| { | |
| "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": 0.6, | |
| "eval_loss": 0.7348268628120422, | |
| "eval_runtime": 2.9006, | |
| "eval_samples_per_second": 34.475, | |
| "eval_steps_per_second": 4.482, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6044729948043823, | |
| "eval_runtime": 2.9185, | |
| "eval_samples_per_second": 34.264, | |
| "eval_steps_per_second": 4.454, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.9239120483398438, | |
| "eval_runtime": 2.9442, | |
| "eval_samples_per_second": 33.965, | |
| "eval_steps_per_second": 4.415, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.62, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6379037499427795, | |
| "eval_runtime": 2.9811, | |
| "eval_samples_per_second": 33.545, | |
| "eval_steps_per_second": 4.361, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5724024176597595, | |
| "eval_runtime": 2.9944, | |
| "eval_samples_per_second": 33.396, | |
| "eval_steps_per_second": 4.341, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 4, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 1.208309292793274, | |
| "eval_runtime": 3.0023, | |
| "eval_samples_per_second": 33.308, | |
| "eval_steps_per_second": 4.33, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 4, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.3074427843093872, | |
| "eval_runtime": 3.0095, | |
| "eval_samples_per_second": 33.228, | |
| "eval_steps_per_second": 4.32, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 4, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 1.1625832319259644, | |
| "eval_runtime": 3.0155, | |
| "eval_samples_per_second": 33.162, | |
| "eval_steps_per_second": 4.311, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 4, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 1.0019404888153076, | |
| "eval_runtime": 3.0171, | |
| "eval_samples_per_second": 33.144, | |
| "eval_steps_per_second": 4.309, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 4, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6239729523658752, | |
| "eval_runtime": 3.019, | |
| "eval_samples_per_second": 33.124, | |
| "eval_steps_per_second": 4.306, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 1.0828949213027954, | |
| "eval_runtime": 3.0193, | |
| "eval_samples_per_second": 33.12, | |
| "eval_steps_per_second": 4.306, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.8052525520324707, | |
| "eval_runtime": 3.0244, | |
| "eval_samples_per_second": 33.064, | |
| "eval_steps_per_second": 4.298, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 1.152608871459961, | |
| "eval_runtime": 3.0194, | |
| "eval_samples_per_second": 33.119, | |
| "eval_steps_per_second": 4.305, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 1.2005667686462402, | |
| "eval_runtime": 3.0191, | |
| "eval_samples_per_second": 33.122, | |
| "eval_steps_per_second": 4.306, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.1382259130477905, | |
| "eval_runtime": 3.0159, | |
| "eval_samples_per_second": 33.157, | |
| "eval_steps_per_second": 4.31, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 1.134488821029663, | |
| "eval_runtime": 3.0167, | |
| "eval_samples_per_second": 33.148, | |
| "eval_steps_per_second": 4.309, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.5028630495071411, | |
| "eval_runtime": 3.0173, | |
| "eval_samples_per_second": 33.142, | |
| "eval_steps_per_second": 4.308, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.3780053853988647, | |
| "eval_runtime": 3.0177, | |
| "eval_samples_per_second": 33.138, | |
| "eval_steps_per_second": 4.308, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 1.1811206340789795, | |
| "eval_runtime": 3.0193, | |
| "eval_samples_per_second": 33.12, | |
| "eval_steps_per_second": 4.306, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.037500000000000006, | |
| "loss": 1.3151, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 1.246147871017456, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 1.2269139289855957, | |
| "eval_runtime": 3.0181, | |
| "eval_samples_per_second": 33.133, | |
| "eval_steps_per_second": 4.307, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 1.151503324508667, | |
| "eval_runtime": 3.018, | |
| "eval_samples_per_second": 33.135, | |
| "eval_steps_per_second": 4.307, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.9943810105323792, | |
| "eval_runtime": 3.0183, | |
| "eval_samples_per_second": 33.131, | |
| "eval_steps_per_second": 4.307, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.2708048820495605, | |
| "eval_runtime": 3.02, | |
| "eval_samples_per_second": 33.112, | |
| "eval_steps_per_second": 4.305, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 1.5816699266433716, | |
| "eval_runtime": 3.0182, | |
| "eval_samples_per_second": 33.132, | |
| "eval_steps_per_second": 4.307, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 1.093419075012207, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.4179078340530396, | |
| "eval_runtime": 3.017, | |
| "eval_samples_per_second": 33.146, | |
| "eval_steps_per_second": 4.309, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 1.4260410070419312, | |
| "eval_runtime": 3.0174, | |
| "eval_samples_per_second": 33.141, | |
| "eval_steps_per_second": 4.308, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 1.381821870803833, | |
| "eval_runtime": 3.0154, | |
| "eval_samples_per_second": 33.163, | |
| "eval_steps_per_second": 4.311, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 1.7165961265563965, | |
| "eval_runtime": 3.0194, | |
| "eval_samples_per_second": 33.119, | |
| "eval_steps_per_second": 4.305, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 1.1709787845611572, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 1.0659561157226562, | |
| "eval_runtime": 3.0169, | |
| "eval_samples_per_second": 33.146, | |
| "eval_steps_per_second": 4.309, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 1.012652039527893, | |
| "eval_runtime": 3.0164, | |
| "eval_samples_per_second": 33.153, | |
| "eval_steps_per_second": 4.31, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.9809643030166626, | |
| "eval_runtime": 3.0193, | |
| "eval_samples_per_second": 33.121, | |
| "eval_steps_per_second": 4.306, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 1.1077015399932861, | |
| "eval_runtime": 3.0155, | |
| "eval_samples_per_second": 33.162, | |
| "eval_steps_per_second": 4.311, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 1.0629407167434692, | |
| "eval_runtime": 3.0168, | |
| "eval_samples_per_second": 33.147, | |
| "eval_steps_per_second": 4.309, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 1.5933445692062378, | |
| "eval_runtime": 3.035, | |
| "eval_samples_per_second": 32.949, | |
| "eval_steps_per_second": 4.283, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 1.1321765184402466, | |
| "eval_runtime": 3.0282, | |
| "eval_samples_per_second": 33.023, | |
| "eval_steps_per_second": 4.293, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 1.0734666585922241, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.025, | |
| "loss": 0.6791, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.894046425819397, | |
| "eval_runtime": 3.0277, | |
| "eval_samples_per_second": 33.028, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.9348558783531189, | |
| "eval_runtime": 3.0186, | |
| "eval_samples_per_second": 33.128, | |
| "eval_steps_per_second": 4.307, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.8961507678031921, | |
| "eval_runtime": 3.0195, | |
| "eval_samples_per_second": 33.118, | |
| "eval_steps_per_second": 4.305, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 1.0662699937820435, | |
| "eval_runtime": 3.0298, | |
| "eval_samples_per_second": 33.006, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.9680778980255127, | |
| "eval_runtime": 3.0292, | |
| "eval_samples_per_second": 33.012, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7694079875946045, | |
| "eval_runtime": 3.0181, | |
| "eval_samples_per_second": 33.134, | |
| "eval_steps_per_second": 4.307, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 1.0310795307159424, | |
| "eval_runtime": 3.0271, | |
| "eval_samples_per_second": 33.035, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7406817674636841, | |
| "eval_runtime": 3.0217, | |
| "eval_samples_per_second": 33.094, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.686066210269928, | |
| "eval_runtime": 3.0204, | |
| "eval_samples_per_second": 33.108, | |
| "eval_steps_per_second": 4.304, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.9919649362564087, | |
| "eval_runtime": 3.0244, | |
| "eval_samples_per_second": 33.064, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7187044620513916, | |
| "eval_runtime": 3.0245, | |
| "eval_samples_per_second": 33.064, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7602307200431824, | |
| "eval_runtime": 3.0264, | |
| "eval_samples_per_second": 33.043, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.728481650352478, | |
| "eval_runtime": 3.0237, | |
| "eval_samples_per_second": 33.072, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.8232598304748535, | |
| "eval_runtime": 3.0259, | |
| "eval_samples_per_second": 33.049, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7931870818138123, | |
| "eval_runtime": 3.0274, | |
| "eval_samples_per_second": 33.032, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.8861235976219177, | |
| "eval_runtime": 3.0352, | |
| "eval_samples_per_second": 32.947, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.7876631021499634, | |
| "eval_runtime": 3.0388, | |
| "eval_samples_per_second": 32.908, | |
| "eval_steps_per_second": 4.278, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7689203023910522, | |
| "eval_runtime": 3.0331, | |
| "eval_samples_per_second": 32.97, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.791912317276001, | |
| "eval_runtime": 3.0297, | |
| "eval_samples_per_second": 33.006, | |
| "eval_steps_per_second": 4.291, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7441076636314392, | |
| "eval_runtime": 3.0301, | |
| "eval_samples_per_second": 33.003, | |
| "eval_steps_per_second": 4.29, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0125, | |
| "loss": 0.3594, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.8327054381370544, | |
| "eval_runtime": 3.0351, | |
| "eval_samples_per_second": 32.948, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.641401469707489, | |
| "eval_runtime": 3.0414, | |
| "eval_samples_per_second": 32.88, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6702091097831726, | |
| "eval_runtime": 3.0349, | |
| "eval_samples_per_second": 32.95, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6862046718597412, | |
| "eval_runtime": 3.0397, | |
| "eval_samples_per_second": 32.898, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6349497437477112, | |
| "eval_runtime": 3.0377, | |
| "eval_samples_per_second": 32.92, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6800190210342407, | |
| "eval_runtime": 3.0362, | |
| "eval_samples_per_second": 32.936, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7005452513694763, | |
| "eval_runtime": 3.035, | |
| "eval_samples_per_second": 32.949, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.705758273601532, | |
| "eval_runtime": 3.0397, | |
| "eval_samples_per_second": 32.898, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6880057454109192, | |
| "eval_runtime": 3.0368, | |
| "eval_samples_per_second": 32.929, | |
| "eval_steps_per_second": 4.281, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.6774427890777588, | |
| "eval_runtime": 3.035, | |
| "eval_samples_per_second": 32.949, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6815641522407532, | |
| "eval_runtime": 3.0352, | |
| "eval_samples_per_second": 32.947, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.7138052582740784, | |
| "eval_runtime": 3.0374, | |
| "eval_samples_per_second": 32.923, | |
| "eval_steps_per_second": 4.28, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6311395168304443, | |
| "eval_runtime": 3.0362, | |
| "eval_samples_per_second": 32.936, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6578870415687561, | |
| "eval_runtime": 3.0358, | |
| "eval_samples_per_second": 32.94, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6955811381340027, | |
| "eval_runtime": 3.0392, | |
| "eval_samples_per_second": 32.903, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6341457366943359, | |
| "eval_runtime": 3.0349, | |
| "eval_samples_per_second": 32.95, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6722196936607361, | |
| "eval_runtime": 3.0452, | |
| "eval_samples_per_second": 32.838, | |
| "eval_steps_per_second": 4.269, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6459449529647827, | |
| "eval_runtime": 3.0323, | |
| "eval_samples_per_second": 32.978, | |
| "eval_steps_per_second": 4.287, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6350699663162231, | |
| "eval_runtime": 3.0361, | |
| "eval_samples_per_second": 32.937, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6435720324516296, | |
| "eval_runtime": 3.0398, | |
| "eval_samples_per_second": 32.897, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2323, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6392772793769836, | |
| "eval_runtime": 3.0389, | |
| "eval_samples_per_second": 32.907, | |
| "eval_steps_per_second": 4.278, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.6464771995544434, | |
| "train_runtime": 1617.1485, | |
| "train_samples_per_second": 19.788, | |
| "train_steps_per_second": 1.237 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |