| { | |
| "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": 0.53, | |
| "eval_loss": 0.6182658672332764, | |
| "eval_runtime": 2.964, | |
| "eval_samples_per_second": 33.738, | |
| "eval_steps_per_second": 4.386, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.53, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.41887640953063965, | |
| "eval_runtime": 3.0102, | |
| "eval_samples_per_second": 33.22, | |
| "eval_steps_per_second": 4.319, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.43512362241744995, | |
| "eval_runtime": 3.0581, | |
| "eval_samples_per_second": 32.7, | |
| "eval_steps_per_second": 4.251, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.41805171966552734, | |
| "eval_runtime": 3.0736, | |
| "eval_samples_per_second": 32.536, | |
| "eval_steps_per_second": 4.23, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.41053706407546997, | |
| "eval_runtime": 3.0847, | |
| "eval_samples_per_second": 32.418, | |
| "eval_steps_per_second": 4.214, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.41397789120674133, | |
| "eval_runtime": 3.0887, | |
| "eval_samples_per_second": 32.376, | |
| "eval_steps_per_second": 4.209, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 5, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.40520820021629333, | |
| "eval_runtime": 3.0892, | |
| "eval_samples_per_second": 32.371, | |
| "eval_steps_per_second": 4.208, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.43216773867607117, | |
| "eval_runtime": 3.0856, | |
| "eval_samples_per_second": 32.408, | |
| "eval_steps_per_second": 4.213, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4364350438117981, | |
| "eval_runtime": 3.0898, | |
| "eval_samples_per_second": 32.365, | |
| "eval_steps_per_second": 4.207, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.42465832829475403, | |
| "eval_runtime": 3.0783, | |
| "eval_samples_per_second": 32.485, | |
| "eval_steps_per_second": 4.223, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.42610007524490356, | |
| "eval_runtime": 3.0769, | |
| "eval_samples_per_second": 32.5, | |
| "eval_steps_per_second": 4.225, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.41756904125213623, | |
| "eval_runtime": 3.0756, | |
| "eval_samples_per_second": 32.514, | |
| "eval_steps_per_second": 4.227, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.4107968211174011, | |
| "eval_runtime": 3.0727, | |
| "eval_samples_per_second": 32.544, | |
| "eval_steps_per_second": 4.231, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.43049681186676025, | |
| "eval_runtime": 3.0737, | |
| "eval_samples_per_second": 32.534, | |
| "eval_steps_per_second": 4.229, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.40635767579078674, | |
| "eval_runtime": 3.0804, | |
| "eval_samples_per_second": 32.463, | |
| "eval_steps_per_second": 4.22, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.40317821502685547, | |
| "eval_runtime": 3.0757, | |
| "eval_samples_per_second": 32.513, | |
| "eval_steps_per_second": 4.227, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.40984559059143066, | |
| "eval_runtime": 3.0757, | |
| "eval_samples_per_second": 32.513, | |
| "eval_steps_per_second": 4.227, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.41323602199554443, | |
| "eval_runtime": 3.0748, | |
| "eval_samples_per_second": 32.522, | |
| "eval_steps_per_second": 4.228, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.3924804627895355, | |
| "eval_runtime": 3.0886, | |
| "eval_samples_per_second": 32.377, | |
| "eval_steps_per_second": 4.209, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.7171, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.3957255482673645, | |
| "eval_runtime": 3.0914, | |
| "eval_samples_per_second": 32.348, | |
| "eval_steps_per_second": 4.205, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.42917293310165405, | |
| "eval_runtime": 3.0689, | |
| "eval_samples_per_second": 32.585, | |
| "eval_steps_per_second": 4.236, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.40250298380851746, | |
| "eval_runtime": 3.0667, | |
| "eval_samples_per_second": 32.608, | |
| "eval_steps_per_second": 4.239, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.3997068703174591, | |
| "eval_runtime": 3.0617, | |
| "eval_samples_per_second": 32.662, | |
| "eval_steps_per_second": 4.246, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.4115046560764313, | |
| "eval_runtime": 3.0625, | |
| "eval_samples_per_second": 32.653, | |
| "eval_steps_per_second": 4.245, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.40437012910842896, | |
| "eval_runtime": 3.062, | |
| "eval_samples_per_second": 32.659, | |
| "eval_steps_per_second": 4.246, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.409763902425766, | |
| "eval_runtime": 3.0623, | |
| "eval_samples_per_second": 32.655, | |
| "eval_steps_per_second": 4.245, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.40510663390159607, | |
| "eval_runtime": 3.0617, | |
| "eval_samples_per_second": 32.661, | |
| "eval_steps_per_second": 4.246, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.42438602447509766, | |
| "eval_runtime": 3.0608, | |
| "eval_samples_per_second": 32.671, | |
| "eval_steps_per_second": 4.247, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.4031755328178406, | |
| "eval_runtime": 3.0609, | |
| "eval_samples_per_second": 32.67, | |
| "eval_steps_per_second": 4.247, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.413577675819397, | |
| "eval_runtime": 3.0625, | |
| "eval_samples_per_second": 32.653, | |
| "eval_steps_per_second": 4.245, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.3992563486099243, | |
| "eval_runtime": 3.064, | |
| "eval_samples_per_second": 32.637, | |
| "eval_steps_per_second": 4.243, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.417043000459671, | |
| "eval_runtime": 3.0628, | |
| "eval_samples_per_second": 32.65, | |
| "eval_steps_per_second": 4.245, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.40380868315696716, | |
| "eval_runtime": 3.0665, | |
| "eval_samples_per_second": 32.611, | |
| "eval_steps_per_second": 4.239, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.42510226368904114, | |
| "eval_runtime": 3.0667, | |
| "eval_samples_per_second": 32.608, | |
| "eval_steps_per_second": 4.239, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4078834056854248, | |
| "eval_runtime": 3.0638, | |
| "eval_samples_per_second": 32.639, | |
| "eval_steps_per_second": 4.243, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4119352102279663, | |
| "eval_runtime": 3.0725, | |
| "eval_samples_per_second": 32.547, | |
| "eval_steps_per_second": 4.231, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4074689745903015, | |
| "eval_runtime": 3.0606, | |
| "eval_samples_per_second": 32.674, | |
| "eval_steps_per_second": 4.248, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 27, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.44055667519569397, | |
| "eval_runtime": 3.0609, | |
| "eval_samples_per_second": 32.67, | |
| "eval_steps_per_second": 4.247, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4081181287765503, | |
| "eval_runtime": 3.061, | |
| "eval_samples_per_second": 32.669, | |
| "eval_steps_per_second": 4.247, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.4731, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4190601110458374, | |
| "eval_runtime": 3.0627, | |
| "eval_samples_per_second": 32.651, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4217339754104614, | |
| "eval_runtime": 3.0615, | |
| "eval_samples_per_second": 32.663, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.39827075600624084, | |
| "eval_runtime": 3.0673, | |
| "eval_samples_per_second": 32.602, | |
| "eval_steps_per_second": 4.238, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.40923720598220825, | |
| "eval_runtime": 3.0598, | |
| "eval_samples_per_second": 32.682, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.42479878664016724, | |
| "eval_runtime": 3.0596, | |
| "eval_samples_per_second": 32.684, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4218236804008484, | |
| "eval_runtime": 3.059, | |
| "eval_samples_per_second": 32.691, | |
| "eval_steps_per_second": 4.25, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4371417164802551, | |
| "eval_runtime": 3.0613, | |
| "eval_samples_per_second": 32.666, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4098566472530365, | |
| "eval_runtime": 3.0607, | |
| "eval_samples_per_second": 32.672, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.4299997091293335, | |
| "eval_runtime": 3.0624, | |
| "eval_samples_per_second": 32.654, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4093553125858307, | |
| "eval_runtime": 3.0616, | |
| "eval_samples_per_second": 32.662, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4205920100212097, | |
| "eval_runtime": 3.062, | |
| "eval_samples_per_second": 32.658, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4240824580192566, | |
| "eval_runtime": 3.0617, | |
| "eval_samples_per_second": 32.662, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4252602756023407, | |
| "eval_runtime": 3.0627, | |
| "eval_samples_per_second": 32.651, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.4116606116294861, | |
| "eval_runtime": 3.0605, | |
| "eval_samples_per_second": 32.674, | |
| "eval_steps_per_second": 4.248, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.417370080947876, | |
| "eval_runtime": 3.0649, | |
| "eval_samples_per_second": 32.628, | |
| "eval_steps_per_second": 4.242, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.41312336921691895, | |
| "eval_runtime": 3.0627, | |
| "eval_samples_per_second": 32.651, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.42308780550956726, | |
| "eval_runtime": 3.0629, | |
| "eval_samples_per_second": 32.648, | |
| "eval_steps_per_second": 4.244, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4059382379055023, | |
| "eval_runtime": 3.0606, | |
| "eval_samples_per_second": 32.674, | |
| "eval_steps_per_second": 4.248, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.4168393313884735, | |
| "eval_runtime": 3.0634, | |
| "eval_samples_per_second": 32.643, | |
| "eval_steps_per_second": 4.244, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.42363443970680237, | |
| "eval_runtime": 3.0615, | |
| "eval_samples_per_second": 32.664, | |
| "eval_steps_per_second": 4.246, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.4204, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.400055855512619, | |
| "eval_runtime": 3.0593, | |
| "eval_samples_per_second": 32.687, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41580745577812195, | |
| "eval_runtime": 3.0632, | |
| "eval_samples_per_second": 32.645, | |
| "eval_steps_per_second": 4.244, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.43029120564460754, | |
| "eval_runtime": 3.0605, | |
| "eval_samples_per_second": 32.675, | |
| "eval_steps_per_second": 4.248, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.41548973321914673, | |
| "eval_runtime": 3.0606, | |
| "eval_samples_per_second": 32.673, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.41950705647468567, | |
| "eval_runtime": 3.0588, | |
| "eval_samples_per_second": 32.692, | |
| "eval_steps_per_second": 4.25, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.43146055936813354, | |
| "eval_runtime": 3.0608, | |
| "eval_samples_per_second": 32.671, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4239749610424042, | |
| "eval_runtime": 3.0598, | |
| "eval_samples_per_second": 32.682, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.4191063344478607, | |
| "eval_runtime": 3.0607, | |
| "eval_samples_per_second": 32.672, | |
| "eval_steps_per_second": 4.247, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.42141807079315186, | |
| "eval_runtime": 3.0592, | |
| "eval_samples_per_second": 32.688, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41697102785110474, | |
| "eval_runtime": 3.0626, | |
| "eval_samples_per_second": 32.652, | |
| "eval_steps_per_second": 4.245, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.41583549976348877, | |
| "eval_runtime": 3.0593, | |
| "eval_samples_per_second": 32.687, | |
| "eval_steps_per_second": 4.249, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.42303329706192017, | |
| "eval_runtime": 3.0591, | |
| "eval_samples_per_second": 32.69, | |
| "eval_steps_per_second": 4.25, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.41061896085739136, | |
| "eval_runtime": 3.0576, | |
| "eval_samples_per_second": 32.706, | |
| "eval_steps_per_second": 4.252, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.42553210258483887, | |
| "eval_runtime": 3.056, | |
| "eval_samples_per_second": 32.723, | |
| "eval_steps_per_second": 4.254, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4223473072052002, | |
| "eval_runtime": 3.057, | |
| "eval_samples_per_second": 32.712, | |
| "eval_steps_per_second": 4.253, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.41237103939056396, | |
| "eval_runtime": 3.0557, | |
| "eval_samples_per_second": 32.725, | |
| "eval_steps_per_second": 4.254, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.4114343225955963, | |
| "eval_runtime": 3.0557, | |
| "eval_samples_per_second": 32.726, | |
| "eval_steps_per_second": 4.254, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4114573001861572, | |
| "eval_runtime": 3.0539, | |
| "eval_samples_per_second": 32.745, | |
| "eval_steps_per_second": 4.257, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.4135931432247162, | |
| "eval_runtime": 3.0553, | |
| "eval_samples_per_second": 32.73, | |
| "eval_steps_per_second": 4.255, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41504940390586853, | |
| "eval_runtime": 3.0541, | |
| "eval_samples_per_second": 32.743, | |
| "eval_steps_per_second": 4.257, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.3939, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.41364365816116333, | |
| "eval_runtime": 3.0573, | |
| "eval_samples_per_second": 32.709, | |
| "eval_steps_per_second": 4.252, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 37, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.5011474990844726, | |
| "train_runtime": 1641.5222, | |
| "train_samples_per_second": 19.494, | |
| "train_steps_per_second": 1.218 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |