| { | |
| "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": 0.4, | |
| "eval_loss": 0.6447609663009644, | |
| "eval_runtime": 2.9283, | |
| "eval_samples_per_second": 34.15, | |
| "eval_steps_per_second": 4.44, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.4, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7949573993682861, | |
| "eval_runtime": 2.9849, | |
| "eval_samples_per_second": 33.502, | |
| "eval_steps_per_second": 4.355, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.6180824041366577, | |
| "eval_runtime": 3.0424, | |
| "eval_samples_per_second": 32.869, | |
| "eval_steps_per_second": 4.273, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.5600907206535339, | |
| "eval_runtime": 3.0801, | |
| "eval_samples_per_second": 32.467, | |
| "eval_steps_per_second": 4.221, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.5816247463226318, | |
| "eval_runtime": 3.1022, | |
| "eval_samples_per_second": 32.235, | |
| "eval_steps_per_second": 4.191, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.43, | |
| "eval_loss": 0.5956963300704956, | |
| "eval_runtime": 3.123, | |
| "eval_samples_per_second": 32.02, | |
| "eval_steps_per_second": 4.163, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.5331138968467712, | |
| "eval_runtime": 3.1336, | |
| "eval_samples_per_second": 31.912, | |
| "eval_steps_per_second": 4.149, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.550652027130127, | |
| "eval_runtime": 3.1424, | |
| "eval_samples_per_second": 31.822, | |
| "eval_steps_per_second": 4.137, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5438417196273804, | |
| "eval_runtime": 3.1439, | |
| "eval_samples_per_second": 31.808, | |
| "eval_steps_per_second": 4.135, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5454716682434082, | |
| "eval_runtime": 3.1529, | |
| "eval_samples_per_second": 31.717, | |
| "eval_steps_per_second": 4.123, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5140843391418457, | |
| "eval_runtime": 3.1585, | |
| "eval_samples_per_second": 31.661, | |
| "eval_steps_per_second": 4.116, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5018832683563232, | |
| "eval_runtime": 3.1501, | |
| "eval_samples_per_second": 31.745, | |
| "eval_steps_per_second": 4.127, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6824415326118469, | |
| "eval_runtime": 3.1461, | |
| "eval_samples_per_second": 31.786, | |
| "eval_steps_per_second": 4.132, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 11, | |
| "best_eval_accuracy": 0.71, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5734787583351135, | |
| "eval_runtime": 3.1445, | |
| "eval_samples_per_second": 31.802, | |
| "eval_steps_per_second": 4.134, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5578454732894897, | |
| "eval_runtime": 3.1437, | |
| "eval_samples_per_second": 31.809, | |
| "eval_steps_per_second": 4.135, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.560732901096344, | |
| "eval_runtime": 3.1447, | |
| "eval_samples_per_second": 31.799, | |
| "eval_steps_per_second": 4.134, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.5974356532096863, | |
| "eval_runtime": 3.1435, | |
| "eval_samples_per_second": 31.812, | |
| "eval_steps_per_second": 4.136, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.8101874589920044, | |
| "eval_runtime": 3.141, | |
| "eval_samples_per_second": 31.837, | |
| "eval_steps_per_second": 4.139, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.6756836175918579, | |
| "eval_runtime": 3.14, | |
| "eval_samples_per_second": 31.847, | |
| "eval_steps_per_second": 4.14, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 13, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.7598, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.74, | |
| "eval_loss": 0.5266308188438416, | |
| "eval_runtime": 3.1409, | |
| "eval_samples_per_second": 31.838, | |
| "eval_steps_per_second": 4.139, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6270546913146973, | |
| "eval_runtime": 3.1418, | |
| "eval_samples_per_second": 31.829, | |
| "eval_steps_per_second": 4.138, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6341220736503601, | |
| "eval_runtime": 3.1406, | |
| "eval_samples_per_second": 31.842, | |
| "eval_steps_per_second": 4.139, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6874021887779236, | |
| "eval_runtime": 3.1434, | |
| "eval_samples_per_second": 31.812, | |
| "eval_steps_per_second": 4.136, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.5264467000961304, | |
| "eval_runtime": 3.1446, | |
| "eval_samples_per_second": 31.801, | |
| "eval_steps_per_second": 4.134, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.5148030519485474, | |
| "eval_runtime": 3.1438, | |
| "eval_samples_per_second": 31.809, | |
| "eval_steps_per_second": 4.135, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.74, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.77, | |
| "eval_loss": 0.5760267972946167, | |
| "eval_runtime": 3.1417, | |
| "eval_samples_per_second": 31.83, | |
| "eval_steps_per_second": 4.138, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6580688953399658, | |
| "eval_runtime": 3.1419, | |
| "eval_samples_per_second": 31.828, | |
| "eval_steps_per_second": 4.138, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.6479081511497498, | |
| "eval_runtime": 3.145, | |
| "eval_samples_per_second": 31.797, | |
| "eval_steps_per_second": 4.134, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.696037232875824, | |
| "eval_runtime": 3.1444, | |
| "eval_samples_per_second": 31.803, | |
| "eval_steps_per_second": 4.134, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.6919419765472412, | |
| "eval_runtime": 3.1427, | |
| "eval_samples_per_second": 31.82, | |
| "eval_steps_per_second": 4.137, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.642117977142334, | |
| "eval_runtime": 3.1402, | |
| "eval_samples_per_second": 31.845, | |
| "eval_steps_per_second": 4.14, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5680819153785706, | |
| "eval_runtime": 3.1395, | |
| "eval_samples_per_second": 31.852, | |
| "eval_steps_per_second": 4.141, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5631085634231567, | |
| "eval_runtime": 3.1383, | |
| "eval_samples_per_second": 31.864, | |
| "eval_steps_per_second": 4.142, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5676264762878418, | |
| "eval_runtime": 3.1421, | |
| "eval_samples_per_second": 31.826, | |
| "eval_steps_per_second": 4.137, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5388873815536499, | |
| "eval_runtime": 3.1413, | |
| "eval_samples_per_second": 31.834, | |
| "eval_steps_per_second": 4.138, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6266981363296509, | |
| "eval_runtime": 3.1411, | |
| "eval_samples_per_second": 31.836, | |
| "eval_steps_per_second": 4.139, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6107380390167236, | |
| "eval_runtime": 3.1415, | |
| "eval_samples_per_second": 31.832, | |
| "eval_steps_per_second": 4.138, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.53592449426651, | |
| "eval_runtime": 3.1409, | |
| "eval_samples_per_second": 31.838, | |
| "eval_steps_per_second": 4.139, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5741092562675476, | |
| "eval_runtime": 3.1433, | |
| "eval_samples_per_second": 31.814, | |
| "eval_steps_per_second": 4.136, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.4266, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5928249955177307, | |
| "eval_runtime": 3.1459, | |
| "eval_samples_per_second": 31.788, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5306583046913147, | |
| "eval_runtime": 3.1398, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5909176468849182, | |
| "eval_runtime": 3.1393, | |
| "eval_samples_per_second": 31.854, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5732528567314148, | |
| "eval_runtime": 3.1401, | |
| "eval_samples_per_second": 31.846, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5560975074768066, | |
| "eval_runtime": 3.1413, | |
| "eval_samples_per_second": 31.834, | |
| "eval_steps_per_second": 4.138, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5600245594978333, | |
| "eval_runtime": 3.1394, | |
| "eval_samples_per_second": 31.854, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5228108763694763, | |
| "eval_runtime": 3.1374, | |
| "eval_samples_per_second": 31.874, | |
| "eval_steps_per_second": 4.144, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5382905602455139, | |
| "eval_runtime": 3.1397, | |
| "eval_samples_per_second": 31.85, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5643246173858643, | |
| "eval_runtime": 3.1398, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5492831468582153, | |
| "eval_runtime": 3.1396, | |
| "eval_samples_per_second": 31.851, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.5576255321502686, | |
| "eval_runtime": 3.1404, | |
| "eval_samples_per_second": 31.843, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5543303489685059, | |
| "eval_runtime": 3.14, | |
| "eval_samples_per_second": 31.847, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.56154465675354, | |
| "eval_runtime": 3.1398, | |
| "eval_samples_per_second": 31.849, | |
| "eval_steps_per_second": 4.14, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5357893109321594, | |
| "eval_runtime": 3.1385, | |
| "eval_samples_per_second": 31.862, | |
| "eval_steps_per_second": 4.142, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5404526591300964, | |
| "eval_runtime": 3.1375, | |
| "eval_samples_per_second": 31.873, | |
| "eval_steps_per_second": 4.143, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.53267902135849, | |
| "eval_runtime": 3.138, | |
| "eval_samples_per_second": 31.868, | |
| "eval_steps_per_second": 4.143, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5644922852516174, | |
| "eval_runtime": 3.1383, | |
| "eval_samples_per_second": 31.864, | |
| "eval_steps_per_second": 4.142, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5240322947502136, | |
| "eval_runtime": 3.1384, | |
| "eval_samples_per_second": 31.863, | |
| "eval_steps_per_second": 4.142, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5401843786239624, | |
| "eval_runtime": 3.1408, | |
| "eval_samples_per_second": 31.839, | |
| "eval_steps_per_second": 4.139, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.549487829208374, | |
| "eval_runtime": 3.1454, | |
| "eval_samples_per_second": 31.793, | |
| "eval_steps_per_second": 4.133, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.3249, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5624470710754395, | |
| "eval_runtime": 3.142, | |
| "eval_samples_per_second": 31.827, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5512508749961853, | |
| "eval_runtime": 3.1363, | |
| "eval_samples_per_second": 31.885, | |
| "eval_steps_per_second": 4.145, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5536577701568604, | |
| "eval_runtime": 3.1364, | |
| "eval_samples_per_second": 31.884, | |
| "eval_steps_per_second": 4.145, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5443692207336426, | |
| "eval_runtime": 3.1379, | |
| "eval_samples_per_second": 31.868, | |
| "eval_steps_per_second": 4.143, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5552690029144287, | |
| "eval_runtime": 3.1344, | |
| "eval_samples_per_second": 31.904, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.522087812423706, | |
| "eval_runtime": 3.1353, | |
| "eval_samples_per_second": 31.895, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5135818123817444, | |
| "eval_runtime": 3.1354, | |
| "eval_samples_per_second": 31.894, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5231194496154785, | |
| "eval_runtime": 3.1365, | |
| "eval_samples_per_second": 31.883, | |
| "eval_steps_per_second": 4.145, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5305217504501343, | |
| "eval_runtime": 3.1343, | |
| "eval_samples_per_second": 31.905, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5278050303459167, | |
| "eval_runtime": 3.1328, | |
| "eval_samples_per_second": 31.92, | |
| "eval_steps_per_second": 4.15, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5439716577529907, | |
| "eval_runtime": 3.1334, | |
| "eval_samples_per_second": 31.915, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5410908460617065, | |
| "eval_runtime": 3.1331, | |
| "eval_samples_per_second": 31.918, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.5346353650093079, | |
| "eval_runtime": 3.1316, | |
| "eval_samples_per_second": 31.933, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5240665078163147, | |
| "eval_runtime": 3.1322, | |
| "eval_samples_per_second": 31.926, | |
| "eval_steps_per_second": 4.15, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5424520969390869, | |
| "eval_runtime": 3.1337, | |
| "eval_samples_per_second": 31.911, | |
| "eval_steps_per_second": 4.148, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.521338701248169, | |
| "eval_runtime": 3.1447, | |
| "eval_samples_per_second": 31.799, | |
| "eval_steps_per_second": 4.134, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.540505588054657, | |
| "eval_runtime": 3.1395, | |
| "eval_samples_per_second": 31.853, | |
| "eval_steps_per_second": 4.141, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.525080144405365, | |
| "eval_runtime": 3.1248, | |
| "eval_samples_per_second": 32.002, | |
| "eval_steps_per_second": 4.16, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5299803614616394, | |
| "eval_runtime": 3.122, | |
| "eval_samples_per_second": 32.031, | |
| "eval_steps_per_second": 4.164, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5285055637359619, | |
| "eval_runtime": 3.129, | |
| "eval_samples_per_second": 31.959, | |
| "eval_steps_per_second": 4.155, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2946, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5294349789619446, | |
| "eval_runtime": 3.1201, | |
| "eval_samples_per_second": 32.05, | |
| "eval_steps_per_second": 4.166, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 25, | |
| "best_eval_accuracy": 0.77, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.4514722671508789, | |
| "train_runtime": 1665.4633, | |
| "train_samples_per_second": 19.214, | |
| "train_steps_per_second": 1.201 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |