| { | |
| "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": 0.7262819409370422, | |
| "eval_runtime": 2.9351, | |
| "eval_samples_per_second": 34.071, | |
| "eval_steps_per_second": 4.429, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.4, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6114884614944458, | |
| "eval_runtime": 2.9878, | |
| "eval_samples_per_second": 33.47, | |
| "eval_steps_per_second": 4.351, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5427446365356445, | |
| "eval_runtime": 3.0364, | |
| "eval_samples_per_second": 32.934, | |
| "eval_steps_per_second": 4.281, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5318807363510132, | |
| "eval_runtime": 3.0643, | |
| "eval_samples_per_second": 32.634, | |
| "eval_steps_per_second": 4.242, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.5817937254905701, | |
| "eval_runtime": 3.0801, | |
| "eval_samples_per_second": 32.467, | |
| "eval_steps_per_second": 4.221, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5093362331390381, | |
| "eval_runtime": 3.0872, | |
| "eval_samples_per_second": 32.391, | |
| "eval_steps_per_second": 4.211, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 5, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.7841365337371826, | |
| "eval_runtime": 3.0946, | |
| "eval_samples_per_second": 32.315, | |
| "eval_steps_per_second": 4.201, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 5, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7628840804100037, | |
| "eval_runtime": 3.096, | |
| "eval_samples_per_second": 32.3, | |
| "eval_steps_per_second": 4.199, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 5, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5873813629150391, | |
| "eval_runtime": 3.1038, | |
| "eval_samples_per_second": 32.218, | |
| "eval_steps_per_second": 4.188, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5227637887001038, | |
| "eval_runtime": 3.1015, | |
| "eval_samples_per_second": 32.242, | |
| "eval_steps_per_second": 4.192, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.843931257724762, | |
| "eval_runtime": 3.1122, | |
| "eval_samples_per_second": 32.132, | |
| "eval_steps_per_second": 4.177, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.8242893218994141, | |
| "eval_runtime": 3.1122, | |
| "eval_samples_per_second": 32.131, | |
| "eval_steps_per_second": 4.177, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5669903755187988, | |
| "eval_runtime": 3.112, | |
| "eval_samples_per_second": 32.134, | |
| "eval_steps_per_second": 4.177, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5600945353507996, | |
| "eval_runtime": 3.1143, | |
| "eval_samples_per_second": 32.11, | |
| "eval_steps_per_second": 4.174, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6452310085296631, | |
| "eval_runtime": 3.1101, | |
| "eval_samples_per_second": 32.153, | |
| "eval_steps_per_second": 4.18, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.523855984210968, | |
| "eval_runtime": 3.1099, | |
| "eval_samples_per_second": 32.155, | |
| "eval_steps_per_second": 4.18, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.7315166592597961, | |
| "eval_runtime": 3.1123, | |
| "eval_samples_per_second": 32.13, | |
| "eval_steps_per_second": 4.177, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6651241779327393, | |
| "eval_runtime": 3.1136, | |
| "eval_samples_per_second": 32.117, | |
| "eval_steps_per_second": 4.175, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.9039706587791443, | |
| "eval_runtime": 3.1111, | |
| "eval_samples_per_second": 32.143, | |
| "eval_steps_per_second": 4.179, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.037500000000000006, | |
| "loss": 1.3727, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5786058306694031, | |
| "eval_runtime": 3.1123, | |
| "eval_samples_per_second": 32.131, | |
| "eval_steps_per_second": 4.177, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7332864999771118, | |
| "eval_runtime": 3.1117, | |
| "eval_samples_per_second": 32.137, | |
| "eval_steps_per_second": 4.178, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7584354877471924, | |
| "eval_runtime": 3.1101, | |
| "eval_samples_per_second": 32.154, | |
| "eval_steps_per_second": 4.18, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.9900915026664734, | |
| "eval_runtime": 3.1096, | |
| "eval_samples_per_second": 32.158, | |
| "eval_steps_per_second": 4.181, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5710880160331726, | |
| "eval_runtime": 3.1075, | |
| "eval_samples_per_second": 32.18, | |
| "eval_steps_per_second": 4.183, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5870327353477478, | |
| "eval_runtime": 3.112, | |
| "eval_samples_per_second": 32.134, | |
| "eval_steps_per_second": 4.177, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5832374691963196, | |
| "eval_runtime": 3.1121, | |
| "eval_samples_per_second": 32.132, | |
| "eval_steps_per_second": 4.177, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.9776791930198669, | |
| "eval_runtime": 3.1127, | |
| "eval_samples_per_second": 32.126, | |
| "eval_steps_per_second": 4.176, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6447842121124268, | |
| "eval_runtime": 3.1119, | |
| "eval_samples_per_second": 32.135, | |
| "eval_steps_per_second": 4.178, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.8738675117492676, | |
| "eval_runtime": 3.1117, | |
| "eval_samples_per_second": 32.136, | |
| "eval_steps_per_second": 4.178, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6709777116775513, | |
| "eval_runtime": 3.112, | |
| "eval_samples_per_second": 32.134, | |
| "eval_steps_per_second": 4.177, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5918809771537781, | |
| "eval_runtime": 3.11, | |
| "eval_samples_per_second": 32.154, | |
| "eval_steps_per_second": 4.18, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7616005539894104, | |
| "eval_runtime": 3.1114, | |
| "eval_samples_per_second": 32.14, | |
| "eval_steps_per_second": 4.178, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5836986303329468, | |
| "eval_runtime": 3.1103, | |
| "eval_samples_per_second": 32.151, | |
| "eval_steps_per_second": 4.18, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 1.0103265047073364, | |
| "eval_runtime": 3.1122, | |
| "eval_samples_per_second": 32.131, | |
| "eval_steps_per_second": 4.177, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.7007672786712646, | |
| "eval_runtime": 3.1105, | |
| "eval_samples_per_second": 32.149, | |
| "eval_steps_per_second": 4.179, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 1.016095519065857, | |
| "eval_runtime": 3.1114, | |
| "eval_samples_per_second": 32.14, | |
| "eval_steps_per_second": 4.178, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6910789012908936, | |
| "eval_runtime": 3.1125, | |
| "eval_samples_per_second": 32.128, | |
| "eval_steps_per_second": 4.177, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6450607776641846, | |
| "eval_runtime": 3.1144, | |
| "eval_samples_per_second": 32.109, | |
| "eval_steps_per_second": 4.174, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.7189579010009766, | |
| "eval_runtime": 3.1147, | |
| "eval_samples_per_second": 32.106, | |
| "eval_steps_per_second": 4.174, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.025, | |
| "loss": 0.7534, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.516425609588623, | |
| "eval_runtime": 3.1107, | |
| "eval_samples_per_second": 32.147, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4994637668132782, | |
| "eval_runtime": 3.1112, | |
| "eval_samples_per_second": 32.142, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.584023654460907, | |
| "eval_runtime": 3.1124, | |
| "eval_samples_per_second": 32.129, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.7394511699676514, | |
| "eval_runtime": 3.1125, | |
| "eval_samples_per_second": 32.129, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6374080181121826, | |
| "eval_runtime": 3.1131, | |
| "eval_samples_per_second": 32.122, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.7466642260551453, | |
| "eval_runtime": 3.1096, | |
| "eval_samples_per_second": 32.158, | |
| "eval_steps_per_second": 4.181, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.6875578165054321, | |
| "eval_runtime": 3.1103, | |
| "eval_samples_per_second": 32.151, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5958933234214783, | |
| "eval_runtime": 3.11, | |
| "eval_samples_per_second": 32.155, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5625195503234863, | |
| "eval_runtime": 3.1122, | |
| "eval_samples_per_second": 32.131, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6837201714515686, | |
| "eval_runtime": 3.1116, | |
| "eval_samples_per_second": 32.137, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 36, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6766077280044556, | |
| "eval_runtime": 3.1132, | |
| "eval_samples_per_second": 32.122, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6265960335731506, | |
| "eval_runtime": 3.1128, | |
| "eval_samples_per_second": 32.125, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.6641672253608704, | |
| "eval_runtime": 3.1146, | |
| "eval_samples_per_second": 32.106, | |
| "eval_steps_per_second": 4.174, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.6202273368835449, | |
| "eval_runtime": 3.1106, | |
| "eval_samples_per_second": 32.148, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6397892236709595, | |
| "eval_runtime": 3.1097, | |
| "eval_samples_per_second": 32.157, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6689194440841675, | |
| "eval_runtime": 3.1153, | |
| "eval_samples_per_second": 32.1, | |
| "eval_steps_per_second": 4.173, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6628773212432861, | |
| "eval_runtime": 3.1124, | |
| "eval_samples_per_second": 32.13, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5903107523918152, | |
| "eval_runtime": 3.1113, | |
| "eval_samples_per_second": 32.141, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 49, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.6132528185844421, | |
| "eval_runtime": 3.1127, | |
| "eval_samples_per_second": 32.127, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6884872317314148, | |
| "eval_runtime": 3.1145, | |
| "eval_samples_per_second": 32.108, | |
| "eval_steps_per_second": 4.174, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0125, | |
| "loss": 0.4477, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5950008630752563, | |
| "eval_runtime": 3.1114, | |
| "eval_samples_per_second": 32.14, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5714966058731079, | |
| "eval_runtime": 3.1122, | |
| "eval_samples_per_second": 32.131, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6111324429512024, | |
| "eval_runtime": 3.1175, | |
| "eval_samples_per_second": 32.077, | |
| "eval_steps_per_second": 4.17, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6023278832435608, | |
| "eval_runtime": 3.113, | |
| "eval_samples_per_second": 32.124, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5792566537857056, | |
| "eval_runtime": 3.1121, | |
| "eval_samples_per_second": 32.133, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5727165341377258, | |
| "eval_runtime": 3.1107, | |
| "eval_samples_per_second": 32.147, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5606411099433899, | |
| "eval_runtime": 3.1095, | |
| "eval_samples_per_second": 32.159, | |
| "eval_steps_per_second": 4.181, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5970374345779419, | |
| "eval_runtime": 3.1149, | |
| "eval_samples_per_second": 32.103, | |
| "eval_steps_per_second": 4.173, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5601862072944641, | |
| "eval_runtime": 3.1113, | |
| "eval_samples_per_second": 32.14, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.578113317489624, | |
| "eval_runtime": 3.1132, | |
| "eval_samples_per_second": 32.121, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.6141662001609802, | |
| "eval_runtime": 3.1126, | |
| "eval_samples_per_second": 32.127, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5757777690887451, | |
| "eval_runtime": 3.1115, | |
| "eval_samples_per_second": 32.139, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5649742484092712, | |
| "eval_runtime": 3.1127, | |
| "eval_samples_per_second": 32.127, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5823287963867188, | |
| "eval_runtime": 3.112, | |
| "eval_samples_per_second": 32.134, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5546900033950806, | |
| "eval_runtime": 3.111, | |
| "eval_samples_per_second": 32.144, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5636573433876038, | |
| "eval_runtime": 3.1109, | |
| "eval_samples_per_second": 32.145, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5805944204330444, | |
| "eval_runtime": 3.1125, | |
| "eval_samples_per_second": 32.129, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5601564645767212, | |
| "eval_runtime": 3.1215, | |
| "eval_samples_per_second": 32.036, | |
| "eval_steps_per_second": 4.165, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.570804238319397, | |
| "eval_runtime": 3.1168, | |
| "eval_samples_per_second": 32.084, | |
| "eval_steps_per_second": 4.171, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5623500347137451, | |
| "eval_runtime": 3.11, | |
| "eval_samples_per_second": 32.155, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.3287, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5641293525695801, | |
| "eval_runtime": 3.1089, | |
| "eval_samples_per_second": 32.166, | |
| "eval_steps_per_second": 4.182, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 57, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.7256226806640625, | |
| "train_runtime": 1655.9889, | |
| "train_samples_per_second": 19.324, | |
| "train_steps_per_second": 1.208 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |