| { | |
| "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": 0.45, | |
| "eval_loss": 0.46248623728752136, | |
| "eval_runtime": 2.8961, | |
| "eval_samples_per_second": 34.53, | |
| "eval_steps_per_second": 4.489, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.45, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.48594656586647034, | |
| "eval_runtime": 2.9335, | |
| "eval_samples_per_second": 34.089, | |
| "eval_steps_per_second": 4.432, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4227062165737152, | |
| "eval_runtime": 2.9668, | |
| "eval_samples_per_second": 33.707, | |
| "eval_steps_per_second": 4.382, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.4247153103351593, | |
| "eval_runtime": 2.9923, | |
| "eval_samples_per_second": 33.419, | |
| "eval_steps_per_second": 4.344, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.44806617498397827, | |
| "eval_runtime": 3.0015, | |
| "eval_samples_per_second": 33.317, | |
| "eval_steps_per_second": 4.331, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.4309893846511841, | |
| "eval_runtime": 3.023, | |
| "eval_samples_per_second": 33.079, | |
| "eval_steps_per_second": 4.3, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4267328679561615, | |
| "eval_runtime": 3.0288, | |
| "eval_samples_per_second": 33.017, | |
| "eval_steps_per_second": 4.292, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.42460113763809204, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.426717609167099, | |
| "eval_runtime": 3.0248, | |
| "eval_samples_per_second": 33.06, | |
| "eval_steps_per_second": 4.298, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4260082542896271, | |
| "eval_runtime": 3.0266, | |
| "eval_samples_per_second": 33.04, | |
| "eval_steps_per_second": 4.295, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.42262589931488037, | |
| "eval_runtime": 3.027, | |
| "eval_samples_per_second": 33.037, | |
| "eval_steps_per_second": 4.295, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.427063524723053, | |
| "eval_runtime": 3.0293, | |
| "eval_samples_per_second": 33.011, | |
| "eval_steps_per_second": 4.291, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.42662951350212097, | |
| "eval_runtime": 3.0315, | |
| "eval_samples_per_second": 32.987, | |
| "eval_steps_per_second": 4.288, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.42443209886550903, | |
| "eval_runtime": 3.0363, | |
| "eval_samples_per_second": 32.935, | |
| "eval_steps_per_second": 4.282, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.42530936002731323, | |
| "eval_runtime": 3.0344, | |
| "eval_samples_per_second": 32.956, | |
| "eval_steps_per_second": 4.284, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4255802631378174, | |
| "eval_runtime": 3.0396, | |
| "eval_samples_per_second": 32.9, | |
| "eval_steps_per_second": 4.277, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4265216588973999, | |
| "eval_runtime": 3.0416, | |
| "eval_samples_per_second": 32.877, | |
| "eval_steps_per_second": 4.274, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.42613503336906433, | |
| "eval_runtime": 3.0532, | |
| "eval_samples_per_second": 32.753, | |
| "eval_steps_per_second": 4.258, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.4262290298938751, | |
| "eval_runtime": 3.0418, | |
| "eval_samples_per_second": 32.875, | |
| "eval_steps_per_second": 4.274, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.037500000000000006, | |
| "loss": 1.4009, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4260247051715851, | |
| "eval_runtime": 3.0426, | |
| "eval_samples_per_second": 32.867, | |
| "eval_steps_per_second": 4.273, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.4285390377044678, | |
| "eval_runtime": 3.0444, | |
| "eval_samples_per_second": 32.848, | |
| "eval_steps_per_second": 4.27, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.4259931147098541, | |
| "eval_runtime": 3.0428, | |
| "eval_samples_per_second": 32.865, | |
| "eval_steps_per_second": 4.272, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.42451244592666626, | |
| "eval_runtime": 3.0442, | |
| "eval_samples_per_second": 32.849, | |
| "eval_steps_per_second": 4.27, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.42513129115104675, | |
| "eval_runtime": 3.0467, | |
| "eval_samples_per_second": 32.823, | |
| "eval_steps_per_second": 4.267, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.42705675959587097, | |
| "eval_runtime": 3.0463, | |
| "eval_samples_per_second": 32.827, | |
| "eval_steps_per_second": 4.268, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.4261193871498108, | |
| "eval_runtime": 3.0479, | |
| "eval_samples_per_second": 32.81, | |
| "eval_steps_per_second": 4.265, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.42566367983818054, | |
| "eval_runtime": 3.0456, | |
| "eval_samples_per_second": 32.834, | |
| "eval_steps_per_second": 4.268, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.42552897334098816, | |
| "eval_runtime": 3.0459, | |
| "eval_samples_per_second": 32.831, | |
| "eval_steps_per_second": 4.268, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.42543452978134155, | |
| "eval_runtime": 3.0492, | |
| "eval_samples_per_second": 32.796, | |
| "eval_steps_per_second": 4.263, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.4259891211986542, | |
| "eval_runtime": 3.0481, | |
| "eval_samples_per_second": 32.808, | |
| "eval_steps_per_second": 4.265, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.4255729913711548, | |
| "eval_runtime": 3.0503, | |
| "eval_samples_per_second": 32.784, | |
| "eval_steps_per_second": 4.262, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.4256853759288788, | |
| "eval_runtime": 3.0494, | |
| "eval_samples_per_second": 32.793, | |
| "eval_steps_per_second": 4.263, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.42549124360084534, | |
| "eval_runtime": 3.0483, | |
| "eval_samples_per_second": 32.806, | |
| "eval_steps_per_second": 4.265, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.4255514144897461, | |
| "eval_runtime": 3.0467, | |
| "eval_samples_per_second": 32.823, | |
| "eval_steps_per_second": 4.267, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4261658191680908, | |
| "eval_runtime": 3.0508, | |
| "eval_samples_per_second": 32.778, | |
| "eval_steps_per_second": 4.261, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4256916344165802, | |
| "eval_runtime": 3.0492, | |
| "eval_samples_per_second": 32.795, | |
| "eval_steps_per_second": 4.263, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.42672908306121826, | |
| "eval_runtime": 3.0502, | |
| "eval_samples_per_second": 32.785, | |
| "eval_steps_per_second": 4.262, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.4258723556995392, | |
| "eval_runtime": 3.0474, | |
| "eval_samples_per_second": 32.815, | |
| "eval_steps_per_second": 4.266, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.4254632890224457, | |
| "eval_runtime": 3.0495, | |
| "eval_samples_per_second": 32.792, | |
| "eval_steps_per_second": 4.263, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.025, | |
| "loss": 0.9833, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.42543506622314453, | |
| "eval_runtime": 3.0505, | |
| "eval_samples_per_second": 32.782, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.4257011413574219, | |
| "eval_runtime": 3.0503, | |
| "eval_samples_per_second": 32.783, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.42543143033981323, | |
| "eval_runtime": 3.0485, | |
| "eval_samples_per_second": 32.803, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.42607200145721436, | |
| "eval_runtime": 3.0551, | |
| "eval_samples_per_second": 32.732, | |
| "eval_steps_per_second": 4.255, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.4259500503540039, | |
| "eval_runtime": 3.0508, | |
| "eval_samples_per_second": 32.779, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.42568397521972656, | |
| "eval_runtime": 3.0487, | |
| "eval_samples_per_second": 32.801, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.4254191517829895, | |
| "eval_runtime": 3.0504, | |
| "eval_samples_per_second": 32.782, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.42553916573524475, | |
| "eval_runtime": 3.0517, | |
| "eval_samples_per_second": 32.769, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.42568346858024597, | |
| "eval_runtime": 3.05, | |
| "eval_samples_per_second": 32.787, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.42611098289489746, | |
| "eval_runtime": 3.0501, | |
| "eval_samples_per_second": 32.786, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.42508599162101746, | |
| "eval_runtime": 3.0503, | |
| "eval_samples_per_second": 32.784, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4258117377758026, | |
| "eval_runtime": 3.0511, | |
| "eval_samples_per_second": 32.775, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.42553529143333435, | |
| "eval_runtime": 3.0496, | |
| "eval_samples_per_second": 32.792, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.42570650577545166, | |
| "eval_runtime": 3.0501, | |
| "eval_samples_per_second": 32.786, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.4255848228931427, | |
| "eval_runtime": 3.0507, | |
| "eval_samples_per_second": 32.78, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4257044196128845, | |
| "eval_runtime": 3.0575, | |
| "eval_samples_per_second": 32.706, | |
| "eval_steps_per_second": 4.252, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.4256804585456848, | |
| "eval_runtime": 3.0497, | |
| "eval_samples_per_second": 32.79, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.4257115423679352, | |
| "eval_runtime": 3.0519, | |
| "eval_samples_per_second": 32.766, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4257067143917084, | |
| "eval_runtime": 3.0511, | |
| "eval_samples_per_second": 32.775, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.42552703619003296, | |
| "eval_runtime": 3.052, | |
| "eval_samples_per_second": 32.765, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0125, | |
| "loss": 0.7428, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.42590823769569397, | |
| "eval_runtime": 3.0534, | |
| "eval_samples_per_second": 32.75, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.4257360100746155, | |
| "eval_runtime": 3.0514, | |
| "eval_samples_per_second": 32.772, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.4255290627479553, | |
| "eval_runtime": 3.0557, | |
| "eval_samples_per_second": 32.725, | |
| "eval_steps_per_second": 4.254, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.42556220293045044, | |
| "eval_runtime": 3.05, | |
| "eval_samples_per_second": 32.787, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4258395731449127, | |
| "eval_runtime": 3.0496, | |
| "eval_samples_per_second": 32.791, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4258385896682739, | |
| "eval_runtime": 3.0514, | |
| "eval_samples_per_second": 32.771, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4258986711502075, | |
| "eval_runtime": 3.0504, | |
| "eval_samples_per_second": 32.782, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.4259600341320038, | |
| "eval_runtime": 3.0516, | |
| "eval_samples_per_second": 32.77, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.4256523549556732, | |
| "eval_runtime": 3.0526, | |
| "eval_samples_per_second": 32.759, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.35, | |
| "eval_loss": 0.42585432529449463, | |
| "eval_runtime": 3.0509, | |
| "eval_samples_per_second": 32.777, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.4259151518344879, | |
| "eval_runtime": 3.0508, | |
| "eval_samples_per_second": 32.779, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4259404242038727, | |
| "eval_runtime": 3.052, | |
| "eval_samples_per_second": 32.766, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.42599546909332275, | |
| "eval_runtime": 3.0519, | |
| "eval_samples_per_second": 32.766, | |
| "eval_steps_per_second": 4.26, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.42574265599250793, | |
| "eval_runtime": 3.0522, | |
| "eval_samples_per_second": 32.764, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.4257790446281433, | |
| "eval_runtime": 3.0506, | |
| "eval_samples_per_second": 32.781, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4257981479167938, | |
| "eval_runtime": 3.0529, | |
| "eval_samples_per_second": 32.755, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4257836639881134, | |
| "eval_runtime": 3.0536, | |
| "eval_samples_per_second": 32.748, | |
| "eval_steps_per_second": 4.257, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.42581596970558167, | |
| "eval_runtime": 3.0529, | |
| "eval_samples_per_second": 32.756, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.4258066415786743, | |
| "eval_runtime": 3.0527, | |
| "eval_samples_per_second": 32.758, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4258076846599579, | |
| "eval_runtime": 3.0537, | |
| "eval_samples_per_second": 32.747, | |
| "eval_steps_per_second": 4.257, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.6138, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4258350431919098, | |
| "eval_runtime": 3.0537, | |
| "eval_samples_per_second": 32.747, | |
| "eval_steps_per_second": 4.257, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.9352073516845704, | |
| "train_runtime": 1633.4708, | |
| "train_samples_per_second": 19.59, | |
| "train_steps_per_second": 1.224 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |