| { | |
| "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": 0.65, | |
| "eval_loss": 0.5447632074356079, | |
| "eval_runtime": 2.9004, | |
| "eval_samples_per_second": 34.478, | |
| "eval_steps_per_second": 4.482, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5351557731628418, | |
| "eval_runtime": 2.9353, | |
| "eval_samples_per_second": 34.068, | |
| "eval_steps_per_second": 4.429, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5466512441635132, | |
| "eval_runtime": 2.9987, | |
| "eval_samples_per_second": 33.348, | |
| "eval_steps_per_second": 4.335, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5432061553001404, | |
| "eval_runtime": 3.0354, | |
| "eval_samples_per_second": 32.945, | |
| "eval_steps_per_second": 4.283, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5446352958679199, | |
| "eval_runtime": 3.0683, | |
| "eval_samples_per_second": 32.591, | |
| "eval_steps_per_second": 4.237, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5419210195541382, | |
| "eval_runtime": 3.0777, | |
| "eval_samples_per_second": 32.492, | |
| "eval_steps_per_second": 4.224, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5364170074462891, | |
| "eval_runtime": 3.075, | |
| "eval_samples_per_second": 32.52, | |
| "eval_steps_per_second": 4.228, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5399757027626038, | |
| "eval_runtime": 3.0939, | |
| "eval_samples_per_second": 32.322, | |
| "eval_steps_per_second": 4.202, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5460018515586853, | |
| "eval_runtime": 3.0861, | |
| "eval_samples_per_second": 32.404, | |
| "eval_steps_per_second": 4.212, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.547890305519104, | |
| "eval_runtime": 3.0888, | |
| "eval_samples_per_second": 32.375, | |
| "eval_steps_per_second": 4.209, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.542870819568634, | |
| "eval_runtime": 3.0902, | |
| "eval_samples_per_second": 32.361, | |
| "eval_steps_per_second": 4.207, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5363436341285706, | |
| "eval_runtime": 3.0944, | |
| "eval_samples_per_second": 32.316, | |
| "eval_steps_per_second": 4.201, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5431585907936096, | |
| "eval_runtime": 3.0952, | |
| "eval_samples_per_second": 32.308, | |
| "eval_steps_per_second": 4.2, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5446100831031799, | |
| "eval_runtime": 3.0959, | |
| "eval_samples_per_second": 32.301, | |
| "eval_steps_per_second": 4.199, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5619304180145264, | |
| "eval_runtime": 3.0987, | |
| "eval_samples_per_second": 32.272, | |
| "eval_steps_per_second": 4.195, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5399982929229736, | |
| "eval_runtime": 3.0957, | |
| "eval_samples_per_second": 32.303, | |
| "eval_steps_per_second": 4.199, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5394828915596008, | |
| "eval_runtime": 3.0984, | |
| "eval_samples_per_second": 32.274, | |
| "eval_steps_per_second": 4.196, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5439099073410034, | |
| "eval_runtime": 3.0994, | |
| "eval_samples_per_second": 32.264, | |
| "eval_steps_per_second": 4.194, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5420183539390564, | |
| "eval_runtime": 3.0999, | |
| "eval_samples_per_second": 32.259, | |
| "eval_steps_per_second": 4.194, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00075, | |
| "loss": 0.6126, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5401513576507568, | |
| "eval_runtime": 3.0996, | |
| "eval_samples_per_second": 32.262, | |
| "eval_steps_per_second": 4.194, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5430670976638794, | |
| "eval_runtime": 3.1006, | |
| "eval_samples_per_second": 32.252, | |
| "eval_steps_per_second": 4.193, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5421171188354492, | |
| "eval_runtime": 3.1013, | |
| "eval_samples_per_second": 32.245, | |
| "eval_steps_per_second": 4.192, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5431703329086304, | |
| "eval_runtime": 3.1002, | |
| "eval_samples_per_second": 32.256, | |
| "eval_steps_per_second": 4.193, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5438470244407654, | |
| "eval_runtime": 3.0981, | |
| "eval_samples_per_second": 32.277, | |
| "eval_steps_per_second": 4.196, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5364149212837219, | |
| "eval_runtime": 3.1003, | |
| "eval_samples_per_second": 32.255, | |
| "eval_steps_per_second": 4.193, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.541420578956604, | |
| "eval_runtime": 3.0998, | |
| "eval_samples_per_second": 32.26, | |
| "eval_steps_per_second": 4.194, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5394607782363892, | |
| "eval_runtime": 3.0995, | |
| "eval_samples_per_second": 32.263, | |
| "eval_steps_per_second": 4.194, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5440071225166321, | |
| "eval_runtime": 3.1009, | |
| "eval_samples_per_second": 32.249, | |
| "eval_steps_per_second": 4.192, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5445890426635742, | |
| "eval_runtime": 3.1002, | |
| "eval_samples_per_second": 32.256, | |
| "eval_steps_per_second": 4.193, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.5472090840339661, | |
| "eval_runtime": 3.1018, | |
| "eval_samples_per_second": 32.239, | |
| "eval_steps_per_second": 4.191, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5418501496315002, | |
| "eval_runtime": 3.0989, | |
| "eval_samples_per_second": 32.27, | |
| "eval_steps_per_second": 4.195, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5413060188293457, | |
| "eval_runtime": 3.1001, | |
| "eval_samples_per_second": 32.257, | |
| "eval_steps_per_second": 4.193, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5530202388763428, | |
| "eval_runtime": 3.1009, | |
| "eval_samples_per_second": 32.249, | |
| "eval_steps_per_second": 4.192, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5460859537124634, | |
| "eval_runtime": 3.1, | |
| "eval_samples_per_second": 32.258, | |
| "eval_steps_per_second": 4.194, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5440330505371094, | |
| "eval_runtime": 3.1019, | |
| "eval_samples_per_second": 32.238, | |
| "eval_steps_per_second": 4.191, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5436714291572571, | |
| "eval_runtime": 3.1024, | |
| "eval_samples_per_second": 32.234, | |
| "eval_steps_per_second": 4.19, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.543533444404602, | |
| "eval_runtime": 3.1035, | |
| "eval_samples_per_second": 32.221, | |
| "eval_steps_per_second": 4.189, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5481903553009033, | |
| "eval_runtime": 3.1013, | |
| "eval_samples_per_second": 32.244, | |
| "eval_steps_per_second": 4.192, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5449146628379822, | |
| "eval_runtime": 3.1036, | |
| "eval_samples_per_second": 32.221, | |
| "eval_steps_per_second": 4.189, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0005, | |
| "loss": 0.6037, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5441970229148865, | |
| "eval_runtime": 3.1022, | |
| "eval_samples_per_second": 32.235, | |
| "eval_steps_per_second": 4.191, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5376811027526855, | |
| "eval_runtime": 3.1035, | |
| "eval_samples_per_second": 32.222, | |
| "eval_steps_per_second": 4.189, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5410917401313782, | |
| "eval_runtime": 3.1029, | |
| "eval_samples_per_second": 32.228, | |
| "eval_steps_per_second": 4.19, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.5482169985771179, | |
| "eval_runtime": 3.1058, | |
| "eval_samples_per_second": 32.198, | |
| "eval_steps_per_second": 4.186, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5494033694267273, | |
| "eval_runtime": 3.1008, | |
| "eval_samples_per_second": 32.25, | |
| "eval_steps_per_second": 4.193, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.5510344505310059, | |
| "eval_runtime": 3.1036, | |
| "eval_samples_per_second": 32.221, | |
| "eval_steps_per_second": 4.189, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5471860766410828, | |
| "eval_runtime": 3.1029, | |
| "eval_samples_per_second": 32.228, | |
| "eval_steps_per_second": 4.19, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5415754318237305, | |
| "eval_runtime": 3.1057, | |
| "eval_samples_per_second": 32.199, | |
| "eval_steps_per_second": 4.186, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5396535396575928, | |
| "eval_runtime": 3.1031, | |
| "eval_samples_per_second": 32.226, | |
| "eval_steps_per_second": 4.189, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5417464375495911, | |
| "eval_runtime": 3.1027, | |
| "eval_samples_per_second": 32.23, | |
| "eval_steps_per_second": 4.19, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5390459895133972, | |
| "eval_runtime": 3.1051, | |
| "eval_samples_per_second": 32.205, | |
| "eval_steps_per_second": 4.187, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5389121174812317, | |
| "eval_runtime": 3.1044, | |
| "eval_samples_per_second": 32.213, | |
| "eval_steps_per_second": 4.188, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5365983843803406, | |
| "eval_runtime": 3.1052, | |
| "eval_samples_per_second": 32.204, | |
| "eval_steps_per_second": 4.187, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5367568731307983, | |
| "eval_runtime": 3.1042, | |
| "eval_samples_per_second": 32.215, | |
| "eval_steps_per_second": 4.188, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5392716526985168, | |
| "eval_runtime": 3.1024, | |
| "eval_samples_per_second": 32.233, | |
| "eval_steps_per_second": 4.19, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5378238558769226, | |
| "eval_runtime": 3.1049, | |
| "eval_samples_per_second": 32.207, | |
| "eval_steps_per_second": 4.187, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5391019582748413, | |
| "eval_runtime": 3.106, | |
| "eval_samples_per_second": 32.196, | |
| "eval_steps_per_second": 4.185, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5383239388465881, | |
| "eval_runtime": 3.1054, | |
| "eval_samples_per_second": 32.202, | |
| "eval_steps_per_second": 4.186, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5379219055175781, | |
| "eval_runtime": 3.1022, | |
| "eval_samples_per_second": 32.235, | |
| "eval_steps_per_second": 4.191, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5381463766098022, | |
| "eval_runtime": 3.1059, | |
| "eval_samples_per_second": 32.197, | |
| "eval_steps_per_second": 4.186, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.00025, | |
| "loss": 0.6021, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5409640669822693, | |
| "eval_runtime": 3.1098, | |
| "eval_samples_per_second": 32.156, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5401256084442139, | |
| "eval_runtime": 3.1056, | |
| "eval_samples_per_second": 32.2, | |
| "eval_steps_per_second": 4.186, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5403258800506592, | |
| "eval_runtime": 3.1087, | |
| "eval_samples_per_second": 32.167, | |
| "eval_steps_per_second": 4.182, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5411276817321777, | |
| "eval_runtime": 3.103, | |
| "eval_samples_per_second": 32.226, | |
| "eval_steps_per_second": 4.189, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5414915680885315, | |
| "eval_runtime": 3.1078, | |
| "eval_samples_per_second": 32.177, | |
| "eval_steps_per_second": 4.183, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5414855480194092, | |
| "eval_runtime": 3.1111, | |
| "eval_samples_per_second": 32.143, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5408964157104492, | |
| "eval_runtime": 3.1087, | |
| "eval_samples_per_second": 32.168, | |
| "eval_steps_per_second": 4.182, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5418642163276672, | |
| "eval_runtime": 3.1104, | |
| "eval_samples_per_second": 32.15, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5401394367218018, | |
| "eval_runtime": 3.1106, | |
| "eval_samples_per_second": 32.148, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5424190163612366, | |
| "eval_runtime": 3.1114, | |
| "eval_samples_per_second": 32.139, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5420389175415039, | |
| "eval_runtime": 3.1113, | |
| "eval_samples_per_second": 32.141, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.541526734828949, | |
| "eval_runtime": 3.1099, | |
| "eval_samples_per_second": 32.155, | |
| "eval_steps_per_second": 4.18, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5391311049461365, | |
| "eval_runtime": 3.1129, | |
| "eval_samples_per_second": 32.124, | |
| "eval_steps_per_second": 4.176, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5395915508270264, | |
| "eval_runtime": 3.1177, | |
| "eval_samples_per_second": 32.075, | |
| "eval_steps_per_second": 4.17, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5395561456680298, | |
| "eval_runtime": 3.1126, | |
| "eval_samples_per_second": 32.128, | |
| "eval_steps_per_second": 4.177, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.54047030210495, | |
| "eval_runtime": 3.1118, | |
| "eval_samples_per_second": 32.135, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5403570532798767, | |
| "eval_runtime": 3.1117, | |
| "eval_samples_per_second": 32.137, | |
| "eval_steps_per_second": 4.178, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5400457978248596, | |
| "eval_runtime": 3.122, | |
| "eval_samples_per_second": 32.031, | |
| "eval_steps_per_second": 4.164, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5401412844657898, | |
| "eval_runtime": 3.1203, | |
| "eval_samples_per_second": 32.048, | |
| "eval_steps_per_second": 4.166, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5402973890304565, | |
| "eval_runtime": 3.1111, | |
| "eval_samples_per_second": 32.143, | |
| "eval_steps_per_second": 4.179, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.5946, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5402593612670898, | |
| "eval_runtime": 3.1254, | |
| "eval_samples_per_second": 31.995, | |
| "eval_steps_per_second": 4.159, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.6032544250488281, | |
| "train_runtime": 1653.7835, | |
| "train_samples_per_second": 19.35, | |
| "train_steps_per_second": 1.209 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |