| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 80.0, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2910219132900238, | |
| "eval_runtime": 2.9634, | |
| "eval_samples_per_second": 33.745, | |
| "eval_steps_per_second": 4.387, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2910943627357483, | |
| "eval_runtime": 3.0131, | |
| "eval_samples_per_second": 33.188, | |
| "eval_steps_per_second": 4.314, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.2874543070793152, | |
| "eval_runtime": 3.065, | |
| "eval_samples_per_second": 32.626, | |
| "eval_steps_per_second": 4.241, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2908875346183777, | |
| "eval_runtime": 3.0986, | |
| "eval_samples_per_second": 32.272, | |
| "eval_steps_per_second": 4.195, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2935199439525604, | |
| "eval_runtime": 3.1198, | |
| "eval_samples_per_second": 32.053, | |
| "eval_steps_per_second": 4.167, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.29770228266716003, | |
| "eval_runtime": 3.1385, | |
| "eval_samples_per_second": 31.863, | |
| "eval_steps_per_second": 4.142, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.285401850938797, | |
| "eval_runtime": 3.1456, | |
| "eval_samples_per_second": 31.79, | |
| "eval_steps_per_second": 4.133, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.29004549980163574, | |
| "eval_runtime": 3.1554, | |
| "eval_samples_per_second": 31.692, | |
| "eval_steps_per_second": 4.12, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.29845982789993286, | |
| "eval_runtime": 3.157, | |
| "eval_samples_per_second": 31.676, | |
| "eval_steps_per_second": 4.118, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.29055431485176086, | |
| "eval_runtime": 3.1623, | |
| "eval_samples_per_second": 31.623, | |
| "eval_steps_per_second": 4.111, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2978856861591339, | |
| "eval_runtime": 3.1644, | |
| "eval_samples_per_second": 31.601, | |
| "eval_steps_per_second": 4.108, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.28910714387893677, | |
| "eval_runtime": 3.1651, | |
| "eval_samples_per_second": 31.594, | |
| "eval_steps_per_second": 4.107, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.28845351934432983, | |
| "eval_runtime": 3.1653, | |
| "eval_samples_per_second": 31.593, | |
| "eval_steps_per_second": 4.107, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2904435992240906, | |
| "eval_runtime": 3.1679, | |
| "eval_samples_per_second": 31.567, | |
| "eval_steps_per_second": 4.104, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.30564746260643005, | |
| "eval_runtime": 3.1686, | |
| "eval_samples_per_second": 31.56, | |
| "eval_steps_per_second": 4.103, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.286005437374115, | |
| "eval_runtime": 3.1691, | |
| "eval_samples_per_second": 31.555, | |
| "eval_steps_per_second": 4.102, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2886832356452942, | |
| "eval_runtime": 3.1705, | |
| "eval_samples_per_second": 31.541, | |
| "eval_steps_per_second": 4.1, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.296792596578598, | |
| "eval_runtime": 3.1687, | |
| "eval_samples_per_second": 31.559, | |
| "eval_steps_per_second": 4.103, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.29268431663513184, | |
| "eval_runtime": 3.172, | |
| "eval_samples_per_second": 31.526, | |
| "eval_steps_per_second": 4.098, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00075, | |
| "loss": 0.4646, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2887031137943268, | |
| "eval_runtime": 3.1725, | |
| "eval_samples_per_second": 31.52, | |
| "eval_steps_per_second": 4.098, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2917368412017822, | |
| "eval_runtime": 3.1728, | |
| "eval_samples_per_second": 31.518, | |
| "eval_steps_per_second": 4.097, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.29400041699409485, | |
| "eval_runtime": 3.1736, | |
| "eval_samples_per_second": 31.509, | |
| "eval_steps_per_second": 4.096, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2913835346698761, | |
| "eval_runtime": 3.1725, | |
| "eval_samples_per_second": 31.521, | |
| "eval_steps_per_second": 4.098, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28750455379486084, | |
| "eval_runtime": 3.1724, | |
| "eval_samples_per_second": 31.522, | |
| "eval_steps_per_second": 4.098, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2928391695022583, | |
| "eval_runtime": 3.174, | |
| "eval_samples_per_second": 31.506, | |
| "eval_steps_per_second": 4.096, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.2887367010116577, | |
| "eval_runtime": 3.1711, | |
| "eval_samples_per_second": 31.535, | |
| "eval_steps_per_second": 4.1, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28706756234169006, | |
| "eval_runtime": 3.1728, | |
| "eval_samples_per_second": 31.517, | |
| "eval_steps_per_second": 4.097, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.29247915744781494, | |
| "eval_runtime": 3.1743, | |
| "eval_samples_per_second": 31.503, | |
| "eval_steps_per_second": 4.095, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2962961494922638, | |
| "eval_runtime": 3.1784, | |
| "eval_samples_per_second": 31.462, | |
| "eval_steps_per_second": 4.09, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.29220831394195557, | |
| "eval_runtime": 3.1748, | |
| "eval_samples_per_second": 31.498, | |
| "eval_steps_per_second": 4.095, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.29017964005470276, | |
| "eval_runtime": 3.1789, | |
| "eval_samples_per_second": 31.458, | |
| "eval_steps_per_second": 4.089, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28849852085113525, | |
| "eval_runtime": 3.174, | |
| "eval_samples_per_second": 31.506, | |
| "eval_steps_per_second": 4.096, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.29404839873313904, | |
| "eval_runtime": 3.1757, | |
| "eval_samples_per_second": 31.489, | |
| "eval_steps_per_second": 4.094, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.2911713719367981, | |
| "eval_runtime": 3.1765, | |
| "eval_samples_per_second": 31.481, | |
| "eval_steps_per_second": 4.092, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2879158556461334, | |
| "eval_runtime": 3.1778, | |
| "eval_samples_per_second": 31.468, | |
| "eval_steps_per_second": 4.091, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2879682183265686, | |
| "eval_runtime": 3.1764, | |
| "eval_samples_per_second": 31.482, | |
| "eval_steps_per_second": 4.093, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.2944921553134918, | |
| "eval_runtime": 3.1777, | |
| "eval_samples_per_second": 31.47, | |
| "eval_steps_per_second": 4.091, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2917746603488922, | |
| "eval_runtime": 3.1773, | |
| "eval_samples_per_second": 31.474, | |
| "eval_steps_per_second": 4.092, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2887328267097473, | |
| "eval_runtime": 3.1751, | |
| "eval_samples_per_second": 31.495, | |
| "eval_steps_per_second": 4.094, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0005, | |
| "loss": 0.4656, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2873851954936981, | |
| "eval_runtime": 3.1756, | |
| "eval_samples_per_second": 31.49, | |
| "eval_steps_per_second": 4.094, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.28980785608291626, | |
| "eval_runtime": 3.1798, | |
| "eval_samples_per_second": 31.448, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2896897494792938, | |
| "eval_runtime": 3.1763, | |
| "eval_samples_per_second": 31.483, | |
| "eval_steps_per_second": 4.093, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.29236462712287903, | |
| "eval_runtime": 3.1769, | |
| "eval_samples_per_second": 31.477, | |
| "eval_steps_per_second": 4.092, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2897891700267792, | |
| "eval_runtime": 3.177, | |
| "eval_samples_per_second": 31.477, | |
| "eval_steps_per_second": 4.092, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.292123407125473, | |
| "eval_runtime": 3.1798, | |
| "eval_samples_per_second": 31.449, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.2894897758960724, | |
| "eval_runtime": 3.1767, | |
| "eval_samples_per_second": 31.479, | |
| "eval_steps_per_second": 4.092, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2862450182437897, | |
| "eval_runtime": 3.1754, | |
| "eval_samples_per_second": 31.492, | |
| "eval_steps_per_second": 4.094, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.28694215416908264, | |
| "eval_runtime": 3.1812, | |
| "eval_samples_per_second": 31.435, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28551337122917175, | |
| "eval_runtime": 3.1793, | |
| "eval_samples_per_second": 31.453, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2858518958091736, | |
| "eval_runtime": 3.1778, | |
| "eval_samples_per_second": 31.468, | |
| "eval_steps_per_second": 4.091, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28994351625442505, | |
| "eval_runtime": 3.1813, | |
| "eval_samples_per_second": 31.434, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.285138875246048, | |
| "eval_runtime": 3.1778, | |
| "eval_samples_per_second": 31.468, | |
| "eval_steps_per_second": 4.091, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.2851716876029968, | |
| "eval_runtime": 3.1789, | |
| "eval_samples_per_second": 31.457, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2886792719364166, | |
| "eval_runtime": 3.1798, | |
| "eval_samples_per_second": 31.449, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2869800329208374, | |
| "eval_runtime": 3.1792, | |
| "eval_samples_per_second": 31.455, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.289546400308609, | |
| "eval_runtime": 3.1807, | |
| "eval_samples_per_second": 31.439, | |
| "eval_steps_per_second": 4.087, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2893470823764801, | |
| "eval_runtime": 3.1797, | |
| "eval_samples_per_second": 31.449, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2891067862510681, | |
| "eval_runtime": 3.1796, | |
| "eval_samples_per_second": 31.45, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28904733061790466, | |
| "eval_runtime": 3.1792, | |
| "eval_samples_per_second": 31.455, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.00025, | |
| "loss": 0.4637, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28896650671958923, | |
| "eval_runtime": 3.179, | |
| "eval_samples_per_second": 31.457, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2883150577545166, | |
| "eval_runtime": 3.1809, | |
| "eval_samples_per_second": 31.438, | |
| "eval_steps_per_second": 4.087, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2882067561149597, | |
| "eval_runtime": 3.1797, | |
| "eval_samples_per_second": 31.449, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2883051633834839, | |
| "eval_runtime": 3.1845, | |
| "eval_samples_per_second": 31.402, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28842997550964355, | |
| "eval_runtime": 3.1787, | |
| "eval_samples_per_second": 31.459, | |
| "eval_steps_per_second": 4.09, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2875995337963104, | |
| "eval_runtime": 3.1796, | |
| "eval_samples_per_second": 31.45, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28709518909454346, | |
| "eval_runtime": 3.1808, | |
| "eval_samples_per_second": 31.438, | |
| "eval_steps_per_second": 4.087, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2878973186016083, | |
| "eval_runtime": 3.1941, | |
| "eval_samples_per_second": 31.308, | |
| "eval_steps_per_second": 4.07, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28794804215431213, | |
| "eval_runtime": 3.1917, | |
| "eval_samples_per_second": 31.332, | |
| "eval_steps_per_second": 4.073, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.287733256816864, | |
| "eval_runtime": 3.1855, | |
| "eval_samples_per_second": 31.392, | |
| "eval_steps_per_second": 4.081, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2870822250843048, | |
| "eval_runtime": 3.1924, | |
| "eval_samples_per_second": 31.325, | |
| "eval_steps_per_second": 4.072, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2874806523323059, | |
| "eval_runtime": 3.1929, | |
| "eval_samples_per_second": 31.319, | |
| "eval_steps_per_second": 4.072, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2870163321495056, | |
| "eval_runtime": 3.1935, | |
| "eval_samples_per_second": 31.314, | |
| "eval_steps_per_second": 4.071, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2874850928783417, | |
| "eval_runtime": 3.1878, | |
| "eval_samples_per_second": 31.369, | |
| "eval_steps_per_second": 4.078, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2879111170768738, | |
| "eval_runtime": 3.1939, | |
| "eval_samples_per_second": 31.31, | |
| "eval_steps_per_second": 4.07, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2886711359024048, | |
| "eval_runtime": 3.1977, | |
| "eval_samples_per_second": 31.272, | |
| "eval_steps_per_second": 4.065, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28828713297843933, | |
| "eval_runtime": 3.1974, | |
| "eval_samples_per_second": 31.275, | |
| "eval_steps_per_second": 4.066, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28815144300460815, | |
| "eval_runtime": 3.1962, | |
| "eval_samples_per_second": 31.287, | |
| "eval_steps_per_second": 4.067, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28827354311943054, | |
| "eval_runtime": 3.197, | |
| "eval_samples_per_second": 31.279, | |
| "eval_steps_per_second": 4.066, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2883530855178833, | |
| "eval_runtime": 3.1937, | |
| "eval_samples_per_second": 31.311, | |
| "eval_steps_per_second": 4.07, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4587, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28832393884658813, | |
| "eval_runtime": 3.1941, | |
| "eval_samples_per_second": 31.308, | |
| "eval_steps_per_second": 4.07, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.46315200805664064, | |
| "train_runtime": 1695.4283, | |
| "train_samples_per_second": 18.874, | |
| "train_steps_per_second": 1.18 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |