval_steps_per_second": 18.477, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_loss": 2.375304698944092, | |
| "eval_runtime": 0.2754, | |
| "eval_samples_per_second": 530.218, | |
| "eval_steps_per_second": 18.158, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_loss": 2.302351474761963, | |
| "eval_runtime": 0.2976, | |
| "eval_samples_per_second": 490.644, | |
| "eval_steps_per_second": 16.803, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_loss": 2.2706165313720703, | |
| "eval_runtime": 0.3674, | |
| "eval_samples_per_second": 397.413, | |
| "eval_steps_per_second": 13.61, | |
| "step": 885 | |
| } | |
| ], | |
| "max_steps": 900, | |
| "num_train_epochs": 60, | |
| "total_flos": 196520659812864.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |