| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 80.0, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.8357499837875366, | |
| "eval_runtime": 2.905, | |
| "eval_samples_per_second": 34.423, | |
| "eval_steps_per_second": 4.475, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.4, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6364213228225708, | |
| "eval_runtime": 2.9225, | |
| "eval_samples_per_second": 34.218, | |
| "eval_steps_per_second": 4.448, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.7512836456298828, | |
| "eval_runtime": 2.9503, | |
| "eval_samples_per_second": 33.895, | |
| "eval_steps_per_second": 4.406, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.5949835181236267, | |
| "eval_runtime": 2.9767, | |
| "eval_samples_per_second": 33.594, | |
| "eval_steps_per_second": 4.367, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.6111248135566711, | |
| "eval_runtime": 3.0094, | |
| "eval_samples_per_second": 33.23, | |
| "eval_steps_per_second": 4.32, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.7314449548721313, | |
| "eval_runtime": 3.0154, | |
| "eval_samples_per_second": 33.163, | |
| "eval_steps_per_second": 4.311, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6188446283340454, | |
| "eval_runtime": 3.0184, | |
| "eval_samples_per_second": 33.13, | |
| "eval_steps_per_second": 4.307, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 1.202831745147705, | |
| "eval_runtime": 3.0198, | |
| "eval_samples_per_second": 33.114, | |
| "eval_steps_per_second": 4.305, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.630307674407959, | |
| "eval_runtime": 3.0221, | |
| "eval_samples_per_second": 33.09, | |
| "eval_steps_per_second": 4.302, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.8704754114151001, | |
| "eval_runtime": 3.0231, | |
| "eval_samples_per_second": 33.078, | |
| "eval_steps_per_second": 4.3, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5481266379356384, | |
| "eval_runtime": 3.0256, | |
| "eval_samples_per_second": 33.052, | |
| "eval_steps_per_second": 4.297, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.8700436949729919, | |
| "eval_runtime": 3.0239, | |
| "eval_samples_per_second": 33.07, | |
| "eval_steps_per_second": 4.299, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.7616304159164429, | |
| "eval_runtime": 3.027, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.295, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7384722232818604, | |
| "eval_runtime": 3.0286, | |
| "eval_samples_per_second": 33.019, | |
| "eval_steps_per_second": 4.292, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.8501294851303101, | |
| "eval_runtime": 3.0347, | |
| "eval_samples_per_second": 32.952, | |
| "eval_steps_per_second": 4.284, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.695446252822876, | |
| "eval_runtime": 3.0254, | |
| "eval_samples_per_second": 33.053, | |
| "eval_steps_per_second": 4.297, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.6255203485488892, | |
| "eval_runtime": 3.021, | |
| "eval_samples_per_second": 33.102, | |
| "eval_steps_per_second": 4.303, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.6264253854751587, | |
| "eval_runtime": 3.0191, | |
| "eval_samples_per_second": 33.123, | |
| "eval_steps_per_second": 4.306, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.6275391578674316, | |
| "eval_runtime": 3.0206, | |
| "eval_samples_per_second": 33.106, | |
| "eval_steps_per_second": 4.304, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.037500000000000006, | |
| "loss": 1.5048, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6259335279464722, | |
| "eval_runtime": 3.0208, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.6269640326499939, | |
| "eval_runtime": 3.0178, | |
| "eval_samples_per_second": 33.137, | |
| "eval_steps_per_second": 4.308, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.6274638772010803, | |
| "eval_runtime": 3.0203, | |
| "eval_samples_per_second": 33.11, | |
| "eval_steps_per_second": 4.304, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6249264478683472, | |
| "eval_runtime": 3.0198, | |
| "eval_samples_per_second": 33.115, | |
| "eval_steps_per_second": 4.305, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.6269152164459229, | |
| "eval_runtime": 3.0189, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.6253538727760315, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.6265491247177124, | |
| "eval_runtime": 3.0189, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6261975169181824, | |
| "eval_runtime": 3.0184, | |
| "eval_samples_per_second": 33.13, | |
| "eval_steps_per_second": 4.307, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.6247080564498901, | |
| "eval_runtime": 3.0202, | |
| "eval_samples_per_second": 33.11, | |
| "eval_steps_per_second": 4.304, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6241124272346497, | |
| "eval_runtime": 3.0185, | |
| "eval_samples_per_second": 33.129, | |
| "eval_steps_per_second": 4.307, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.6246690154075623, | |
| "eval_runtime": 3.0184, | |
| "eval_samples_per_second": 33.13, | |
| "eval_steps_per_second": 4.307, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.6262466907501221, | |
| "eval_runtime": 3.0193, | |
| "eval_samples_per_second": 33.121, | |
| "eval_steps_per_second": 4.306, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6260530352592468, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.6261406540870667, | |
| "eval_runtime": 3.0185, | |
| "eval_samples_per_second": 33.129, | |
| "eval_steps_per_second": 4.307, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.6263586282730103, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.6266360282897949, | |
| "eval_runtime": 3.0196, | |
| "eval_samples_per_second": 33.117, | |
| "eval_steps_per_second": 4.305, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.626537561416626, | |
| "eval_runtime": 3.0175, | |
| "eval_samples_per_second": 33.141, | |
| "eval_steps_per_second": 4.308, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.6262317895889282, | |
| "eval_runtime": 3.0187, | |
| "eval_samples_per_second": 33.127, | |
| "eval_steps_per_second": 4.306, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.6264010667800903, | |
| "eval_runtime": 3.0289, | |
| "eval_samples_per_second": 33.016, | |
| "eval_steps_per_second": 4.292, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.6263816356658936, | |
| "eval_runtime": 3.0275, | |
| "eval_samples_per_second": 33.031, | |
| "eval_steps_per_second": 4.294, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.025, | |
| "loss": 1.2203, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6261613965034485, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.6262817978858948, | |
| "eval_runtime": 3.0154, | |
| "eval_samples_per_second": 33.163, | |
| "eval_steps_per_second": 4.311, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6261902451515198, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.6264880895614624, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6262249946594238, | |
| "eval_runtime": 3.0162, | |
| "eval_samples_per_second": 33.154, | |
| "eval_steps_per_second": 4.31, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6261861324310303, | |
| "eval_runtime": 3.0154, | |
| "eval_samples_per_second": 33.163, | |
| "eval_steps_per_second": 4.311, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.6262775659561157, | |
| "eval_runtime": 3.0142, | |
| "eval_samples_per_second": 33.177, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6262478828430176, | |
| "eval_runtime": 3.014, | |
| "eval_samples_per_second": 33.178, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.6262679100036621, | |
| "eval_runtime": 3.0142, | |
| "eval_samples_per_second": 33.176, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.39, | |
| "eval_loss": 0.6264752149581909, | |
| "eval_runtime": 3.014, | |
| "eval_samples_per_second": 33.178, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6261715292930603, | |
| "eval_runtime": 3.0114, | |
| "eval_samples_per_second": 33.207, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.626240611076355, | |
| "eval_runtime": 3.0123, | |
| "eval_samples_per_second": 33.197, | |
| "eval_steps_per_second": 4.316, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.6261351108551025, | |
| "eval_runtime": 3.0136, | |
| "eval_samples_per_second": 33.183, | |
| "eval_steps_per_second": 4.314, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.6261637210845947, | |
| "eval_runtime": 3.0144, | |
| "eval_samples_per_second": 33.174, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.6261104345321655, | |
| "eval_runtime": 3.014, | |
| "eval_samples_per_second": 33.179, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6260131597518921, | |
| "eval_runtime": 3.0129, | |
| "eval_samples_per_second": 33.19, | |
| "eval_steps_per_second": 4.315, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6261314749717712, | |
| "eval_runtime": 3.0129, | |
| "eval_samples_per_second": 33.191, | |
| "eval_steps_per_second": 4.315, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.6263273358345032, | |
| "eval_runtime": 3.0142, | |
| "eval_samples_per_second": 33.176, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.6263743042945862, | |
| "eval_runtime": 3.0121, | |
| "eval_samples_per_second": 33.199, | |
| "eval_steps_per_second": 4.316, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.6262876391410828, | |
| "eval_runtime": 3.0151, | |
| "eval_samples_per_second": 33.167, | |
| "eval_steps_per_second": 4.312, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0125, | |
| "loss": 0.9516, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.6262710690498352, | |
| "eval_runtime": 3.0132, | |
| "eval_samples_per_second": 33.187, | |
| "eval_steps_per_second": 4.314, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.6263102889060974, | |
| "eval_runtime": 3.0116, | |
| "eval_samples_per_second": 33.205, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6261453628540039, | |
| "eval_runtime": 3.0117, | |
| "eval_samples_per_second": 33.203, | |
| "eval_steps_per_second": 4.316, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6262593865394592, | |
| "eval_runtime": 3.0153, | |
| "eval_samples_per_second": 33.164, | |
| "eval_steps_per_second": 4.311, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6261275410652161, | |
| "eval_runtime": 3.019, | |
| "eval_samples_per_second": 33.123, | |
| "eval_steps_per_second": 4.306, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.6263031959533691, | |
| "eval_runtime": 3.0189, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.39, | |
| "eval_loss": 0.6264517307281494, | |
| "eval_runtime": 3.0111, | |
| "eval_samples_per_second": 33.21, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6262269020080566, | |
| "eval_runtime": 3.0148, | |
| "eval_samples_per_second": 33.17, | |
| "eval_steps_per_second": 4.312, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.6263662576675415, | |
| "eval_runtime": 3.0138, | |
| "eval_samples_per_second": 33.18, | |
| "eval_steps_per_second": 4.313, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.6261860132217407, | |
| "eval_runtime": 3.0135, | |
| "eval_samples_per_second": 33.184, | |
| "eval_steps_per_second": 4.314, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.6262713074684143, | |
| "eval_runtime": 3.013, | |
| "eval_samples_per_second": 33.189, | |
| "eval_steps_per_second": 4.315, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6261432766914368, | |
| "eval_runtime": 3.0111, | |
| "eval_samples_per_second": 33.21, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.6263407468795776, | |
| "eval_runtime": 3.0159, | |
| "eval_samples_per_second": 33.158, | |
| "eval_steps_per_second": 4.311, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6262317895889282, | |
| "eval_runtime": 3.0131, | |
| "eval_samples_per_second": 33.189, | |
| "eval_steps_per_second": 4.315, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.6263194680213928, | |
| "eval_runtime": 3.0137, | |
| "eval_samples_per_second": 33.182, | |
| "eval_steps_per_second": 4.314, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6262105703353882, | |
| "eval_runtime": 3.0116, | |
| "eval_samples_per_second": 33.204, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.6263117790222168, | |
| "eval_runtime": 3.0104, | |
| "eval_samples_per_second": 33.218, | |
| "eval_steps_per_second": 4.318, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.6263181567192078, | |
| "eval_runtime": 3.0116, | |
| "eval_samples_per_second": 33.205, | |
| "eval_steps_per_second": 4.317, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.6262911558151245, | |
| "eval_runtime": 3.0137, | |
| "eval_samples_per_second": 33.182, | |
| "eval_steps_per_second": 4.314, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.6263221502304077, | |
| "eval_runtime": 3.0254, | |
| "eval_samples_per_second": 33.053, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.7734, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.6263124346733093, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 1.112525405883789, | |
| "train_runtime": 1617.2068, | |
| "train_samples_per_second": 19.787, | |
| "train_steps_per_second": 1.237 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |