| { | |
| "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": 0.36, | |
| "eval_loss": 0.6900678277015686, | |
| "eval_runtime": 2.9568, | |
| "eval_samples_per_second": 33.82, | |
| "eval_steps_per_second": 4.397, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.36, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.7172253131866455, | |
| "eval_runtime": 3.0289, | |
| "eval_samples_per_second": 33.016, | |
| "eval_steps_per_second": 4.292, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.56, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.6264107823371887, | |
| "eval_runtime": 3.0763, | |
| "eval_samples_per_second": 32.507, | |
| "eval_steps_per_second": 4.226, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.58, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5863716006278992, | |
| "eval_runtime": 3.1066, | |
| "eval_samples_per_second": 32.189, | |
| "eval_steps_per_second": 4.185, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.5716881155967712, | |
| "eval_runtime": 3.1333, | |
| "eval_samples_per_second": 31.915, | |
| "eval_steps_per_second": 4.149, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.6748393774032593, | |
| "eval_runtime": 3.1391, | |
| "eval_samples_per_second": 31.856, | |
| "eval_steps_per_second": 4.141, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 3, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5271764993667603, | |
| "eval_runtime": 3.1483, | |
| "eval_samples_per_second": 31.763, | |
| "eval_steps_per_second": 4.129, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.565096914768219, | |
| "eval_runtime": 3.1481, | |
| "eval_samples_per_second": 31.765, | |
| "eval_steps_per_second": 4.129, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.578468918800354, | |
| "eval_runtime": 3.1568, | |
| "eval_samples_per_second": 31.677, | |
| "eval_steps_per_second": 4.118, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5772669911384583, | |
| "eval_runtime": 3.1558, | |
| "eval_samples_per_second": 31.688, | |
| "eval_steps_per_second": 4.119, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5286800265312195, | |
| "eval_runtime": 3.1451, | |
| "eval_samples_per_second": 31.795, | |
| "eval_steps_per_second": 4.133, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5612473487854004, | |
| "eval_runtime": 3.1564, | |
| "eval_samples_per_second": 31.682, | |
| "eval_steps_per_second": 4.119, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5733940005302429, | |
| "eval_runtime": 3.1466, | |
| "eval_samples_per_second": 31.78, | |
| "eval_steps_per_second": 4.131, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5195974111557007, | |
| "eval_runtime": 3.1431, | |
| "eval_samples_per_second": 31.815, | |
| "eval_steps_per_second": 4.136, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5490823984146118, | |
| "eval_runtime": 3.1425, | |
| "eval_samples_per_second": 31.821, | |
| "eval_steps_per_second": 4.137, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5136759281158447, | |
| "eval_runtime": 3.1402, | |
| "eval_samples_per_second": 31.845, | |
| "eval_steps_per_second": 4.14, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.533294141292572, | |
| "eval_runtime": 3.1412, | |
| "eval_samples_per_second": 31.835, | |
| "eval_steps_per_second": 4.139, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5518142580986023, | |
| "eval_runtime": 3.1392, | |
| "eval_samples_per_second": 31.855, | |
| "eval_steps_per_second": 4.141, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5221859216690063, | |
| "eval_runtime": 3.1382, | |
| "eval_samples_per_second": 31.865, | |
| "eval_steps_per_second": 4.142, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.7077, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4975990355014801, | |
| "eval_runtime": 3.1365, | |
| "eval_samples_per_second": 31.883, | |
| "eval_steps_per_second": 4.145, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.4995178282260895, | |
| "eval_runtime": 3.1409, | |
| "eval_samples_per_second": 31.838, | |
| "eval_steps_per_second": 4.139, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5836873650550842, | |
| "eval_runtime": 3.1414, | |
| "eval_samples_per_second": 31.833, | |
| "eval_steps_per_second": 4.138, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5800585150718689, | |
| "eval_runtime": 3.1419, | |
| "eval_samples_per_second": 31.828, | |
| "eval_steps_per_second": 4.138, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5376842617988586, | |
| "eval_runtime": 3.1413, | |
| "eval_samples_per_second": 31.834, | |
| "eval_steps_per_second": 4.138, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5508807897567749, | |
| "eval_runtime": 3.1403, | |
| "eval_samples_per_second": 31.844, | |
| "eval_steps_per_second": 4.14, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5863495469093323, | |
| "eval_runtime": 3.1415, | |
| "eval_samples_per_second": 31.831, | |
| "eval_steps_per_second": 4.138, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5980372428894043, | |
| "eval_runtime": 3.144, | |
| "eval_samples_per_second": 31.806, | |
| "eval_steps_per_second": 4.135, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6481872797012329, | |
| "eval_runtime": 3.1402, | |
| "eval_samples_per_second": 31.845, | |
| "eval_steps_per_second": 4.14, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5850781798362732, | |
| "eval_runtime": 3.144, | |
| "eval_samples_per_second": 31.807, | |
| "eval_steps_per_second": 4.135, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6651497483253479, | |
| "eval_runtime": 3.1422, | |
| "eval_samples_per_second": 31.825, | |
| "eval_steps_per_second": 4.137, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5496576428413391, | |
| "eval_runtime": 3.1415, | |
| "eval_samples_per_second": 31.832, | |
| "eval_steps_per_second": 4.138, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 30, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5907086133956909, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.787, | |
| "eval_steps_per_second": 4.132, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5805070400238037, | |
| "eval_runtime": 3.1454, | |
| "eval_samples_per_second": 31.792, | |
| "eval_steps_per_second": 4.133, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5844395160675049, | |
| "eval_runtime": 3.1444, | |
| "eval_samples_per_second": 31.803, | |
| "eval_steps_per_second": 4.134, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5749659538269043, | |
| "eval_runtime": 3.1443, | |
| "eval_samples_per_second": 31.803, | |
| "eval_steps_per_second": 4.134, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6175033450126648, | |
| "eval_runtime": 3.1415, | |
| "eval_samples_per_second": 31.832, | |
| "eval_steps_per_second": 4.138, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5753803849220276, | |
| "eval_runtime": 3.1437, | |
| "eval_samples_per_second": 31.809, | |
| "eval_steps_per_second": 4.135, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.575753390789032, | |
| "eval_runtime": 3.1433, | |
| "eval_samples_per_second": 31.814, | |
| "eval_steps_per_second": 4.136, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6012661457061768, | |
| "eval_runtime": 3.1421, | |
| "eval_samples_per_second": 31.826, | |
| "eval_steps_per_second": 4.137, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.4491, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5383523106575012, | |
| "eval_runtime": 3.1442, | |
| "eval_samples_per_second": 31.805, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5930788516998291, | |
| "eval_runtime": 3.1439, | |
| "eval_samples_per_second": 31.807, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6029644012451172, | |
| "eval_runtime": 3.1452, | |
| "eval_samples_per_second": 31.794, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5629833936691284, | |
| "eval_runtime": 3.1473, | |
| "eval_samples_per_second": 31.773, | |
| "eval_steps_per_second": 4.13, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5599313974380493, | |
| "eval_runtime": 3.146, | |
| "eval_samples_per_second": 31.787, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5798678994178772, | |
| "eval_runtime": 3.1489, | |
| "eval_samples_per_second": 31.757, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5544764995574951, | |
| "eval_runtime": 3.1471, | |
| "eval_samples_per_second": 31.775, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5642539262771606, | |
| "eval_runtime": 3.1464, | |
| "eval_samples_per_second": 31.782, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.584467887878418, | |
| "eval_runtime": 3.149, | |
| "eval_samples_per_second": 31.757, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5780852437019348, | |
| "eval_runtime": 3.1469, | |
| "eval_samples_per_second": 31.777, | |
| "eval_steps_per_second": 4.131, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5623089671134949, | |
| "eval_runtime": 3.1534, | |
| "eval_samples_per_second": 31.712, | |
| "eval_steps_per_second": 4.123, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5528396368026733, | |
| "eval_runtime": 3.1495, | |
| "eval_samples_per_second": 31.751, | |
| "eval_steps_per_second": 4.128, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.544218122959137, | |
| "eval_runtime": 3.1486, | |
| "eval_samples_per_second": 31.76, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5498414039611816, | |
| "eval_runtime": 3.1488, | |
| "eval_samples_per_second": 31.759, | |
| "eval_steps_per_second": 4.129, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5390793681144714, | |
| "eval_runtime": 3.1542, | |
| "eval_samples_per_second": 31.704, | |
| "eval_steps_per_second": 4.121, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5569570660591125, | |
| "eval_runtime": 3.154, | |
| "eval_samples_per_second": 31.706, | |
| "eval_steps_per_second": 4.122, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5728991031646729, | |
| "eval_runtime": 3.1411, | |
| "eval_samples_per_second": 31.836, | |
| "eval_steps_per_second": 4.139, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5351668000221252, | |
| "eval_runtime": 3.1381, | |
| "eval_samples_per_second": 31.866, | |
| "eval_steps_per_second": 4.143, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5537936687469482, | |
| "eval_runtime": 3.1402, | |
| "eval_samples_per_second": 31.845, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5563415884971619, | |
| "eval_runtime": 3.137, | |
| "eval_samples_per_second": 31.878, | |
| "eval_steps_per_second": 4.144, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.3353, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5703515410423279, | |
| "eval_runtime": 3.1319, | |
| "eval_samples_per_second": 31.93, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5726474523544312, | |
| "eval_runtime": 3.134, | |
| "eval_samples_per_second": 31.908, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5693960785865784, | |
| "eval_runtime": 3.1339, | |
| "eval_samples_per_second": 31.91, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5714474320411682, | |
| "eval_runtime": 3.132, | |
| "eval_samples_per_second": 31.928, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5551468729972839, | |
| "eval_runtime": 3.1362, | |
| "eval_samples_per_second": 31.886, | |
| "eval_steps_per_second": 4.145, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5547963380813599, | |
| "eval_runtime": 3.136, | |
| "eval_samples_per_second": 31.888, | |
| "eval_steps_per_second": 4.145, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5430014133453369, | |
| "eval_runtime": 3.1344, | |
| "eval_samples_per_second": 31.904, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5448939800262451, | |
| "eval_runtime": 3.1352, | |
| "eval_samples_per_second": 31.895, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5461122393608093, | |
| "eval_runtime": 3.134, | |
| "eval_samples_per_second": 31.908, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5440198183059692, | |
| "eval_runtime": 3.1319, | |
| "eval_samples_per_second": 31.93, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5590041875839233, | |
| "eval_runtime": 3.1357, | |
| "eval_samples_per_second": 31.891, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5390868186950684, | |
| "eval_runtime": 3.1403, | |
| "eval_samples_per_second": 31.844, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5516095757484436, | |
| "eval_runtime": 3.14, | |
| "eval_samples_per_second": 31.847, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.547356128692627, | |
| "eval_runtime": 3.1285, | |
| "eval_samples_per_second": 31.964, | |
| "eval_steps_per_second": 4.155, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5476573705673218, | |
| "eval_runtime": 3.1324, | |
| "eval_samples_per_second": 31.924, | |
| "eval_steps_per_second": 4.15, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5372492074966431, | |
| "eval_runtime": 3.1315, | |
| "eval_samples_per_second": 31.933, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5445277690887451, | |
| "eval_runtime": 3.1318, | |
| "eval_samples_per_second": 31.93, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5421162247657776, | |
| "eval_runtime": 3.1315, | |
| "eval_samples_per_second": 31.933, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5376150012016296, | |
| "eval_runtime": 3.1315, | |
| "eval_samples_per_second": 31.934, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5357780456542969, | |
| "eval_runtime": 3.1264, | |
| "eval_samples_per_second": 31.986, | |
| "eval_steps_per_second": 4.158, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.3108, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5369434952735901, | |
| "eval_runtime": 3.1263, | |
| "eval_samples_per_second": 31.986, | |
| "eval_steps_per_second": 4.158, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 31, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.4507393341064453, | |
| "train_runtime": 1667.678, | |
| "train_samples_per_second": 19.188, | |
| "train_steps_per_second": 1.199 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |