| { | |
| "best_metric": 1.0, | |
| "best_model_checkpoint": "ai_detector_v3/checkpoint-1709", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 2279, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4619883040935671e-06, | |
| "loss": 1.3185, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.9239766081871343e-06, | |
| "loss": 0.9483, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.3859649122807014e-06, | |
| "loss": 0.591, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.8479532163742686e-06, | |
| "loss": 0.3762, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 7.3099415204678366e-06, | |
| "loss": 0.2428, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.771929824561403e-06, | |
| "loss": 0.1249, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.023391812865497e-05, | |
| "loss": 0.0659, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.1695906432748537e-05, | |
| "loss": 0.0381, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.3157894736842106e-05, | |
| "loss": 0.0321, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.4619883040935673e-05, | |
| "loss": 0.0301, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.608187134502924e-05, | |
| "loss": 0.0395, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7543859649122806e-05, | |
| "loss": 0.0197, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9005847953216373e-05, | |
| "loss": 0.0226, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.046783625730994e-05, | |
| "loss": 0.0233, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.1929824561403507e-05, | |
| "loss": 0.0098, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.3391812865497074e-05, | |
| "loss": 0.0188, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.485380116959064e-05, | |
| "loss": 0.014, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.6315789473684212e-05, | |
| "loss": 0.0354, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 0.0126, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9239766081871346e-05, | |
| "loss": 0.0063, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.0701754385964913e-05, | |
| "loss": 0.0325, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.216374269005848e-05, | |
| "loss": 0.0139, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.362573099415205e-05, | |
| "loss": 0.0258, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.508771929824561e-05, | |
| "loss": 0.032, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.654970760233918e-05, | |
| "loss": 0.0188, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.8011695906432746e-05, | |
| "loss": 0.0005, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.9473684210526316e-05, | |
| "loss": 0.0124, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.093567251461988e-05, | |
| "loss": 0.0615, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.239766081871345e-05, | |
| "loss": 0.0297, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.3859649122807014e-05, | |
| "loss": 0.0387, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.5321637426900585e-05, | |
| "loss": 0.0151, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.678362573099415e-05, | |
| "loss": 0.0329, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.824561403508772e-05, | |
| "loss": 0.0155, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.970760233918128e-05, | |
| "loss": 0.019, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.986979166666667e-05, | |
| "loss": 0.0359, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.970703125e-05, | |
| "loss": 0.0117, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.954427083333333e-05, | |
| "loss": 0.0169, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.938151041666667e-05, | |
| "loss": 0.0084, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.921875e-05, | |
| "loss": 0.0178, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.905598958333333e-05, | |
| "loss": 0.0063, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.889322916666667e-05, | |
| "loss": 0.034, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.873046875e-05, | |
| "loss": 0.0216, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.856770833333333e-05, | |
| "loss": 0.003, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.840494791666667e-05, | |
| "loss": 0.0081, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.82421875e-05, | |
| "loss": 0.009, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.807942708333333e-05, | |
| "loss": 0.0015, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.791666666666667e-05, | |
| "loss": 0.0056, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.775390625e-05, | |
| "loss": 0.0179, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.759114583333334e-05, | |
| "loss": 0.0051, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.742838541666667e-05, | |
| "loss": 0.0005, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.7265625000000005e-05, | |
| "loss": 0.0244, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.7102864583333336e-05, | |
| "loss": 0.0008, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.6940104166666674e-05, | |
| "loss": 0.0074, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.6777343750000005e-05, | |
| "loss": 0.0083, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.6614583333333336e-05, | |
| "loss": 0.0139, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.6451822916666674e-05, | |
| "loss": 0.0243, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9985, | |
| "eval_loss": 0.00362112233415246, | |
| "eval_runtime": 132.255, | |
| "eval_samples_per_second": 15.122, | |
| "eval_steps_per_second": 0.945, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.6289062500000005e-05, | |
| "loss": 0.0081, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.6126302083333335e-05, | |
| "loss": 0.0097, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.596354166666667e-05, | |
| "loss": 0.0084, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.5800781250000004e-05, | |
| "loss": 0.0032, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.5638020833333335e-05, | |
| "loss": 0.017, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.547526041666667e-05, | |
| "loss": 0.0099, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.5312500000000004e-05, | |
| "loss": 0.0126, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.5149739583333335e-05, | |
| "loss": 0.0001, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.498697916666667e-05, | |
| "loss": 0.0009, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.482421875e-05, | |
| "loss": 0.0134, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.4661458333333334e-05, | |
| "loss": 0.0115, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.449869791666667e-05, | |
| "loss": 0.0019, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.43359375e-05, | |
| "loss": 0.0005, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.4173177083333334e-05, | |
| "loss": 0.0229, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.401041666666667e-05, | |
| "loss": 0.0052, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.384765625e-05, | |
| "loss": 0.0088, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.368489583333333e-05, | |
| "loss": 0.0252, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.352213541666667e-05, | |
| "loss": 0.0143, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.3359375e-05, | |
| "loss": 0.0051, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.319661458333333e-05, | |
| "loss": 0.0282, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.303385416666667e-05, | |
| "loss": 0.0107, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.287109375e-05, | |
| "loss": 0.022, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.270833333333333e-05, | |
| "loss": 0.0206, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.254557291666667e-05, | |
| "loss": 0.0218, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.23828125e-05, | |
| "loss": 0.0038, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.222005208333333e-05, | |
| "loss": 0.0176, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.205729166666667e-05, | |
| "loss": 0.0154, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.189453125e-05, | |
| "loss": 0.0066, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.173177083333333e-05, | |
| "loss": 0.013, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.156901041666667e-05, | |
| "loss": 0.0027, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.140625e-05, | |
| "loss": 0.0085, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.124348958333333e-05, | |
| "loss": 0.0099, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.108072916666667e-05, | |
| "loss": 0.0059, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.091796875e-05, | |
| "loss": 0.0067, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.075520833333333e-05, | |
| "loss": 0.0073, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.059244791666667e-05, | |
| "loss": 0.006, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.04296875e-05, | |
| "loss": 0.0022, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.026692708333333e-05, | |
| "loss": 0.0007, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.010416666666667e-05, | |
| "loss": 0.0031, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.994140625e-05, | |
| "loss": 0.0107, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.9778645833333337e-05, | |
| "loss": 0.0582, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.961588541666667e-05, | |
| "loss": 0.003, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.9453125000000005e-05, | |
| "loss": 0.0234, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.9290364583333336e-05, | |
| "loss": 0.0026, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.912760416666667e-05, | |
| "loss": 0.0067, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.8964843750000005e-05, | |
| "loss": 0.0079, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 3.8802083333333336e-05, | |
| "loss": 0.0045, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.8639322916666673e-05, | |
| "loss": 0.0007, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.8476562500000004e-05, | |
| "loss": 0.0058, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.8313802083333335e-05, | |
| "loss": 0.0004, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.815104166666667e-05, | |
| "loss": 0.0005, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.7988281250000004e-05, | |
| "loss": 0.01, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.7825520833333335e-05, | |
| "loss": 0.0002, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.766276041666667e-05, | |
| "loss": 0.0137, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.0002, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.7337239583333334e-05, | |
| "loss": 0.0067, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.717447916666667e-05, | |
| "loss": 0.0002, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9995, | |
| "eval_loss": 0.0011123190633952618, | |
| "eval_runtime": 135.6796, | |
| "eval_samples_per_second": 14.741, | |
| "eval_steps_per_second": 0.921, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.701171875e-05, | |
| "loss": 0.0006, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.6848958333333334e-05, | |
| "loss": 0.0001, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.668619791666667e-05, | |
| "loss": 0.0001, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.65234375e-05, | |
| "loss": 0.0002, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.6360677083333334e-05, | |
| "loss": 0.0001, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.619791666666667e-05, | |
| "loss": 0.004, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.603515625e-05, | |
| "loss": 0.0038, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.587239583333333e-05, | |
| "loss": 0.0234, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.570963541666667e-05, | |
| "loss": 0.0009, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.5546875e-05, | |
| "loss": 0.0022, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.538411458333333e-05, | |
| "loss": 0.0001, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.522135416666667e-05, | |
| "loss": 0.0003, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.505859375e-05, | |
| "loss": 0.0071, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.489583333333333e-05, | |
| "loss": 0.0004, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.473307291666667e-05, | |
| "loss": 0.0008, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.45703125e-05, | |
| "loss": 0.0048, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.440755208333333e-05, | |
| "loss": 0.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.424479166666667e-05, | |
| "loss": 0.0, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.408203125e-05, | |
| "loss": 0.0, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.391927083333333e-05, | |
| "loss": 0.0, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.375651041666667e-05, | |
| "loss": 0.0, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.359375e-05, | |
| "loss": 0.0002, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 3.343098958333333e-05, | |
| "loss": 0.0001, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.326822916666667e-05, | |
| "loss": 0.0019, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.310546875e-05, | |
| "loss": 0.0001, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.294270833333333e-05, | |
| "loss": 0.0, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.277994791666667e-05, | |
| "loss": 0.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.26171875e-05, | |
| "loss": 0.0, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.245442708333333e-05, | |
| "loss": 0.0, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.229166666666667e-05, | |
| "loss": 0.0, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.212890625e-05, | |
| "loss": 0.0, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.196614583333333e-05, | |
| "loss": 0.0022, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.180338541666667e-05, | |
| "loss": 0.0152, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.1640625e-05, | |
| "loss": 0.0022, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.1477864583333336e-05, | |
| "loss": 0.0002, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.131510416666667e-05, | |
| "loss": 0.0008, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.1152343750000005e-05, | |
| "loss": 0.0003, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.0989583333333336e-05, | |
| "loss": 0.0051, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.082682291666667e-05, | |
| "loss": 0.0, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.0664062500000004e-05, | |
| "loss": 0.0, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 3.0501302083333332e-05, | |
| "loss": 0.0035, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.033854166666667e-05, | |
| "loss": 0.0084, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.017578125e-05, | |
| "loss": 0.0002, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.001302083333333e-05, | |
| "loss": 0.0168, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.985026041666667e-05, | |
| "loss": 0.0019, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.96875e-05, | |
| "loss": 0.0002, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 2.9524739583333334e-05, | |
| "loss": 0.0005, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.936197916666667e-05, | |
| "loss": 0.001, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.9199218750000003e-05, | |
| "loss": 0.0, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.9036458333333334e-05, | |
| "loss": 0.0, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.887369791666667e-05, | |
| "loss": 0.0, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 2.8710937500000002e-05, | |
| "loss": 0.002, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.8548177083333333e-05, | |
| "loss": 0.0, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.838541666666667e-05, | |
| "loss": 0.0, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.8222656250000002e-05, | |
| "loss": 0.0, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.8059895833333333e-05, | |
| "loss": 0.0, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.789713541666667e-05, | |
| "loss": 0.0006, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 1.0, | |
| "eval_loss": 9.786892042029649e-05, | |
| "eval_runtime": 133.229, | |
| "eval_samples_per_second": 15.012, | |
| "eval_steps_per_second": 0.938, | |
| "step": 1709 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2.7734375e-05, | |
| "loss": 0.0017, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 2.7571614583333333e-05, | |
| "loss": 0.001, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.740885416666667e-05, | |
| "loss": 0.0195, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.724609375e-05, | |
| "loss": 0.0001, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.7083333333333332e-05, | |
| "loss": 0.0141, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.692057291666667e-05, | |
| "loss": 0.0002, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.67578125e-05, | |
| "loss": 0.0072, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.659505208333333e-05, | |
| "loss": 0.0, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.643229166666667e-05, | |
| "loss": 0.0001, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.626953125e-05, | |
| "loss": 0.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.610677083333333e-05, | |
| "loss": 0.0, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.594401041666667e-05, | |
| "loss": 0.0001, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.578125e-05, | |
| "loss": 0.0098, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 2.5618489583333334e-05, | |
| "loss": 0.0014, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 2.545572916666667e-05, | |
| "loss": 0.0003, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 2.5292968750000003e-05, | |
| "loss": 0.0021, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 2.5130208333333334e-05, | |
| "loss": 0.0, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 2.4967447916666668e-05, | |
| "loss": 0.0005, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 2.4804687500000002e-05, | |
| "loss": 0.0054, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 2.4641927083333337e-05, | |
| "loss": 0.0004, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 2.4479166666666668e-05, | |
| "loss": 0.0, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 2.4316406250000002e-05, | |
| "loss": 0.0, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 2.4153645833333336e-05, | |
| "loss": 0.0, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 2.3990885416666667e-05, | |
| "loss": 0.0, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 2.3828125e-05, | |
| "loss": 0.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 2.3665364583333336e-05, | |
| "loss": 0.0053, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 2.3502604166666667e-05, | |
| "loss": 0.0025, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 2.333984375e-05, | |
| "loss": 0.0002, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 2.3177083333333335e-05, | |
| "loss": 0.0001, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 2.3014322916666666e-05, | |
| "loss": 0.0019, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 2.28515625e-05, | |
| "loss": 0.0, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 2.2688802083333335e-05, | |
| "loss": 0.0001, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.2526041666666666e-05, | |
| "loss": 0.0031, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.236328125e-05, | |
| "loss": 0.0003, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 2.2200520833333335e-05, | |
| "loss": 0.0002, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 2.2037760416666665e-05, | |
| "loss": 0.0109, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 2.1875e-05, | |
| "loss": 0.0001, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.1712239583333334e-05, | |
| "loss": 0.0014, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 2.154947916666667e-05, | |
| "loss": 0.0001, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 2.138671875e-05, | |
| "loss": 0.0013, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.1223958333333334e-05, | |
| "loss": 0.0, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.1061197916666668e-05, | |
| "loss": 0.0, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.0898437500000002e-05, | |
| "loss": 0.0, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 2.0735677083333337e-05, | |
| "loss": 0.0, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 2.0572916666666668e-05, | |
| "loss": 0.0, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 2.0410156250000002e-05, | |
| "loss": 0.0, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 2.0247395833333336e-05, | |
| "loss": 0.0195, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.0084635416666667e-05, | |
| "loss": 0.02, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 1.9921875e-05, | |
| "loss": 0.0, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 1.9759114583333336e-05, | |
| "loss": 0.0, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 1.9596354166666667e-05, | |
| "loss": 0.0, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 1.943359375e-05, | |
| "loss": 0.0, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.9270833333333335e-05, | |
| "loss": 0.0008, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.9108072916666666e-05, | |
| "loss": 0.0, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.89453125e-05, | |
| "loss": 0.0002, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.8782552083333335e-05, | |
| "loss": 0.0055, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.8619791666666666e-05, | |
| "loss": 0.0002, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 1.0, | |
| "eval_loss": 4.17689407186117e-05, | |
| "eval_runtime": 133.2569, | |
| "eval_samples_per_second": 15.009, | |
| "eval_steps_per_second": 0.938, | |
| "step": 2279 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3414, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "save_steps": 500, | |
| "total_flos": 1.1427169361682235e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |