| { | |
| "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": 0.48, | |
| "eval_loss": 0.6259328126907349, | |
| "eval_runtime": 2.8974, | |
| "eval_samples_per_second": 34.514, | |
| "eval_steps_per_second": 4.487, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.48, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.7321136593818665, | |
| "eval_runtime": 2.9205, | |
| "eval_samples_per_second": 34.241, | |
| "eval_steps_per_second": 4.451, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.7952651381492615, | |
| "eval_runtime": 2.9672, | |
| "eval_samples_per_second": 33.702, | |
| "eval_steps_per_second": 4.381, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6992718577384949, | |
| "eval_runtime": 2.9932, | |
| "eval_samples_per_second": 33.409, | |
| "eval_steps_per_second": 4.343, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5882292985916138, | |
| "eval_runtime": 3.0071, | |
| "eval_samples_per_second": 33.255, | |
| "eval_steps_per_second": 4.323, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5895842909812927, | |
| "eval_runtime": 3.0189, | |
| "eval_samples_per_second": 33.125, | |
| "eval_steps_per_second": 4.306, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6143475770950317, | |
| "eval_runtime": 3.0217, | |
| "eval_samples_per_second": 33.093, | |
| "eval_steps_per_second": 4.302, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.706989586353302, | |
| "eval_runtime": 3.0244, | |
| "eval_samples_per_second": 33.064, | |
| "eval_steps_per_second": 4.298, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6440533995628357, | |
| "eval_runtime": 3.0216, | |
| "eval_samples_per_second": 33.095, | |
| "eval_steps_per_second": 4.302, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.704773485660553, | |
| "eval_runtime": 3.0257, | |
| "eval_samples_per_second": 33.05, | |
| "eval_steps_per_second": 4.296, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5610386729240417, | |
| "eval_runtime": 3.0256, | |
| "eval_samples_per_second": 33.052, | |
| "eval_steps_per_second": 4.297, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6845036149024963, | |
| "eval_runtime": 3.0242, | |
| "eval_samples_per_second": 33.067, | |
| "eval_steps_per_second": 4.299, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7743116617202759, | |
| "eval_runtime": 3.0303, | |
| "eval_samples_per_second": 33.0, | |
| "eval_steps_per_second": 4.29, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7745197415351868, | |
| "eval_runtime": 3.0249, | |
| "eval_samples_per_second": 33.059, | |
| "eval_steps_per_second": 4.298, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 10, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7992074489593506, | |
| "eval_runtime": 3.033, | |
| "eval_samples_per_second": 32.97, | |
| "eval_steps_per_second": 4.286, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7165772914886475, | |
| "eval_runtime": 3.0399, | |
| "eval_samples_per_second": 32.896, | |
| "eval_steps_per_second": 4.276, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.7013249397277832, | |
| "eval_runtime": 3.0281, | |
| "eval_samples_per_second": 33.024, | |
| "eval_steps_per_second": 4.293, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.8815276622772217, | |
| "eval_runtime": 3.04, | |
| "eval_samples_per_second": 32.894, | |
| "eval_steps_per_second": 4.276, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7997011542320251, | |
| "eval_runtime": 3.0324, | |
| "eval_samples_per_second": 32.977, | |
| "eval_steps_per_second": 4.287, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.6923, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.741098940372467, | |
| "eval_runtime": 3.0327, | |
| "eval_samples_per_second": 32.974, | |
| "eval_steps_per_second": 4.287, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7322430610656738, | |
| "eval_runtime": 3.0313, | |
| "eval_samples_per_second": 32.989, | |
| "eval_steps_per_second": 4.289, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.8924081325531006, | |
| "eval_runtime": 3.0335, | |
| "eval_samples_per_second": 32.965, | |
| "eval_steps_per_second": 4.285, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.72379469871521, | |
| "eval_runtime": 3.0331, | |
| "eval_samples_per_second": 32.97, | |
| "eval_steps_per_second": 4.286, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7784560322761536, | |
| "eval_runtime": 3.0334, | |
| "eval_samples_per_second": 32.966, | |
| "eval_steps_per_second": 4.286, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6886314153671265, | |
| "eval_runtime": 3.034, | |
| "eval_samples_per_second": 32.96, | |
| "eval_steps_per_second": 4.285, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7781857848167419, | |
| "eval_runtime": 3.0361, | |
| "eval_samples_per_second": 32.937, | |
| "eval_steps_per_second": 4.282, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7322260141372681, | |
| "eval_runtime": 3.0417, | |
| "eval_samples_per_second": 32.877, | |
| "eval_steps_per_second": 4.274, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7590029239654541, | |
| "eval_runtime": 3.0373, | |
| "eval_samples_per_second": 32.924, | |
| "eval_steps_per_second": 4.28, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7170096635818481, | |
| "eval_runtime": 3.036, | |
| "eval_samples_per_second": 32.938, | |
| "eval_steps_per_second": 4.282, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7993258833885193, | |
| "eval_runtime": 3.0342, | |
| "eval_samples_per_second": 32.958, | |
| "eval_steps_per_second": 4.285, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.746490478515625, | |
| "eval_runtime": 3.0396, | |
| "eval_samples_per_second": 32.899, | |
| "eval_steps_per_second": 4.277, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6626662611961365, | |
| "eval_runtime": 3.0423, | |
| "eval_samples_per_second": 32.869, | |
| "eval_steps_per_second": 4.273, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.71281898021698, | |
| "eval_runtime": 3.0422, | |
| "eval_samples_per_second": 32.871, | |
| "eval_steps_per_second": 4.273, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6698910593986511, | |
| "eval_runtime": 3.0416, | |
| "eval_samples_per_second": 32.877, | |
| "eval_steps_per_second": 4.274, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6974368095397949, | |
| "eval_runtime": 3.0432, | |
| "eval_samples_per_second": 32.861, | |
| "eval_steps_per_second": 4.272, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6626088619232178, | |
| "eval_runtime": 3.0454, | |
| "eval_samples_per_second": 32.837, | |
| "eval_steps_per_second": 4.269, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6843433976173401, | |
| "eval_runtime": 3.0433, | |
| "eval_samples_per_second": 32.859, | |
| "eval_steps_per_second": 4.272, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6846016049385071, | |
| "eval_runtime": 3.0416, | |
| "eval_samples_per_second": 32.878, | |
| "eval_steps_per_second": 4.274, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7097976207733154, | |
| "eval_runtime": 3.0459, | |
| "eval_samples_per_second": 32.831, | |
| "eval_steps_per_second": 4.268, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.2907, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.684529185295105, | |
| "eval_runtime": 3.046, | |
| "eval_samples_per_second": 32.83, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6782032251358032, | |
| "eval_runtime": 3.044, | |
| "eval_samples_per_second": 32.851, | |
| "eval_steps_per_second": 4.271, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6635347008705139, | |
| "eval_runtime": 3.0457, | |
| "eval_samples_per_second": 32.833, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5903083682060242, | |
| "eval_runtime": 3.0458, | |
| "eval_samples_per_second": 32.832, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6071918606758118, | |
| "eval_runtime": 3.0441, | |
| "eval_samples_per_second": 32.851, | |
| "eval_steps_per_second": 4.271, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5961195230484009, | |
| "eval_runtime": 3.0477, | |
| "eval_samples_per_second": 32.811, | |
| "eval_steps_per_second": 4.265, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6114852428436279, | |
| "eval_runtime": 3.0485, | |
| "eval_samples_per_second": 32.803, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6239812970161438, | |
| "eval_runtime": 3.0453, | |
| "eval_samples_per_second": 32.837, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6327325701713562, | |
| "eval_runtime": 3.0465, | |
| "eval_samples_per_second": 32.824, | |
| "eval_steps_per_second": 4.267, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.693472146987915, | |
| "eval_runtime": 3.0485, | |
| "eval_samples_per_second": 32.803, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5863702893257141, | |
| "eval_runtime": 3.0495, | |
| "eval_samples_per_second": 32.793, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.577936053276062, | |
| "eval_runtime": 3.0454, | |
| "eval_samples_per_second": 32.836, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6012981534004211, | |
| "eval_runtime": 3.0456, | |
| "eval_samples_per_second": 32.834, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5664993524551392, | |
| "eval_runtime": 3.0459, | |
| "eval_samples_per_second": 32.831, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.75, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.5744893550872803, | |
| "eval_runtime": 3.0448, | |
| "eval_samples_per_second": 32.842, | |
| "eval_steps_per_second": 4.27, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 53, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.6107723712921143, | |
| "eval_runtime": 3.0477, | |
| "eval_samples_per_second": 32.812, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 53, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.584411084651947, | |
| "eval_runtime": 3.045, | |
| "eval_samples_per_second": 32.841, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 53, | |
| "best_eval_accuracy": 0.76, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.5647205114364624, | |
| "eval_runtime": 3.0572, | |
| "eval_samples_per_second": 32.71, | |
| "eval_steps_per_second": 4.252, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.76, | |
| "eval_loss": 0.584377110004425, | |
| "eval_runtime": 3.0552, | |
| "eval_samples_per_second": 32.731, | |
| "eval_steps_per_second": 4.255, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5720120668411255, | |
| "eval_runtime": 3.0488, | |
| "eval_samples_per_second": 32.8, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.2156, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5814980864524841, | |
| "eval_runtime": 3.0453, | |
| "eval_samples_per_second": 32.838, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5615140795707703, | |
| "eval_runtime": 3.045, | |
| "eval_samples_per_second": 32.841, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.75, | |
| "eval_loss": 0.5819821953773499, | |
| "eval_runtime": 3.0442, | |
| "eval_samples_per_second": 32.85, | |
| "eval_steps_per_second": 4.27, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5712084174156189, | |
| "eval_runtime": 3.0434, | |
| "eval_samples_per_second": 32.858, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5682441592216492, | |
| "eval_runtime": 3.0401, | |
| "eval_samples_per_second": 32.893, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6267381310462952, | |
| "eval_runtime": 3.0478, | |
| "eval_samples_per_second": 32.811, | |
| "eval_steps_per_second": 4.265, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5814998745918274, | |
| "eval_runtime": 3.0406, | |
| "eval_samples_per_second": 32.888, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.617068886756897, | |
| "eval_runtime": 3.0363, | |
| "eval_samples_per_second": 32.935, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5554370284080505, | |
| "eval_runtime": 3.0371, | |
| "eval_samples_per_second": 32.926, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6060196161270142, | |
| "eval_runtime": 3.0399, | |
| "eval_samples_per_second": 32.896, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5574981570243835, | |
| "eval_runtime": 3.0343, | |
| "eval_samples_per_second": 32.956, | |
| "eval_steps_per_second": 4.284, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5885199904441833, | |
| "eval_runtime": 3.0375, | |
| "eval_samples_per_second": 32.922, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5570526123046875, | |
| "eval_runtime": 3.0354, | |
| "eval_samples_per_second": 32.945, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5845230221748352, | |
| "eval_runtime": 3.0345, | |
| "eval_samples_per_second": 32.954, | |
| "eval_steps_per_second": 4.284, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5710142254829407, | |
| "eval_runtime": 3.0367, | |
| "eval_samples_per_second": 32.93, | |
| "eval_steps_per_second": 4.281, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5680321455001831, | |
| "eval_runtime": 3.0337, | |
| "eval_samples_per_second": 32.963, | |
| "eval_steps_per_second": 4.285, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5799435973167419, | |
| "eval_runtime": 3.0373, | |
| "eval_samples_per_second": 32.924, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.563614010810852, | |
| "eval_runtime": 3.0394, | |
| "eval_samples_per_second": 32.901, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5738111734390259, | |
| "eval_runtime": 3.0401, | |
| "eval_samples_per_second": 32.893, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5750021934509277, | |
| "eval_runtime": 3.0413, | |
| "eval_samples_per_second": 32.881, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.194, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5758109092712402, | |
| "eval_runtime": 3.0331, | |
| "eval_samples_per_second": 32.97, | |
| "eval_steps_per_second": 4.286, | |
| "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": 0.3481477241516113, | |
| "train_runtime": 1624.5487, | |
| "train_samples_per_second": 19.698, | |
| "train_steps_per_second": 1.231 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |