| { | |
| "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": 0.66, | |
| "eval_loss": 0.5415894389152527, | |
| "eval_runtime": 2.9244, | |
| "eval_samples_per_second": 34.195, | |
| "eval_steps_per_second": 4.445, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5393697023391724, | |
| "eval_runtime": 2.9776, | |
| "eval_samples_per_second": 33.584, | |
| "eval_steps_per_second": 4.366, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5375608801841736, | |
| "eval_runtime": 3.0286, | |
| "eval_samples_per_second": 33.018, | |
| "eval_steps_per_second": 4.292, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5475878119468689, | |
| "eval_runtime": 3.06, | |
| "eval_samples_per_second": 32.68, | |
| "eval_steps_per_second": 4.248, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5370528101921082, | |
| "eval_runtime": 3.0811, | |
| "eval_samples_per_second": 32.456, | |
| "eval_steps_per_second": 4.219, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5441551208496094, | |
| "eval_runtime": 3.0936, | |
| "eval_samples_per_second": 32.324, | |
| "eval_steps_per_second": 4.202, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5412868857383728, | |
| "eval_runtime": 3.1033, | |
| "eval_samples_per_second": 32.224, | |
| "eval_steps_per_second": 4.189, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5381051898002625, | |
| "eval_runtime": 3.1116, | |
| "eval_samples_per_second": 32.138, | |
| "eval_steps_per_second": 4.178, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5366445183753967, | |
| "eval_runtime": 3.1179, | |
| "eval_samples_per_second": 32.073, | |
| "eval_steps_per_second": 4.17, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5402125120162964, | |
| "eval_runtime": 3.1204, | |
| "eval_samples_per_second": 32.047, | |
| "eval_steps_per_second": 4.166, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5405026078224182, | |
| "eval_runtime": 3.1198, | |
| "eval_samples_per_second": 32.053, | |
| "eval_steps_per_second": 4.167, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5396003127098083, | |
| "eval_runtime": 3.1242, | |
| "eval_samples_per_second": 32.009, | |
| "eval_steps_per_second": 4.161, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5378925204277039, | |
| "eval_runtime": 3.1244, | |
| "eval_samples_per_second": 32.006, | |
| "eval_steps_per_second": 4.161, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5375378727912903, | |
| "eval_runtime": 3.1264, | |
| "eval_samples_per_second": 31.985, | |
| "eval_steps_per_second": 4.158, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5392658114433289, | |
| "eval_runtime": 3.1259, | |
| "eval_samples_per_second": 31.991, | |
| "eval_steps_per_second": 4.159, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5370680689811707, | |
| "eval_runtime": 3.1284, | |
| "eval_samples_per_second": 31.965, | |
| "eval_steps_per_second": 4.155, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5286482572555542, | |
| "eval_runtime": 3.128, | |
| "eval_samples_per_second": 31.969, | |
| "eval_steps_per_second": 4.156, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5312860608100891, | |
| "eval_runtime": 3.1331, | |
| "eval_samples_per_second": 31.917, | |
| "eval_steps_per_second": 4.149, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.5426998734474182, | |
| "eval_runtime": 3.1294, | |
| "eval_samples_per_second": 31.955, | |
| "eval_steps_per_second": 4.154, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.00075, | |
| "loss": 0.616, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5469082593917847, | |
| "eval_runtime": 3.1318, | |
| "eval_samples_per_second": 31.931, | |
| "eval_steps_per_second": 4.151, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5347804427146912, | |
| "eval_runtime": 3.1345, | |
| "eval_samples_per_second": 31.903, | |
| "eval_steps_per_second": 4.147, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5352023243904114, | |
| "eval_runtime": 3.1312, | |
| "eval_samples_per_second": 31.936, | |
| "eval_steps_per_second": 4.152, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5434411764144897, | |
| "eval_runtime": 3.1342, | |
| "eval_samples_per_second": 31.906, | |
| "eval_steps_per_second": 4.148, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.543671190738678, | |
| "eval_runtime": 3.1348, | |
| "eval_samples_per_second": 31.9, | |
| "eval_steps_per_second": 4.147, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5343627333641052, | |
| "eval_runtime": 3.1317, | |
| "eval_samples_per_second": 31.932, | |
| "eval_steps_per_second": 4.151, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5343685746192932, | |
| "eval_runtime": 3.132, | |
| "eval_samples_per_second": 31.929, | |
| "eval_steps_per_second": 4.151, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5318924188613892, | |
| "eval_runtime": 3.1314, | |
| "eval_samples_per_second": 31.934, | |
| "eval_steps_per_second": 4.151, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5328655242919922, | |
| "eval_runtime": 3.1348, | |
| "eval_samples_per_second": 31.9, | |
| "eval_steps_per_second": 4.147, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5312965512275696, | |
| "eval_runtime": 3.1312, | |
| "eval_samples_per_second": 31.937, | |
| "eval_steps_per_second": 4.152, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5321267247200012, | |
| "eval_runtime": 3.1321, | |
| "eval_samples_per_second": 31.927, | |
| "eval_steps_per_second": 4.151, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5341834425926208, | |
| "eval_runtime": 3.1318, | |
| "eval_samples_per_second": 31.93, | |
| "eval_steps_per_second": 4.151, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5363546013832092, | |
| "eval_runtime": 3.131, | |
| "eval_samples_per_second": 31.938, | |
| "eval_steps_per_second": 4.152, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5350150465965271, | |
| "eval_runtime": 3.1305, | |
| "eval_samples_per_second": 31.944, | |
| "eval_steps_per_second": 4.153, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5382318496704102, | |
| "eval_runtime": 3.1297, | |
| "eval_samples_per_second": 31.952, | |
| "eval_steps_per_second": 4.154, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5329770445823669, | |
| "eval_runtime": 3.1308, | |
| "eval_samples_per_second": 31.941, | |
| "eval_steps_per_second": 4.152, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5360609292984009, | |
| "eval_runtime": 3.129, | |
| "eval_samples_per_second": 31.959, | |
| "eval_steps_per_second": 4.155, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5378895998001099, | |
| "eval_runtime": 3.1316, | |
| "eval_samples_per_second": 31.932, | |
| "eval_steps_per_second": 4.151, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5313640236854553, | |
| "eval_runtime": 3.1315, | |
| "eval_samples_per_second": 31.933, | |
| "eval_steps_per_second": 4.151, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5307939648628235, | |
| "eval_runtime": 3.1306, | |
| "eval_samples_per_second": 31.943, | |
| "eval_steps_per_second": 4.153, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.0005, | |
| "loss": 0.6054, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5347521901130676, | |
| "eval_runtime": 3.132, | |
| "eval_samples_per_second": 31.929, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5373522043228149, | |
| "eval_runtime": 3.1333, | |
| "eval_samples_per_second": 31.915, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5363101363182068, | |
| "eval_runtime": 3.1311, | |
| "eval_samples_per_second": 31.938, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5361413359642029, | |
| "eval_runtime": 3.1442, | |
| "eval_samples_per_second": 31.804, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5333101153373718, | |
| "eval_runtime": 3.1351, | |
| "eval_samples_per_second": 31.896, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5345572829246521, | |
| "eval_runtime": 3.131, | |
| "eval_samples_per_second": 31.939, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5354093313217163, | |
| "eval_runtime": 3.132, | |
| "eval_samples_per_second": 31.929, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5338207483291626, | |
| "eval_runtime": 3.1317, | |
| "eval_samples_per_second": 31.931, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5332033634185791, | |
| "eval_runtime": 3.1333, | |
| "eval_samples_per_second": 31.915, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5333837270736694, | |
| "eval_runtime": 3.1309, | |
| "eval_samples_per_second": 31.94, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5361276268959045, | |
| "eval_runtime": 3.1288, | |
| "eval_samples_per_second": 31.962, | |
| "eval_steps_per_second": 4.155, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5311057567596436, | |
| "eval_runtime": 3.1314, | |
| "eval_samples_per_second": 31.935, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5332194566726685, | |
| "eval_runtime": 3.1329, | |
| "eval_samples_per_second": 31.919, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5312312841415405, | |
| "eval_runtime": 3.1314, | |
| "eval_samples_per_second": 31.935, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5333719253540039, | |
| "eval_runtime": 3.1308, | |
| "eval_samples_per_second": 31.941, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5306155681610107, | |
| "eval_runtime": 3.1331, | |
| "eval_samples_per_second": 31.917, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5325944423675537, | |
| "eval_runtime": 3.1292, | |
| "eval_samples_per_second": 31.957, | |
| "eval_steps_per_second": 4.154, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5336495041847229, | |
| "eval_runtime": 3.1276, | |
| "eval_samples_per_second": 31.973, | |
| "eval_steps_per_second": 4.157, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5360873937606812, | |
| "eval_runtime": 3.1422, | |
| "eval_samples_per_second": 31.825, | |
| "eval_steps_per_second": 4.137, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.5358687043190002, | |
| "eval_runtime": 3.1436, | |
| "eval_samples_per_second": 31.811, | |
| "eval_steps_per_second": 4.135, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.00025, | |
| "loss": 0.5996, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5341857671737671, | |
| "eval_runtime": 3.1334, | |
| "eval_samples_per_second": 31.914, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5345585942268372, | |
| "eval_runtime": 3.1431, | |
| "eval_samples_per_second": 31.816, | |
| "eval_steps_per_second": 4.136, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5333269238471985, | |
| "eval_runtime": 3.1337, | |
| "eval_samples_per_second": 31.912, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5321747660636902, | |
| "eval_runtime": 3.1351, | |
| "eval_samples_per_second": 31.896, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5307139754295349, | |
| "eval_runtime": 3.1326, | |
| "eval_samples_per_second": 31.923, | |
| "eval_steps_per_second": 4.15, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5297897458076477, | |
| "eval_runtime": 3.1367, | |
| "eval_samples_per_second": 31.881, | |
| "eval_steps_per_second": 4.144, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5299915671348572, | |
| "eval_runtime": 3.1344, | |
| "eval_samples_per_second": 31.904, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5306195616722107, | |
| "eval_runtime": 3.1331, | |
| "eval_samples_per_second": 31.917, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5310782790184021, | |
| "eval_runtime": 3.135, | |
| "eval_samples_per_second": 31.898, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5317740440368652, | |
| "eval_runtime": 3.1334, | |
| "eval_samples_per_second": 31.914, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5319586396217346, | |
| "eval_runtime": 3.1355, | |
| "eval_samples_per_second": 31.892, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5320259928703308, | |
| "eval_runtime": 3.1356, | |
| "eval_samples_per_second": 31.892, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5309186577796936, | |
| "eval_runtime": 3.1334, | |
| "eval_samples_per_second": 31.914, | |
| "eval_steps_per_second": 4.149, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5307186245918274, | |
| "eval_runtime": 3.1346, | |
| "eval_samples_per_second": 31.902, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5306009650230408, | |
| "eval_runtime": 3.135, | |
| "eval_samples_per_second": 31.898, | |
| "eval_steps_per_second": 4.147, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5314067006111145, | |
| "eval_runtime": 3.1521, | |
| "eval_samples_per_second": 31.725, | |
| "eval_steps_per_second": 4.124, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5311477184295654, | |
| "eval_runtime": 3.1462, | |
| "eval_samples_per_second": 31.785, | |
| "eval_steps_per_second": 4.132, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5311266779899597, | |
| "eval_runtime": 3.1309, | |
| "eval_samples_per_second": 31.94, | |
| "eval_steps_per_second": 4.152, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.53106290102005, | |
| "eval_runtime": 3.1356, | |
| "eval_samples_per_second": 31.892, | |
| "eval_steps_per_second": 4.146, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5310864448547363, | |
| "eval_runtime": 3.1316, | |
| "eval_samples_per_second": 31.932, | |
| "eval_steps_per_second": 4.151, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.596, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5310752391815186, | |
| "eval_runtime": 3.1362, | |
| "eval_samples_per_second": 31.886, | |
| "eval_steps_per_second": 4.145, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.6042405700683594, | |
| "train_runtime": 1661.2119, | |
| "train_samples_per_second": 19.263, | |
| "train_steps_per_second": 1.204 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |