| { | |
| "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": 0.34, | |
| "eval_loss": 0.9872172474861145, | |
| "eval_runtime": 2.9066, | |
| "eval_samples_per_second": 34.405, | |
| "eval_steps_per_second": 4.473, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.34, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.854668378829956, | |
| "eval_runtime": 2.922, | |
| "eval_samples_per_second": 34.224, | |
| "eval_steps_per_second": 4.449, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.59, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6061574816703796, | |
| "eval_runtime": 2.9748, | |
| "eval_samples_per_second": 33.615, | |
| "eval_steps_per_second": 4.37, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.609697163105011, | |
| "eval_runtime": 3.0021, | |
| "eval_samples_per_second": 33.31, | |
| "eval_steps_per_second": 4.33, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6063556671142578, | |
| "eval_runtime": 3.0192, | |
| "eval_samples_per_second": 33.122, | |
| "eval_steps_per_second": 4.306, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5974360704421997, | |
| "eval_runtime": 3.0236, | |
| "eval_samples_per_second": 33.073, | |
| "eval_steps_per_second": 4.3, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5722998976707458, | |
| "eval_runtime": 3.0321, | |
| "eval_samples_per_second": 32.98, | |
| "eval_steps_per_second": 4.287, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6179049015045166, | |
| "eval_runtime": 3.0319, | |
| "eval_samples_per_second": 32.983, | |
| "eval_steps_per_second": 4.288, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5842037796974182, | |
| "eval_runtime": 3.0474, | |
| "eval_samples_per_second": 32.815, | |
| "eval_steps_per_second": 4.266, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6117396354675293, | |
| "eval_runtime": 3.0455, | |
| "eval_samples_per_second": 32.835, | |
| "eval_steps_per_second": 4.269, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5443994402885437, | |
| "eval_runtime": 3.0418, | |
| "eval_samples_per_second": 32.875, | |
| "eval_steps_per_second": 4.274, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7898240089416504, | |
| "eval_runtime": 3.0445, | |
| "eval_samples_per_second": 32.846, | |
| "eval_steps_per_second": 4.27, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6850593686103821, | |
| "eval_runtime": 3.0447, | |
| "eval_samples_per_second": 32.844, | |
| "eval_steps_per_second": 4.27, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 9, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7715756893157959, | |
| "eval_runtime": 3.0444, | |
| "eval_samples_per_second": 32.847, | |
| "eval_steps_per_second": 4.27, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6750221252441406, | |
| "eval_runtime": 3.0491, | |
| "eval_samples_per_second": 32.796, | |
| "eval_steps_per_second": 4.264, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7645028829574585, | |
| "eval_runtime": 3.051, | |
| "eval_samples_per_second": 32.776, | |
| "eval_steps_per_second": 4.261, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.7337567806243896, | |
| "eval_runtime": 3.0492, | |
| "eval_samples_per_second": 32.795, | |
| "eval_steps_per_second": 4.263, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.8156324625015259, | |
| "eval_runtime": 3.0486, | |
| "eval_samples_per_second": 32.801, | |
| "eval_steps_per_second": 4.264, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7524186968803406, | |
| "eval_runtime": 3.0469, | |
| "eval_samples_per_second": 32.821, | |
| "eval_steps_per_second": 4.267, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.7431, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.8516017198562622, | |
| "eval_runtime": 3.0478, | |
| "eval_samples_per_second": 32.811, | |
| "eval_steps_per_second": 4.265, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.8224111199378967, | |
| "eval_runtime": 3.0473, | |
| "eval_samples_per_second": 32.816, | |
| "eval_steps_per_second": 4.266, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 1.0606846809387207, | |
| "eval_runtime": 3.0487, | |
| "eval_samples_per_second": 32.801, | |
| "eval_steps_per_second": 4.264, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.8977242112159729, | |
| "eval_runtime": 3.0478, | |
| "eval_samples_per_second": 32.81, | |
| "eval_steps_per_second": 4.265, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.7860437631607056, | |
| "eval_runtime": 3.0465, | |
| "eval_samples_per_second": 32.825, | |
| "eval_steps_per_second": 4.267, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.7284977436065674, | |
| "eval_runtime": 3.0473, | |
| "eval_samples_per_second": 32.816, | |
| "eval_steps_per_second": 4.266, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.7096824049949646, | |
| "eval_runtime": 3.0452, | |
| "eval_samples_per_second": 32.839, | |
| "eval_steps_per_second": 4.269, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.7291831374168396, | |
| "eval_runtime": 3.0439, | |
| "eval_samples_per_second": 32.852, | |
| "eval_steps_per_second": 4.271, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7131230235099792, | |
| "eval_runtime": 3.0452, | |
| "eval_samples_per_second": 32.839, | |
| "eval_steps_per_second": 4.269, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.8038756847381592, | |
| "eval_runtime": 3.0474, | |
| "eval_samples_per_second": 32.815, | |
| "eval_steps_per_second": 4.266, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.798814058303833, | |
| "eval_runtime": 3.0483, | |
| "eval_samples_per_second": 32.806, | |
| "eval_steps_per_second": 4.265, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.7809301614761353, | |
| "eval_runtime": 3.0445, | |
| "eval_samples_per_second": 32.846, | |
| "eval_steps_per_second": 4.27, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.7544003129005432, | |
| "eval_runtime": 3.0471, | |
| "eval_samples_per_second": 32.818, | |
| "eval_steps_per_second": 4.266, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.7492163181304932, | |
| "eval_runtime": 3.0436, | |
| "eval_samples_per_second": 32.856, | |
| "eval_steps_per_second": 4.271, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.8205734491348267, | |
| "eval_runtime": 3.0415, | |
| "eval_samples_per_second": 32.878, | |
| "eval_steps_per_second": 4.274, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.640932559967041, | |
| "eval_runtime": 3.0376, | |
| "eval_samples_per_second": 32.921, | |
| "eval_steps_per_second": 4.28, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.7143794894218445, | |
| "eval_runtime": 3.0421, | |
| "eval_samples_per_second": 32.872, | |
| "eval_steps_per_second": 4.273, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.74139004945755, | |
| "eval_runtime": 3.037, | |
| "eval_samples_per_second": 32.927, | |
| "eval_steps_per_second": 4.281, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7422806620597839, | |
| "eval_runtime": 3.0339, | |
| "eval_samples_per_second": 32.961, | |
| "eval_steps_per_second": 4.285, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7765854001045227, | |
| "eval_runtime": 3.0422, | |
| "eval_samples_per_second": 32.871, | |
| "eval_steps_per_second": 4.273, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.3363, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7181565761566162, | |
| "eval_runtime": 3.0356, | |
| "eval_samples_per_second": 32.942, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7375120520591736, | |
| "eval_runtime": 3.0388, | |
| "eval_samples_per_second": 32.908, | |
| "eval_steps_per_second": 4.278, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7235824465751648, | |
| "eval_runtime": 3.033, | |
| "eval_samples_per_second": 32.971, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.7218219041824341, | |
| "eval_runtime": 3.039, | |
| "eval_samples_per_second": 32.906, | |
| "eval_steps_per_second": 4.278, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.732373833656311, | |
| "eval_runtime": 3.0415, | |
| "eval_samples_per_second": 32.879, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7290899157524109, | |
| "eval_runtime": 3.0411, | |
| "eval_samples_per_second": 32.883, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6803033947944641, | |
| "eval_runtime": 3.0409, | |
| "eval_samples_per_second": 32.885, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6636987924575806, | |
| "eval_runtime": 3.0397, | |
| "eval_samples_per_second": 32.898, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7063630819320679, | |
| "eval_runtime": 3.0394, | |
| "eval_samples_per_second": 32.901, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6534314751625061, | |
| "eval_runtime": 3.0351, | |
| "eval_samples_per_second": 32.948, | |
| "eval_steps_per_second": 4.283, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7229902148246765, | |
| "eval_runtime": 3.0356, | |
| "eval_samples_per_second": 32.942, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.733811616897583, | |
| "eval_runtime": 3.0331, | |
| "eval_samples_per_second": 32.97, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6494699716567993, | |
| "eval_runtime": 3.036, | |
| "eval_samples_per_second": 32.938, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.653989851474762, | |
| "eval_runtime": 3.0366, | |
| "eval_samples_per_second": 32.932, | |
| "eval_steps_per_second": 4.281, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.6994405388832092, | |
| "eval_runtime": 3.0342, | |
| "eval_samples_per_second": 32.958, | |
| "eval_steps_per_second": 4.285, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.7039541006088257, | |
| "eval_runtime": 3.0349, | |
| "eval_samples_per_second": 32.95, | |
| "eval_steps_per_second": 4.284, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.6774616837501526, | |
| "eval_runtime": 3.0335, | |
| "eval_samples_per_second": 32.965, | |
| "eval_steps_per_second": 4.285, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6425368785858154, | |
| "eval_runtime": 3.0318, | |
| "eval_samples_per_second": 32.984, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6424228549003601, | |
| "eval_runtime": 3.038, | |
| "eval_samples_per_second": 32.917, | |
| "eval_steps_per_second": 4.279, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6781678795814514, | |
| "eval_runtime": 3.0394, | |
| "eval_samples_per_second": 32.901, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.2375, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6769825220108032, | |
| "eval_runtime": 3.0319, | |
| "eval_samples_per_second": 32.983, | |
| "eval_steps_per_second": 4.288, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7028858661651611, | |
| "eval_runtime": 3.041, | |
| "eval_samples_per_second": 32.884, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6823871731758118, | |
| "eval_runtime": 3.036, | |
| "eval_samples_per_second": 32.938, | |
| "eval_steps_per_second": 4.282, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6846780180931091, | |
| "eval_runtime": 3.0396, | |
| "eval_samples_per_second": 32.899, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6766635775566101, | |
| "eval_runtime": 3.0413, | |
| "eval_samples_per_second": 32.88, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6361604928970337, | |
| "eval_runtime": 3.033, | |
| "eval_samples_per_second": 32.971, | |
| "eval_steps_per_second": 4.286, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6292469501495361, | |
| "eval_runtime": 3.0278, | |
| "eval_samples_per_second": 33.027, | |
| "eval_steps_per_second": 4.294, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6470186710357666, | |
| "eval_runtime": 3.0287, | |
| "eval_samples_per_second": 33.018, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6660796403884888, | |
| "eval_runtime": 3.0283, | |
| "eval_samples_per_second": 33.022, | |
| "eval_steps_per_second": 4.293, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6304946541786194, | |
| "eval_runtime": 3.0292, | |
| "eval_samples_per_second": 33.012, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6491644978523254, | |
| "eval_runtime": 3.0286, | |
| "eval_samples_per_second": 33.019, | |
| "eval_steps_per_second": 4.292, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.652509331703186, | |
| "eval_runtime": 3.0249, | |
| "eval_samples_per_second": 33.058, | |
| "eval_steps_per_second": 4.298, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6339465379714966, | |
| "eval_runtime": 3.0269, | |
| "eval_samples_per_second": 33.037, | |
| "eval_steps_per_second": 4.295, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6621412634849548, | |
| "eval_runtime": 3.0262, | |
| "eval_samples_per_second": 33.045, | |
| "eval_steps_per_second": 4.296, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6562237739562988, | |
| "eval_runtime": 3.0239, | |
| "eval_samples_per_second": 33.07, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6396531462669373, | |
| "eval_runtime": 3.0227, | |
| "eval_samples_per_second": 33.083, | |
| "eval_steps_per_second": 4.301, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6496105194091797, | |
| "eval_runtime": 3.0233, | |
| "eval_samples_per_second": 33.077, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6401783227920532, | |
| "eval_runtime": 3.0238, | |
| "eval_samples_per_second": 33.072, | |
| "eval_steps_per_second": 4.299, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6382293105125427, | |
| "eval_runtime": 3.0235, | |
| "eval_samples_per_second": 33.074, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6407208442687988, | |
| "eval_runtime": 3.0231, | |
| "eval_samples_per_second": 33.079, | |
| "eval_steps_per_second": 4.3, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2102, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6390834450721741, | |
| "eval_runtime": 3.0233, | |
| "eval_samples_per_second": 33.076, | |
| "eval_steps_per_second": 4.3, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 14, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.38176900482177734, | |
| "train_runtime": 1625.2113, | |
| "train_samples_per_second": 19.69, | |
| "train_steps_per_second": 1.231 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |