| { | |
| "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": 0.5, | |
| "eval_loss": 0.4473404288291931, | |
| "eval_runtime": 2.9685, | |
| "eval_samples_per_second": 33.687, | |
| "eval_steps_per_second": 4.379, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.5, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.37501347064971924, | |
| "eval_runtime": 3.0237, | |
| "eval_samples_per_second": 33.072, | |
| "eval_steps_per_second": 4.299, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.3427189886569977, | |
| "eval_runtime": 3.0795, | |
| "eval_samples_per_second": 32.473, | |
| "eval_steps_per_second": 4.221, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2967395484447479, | |
| "eval_runtime": 3.1247, | |
| "eval_samples_per_second": 32.003, | |
| "eval_steps_per_second": 4.16, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.2981172800064087, | |
| "eval_runtime": 3.1551, | |
| "eval_samples_per_second": 31.695, | |
| "eval_steps_per_second": 4.12, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.32641762495040894, | |
| "eval_runtime": 3.1755, | |
| "eval_samples_per_second": 31.491, | |
| "eval_steps_per_second": 4.094, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2917565405368805, | |
| "eval_runtime": 3.1839, | |
| "eval_samples_per_second": 31.408, | |
| "eval_steps_per_second": 4.083, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.30622953176498413, | |
| "eval_runtime": 3.1953, | |
| "eval_samples_per_second": 31.296, | |
| "eval_steps_per_second": 4.068, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2885332703590393, | |
| "eval_runtime": 3.2033, | |
| "eval_samples_per_second": 31.218, | |
| "eval_steps_per_second": 4.058, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.288424015045166, | |
| "eval_runtime": 3.2092, | |
| "eval_samples_per_second": 31.16, | |
| "eval_steps_per_second": 4.051, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.2963046133518219, | |
| "eval_runtime": 3.2093, | |
| "eval_samples_per_second": 31.16, | |
| "eval_steps_per_second": 4.051, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28954315185546875, | |
| "eval_runtime": 3.2079, | |
| "eval_samples_per_second": 31.173, | |
| "eval_steps_per_second": 4.052, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28725749254226685, | |
| "eval_runtime": 3.2054, | |
| "eval_samples_per_second": 31.197, | |
| "eval_steps_per_second": 4.056, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28840315341949463, | |
| "eval_runtime": 3.2071, | |
| "eval_samples_per_second": 31.181, | |
| "eval_steps_per_second": 4.053, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2870934009552002, | |
| "eval_runtime": 3.2042, | |
| "eval_samples_per_second": 31.209, | |
| "eval_steps_per_second": 4.057, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28593483567237854, | |
| "eval_runtime": 3.1986, | |
| "eval_samples_per_second": 31.264, | |
| "eval_steps_per_second": 4.064, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.29119229316711426, | |
| "eval_runtime": 3.2008, | |
| "eval_samples_per_second": 31.242, | |
| "eval_steps_per_second": 4.061, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.2840566337108612, | |
| "eval_runtime": 3.2065, | |
| "eval_samples_per_second": 31.186, | |
| "eval_steps_per_second": 4.054, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28335732221603394, | |
| "eval_runtime": 3.2057, | |
| "eval_samples_per_second": 31.195, | |
| "eval_steps_per_second": 4.055, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.5493, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2825382947921753, | |
| "eval_runtime": 3.2031, | |
| "eval_samples_per_second": 31.22, | |
| "eval_steps_per_second": 4.059, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28468167781829834, | |
| "eval_runtime": 3.2095, | |
| "eval_samples_per_second": 31.158, | |
| "eval_steps_per_second": 4.051, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2782175540924072, | |
| "eval_runtime": 3.2053, | |
| "eval_samples_per_second": 31.198, | |
| "eval_steps_per_second": 4.056, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2759018838405609, | |
| "eval_runtime": 3.2029, | |
| "eval_samples_per_second": 31.221, | |
| "eval_steps_per_second": 4.059, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.27502644062042236, | |
| "eval_runtime": 3.2041, | |
| "eval_samples_per_second": 31.21, | |
| "eval_steps_per_second": 4.057, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 23, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2744516432285309, | |
| "eval_runtime": 3.2088, | |
| "eval_samples_per_second": 31.164, | |
| "eval_steps_per_second": 4.051, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.2720719873905182, | |
| "eval_runtime": 3.2072, | |
| "eval_samples_per_second": 31.18, | |
| "eval_steps_per_second": 4.053, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.2727718949317932, | |
| "eval_runtime": 3.2062, | |
| "eval_samples_per_second": 31.189, | |
| "eval_steps_per_second": 4.055, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2848474383354187, | |
| "eval_runtime": 3.2049, | |
| "eval_samples_per_second": 31.202, | |
| "eval_steps_per_second": 4.056, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.2726779580116272, | |
| "eval_runtime": 3.2066, | |
| "eval_samples_per_second": 31.185, | |
| "eval_steps_per_second": 4.054, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.273853600025177, | |
| "eval_runtime": 3.2057, | |
| "eval_samples_per_second": 31.194, | |
| "eval_steps_per_second": 4.055, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.271501362323761, | |
| "eval_runtime": 3.2048, | |
| "eval_samples_per_second": 31.204, | |
| "eval_steps_per_second": 4.056, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.2950344383716583, | |
| "eval_runtime": 3.2044, | |
| "eval_samples_per_second": 31.207, | |
| "eval_steps_per_second": 4.057, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.27638041973114014, | |
| "eval_runtime": 3.2097, | |
| "eval_samples_per_second": 31.156, | |
| "eval_steps_per_second": 4.05, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.26928141713142395, | |
| "eval_runtime": 3.206, | |
| "eval_samples_per_second": 31.191, | |
| "eval_steps_per_second": 4.055, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2686476409435272, | |
| "eval_runtime": 3.2046, | |
| "eval_samples_per_second": 31.205, | |
| "eval_steps_per_second": 4.057, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.2793463468551636, | |
| "eval_runtime": 3.2054, | |
| "eval_samples_per_second": 31.197, | |
| "eval_steps_per_second": 4.056, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2699900269508362, | |
| "eval_runtime": 3.2061, | |
| "eval_samples_per_second": 31.191, | |
| "eval_steps_per_second": 4.055, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.27441737055778503, | |
| "eval_runtime": 3.2047, | |
| "eval_samples_per_second": 31.205, | |
| "eval_steps_per_second": 4.057, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 24, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.27893561124801636, | |
| "eval_runtime": 3.2019, | |
| "eval_samples_per_second": 31.232, | |
| "eval_steps_per_second": 4.06, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.4987, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.2756670415401459, | |
| "eval_runtime": 3.2056, | |
| "eval_samples_per_second": 31.195, | |
| "eval_steps_per_second": 4.055, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.27051329612731934, | |
| "eval_runtime": 3.206, | |
| "eval_samples_per_second": 31.192, | |
| "eval_steps_per_second": 4.055, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.2836475074291229, | |
| "eval_runtime": 3.2118, | |
| "eval_samples_per_second": 31.135, | |
| "eval_steps_per_second": 4.048, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28077641129493713, | |
| "eval_runtime": 3.2076, | |
| "eval_samples_per_second": 31.176, | |
| "eval_steps_per_second": 4.053, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.27339887619018555, | |
| "eval_runtime": 3.2087, | |
| "eval_samples_per_second": 31.165, | |
| "eval_steps_per_second": 4.052, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2703300416469574, | |
| "eval_runtime": 3.2077, | |
| "eval_samples_per_second": 31.175, | |
| "eval_steps_per_second": 4.053, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2787483334541321, | |
| "eval_runtime": 3.2167, | |
| "eval_samples_per_second": 31.088, | |
| "eval_steps_per_second": 4.041, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2683820128440857, | |
| "eval_runtime": 3.2087, | |
| "eval_samples_per_second": 31.165, | |
| "eval_steps_per_second": 4.051, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.27374398708343506, | |
| "eval_runtime": 3.2081, | |
| "eval_samples_per_second": 31.171, | |
| "eval_steps_per_second": 4.052, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2792060375213623, | |
| "eval_runtime": 3.2041, | |
| "eval_samples_per_second": 31.21, | |
| "eval_steps_per_second": 4.057, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.273674875497818, | |
| "eval_runtime": 3.2047, | |
| "eval_samples_per_second": 31.205, | |
| "eval_steps_per_second": 4.057, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2722756266593933, | |
| "eval_runtime": 3.2013, | |
| "eval_samples_per_second": 31.238, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 45, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2725033164024353, | |
| "eval_runtime": 3.2035, | |
| "eval_samples_per_second": 31.216, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 51, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2722368836402893, | |
| "eval_runtime": 3.207, | |
| "eval_samples_per_second": 31.181, | |
| "eval_steps_per_second": 4.054, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 51, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.28004658222198486, | |
| "eval_runtime": 3.2057, | |
| "eval_samples_per_second": 31.194, | |
| "eval_steps_per_second": 4.055, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 51, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2768723666667938, | |
| "eval_runtime": 3.205, | |
| "eval_samples_per_second": 31.201, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 51, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.27720382809638977, | |
| "eval_runtime": 3.2049, | |
| "eval_samples_per_second": 31.202, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 55, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.27145129442214966, | |
| "eval_runtime": 3.2036, | |
| "eval_samples_per_second": 31.215, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.27938613295555115, | |
| "eval_runtime": 3.2025, | |
| "eval_samples_per_second": 31.226, | |
| "eval_steps_per_second": 4.059, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.277148574590683, | |
| "eval_runtime": 3.2053, | |
| "eval_samples_per_second": 31.198, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.447, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.2797820270061493, | |
| "eval_runtime": 3.2029, | |
| "eval_samples_per_second": 31.221, | |
| "eval_steps_per_second": 4.059, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.2717023193836212, | |
| "eval_runtime": 3.2014, | |
| "eval_samples_per_second": 31.236, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2990992069244385, | |
| "eval_runtime": 3.2016, | |
| "eval_samples_per_second": 31.234, | |
| "eval_steps_per_second": 4.06, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2718846797943115, | |
| "eval_runtime": 3.2014, | |
| "eval_samples_per_second": 31.236, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2761527895927429, | |
| "eval_runtime": 3.2015, | |
| "eval_samples_per_second": 31.235, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2833251953125, | |
| "eval_runtime": 3.2011, | |
| "eval_samples_per_second": 31.24, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.2772473096847534, | |
| "eval_runtime": 3.2001, | |
| "eval_samples_per_second": 31.249, | |
| "eval_steps_per_second": 4.062, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2806840240955353, | |
| "eval_runtime": 3.2008, | |
| "eval_samples_per_second": 31.242, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2740679979324341, | |
| "eval_runtime": 3.1997, | |
| "eval_samples_per_second": 31.253, | |
| "eval_steps_per_second": 4.063, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.27646538615226746, | |
| "eval_runtime": 3.2009, | |
| "eval_samples_per_second": 31.241, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.27863335609436035, | |
| "eval_runtime": 3.2011, | |
| "eval_samples_per_second": 31.239, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2795208692550659, | |
| "eval_runtime": 3.2, | |
| "eval_samples_per_second": 31.25, | |
| "eval_steps_per_second": 4.062, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.27518370747566223, | |
| "eval_runtime": 3.2014, | |
| "eval_samples_per_second": 31.236, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.283783495426178, | |
| "eval_runtime": 3.2004, | |
| "eval_samples_per_second": 31.246, | |
| "eval_steps_per_second": 4.062, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.2762831747531891, | |
| "eval_runtime": 3.2012, | |
| "eval_samples_per_second": 31.239, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.27643781900405884, | |
| "eval_runtime": 3.1996, | |
| "eval_samples_per_second": 31.254, | |
| "eval_steps_per_second": 4.063, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2755553722381592, | |
| "eval_runtime": 3.1982, | |
| "eval_samples_per_second": 31.268, | |
| "eval_steps_per_second": 4.065, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.2737685441970825, | |
| "eval_runtime": 3.2108, | |
| "eval_samples_per_second": 31.145, | |
| "eval_steps_per_second": 4.049, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.27428507804870605, | |
| "eval_runtime": 3.2072, | |
| "eval_samples_per_second": 31.18, | |
| "eval_steps_per_second": 4.053, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2778934836387634, | |
| "eval_runtime": 3.2047, | |
| "eval_samples_per_second": 31.204, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4199, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2766960859298706, | |
| "eval_runtime": 3.1925, | |
| "eval_samples_per_second": 31.324, | |
| "eval_steps_per_second": 4.072, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.4787121353149414, | |
| "train_runtime": 1702.5964, | |
| "train_samples_per_second": 18.795, | |
| "train_steps_per_second": 1.175 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |