| { | |
| "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": 0.61, | |
| "eval_loss": 0.5239093899726868, | |
| "eval_runtime": 2.8956, | |
| "eval_samples_per_second": 34.535, | |
| "eval_steps_per_second": 4.49, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.42313864827156067, | |
| "eval_runtime": 2.9316, | |
| "eval_samples_per_second": 34.111, | |
| "eval_steps_per_second": 4.434, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.4341623783111572, | |
| "eval_runtime": 2.9678, | |
| "eval_samples_per_second": 33.695, | |
| "eval_steps_per_second": 4.38, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.4309338331222534, | |
| "eval_runtime": 2.9919, | |
| "eval_samples_per_second": 33.423, | |
| "eval_steps_per_second": 4.345, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.42618507146835327, | |
| "eval_runtime": 2.9938, | |
| "eval_samples_per_second": 33.402, | |
| "eval_steps_per_second": 4.342, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.42672210931777954, | |
| "eval_runtime": 3.0083, | |
| "eval_samples_per_second": 33.241, | |
| "eval_steps_per_second": 4.321, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.42631956934928894, | |
| "eval_runtime": 3.0171, | |
| "eval_samples_per_second": 33.145, | |
| "eval_steps_per_second": 4.309, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.42677411437034607, | |
| "eval_runtime": 3.0175, | |
| "eval_samples_per_second": 33.14, | |
| "eval_steps_per_second": 4.308, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.42669591307640076, | |
| "eval_runtime": 3.0181, | |
| "eval_samples_per_second": 33.133, | |
| "eval_steps_per_second": 4.307, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.42676296830177307, | |
| "eval_runtime": 3.0209, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4275117814540863, | |
| "eval_runtime": 3.0266, | |
| "eval_samples_per_second": 33.041, | |
| "eval_steps_per_second": 4.295, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.42694392800331116, | |
| "eval_runtime": 3.0194, | |
| "eval_samples_per_second": 33.119, | |
| "eval_steps_per_second": 4.306, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.42673835158348083, | |
| "eval_runtime": 3.0256, | |
| "eval_samples_per_second": 33.051, | |
| "eval_steps_per_second": 4.297, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.4267115294933319, | |
| "eval_runtime": 3.0223, | |
| "eval_samples_per_second": 33.087, | |
| "eval_steps_per_second": 4.301, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.4267922639846802, | |
| "eval_runtime": 3.0218, | |
| "eval_samples_per_second": 33.093, | |
| "eval_steps_per_second": 4.302, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.42681288719177246, | |
| "eval_runtime": 3.0218, | |
| "eval_samples_per_second": 33.093, | |
| "eval_steps_per_second": 4.302, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.42698243260383606, | |
| "eval_runtime": 3.0208, | |
| "eval_samples_per_second": 33.104, | |
| "eval_steps_per_second": 4.304, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.4267401397228241, | |
| "eval_runtime": 3.019, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.4267587959766388, | |
| "eval_runtime": 3.0189, | |
| "eval_samples_per_second": 33.125, | |
| "eval_steps_per_second": 4.306, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.037500000000000006, | |
| "loss": 1.2569, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.4268057644367218, | |
| "eval_runtime": 3.0209, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.4267561435699463, | |
| "eval_runtime": 3.0207, | |
| "eval_samples_per_second": 33.105, | |
| "eval_steps_per_second": 4.304, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.42674654722213745, | |
| "eval_runtime": 3.0209, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.4267425239086151, | |
| "eval_runtime": 3.0182, | |
| "eval_samples_per_second": 33.132, | |
| "eval_steps_per_second": 4.307, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.42674362659454346, | |
| "eval_runtime": 3.018, | |
| "eval_samples_per_second": 33.135, | |
| "eval_steps_per_second": 4.307, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.4267553389072418, | |
| "eval_runtime": 3.0198, | |
| "eval_samples_per_second": 33.115, | |
| "eval_steps_per_second": 4.305, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.42682936787605286, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.42673712968826294, | |
| "eval_runtime": 3.019, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.42678844928741455, | |
| "eval_runtime": 3.0166, | |
| "eval_samples_per_second": 33.15, | |
| "eval_steps_per_second": 4.309, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.42675766348838806, | |
| "eval_runtime": 3.0194, | |
| "eval_samples_per_second": 33.12, | |
| "eval_steps_per_second": 4.306, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.4268340766429901, | |
| "eval_runtime": 3.0188, | |
| "eval_samples_per_second": 33.126, | |
| "eval_steps_per_second": 4.306, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.42684388160705566, | |
| "eval_runtime": 3.0188, | |
| "eval_samples_per_second": 33.126, | |
| "eval_steps_per_second": 4.306, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.42676299810409546, | |
| "eval_runtime": 3.0168, | |
| "eval_samples_per_second": 33.148, | |
| "eval_steps_per_second": 4.309, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4267823100090027, | |
| "eval_runtime": 3.0195, | |
| "eval_samples_per_second": 33.118, | |
| "eval_steps_per_second": 4.305, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.42678964138031006, | |
| "eval_runtime": 3.0178, | |
| "eval_samples_per_second": 33.137, | |
| "eval_steps_per_second": 4.308, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.42680954933166504, | |
| "eval_runtime": 3.0179, | |
| "eval_samples_per_second": 33.136, | |
| "eval_steps_per_second": 4.308, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 12, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.42676040530204773, | |
| "eval_runtime": 3.0184, | |
| "eval_samples_per_second": 33.13, | |
| "eval_steps_per_second": 4.307, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.4267849028110504, | |
| "eval_runtime": 3.0192, | |
| "eval_samples_per_second": 33.122, | |
| "eval_steps_per_second": 4.306, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.4268019497394562, | |
| "eval_runtime": 3.017, | |
| "eval_samples_per_second": 33.145, | |
| "eval_steps_per_second": 4.309, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.42679381370544434, | |
| "eval_runtime": 3.0191, | |
| "eval_samples_per_second": 33.123, | |
| "eval_steps_per_second": 4.306, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.025, | |
| "loss": 0.9505, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.4267384707927704, | |
| "eval_runtime": 3.0439, | |
| "eval_samples_per_second": 32.853, | |
| "eval_steps_per_second": 4.271, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.42674705386161804, | |
| "eval_runtime": 3.0268, | |
| "eval_samples_per_second": 33.038, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.42676541209220886, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.42678970098495483, | |
| "eval_runtime": 3.0222, | |
| "eval_samples_per_second": 33.088, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.4267711937427521, | |
| "eval_runtime": 3.0219, | |
| "eval_samples_per_second": 33.092, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.4267594814300537, | |
| "eval_runtime": 3.0183, | |
| "eval_samples_per_second": 33.131, | |
| "eval_steps_per_second": 4.307, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.42673030495643616, | |
| "eval_runtime": 3.0222, | |
| "eval_samples_per_second": 33.089, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.4267197549343109, | |
| "eval_runtime": 3.0221, | |
| "eval_samples_per_second": 33.089, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.4267454147338867, | |
| "eval_runtime": 3.0238, | |
| "eval_samples_per_second": 33.071, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4267684817314148, | |
| "eval_runtime": 3.0255, | |
| "eval_samples_per_second": 33.052, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.42676082253456116, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4267779588699341, | |
| "eval_runtime": 3.0274, | |
| "eval_samples_per_second": 33.032, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.46, | |
| "eval_loss": 0.42678937315940857, | |
| "eval_runtime": 3.0292, | |
| "eval_samples_per_second": 33.012, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4267615079879761, | |
| "eval_runtime": 3.0317, | |
| "eval_samples_per_second": 32.984, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.426766961812973, | |
| "eval_runtime": 3.0334, | |
| "eval_samples_per_second": 32.966, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.42677998542785645, | |
| "eval_runtime": 3.0347, | |
| "eval_samples_per_second": 32.952, | |
| "eval_steps_per_second": 4.284, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.4267677962779999, | |
| "eval_runtime": 3.0362, | |
| "eval_samples_per_second": 32.936, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.4267272651195526, | |
| "eval_runtime": 3.0368, | |
| "eval_samples_per_second": 32.929, | |
| "eval_steps_per_second": 4.281, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.55, | |
| "eval_loss": 0.42671799659729004, | |
| "eval_runtime": 3.0363, | |
| "eval_samples_per_second": 32.935, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.4267313480377197, | |
| "eval_runtime": 3.0384, | |
| "eval_samples_per_second": 32.912, | |
| "eval_steps_per_second": 4.279, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0125, | |
| "loss": 0.7437, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.42671775817871094, | |
| "eval_runtime": 3.0377, | |
| "eval_samples_per_second": 32.919, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.42675554752349854, | |
| "eval_runtime": 3.0372, | |
| "eval_samples_per_second": 32.925, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.42678001523017883, | |
| "eval_runtime": 3.0378, | |
| "eval_samples_per_second": 32.919, | |
| "eval_steps_per_second": 4.279, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267752468585968, | |
| "eval_runtime": 3.0403, | |
| "eval_samples_per_second": 32.892, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.44, | |
| "eval_loss": 0.4268019199371338, | |
| "eval_runtime": 3.0419, | |
| "eval_samples_per_second": 32.874, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.4267675280570984, | |
| "eval_runtime": 3.0421, | |
| "eval_samples_per_second": 32.872, | |
| "eval_steps_per_second": 4.273, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267730414867401, | |
| "eval_runtime": 3.0427, | |
| "eval_samples_per_second": 32.865, | |
| "eval_steps_per_second": 4.273, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.4267617166042328, | |
| "eval_runtime": 3.0398, | |
| "eval_samples_per_second": 32.897, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4267881512641907, | |
| "eval_runtime": 3.0429, | |
| "eval_samples_per_second": 32.863, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267854392528534, | |
| "eval_runtime": 3.0413, | |
| "eval_samples_per_second": 32.881, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4267839193344116, | |
| "eval_runtime": 3.0528, | |
| "eval_samples_per_second": 32.756, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.42678093910217285, | |
| "eval_runtime": 3.0428, | |
| "eval_samples_per_second": 32.865, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.4267817735671997, | |
| "eval_runtime": 3.0388, | |
| "eval_samples_per_second": 32.908, | |
| "eval_steps_per_second": 4.278, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.4267866611480713, | |
| "eval_runtime": 3.0431, | |
| "eval_samples_per_second": 32.861, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267853796482086, | |
| "eval_runtime": 3.0429, | |
| "eval_samples_per_second": 32.864, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267774820327759, | |
| "eval_runtime": 3.0522, | |
| "eval_samples_per_second": 32.763, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.4267762005329132, | |
| "eval_runtime": 3.043, | |
| "eval_samples_per_second": 32.862, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.42677849531173706, | |
| "eval_runtime": 3.0459, | |
| "eval_samples_per_second": 32.831, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.41, | |
| "eval_loss": 0.4267786741256714, | |
| "eval_runtime": 3.0437, | |
| "eval_samples_per_second": 32.855, | |
| "eval_steps_per_second": 4.271, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.38, | |
| "eval_loss": 0.42677730321884155, | |
| "eval_runtime": 3.0417, | |
| "eval_samples_per_second": 32.876, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.6146, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.37, | |
| "eval_loss": 0.42677754163742065, | |
| "eval_runtime": 3.0421, | |
| "eval_samples_per_second": 32.872, | |
| "eval_steps_per_second": 4.273, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.8914242706298828, | |
| "train_runtime": 1627.2695, | |
| "train_samples_per_second": 19.665, | |
| "train_steps_per_second": 1.229 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |