| { | |
| "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": 0.66, | |
| "eval_loss": 0.6010557413101196, | |
| "eval_runtime": 2.8952, | |
| "eval_samples_per_second": 34.539, | |
| "eval_steps_per_second": 4.49, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5991448163986206, | |
| "eval_runtime": 2.9197, | |
| "eval_samples_per_second": 34.25, | |
| "eval_steps_per_second": 4.453, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5983209609985352, | |
| "eval_runtime": 2.9581, | |
| "eval_samples_per_second": 33.806, | |
| "eval_steps_per_second": 4.395, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6062744855880737, | |
| "eval_runtime": 2.9913, | |
| "eval_samples_per_second": 33.43, | |
| "eval_steps_per_second": 4.346, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5972874164581299, | |
| "eval_runtime": 3.0135, | |
| "eval_samples_per_second": 33.184, | |
| "eval_steps_per_second": 4.314, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6048508286476135, | |
| "eval_runtime": 3.0172, | |
| "eval_samples_per_second": 33.143, | |
| "eval_steps_per_second": 4.309, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6030882596969604, | |
| "eval_runtime": 3.0217, | |
| "eval_samples_per_second": 33.094, | |
| "eval_steps_per_second": 4.302, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6000898480415344, | |
| "eval_runtime": 3.0239, | |
| "eval_samples_per_second": 33.07, | |
| "eval_steps_per_second": 4.299, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5968500971794128, | |
| "eval_runtime": 3.0268, | |
| "eval_samples_per_second": 33.038, | |
| "eval_steps_per_second": 4.295, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6007280945777893, | |
| "eval_runtime": 3.0282, | |
| "eval_samples_per_second": 33.023, | |
| "eval_steps_per_second": 4.293, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6015586853027344, | |
| "eval_runtime": 3.027, | |
| "eval_samples_per_second": 33.036, | |
| "eval_steps_per_second": 4.295, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5992419123649597, | |
| "eval_runtime": 3.0339, | |
| "eval_samples_per_second": 32.961, | |
| "eval_steps_per_second": 4.285, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5967671275138855, | |
| "eval_runtime": 3.0311, | |
| "eval_samples_per_second": 32.992, | |
| "eval_steps_per_second": 4.289, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.596794605255127, | |
| "eval_runtime": 3.0286, | |
| "eval_samples_per_second": 33.018, | |
| "eval_steps_per_second": 4.292, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.600039541721344, | |
| "eval_runtime": 3.0308, | |
| "eval_samples_per_second": 32.995, | |
| "eval_steps_per_second": 4.289, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6000267267227173, | |
| "eval_runtime": 3.0298, | |
| "eval_samples_per_second": 33.006, | |
| "eval_steps_per_second": 4.291, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5882958769798279, | |
| "eval_runtime": 3.029, | |
| "eval_samples_per_second": 33.015, | |
| "eval_steps_per_second": 4.292, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5920267701148987, | |
| "eval_runtime": 3.0314, | |
| "eval_samples_per_second": 32.988, | |
| "eval_steps_per_second": 4.288, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6034604907035828, | |
| "eval_runtime": 3.0306, | |
| "eval_samples_per_second": 32.997, | |
| "eval_steps_per_second": 4.29, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00075, | |
| "loss": 0.6519, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6074941158294678, | |
| "eval_runtime": 3.0321, | |
| "eval_samples_per_second": 32.981, | |
| "eval_steps_per_second": 4.287, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5919292569160461, | |
| "eval_runtime": 3.03, | |
| "eval_samples_per_second": 33.003, | |
| "eval_steps_per_second": 4.29, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5951123237609863, | |
| "eval_runtime": 3.0298, | |
| "eval_samples_per_second": 33.006, | |
| "eval_steps_per_second": 4.291, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.603710412979126, | |
| "eval_runtime": 3.0294, | |
| "eval_samples_per_second": 33.009, | |
| "eval_steps_per_second": 4.291, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6057878732681274, | |
| "eval_runtime": 3.0274, | |
| "eval_samples_per_second": 33.032, | |
| "eval_steps_per_second": 4.294, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5943801999092102, | |
| "eval_runtime": 3.0273, | |
| "eval_samples_per_second": 33.033, | |
| "eval_steps_per_second": 4.294, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5938175320625305, | |
| "eval_runtime": 3.0318, | |
| "eval_samples_per_second": 32.983, | |
| "eval_steps_per_second": 4.288, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5908843278884888, | |
| "eval_runtime": 3.0311, | |
| "eval_samples_per_second": 32.991, | |
| "eval_steps_per_second": 4.289, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5914073586463928, | |
| "eval_runtime": 3.0408, | |
| "eval_samples_per_second": 32.886, | |
| "eval_steps_per_second": 4.275, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5901620388031006, | |
| "eval_runtime": 3.0355, | |
| "eval_samples_per_second": 32.943, | |
| "eval_steps_per_second": 4.283, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5905589461326599, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.073, | |
| "eval_steps_per_second": 4.3, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5935985445976257, | |
| "eval_runtime": 3.023, | |
| "eval_samples_per_second": 33.08, | |
| "eval_steps_per_second": 4.3, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5959981679916382, | |
| "eval_runtime": 3.0241, | |
| "eval_samples_per_second": 33.068, | |
| "eval_steps_per_second": 4.299, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.595251739025116, | |
| "eval_runtime": 3.0223, | |
| "eval_samples_per_second": 33.088, | |
| "eval_steps_per_second": 4.301, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.597002387046814, | |
| "eval_runtime": 3.0261, | |
| "eval_samples_per_second": 33.046, | |
| "eval_steps_per_second": 4.296, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.593684732913971, | |
| "eval_runtime": 3.0258, | |
| "eval_samples_per_second": 33.049, | |
| "eval_steps_per_second": 4.296, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5954169034957886, | |
| "eval_runtime": 3.0248, | |
| "eval_samples_per_second": 33.06, | |
| "eval_steps_per_second": 4.298, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5992526412010193, | |
| "eval_runtime": 3.0254, | |
| "eval_samples_per_second": 33.053, | |
| "eval_steps_per_second": 4.297, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5905333161354065, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5898448824882507, | |
| "eval_runtime": 3.0254, | |
| "eval_samples_per_second": 33.053, | |
| "eval_steps_per_second": 4.297, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0005, | |
| "loss": 0.6395, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5947265028953552, | |
| "eval_runtime": 3.0242, | |
| "eval_samples_per_second": 33.067, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5965713262557983, | |
| "eval_runtime": 3.0253, | |
| "eval_samples_per_second": 33.054, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.595311164855957, | |
| "eval_runtime": 3.0266, | |
| "eval_samples_per_second": 33.04, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.59684818983078, | |
| "eval_runtime": 3.0241, | |
| "eval_samples_per_second": 33.067, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5934141874313354, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.59480881690979, | |
| "eval_runtime": 3.0258, | |
| "eval_samples_per_second": 33.049, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5958330035209656, | |
| "eval_runtime": 3.0235, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5928426384925842, | |
| "eval_runtime": 3.0224, | |
| "eval_samples_per_second": 33.086, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5922430157661438, | |
| "eval_runtime": 3.0243, | |
| "eval_samples_per_second": 33.065, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5928506255149841, | |
| "eval_runtime": 3.0246, | |
| "eval_samples_per_second": 33.063, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5967039465904236, | |
| "eval_runtime": 3.0208, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5907678008079529, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5929586887359619, | |
| "eval_runtime": 3.0225, | |
| "eval_samples_per_second": 33.085, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.590965986251831, | |
| "eval_runtime": 3.0223, | |
| "eval_samples_per_second": 33.088, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5930989980697632, | |
| "eval_runtime": 3.0227, | |
| "eval_samples_per_second": 33.083, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5900382399559021, | |
| "eval_runtime": 3.022, | |
| "eval_samples_per_second": 33.091, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5924661159515381, | |
| "eval_runtime": 3.0221, | |
| "eval_samples_per_second": 33.09, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5937650203704834, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.073, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5963329076766968, | |
| "eval_runtime": 3.0249, | |
| "eval_samples_per_second": 33.059, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5954621434211731, | |
| "eval_runtime": 3.0247, | |
| "eval_samples_per_second": 33.061, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.00025, | |
| "loss": 0.6331, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5935218930244446, | |
| "eval_runtime": 3.0258, | |
| "eval_samples_per_second": 33.049, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5937374830245972, | |
| "eval_runtime": 3.0237, | |
| "eval_samples_per_second": 33.072, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5923505425453186, | |
| "eval_runtime": 3.0211, | |
| "eval_samples_per_second": 33.101, | |
| "eval_steps_per_second": 4.303, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5909485816955566, | |
| "eval_runtime": 3.0219, | |
| "eval_samples_per_second": 33.092, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.589134931564331, | |
| "eval_runtime": 3.026, | |
| "eval_samples_per_second": 33.047, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5880649089813232, | |
| "eval_runtime": 3.0289, | |
| "eval_samples_per_second": 33.015, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5884314775466919, | |
| "eval_runtime": 3.0234, | |
| "eval_samples_per_second": 33.075, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5893000960350037, | |
| "eval_runtime": 3.0241, | |
| "eval_samples_per_second": 33.067, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5899676084518433, | |
| "eval_runtime": 3.028, | |
| "eval_samples_per_second": 33.025, | |
| "eval_steps_per_second": 4.293, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5908388495445251, | |
| "eval_runtime": 3.0265, | |
| "eval_samples_per_second": 33.041, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5911998748779297, | |
| "eval_runtime": 3.0244, | |
| "eval_samples_per_second": 33.065, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5914051532745361, | |
| "eval_runtime": 3.0277, | |
| "eval_samples_per_second": 33.029, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5901281237602234, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5898309350013733, | |
| "eval_runtime": 3.0251, | |
| "eval_samples_per_second": 33.057, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.589576005935669, | |
| "eval_runtime": 3.0338, | |
| "eval_samples_per_second": 32.961, | |
| "eval_steps_per_second": 4.285, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5904554128646851, | |
| "eval_runtime": 3.0252, | |
| "eval_samples_per_second": 33.055, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5900978446006775, | |
| "eval_runtime": 3.0257, | |
| "eval_samples_per_second": 33.05, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5900523662567139, | |
| "eval_runtime": 3.0269, | |
| "eval_samples_per_second": 33.037, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5899617671966553, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.034, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5899820327758789, | |
| "eval_runtime": 3.0253, | |
| "eval_samples_per_second": 33.055, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.6276, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5899689793586731, | |
| "eval_runtime": 3.0275, | |
| "eval_samples_per_second": 33.03, | |
| "eval_steps_per_second": 4.294, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.6380199432373047, | |
| "train_runtime": 1620.376, | |
| "train_samples_per_second": 19.749, | |
| "train_steps_per_second": 1.234 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |