| { | |
| "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": 0.65, | |
| "eval_loss": 0.6077527403831482, | |
| "eval_runtime": 2.8957, | |
| "eval_samples_per_second": 34.534, | |
| "eval_steps_per_second": 4.489, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5962929725646973, | |
| "eval_runtime": 2.9223, | |
| "eval_samples_per_second": 34.22, | |
| "eval_steps_per_second": 4.449, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6125411987304688, | |
| "eval_runtime": 2.9502, | |
| "eval_samples_per_second": 33.896, | |
| "eval_steps_per_second": 4.407, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6042141914367676, | |
| "eval_runtime": 2.9786, | |
| "eval_samples_per_second": 33.573, | |
| "eval_steps_per_second": 4.364, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6065436005592346, | |
| "eval_runtime": 2.9992, | |
| "eval_samples_per_second": 33.342, | |
| "eval_steps_per_second": 4.334, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6019887328147888, | |
| "eval_runtime": 3.0098, | |
| "eval_samples_per_second": 33.225, | |
| "eval_steps_per_second": 4.319, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.598693311214447, | |
| "eval_runtime": 3.016, | |
| "eval_samples_per_second": 33.157, | |
| "eval_steps_per_second": 4.31, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6016180515289307, | |
| "eval_runtime": 3.0167, | |
| "eval_samples_per_second": 33.149, | |
| "eval_steps_per_second": 4.309, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6065845489501953, | |
| "eval_runtime": 3.0194, | |
| "eval_samples_per_second": 33.119, | |
| "eval_steps_per_second": 4.305, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6111646294593811, | |
| "eval_runtime": 3.0214, | |
| "eval_samples_per_second": 33.097, | |
| "eval_steps_per_second": 4.303, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.608534574508667, | |
| "eval_runtime": 3.0216, | |
| "eval_samples_per_second": 33.095, | |
| "eval_steps_per_second": 4.302, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5975815057754517, | |
| "eval_runtime": 3.0222, | |
| "eval_samples_per_second": 33.089, | |
| "eval_steps_per_second": 4.302, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6074051856994629, | |
| "eval_runtime": 3.0229, | |
| "eval_samples_per_second": 33.081, | |
| "eval_steps_per_second": 4.3, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6060487627983093, | |
| "eval_runtime": 3.0206, | |
| "eval_samples_per_second": 33.106, | |
| "eval_steps_per_second": 4.304, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6253820061683655, | |
| "eval_runtime": 3.022, | |
| "eval_samples_per_second": 33.09, | |
| "eval_steps_per_second": 4.302, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6030685305595398, | |
| "eval_runtime": 3.022, | |
| "eval_samples_per_second": 33.091, | |
| "eval_steps_per_second": 4.302, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6011382937431335, | |
| "eval_runtime": 3.0277, | |
| "eval_samples_per_second": 33.028, | |
| "eval_steps_per_second": 4.294, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6062633991241455, | |
| "eval_runtime": 3.0243, | |
| "eval_samples_per_second": 33.066, | |
| "eval_steps_per_second": 4.299, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6030519008636475, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00075, | |
| "loss": 0.6484, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6012805700302124, | |
| "eval_runtime": 3.023, | |
| "eval_samples_per_second": 33.08, | |
| "eval_steps_per_second": 4.3, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6041200160980225, | |
| "eval_runtime": 3.023, | |
| "eval_samples_per_second": 33.08, | |
| "eval_steps_per_second": 4.3, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.603689968585968, | |
| "eval_runtime": 3.0216, | |
| "eval_samples_per_second": 33.095, | |
| "eval_steps_per_second": 4.302, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6045952439308167, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.607201099395752, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.073, | |
| "eval_steps_per_second": 4.299, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.598048985004425, | |
| "eval_runtime": 3.0235, | |
| "eval_samples_per_second": 33.075, | |
| "eval_steps_per_second": 4.3, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6038631796836853, | |
| "eval_runtime": 3.0246, | |
| "eval_samples_per_second": 33.062, | |
| "eval_steps_per_second": 4.298, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6025145053863525, | |
| "eval_runtime": 3.0238, | |
| "eval_samples_per_second": 33.071, | |
| "eval_steps_per_second": 4.299, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6062153577804565, | |
| "eval_runtime": 3.0238, | |
| "eval_samples_per_second": 33.071, | |
| "eval_steps_per_second": 4.299, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6056156754493713, | |
| "eval_runtime": 3.0237, | |
| "eval_samples_per_second": 33.072, | |
| "eval_steps_per_second": 4.299, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6091020703315735, | |
| "eval_runtime": 3.0256, | |
| "eval_samples_per_second": 33.051, | |
| "eval_steps_per_second": 4.297, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.603661060333252, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.073, | |
| "eval_steps_per_second": 4.3, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6037007570266724, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6175075769424438, | |
| "eval_runtime": 3.0227, | |
| "eval_samples_per_second": 33.083, | |
| "eval_steps_per_second": 4.301, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6089429259300232, | |
| "eval_runtime": 3.0249, | |
| "eval_samples_per_second": 33.058, | |
| "eval_steps_per_second": 4.298, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6076489090919495, | |
| "eval_runtime": 3.0243, | |
| "eval_samples_per_second": 33.065, | |
| "eval_steps_per_second": 4.299, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6072561740875244, | |
| "eval_runtime": 3.0248, | |
| "eval_samples_per_second": 33.06, | |
| "eval_steps_per_second": 4.298, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6059485673904419, | |
| "eval_runtime": 3.0219, | |
| "eval_samples_per_second": 33.091, | |
| "eval_steps_per_second": 4.302, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6108887195587158, | |
| "eval_runtime": 3.0227, | |
| "eval_samples_per_second": 33.083, | |
| "eval_steps_per_second": 4.301, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6089987754821777, | |
| "eval_runtime": 3.0258, | |
| "eval_samples_per_second": 33.049, | |
| "eval_steps_per_second": 4.296, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0005, | |
| "loss": 0.6362, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.608008086681366, | |
| "eval_runtime": 3.0289, | |
| "eval_samples_per_second": 33.016, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5993980169296265, | |
| "eval_runtime": 3.0219, | |
| "eval_samples_per_second": 33.092, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6034057140350342, | |
| "eval_runtime": 3.0231, | |
| "eval_samples_per_second": 33.079, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.6113384366035461, | |
| "eval_runtime": 3.0245, | |
| "eval_samples_per_second": 33.063, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6130579113960266, | |
| "eval_runtime": 3.0267, | |
| "eval_samples_per_second": 33.04, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6150393486022949, | |
| "eval_runtime": 3.0244, | |
| "eval_samples_per_second": 33.065, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6114516258239746, | |
| "eval_runtime": 3.0243, | |
| "eval_samples_per_second": 33.066, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6054720878601074, | |
| "eval_runtime": 3.0238, | |
| "eval_samples_per_second": 33.071, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6033183336257935, | |
| "eval_runtime": 3.0267, | |
| "eval_samples_per_second": 33.039, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6046847701072693, | |
| "eval_runtime": 3.0229, | |
| "eval_samples_per_second": 33.081, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6036849617958069, | |
| "eval_runtime": 3.0227, | |
| "eval_samples_per_second": 33.083, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6010092496871948, | |
| "eval_runtime": 3.0225, | |
| "eval_samples_per_second": 33.086, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5987973213195801, | |
| "eval_runtime": 3.0233, | |
| "eval_samples_per_second": 33.077, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5991275906562805, | |
| "eval_runtime": 3.0229, | |
| "eval_samples_per_second": 33.081, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6018882989883423, | |
| "eval_runtime": 3.0234, | |
| "eval_samples_per_second": 33.076, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6002119183540344, | |
| "eval_runtime": 3.0225, | |
| "eval_samples_per_second": 33.085, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6006402373313904, | |
| "eval_runtime": 3.0226, | |
| "eval_samples_per_second": 33.084, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5991740226745605, | |
| "eval_runtime": 3.0248, | |
| "eval_samples_per_second": 33.06, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5991753935813904, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5992391109466553, | |
| "eval_runtime": 3.0225, | |
| "eval_samples_per_second": 33.085, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.00025, | |
| "loss": 0.6341, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6026336550712585, | |
| "eval_runtime": 3.0215, | |
| "eval_samples_per_second": 33.096, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6022050976753235, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6025804281234741, | |
| "eval_runtime": 3.0231, | |
| "eval_samples_per_second": 33.078, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6036398410797119, | |
| "eval_runtime": 3.0233, | |
| "eval_samples_per_second": 33.077, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6039140224456787, | |
| "eval_runtime": 3.0225, | |
| "eval_samples_per_second": 33.085, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6040653586387634, | |
| "eval_runtime": 3.0224, | |
| "eval_samples_per_second": 33.086, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6034075021743774, | |
| "eval_runtime": 3.0223, | |
| "eval_samples_per_second": 33.088, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6048703789710999, | |
| "eval_runtime": 3.0235, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6027147769927979, | |
| "eval_runtime": 3.0195, | |
| "eval_samples_per_second": 33.118, | |
| "eval_steps_per_second": 4.305, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6057494878768921, | |
| "eval_runtime": 3.024, | |
| "eval_samples_per_second": 33.069, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6056165099143982, | |
| "eval_runtime": 3.021, | |
| "eval_samples_per_second": 33.102, | |
| "eval_steps_per_second": 4.303, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6048298478126526, | |
| "eval_runtime": 3.0253, | |
| "eval_samples_per_second": 33.054, | |
| "eval_steps_per_second": 4.297, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6018526554107666, | |
| "eval_runtime": 3.0215, | |
| "eval_samples_per_second": 33.096, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6020650267601013, | |
| "eval_runtime": 3.0323, | |
| "eval_samples_per_second": 32.978, | |
| "eval_steps_per_second": 4.287, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6018134951591492, | |
| "eval_runtime": 3.0272, | |
| "eval_samples_per_second": 33.034, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6026907563209534, | |
| "eval_runtime": 3.021, | |
| "eval_samples_per_second": 33.102, | |
| "eval_steps_per_second": 4.303, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6025264263153076, | |
| "eval_runtime": 3.0229, | |
| "eval_samples_per_second": 33.081, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6021308898925781, | |
| "eval_runtime": 3.0228, | |
| "eval_samples_per_second": 33.082, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6022512316703796, | |
| "eval_runtime": 3.0219, | |
| "eval_samples_per_second": 33.092, | |
| "eval_steps_per_second": 4.302, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6024240255355835, | |
| "eval_runtime": 3.0204, | |
| "eval_samples_per_second": 33.109, | |
| "eval_steps_per_second": 4.304, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.626, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6023900508880615, | |
| "eval_runtime": 3.0377, | |
| "eval_samples_per_second": 32.92, | |
| "eval_steps_per_second": 4.28, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.6361599731445312, | |
| "train_runtime": 1616.4107, | |
| "train_samples_per_second": 19.797, | |
| "train_steps_per_second": 1.237 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |