| { | |
| "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": 0.4, | |
| "eval_loss": 0.6199389696121216, | |
| "eval_runtime": 2.9463, | |
| "eval_samples_per_second": 33.94, | |
| "eval_steps_per_second": 4.412, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.4, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.664335310459137, | |
| "eval_runtime": 2.9957, | |
| "eval_samples_per_second": 33.381, | |
| "eval_steps_per_second": 4.34, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.59, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.5066738724708557, | |
| "eval_runtime": 3.0312, | |
| "eval_samples_per_second": 32.99, | |
| "eval_steps_per_second": 4.289, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.59, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4272070825099945, | |
| "eval_runtime": 3.0601, | |
| "eval_samples_per_second": 32.679, | |
| "eval_steps_per_second": 4.248, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.4341088831424713, | |
| "eval_runtime": 3.0787, | |
| "eval_samples_per_second": 32.482, | |
| "eval_steps_per_second": 4.223, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4487706124782562, | |
| "eval_runtime": 3.0875, | |
| "eval_samples_per_second": 32.388, | |
| "eval_steps_per_second": 4.21, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4091510474681854, | |
| "eval_runtime": 3.0921, | |
| "eval_samples_per_second": 32.34, | |
| "eval_steps_per_second": 4.204, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4563831686973572, | |
| "eval_runtime": 3.0938, | |
| "eval_samples_per_second": 32.322, | |
| "eval_steps_per_second": 4.202, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.43667757511138916, | |
| "eval_runtime": 3.0893, | |
| "eval_samples_per_second": 32.37, | |
| "eval_steps_per_second": 4.208, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.4343050420284271, | |
| "eval_runtime": 3.0861, | |
| "eval_samples_per_second": 32.403, | |
| "eval_steps_per_second": 4.212, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.41208839416503906, | |
| "eval_runtime": 3.0842, | |
| "eval_samples_per_second": 32.423, | |
| "eval_steps_per_second": 4.215, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.4299950897693634, | |
| "eval_runtime": 3.0824, | |
| "eval_samples_per_second": 32.442, | |
| "eval_steps_per_second": 4.217, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.423895925283432, | |
| "eval_runtime": 3.0828, | |
| "eval_samples_per_second": 32.438, | |
| "eval_steps_per_second": 4.217, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.4147571325302124, | |
| "eval_runtime": 3.0847, | |
| "eval_samples_per_second": 32.418, | |
| "eval_steps_per_second": 4.214, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.43110018968582153, | |
| "eval_runtime": 3.0818, | |
| "eval_samples_per_second": 32.449, | |
| "eval_steps_per_second": 4.218, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.41430824995040894, | |
| "eval_runtime": 3.0802, | |
| "eval_samples_per_second": 32.465, | |
| "eval_steps_per_second": 4.22, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.4166063964366913, | |
| "eval_runtime": 3.0812, | |
| "eval_samples_per_second": 32.455, | |
| "eval_steps_per_second": 4.219, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.41197219491004944, | |
| "eval_runtime": 3.0825, | |
| "eval_samples_per_second": 32.441, | |
| "eval_steps_per_second": 4.217, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4121343493461609, | |
| "eval_runtime": 3.0809, | |
| "eval_samples_per_second": 32.458, | |
| "eval_steps_per_second": 4.22, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.6423, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.40662243962287903, | |
| "eval_runtime": 3.0833, | |
| "eval_samples_per_second": 32.433, | |
| "eval_steps_per_second": 4.216, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.404727965593338, | |
| "eval_runtime": 3.0808, | |
| "eval_samples_per_second": 32.459, | |
| "eval_steps_per_second": 4.22, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4215034246444702, | |
| "eval_runtime": 3.0822, | |
| "eval_samples_per_second": 32.444, | |
| "eval_steps_per_second": 4.218, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4074056148529053, | |
| "eval_runtime": 3.0831, | |
| "eval_samples_per_second": 32.435, | |
| "eval_steps_per_second": 4.217, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.40676093101501465, | |
| "eval_runtime": 3.0829, | |
| "eval_samples_per_second": 32.437, | |
| "eval_steps_per_second": 4.217, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4190574586391449, | |
| "eval_runtime": 3.0842, | |
| "eval_samples_per_second": 32.423, | |
| "eval_steps_per_second": 4.215, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.4035128355026245, | |
| "eval_runtime": 3.0849, | |
| "eval_samples_per_second": 32.416, | |
| "eval_steps_per_second": 4.214, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.4227715730667114, | |
| "eval_runtime": 3.0831, | |
| "eval_samples_per_second": 32.435, | |
| "eval_steps_per_second": 4.216, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4241797626018524, | |
| "eval_runtime": 3.082, | |
| "eval_samples_per_second": 32.446, | |
| "eval_steps_per_second": 4.218, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.42384228110313416, | |
| "eval_runtime": 3.0835, | |
| "eval_samples_per_second": 32.431, | |
| "eval_steps_per_second": 4.216, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.47883549332618713, | |
| "eval_runtime": 3.086, | |
| "eval_samples_per_second": 32.405, | |
| "eval_steps_per_second": 4.213, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.42142295837402344, | |
| "eval_runtime": 3.0841, | |
| "eval_samples_per_second": 32.424, | |
| "eval_steps_per_second": 4.215, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.42832931876182556, | |
| "eval_runtime": 3.0875, | |
| "eval_samples_per_second": 32.388, | |
| "eval_steps_per_second": 4.211, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.4221520721912384, | |
| "eval_runtime": 3.0864, | |
| "eval_samples_per_second": 32.4, | |
| "eval_steps_per_second": 4.212, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4233008623123169, | |
| "eval_runtime": 3.086, | |
| "eval_samples_per_second": 32.404, | |
| "eval_steps_per_second": 4.213, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4400712549686432, | |
| "eval_runtime": 3.0875, | |
| "eval_samples_per_second": 32.388, | |
| "eval_steps_per_second": 4.21, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4583715796470642, | |
| "eval_runtime": 3.0883, | |
| "eval_samples_per_second": 32.38, | |
| "eval_steps_per_second": 4.209, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.43615418672561646, | |
| "eval_runtime": 3.0876, | |
| "eval_samples_per_second": 32.388, | |
| "eval_steps_per_second": 4.21, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.3988666534423828, | |
| "eval_runtime": 3.0861, | |
| "eval_samples_per_second": 32.403, | |
| "eval_steps_per_second": 4.212, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4379017651081085, | |
| "eval_runtime": 3.0871, | |
| "eval_samples_per_second": 32.393, | |
| "eval_steps_per_second": 4.211, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.5234, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.40936079621315, | |
| "eval_runtime": 3.0882, | |
| "eval_samples_per_second": 32.382, | |
| "eval_steps_per_second": 4.21, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4683462381362915, | |
| "eval_runtime": 3.0846, | |
| "eval_samples_per_second": 32.419, | |
| "eval_steps_per_second": 4.214, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.43595993518829346, | |
| "eval_runtime": 3.0964, | |
| "eval_samples_per_second": 32.296, | |
| "eval_steps_per_second": 4.198, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.43824779987335205, | |
| "eval_runtime": 3.0945, | |
| "eval_samples_per_second": 32.315, | |
| "eval_steps_per_second": 4.201, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4057212769985199, | |
| "eval_runtime": 3.0836, | |
| "eval_samples_per_second": 32.43, | |
| "eval_steps_per_second": 4.216, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.42997491359710693, | |
| "eval_runtime": 3.0845, | |
| "eval_samples_per_second": 32.42, | |
| "eval_steps_per_second": 4.215, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42525675892829895, | |
| "eval_runtime": 3.0823, | |
| "eval_samples_per_second": 32.443, | |
| "eval_steps_per_second": 4.218, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.43460795283317566, | |
| "eval_runtime": 3.0847, | |
| "eval_samples_per_second": 32.418, | |
| "eval_steps_per_second": 4.214, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4166887402534485, | |
| "eval_runtime": 3.0845, | |
| "eval_samples_per_second": 32.42, | |
| "eval_steps_per_second": 4.215, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.4572359025478363, | |
| "eval_runtime": 3.0805, | |
| "eval_samples_per_second": 32.462, | |
| "eval_steps_per_second": 4.22, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4412834644317627, | |
| "eval_runtime": 3.081, | |
| "eval_samples_per_second": 32.457, | |
| "eval_steps_per_second": 4.219, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4160061776638031, | |
| "eval_runtime": 3.0802, | |
| "eval_samples_per_second": 32.465, | |
| "eval_steps_per_second": 4.22, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.40441954135894775, | |
| "eval_runtime": 3.0786, | |
| "eval_samples_per_second": 32.482, | |
| "eval_steps_per_second": 4.223, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42456939816474915, | |
| "eval_runtime": 3.0814, | |
| "eval_samples_per_second": 32.453, | |
| "eval_steps_per_second": 4.219, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4075394570827484, | |
| "eval_runtime": 3.0794, | |
| "eval_samples_per_second": 32.474, | |
| "eval_steps_per_second": 4.222, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4202047288417816, | |
| "eval_runtime": 3.0797, | |
| "eval_samples_per_second": 32.471, | |
| "eval_steps_per_second": 4.221, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4381832480430603, | |
| "eval_runtime": 3.0788, | |
| "eval_samples_per_second": 32.481, | |
| "eval_steps_per_second": 4.222, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4282267391681671, | |
| "eval_runtime": 3.078, | |
| "eval_samples_per_second": 32.489, | |
| "eval_steps_per_second": 4.224, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.41453811526298523, | |
| "eval_runtime": 3.0841, | |
| "eval_samples_per_second": 32.424, | |
| "eval_steps_per_second": 4.215, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42016276717185974, | |
| "eval_runtime": 3.0831, | |
| "eval_samples_per_second": 32.435, | |
| "eval_steps_per_second": 4.217, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.4334, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.42332953214645386, | |
| "eval_runtime": 3.0858, | |
| "eval_samples_per_second": 32.406, | |
| "eval_steps_per_second": 4.213, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42850321531295776, | |
| "eval_runtime": 3.0831, | |
| "eval_samples_per_second": 32.435, | |
| "eval_steps_per_second": 4.217, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42719802260398865, | |
| "eval_runtime": 3.0793, | |
| "eval_samples_per_second": 32.475, | |
| "eval_steps_per_second": 4.222, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4232694208621979, | |
| "eval_runtime": 3.0794, | |
| "eval_samples_per_second": 32.474, | |
| "eval_steps_per_second": 4.222, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4338792860507965, | |
| "eval_runtime": 3.0814, | |
| "eval_samples_per_second": 32.453, | |
| "eval_steps_per_second": 4.219, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.41708603501319885, | |
| "eval_runtime": 3.0817, | |
| "eval_samples_per_second": 32.449, | |
| "eval_steps_per_second": 4.218, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.40945830941200256, | |
| "eval_runtime": 3.0809, | |
| "eval_samples_per_second": 32.458, | |
| "eval_steps_per_second": 4.219, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4198063910007477, | |
| "eval_runtime": 3.0829, | |
| "eval_samples_per_second": 32.436, | |
| "eval_steps_per_second": 4.217, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.41699719429016113, | |
| "eval_runtime": 3.0779, | |
| "eval_samples_per_second": 32.49, | |
| "eval_steps_per_second": 4.224, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42641326785087585, | |
| "eval_runtime": 3.086, | |
| "eval_samples_per_second": 32.405, | |
| "eval_steps_per_second": 4.213, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.43629851937294006, | |
| "eval_runtime": 3.0831, | |
| "eval_samples_per_second": 32.435, | |
| "eval_steps_per_second": 4.217, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42060795426368713, | |
| "eval_runtime": 3.0701, | |
| "eval_samples_per_second": 32.572, | |
| "eval_steps_per_second": 4.234, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4196779727935791, | |
| "eval_runtime": 3.0659, | |
| "eval_samples_per_second": 32.617, | |
| "eval_steps_per_second": 4.24, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.43019339442253113, | |
| "eval_runtime": 3.067, | |
| "eval_samples_per_second": 32.605, | |
| "eval_steps_per_second": 4.239, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4256959855556488, | |
| "eval_runtime": 3.0648, | |
| "eval_samples_per_second": 32.629, | |
| "eval_steps_per_second": 4.242, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.41873273253440857, | |
| "eval_runtime": 3.0621, | |
| "eval_samples_per_second": 32.657, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.42521288990974426, | |
| "eval_runtime": 3.0634, | |
| "eval_samples_per_second": 32.644, | |
| "eval_steps_per_second": 4.244, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4271906316280365, | |
| "eval_runtime": 3.0628, | |
| "eval_samples_per_second": 32.65, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4203490912914276, | |
| "eval_runtime": 3.0621, | |
| "eval_samples_per_second": 32.658, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.41599762439727783, | |
| "eval_runtime": 3.0609, | |
| "eval_samples_per_second": 32.67, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4063, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4164546728134155, | |
| "eval_runtime": 3.0632, | |
| "eval_samples_per_second": 32.646, | |
| "eval_steps_per_second": 4.244, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 39, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.5013418884277344, | |
| "train_runtime": 1645.3455, | |
| "train_samples_per_second": 19.449, | |
| "train_steps_per_second": 1.216 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |