| { | |
| "best_metric": 0.9365, | |
| "best_model_checkpoint": "dinov2-base-imagenet1k-1-layer-finetuned-galaxy_mnist/checkpoint-593", | |
| "epoch": 29.76, | |
| "eval_steps": 500, | |
| "global_step": 930, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.376344086021506e-07, | |
| "loss": 1.6159, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.0752688172043011e-06, | |
| "loss": 1.2128, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.6129032258064516e-06, | |
| "loss": 0.8401, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_accuracy": 0.7605, | |
| "eval_f1": 0.7542683716303844, | |
| "eval_loss": 0.6220366358757019, | |
| "eval_precision": 0.7579355479379479, | |
| "eval_recall": 0.7605, | |
| "eval_runtime": 17.2083, | |
| "eval_samples_per_second": 116.223, | |
| "eval_steps_per_second": 1.86, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.1505376344086023e-06, | |
| "loss": 0.5859, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.688172043010753e-06, | |
| "loss": 0.4302, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.225806451612903e-06, | |
| "loss": 0.3857, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "eval_accuracy": 0.8875, | |
| "eval_f1": 0.8876656285967213, | |
| "eval_loss": 0.26959437131881714, | |
| "eval_precision": 0.8880762262738604, | |
| "eval_recall": 0.8875, | |
| "eval_runtime": 17.6301, | |
| "eval_samples_per_second": 113.442, | |
| "eval_steps_per_second": 1.815, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.763440860215054e-06, | |
| "loss": 0.3494, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.3010752688172045e-06, | |
| "loss": 0.3157, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.838709677419355e-06, | |
| "loss": 0.3144, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_accuracy": 0.9015, | |
| "eval_f1": 0.9010338577553875, | |
| "eval_loss": 0.24912068247795105, | |
| "eval_precision": 0.9028059173186274, | |
| "eval_recall": 0.9015, | |
| "eval_runtime": 17.2327, | |
| "eval_samples_per_second": 116.059, | |
| "eval_steps_per_second": 1.857, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 4.9581839904420555e-06, | |
| "loss": 0.33, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 4.898446833930705e-06, | |
| "loss": 0.2905, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 4.838709677419355e-06, | |
| "loss": 0.2769, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.913, | |
| "eval_f1": 0.9128412870419806, | |
| "eval_loss": 0.21787521243095398, | |
| "eval_precision": 0.9129249597927858, | |
| "eval_recall": 0.913, | |
| "eval_runtime": 17.4445, | |
| "eval_samples_per_second": 114.649, | |
| "eval_steps_per_second": 1.834, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 4.7789725209080055e-06, | |
| "loss": 0.2902, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 4.719235364396655e-06, | |
| "loss": 0.2569, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 4.659498207885305e-06, | |
| "loss": 0.2858, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "eval_accuracy": 0.9025, | |
| "eval_f1": 0.9020162725126999, | |
| "eval_loss": 0.2454933226108551, | |
| "eval_precision": 0.9069644594388034, | |
| "eval_recall": 0.9025, | |
| "eval_runtime": 17.3381, | |
| "eval_samples_per_second": 115.353, | |
| "eval_steps_per_second": 1.846, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 4.599761051373955e-06, | |
| "loss": 0.2735, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 4.540023894862605e-06, | |
| "loss": 0.2562, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 4.480286738351255e-06, | |
| "loss": 0.2704, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "eval_accuracy": 0.9155, | |
| "eval_f1": 0.9155929734022, | |
| "eval_loss": 0.21211333572864532, | |
| "eval_precision": 0.9233801924802966, | |
| "eval_recall": 0.9155, | |
| "eval_runtime": 16.9571, | |
| "eval_samples_per_second": 117.945, | |
| "eval_steps_per_second": 1.887, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 4.420549581839905e-06, | |
| "loss": 0.2634, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 4.360812425328555e-06, | |
| "loss": 0.2351, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 4.3010752688172045e-06, | |
| "loss": 0.2557, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "eval_accuracy": 0.9155, | |
| "eval_f1": 0.9152474970088074, | |
| "eval_loss": 0.2177189141511917, | |
| "eval_precision": 0.9189605750516177, | |
| "eval_recall": 0.9155, | |
| "eval_runtime": 16.5188, | |
| "eval_samples_per_second": 121.074, | |
| "eval_steps_per_second": 1.937, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 4.241338112305855e-06, | |
| "loss": 0.2624, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 4.181600955794505e-06, | |
| "loss": 0.2364, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 4.121863799283155e-06, | |
| "loss": 0.2416, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 4.062126642771804e-06, | |
| "loss": 0.2069, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9255, | |
| "eval_f1": 0.9254810632455276, | |
| "eval_loss": 0.18643927574157715, | |
| "eval_precision": 0.9256381458578058, | |
| "eval_recall": 0.9255, | |
| "eval_runtime": 16.5437, | |
| "eval_samples_per_second": 120.892, | |
| "eval_steps_per_second": 1.934, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 4.002389486260454e-06, | |
| "loss": 0.2402, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 3.942652329749105e-06, | |
| "loss": 0.2509, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 3.882915173237754e-06, | |
| "loss": 0.2344, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "eval_accuracy": 0.923, | |
| "eval_f1": 0.9230031822087664, | |
| "eval_loss": 0.1893594115972519, | |
| "eval_precision": 0.9236882399086506, | |
| "eval_recall": 0.923, | |
| "eval_runtime": 17.2256, | |
| "eval_samples_per_second": 116.106, | |
| "eval_steps_per_second": 1.858, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 3.823178016726404e-06, | |
| "loss": 0.2187, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 3.763440860215054e-06, | |
| "loss": 0.208, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 0.1996, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "eval_accuracy": 0.9235, | |
| "eval_f1": 0.9233698349343116, | |
| "eval_loss": 0.19926807284355164, | |
| "eval_precision": 0.9259814793932657, | |
| "eval_recall": 0.9235, | |
| "eval_runtime": 17.2364, | |
| "eval_samples_per_second": 116.033, | |
| "eval_steps_per_second": 1.857, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 10.24, | |
| "learning_rate": 3.643966547192354e-06, | |
| "loss": 0.2049, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 3.584229390681004e-06, | |
| "loss": 0.2089, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 3.5244922341696534e-06, | |
| "loss": 0.2011, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "eval_accuracy": 0.928, | |
| "eval_f1": 0.9279233986678691, | |
| "eval_loss": 0.18277481198310852, | |
| "eval_precision": 0.9279656529071406, | |
| "eval_recall": 0.928, | |
| "eval_runtime": 16.5403, | |
| "eval_samples_per_second": 120.917, | |
| "eval_steps_per_second": 1.935, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 11.2, | |
| "learning_rate": 3.4647550776583037e-06, | |
| "loss": 0.1899, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 3.4050179211469536e-06, | |
| "loss": 0.1961, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 3.3452807646356034e-06, | |
| "loss": 0.2229, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.9155, | |
| "eval_f1": 0.9145163167356054, | |
| "eval_loss": 0.23575706779956818, | |
| "eval_precision": 0.9233304225050336, | |
| "eval_recall": 0.9155, | |
| "eval_runtime": 15.8252, | |
| "eval_samples_per_second": 126.38, | |
| "eval_steps_per_second": 2.022, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 3.2855436081242537e-06, | |
| "loss": 0.1956, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 12.48, | |
| "learning_rate": 3.225806451612903e-06, | |
| "loss": 0.2398, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 12.8, | |
| "learning_rate": 3.1660692951015535e-06, | |
| "loss": 0.1792, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "eval_accuracy": 0.9205, | |
| "eval_f1": 0.9205455092354193, | |
| "eval_loss": 0.18970580399036407, | |
| "eval_precision": 0.9214418149774888, | |
| "eval_recall": 0.9205, | |
| "eval_runtime": 16.2468, | |
| "eval_samples_per_second": 123.101, | |
| "eval_steps_per_second": 1.97, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 3.1063321385902034e-06, | |
| "loss": 0.2103, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 3.0465949820788532e-06, | |
| "loss": 0.1909, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 2.9868578255675035e-06, | |
| "loss": 0.1898, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 13.98, | |
| "eval_accuracy": 0.921, | |
| "eval_f1": 0.9208312467941863, | |
| "eval_loss": 0.20168715715408325, | |
| "eval_precision": 0.9216542515530398, | |
| "eval_recall": 0.921, | |
| "eval_runtime": 16.1466, | |
| "eval_samples_per_second": 123.865, | |
| "eval_steps_per_second": 1.982, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 14.08, | |
| "learning_rate": 2.9271206690561534e-06, | |
| "loss": 0.1804, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 14.4, | |
| "learning_rate": 2.867383512544803e-06, | |
| "loss": 0.2052, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 14.72, | |
| "learning_rate": 2.807646356033453e-06, | |
| "loss": 0.1735, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "eval_accuracy": 0.927, | |
| "eval_f1": 0.9269426239594731, | |
| "eval_loss": 0.1954416185617447, | |
| "eval_precision": 0.9269718249595981, | |
| "eval_recall": 0.927, | |
| "eval_runtime": 16.9373, | |
| "eval_samples_per_second": 118.083, | |
| "eval_steps_per_second": 1.889, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.747909199522103e-06, | |
| "loss": 0.1933, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 2.688172043010753e-06, | |
| "loss": 0.1727, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 15.68, | |
| "learning_rate": 2.6284348864994032e-06, | |
| "loss": 0.1835, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.5686977299880527e-06, | |
| "loss": 0.1751, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.9295, | |
| "eval_f1": 0.9294043091590938, | |
| "eval_loss": 0.19178292155265808, | |
| "eval_precision": 0.9299108071265199, | |
| "eval_recall": 0.9295, | |
| "eval_runtime": 17.2166, | |
| "eval_samples_per_second": 116.167, | |
| "eval_steps_per_second": 1.859, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 16.32, | |
| "learning_rate": 2.5089605734767026e-06, | |
| "loss": 0.1912, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 16.64, | |
| "learning_rate": 2.4492234169653525e-06, | |
| "loss": 0.1815, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 2.3894862604540028e-06, | |
| "loss": 0.1732, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 16.99, | |
| "eval_accuracy": 0.922, | |
| "eval_f1": 0.9218903176887637, | |
| "eval_loss": 0.190641850233078, | |
| "eval_precision": 0.9224739140720933, | |
| "eval_recall": 0.922, | |
| "eval_runtime": 17.563, | |
| "eval_samples_per_second": 113.876, | |
| "eval_steps_per_second": 1.822, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 17.28, | |
| "learning_rate": 2.3297491039426526e-06, | |
| "loss": 0.1725, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 2.2700119474313025e-06, | |
| "loss": 0.1766, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 17.92, | |
| "learning_rate": 2.2102747909199524e-06, | |
| "loss": 0.1738, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 17.98, | |
| "eval_accuracy": 0.931, | |
| "eval_f1": 0.9309560007388311, | |
| "eval_loss": 0.18456581234931946, | |
| "eval_precision": 0.9316696136336309, | |
| "eval_recall": 0.931, | |
| "eval_runtime": 17.21, | |
| "eval_samples_per_second": 116.212, | |
| "eval_steps_per_second": 1.859, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 2.1505376344086023e-06, | |
| "loss": 0.1638, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 18.56, | |
| "learning_rate": 2.0908004778972526e-06, | |
| "loss": 0.1695, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 18.88, | |
| "learning_rate": 2.031063321385902e-06, | |
| "loss": 0.1694, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 18.98, | |
| "eval_accuracy": 0.9365, | |
| "eval_f1": 0.9365419334555197, | |
| "eval_loss": 0.18749217689037323, | |
| "eval_precision": 0.9367142468929726, | |
| "eval_recall": 0.9365, | |
| "eval_runtime": 17.245, | |
| "eval_samples_per_second": 115.976, | |
| "eval_steps_per_second": 1.856, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 19.2, | |
| "learning_rate": 1.9713261648745523e-06, | |
| "loss": 0.1627, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 19.52, | |
| "learning_rate": 1.911589008363202e-06, | |
| "loss": 0.156, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 19.84, | |
| "learning_rate": 1.8518518518518519e-06, | |
| "loss": 0.1723, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.9285, | |
| "eval_f1": 0.9283564470960457, | |
| "eval_loss": 0.19414418935775757, | |
| "eval_precision": 0.9293172724622787, | |
| "eval_recall": 0.9285, | |
| "eval_runtime": 17.3836, | |
| "eval_samples_per_second": 115.051, | |
| "eval_steps_per_second": 1.841, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 20.16, | |
| "learning_rate": 1.792114695340502e-06, | |
| "loss": 0.1604, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 20.48, | |
| "learning_rate": 1.7323775388291518e-06, | |
| "loss": 0.1688, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 20.8, | |
| "learning_rate": 1.6726403823178017e-06, | |
| "loss": 0.1574, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 20.99, | |
| "eval_accuracy": 0.9335, | |
| "eval_f1": 0.933604382541505, | |
| "eval_loss": 0.19053442776203156, | |
| "eval_precision": 0.933748843513901, | |
| "eval_recall": 0.9335, | |
| "eval_runtime": 17.3493, | |
| "eval_samples_per_second": 115.278, | |
| "eval_steps_per_second": 1.844, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 21.12, | |
| "learning_rate": 1.6129032258064516e-06, | |
| "loss": 0.1514, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 21.44, | |
| "learning_rate": 1.5531660692951017e-06, | |
| "loss": 0.1396, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 21.76, | |
| "learning_rate": 1.4934289127837518e-06, | |
| "loss": 0.1485, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 21.98, | |
| "eval_accuracy": 0.9315, | |
| "eval_f1": 0.9314167545216125, | |
| "eval_loss": 0.18693630397319794, | |
| "eval_precision": 0.9313446931532684, | |
| "eval_recall": 0.9315, | |
| "eval_runtime": 20.8097, | |
| "eval_samples_per_second": 96.109, | |
| "eval_steps_per_second": 1.538, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 22.08, | |
| "learning_rate": 1.4336917562724014e-06, | |
| "loss": 0.1316, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 22.4, | |
| "learning_rate": 1.3739545997610515e-06, | |
| "loss": 0.1435, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "learning_rate": 1.3142174432497016e-06, | |
| "loss": 0.1537, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 22.98, | |
| "eval_accuracy": 0.936, | |
| "eval_f1": 0.9359520842741622, | |
| "eval_loss": 0.18298617005348206, | |
| "eval_precision": 0.9359672973348017, | |
| "eval_recall": 0.936, | |
| "eval_runtime": 17.2379, | |
| "eval_samples_per_second": 116.023, | |
| "eval_steps_per_second": 1.856, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 23.04, | |
| "learning_rate": 1.2544802867383513e-06, | |
| "loss": 0.1349, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 23.36, | |
| "learning_rate": 1.1947431302270014e-06, | |
| "loss": 0.1282, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 23.68, | |
| "learning_rate": 1.1350059737156513e-06, | |
| "loss": 0.1408, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 1.0752688172043011e-06, | |
| "loss": 0.1406, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.932, | |
| "eval_f1": 0.9319845343042348, | |
| "eval_loss": 0.19753658771514893, | |
| "eval_precision": 0.9322001623351449, | |
| "eval_recall": 0.932, | |
| "eval_runtime": 17.216, | |
| "eval_samples_per_second": 116.171, | |
| "eval_steps_per_second": 1.859, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 24.32, | |
| "learning_rate": 1.015531660692951e-06, | |
| "loss": 0.1217, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 24.64, | |
| "learning_rate": 9.55794504181601e-07, | |
| "loss": 0.1475, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 24.96, | |
| "learning_rate": 8.96057347670251e-07, | |
| "loss": 0.1326, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 24.99, | |
| "eval_accuracy": 0.9315, | |
| "eval_f1": 0.9315212925629236, | |
| "eval_loss": 0.1918136179447174, | |
| "eval_precision": 0.9316398530410831, | |
| "eval_recall": 0.9315, | |
| "eval_runtime": 17.2178, | |
| "eval_samples_per_second": 116.159, | |
| "eval_steps_per_second": 1.859, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 25.28, | |
| "learning_rate": 8.363201911589009e-07, | |
| "loss": 0.1423, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 25.6, | |
| "learning_rate": 7.765830346475508e-07, | |
| "loss": 0.1462, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 25.92, | |
| "learning_rate": 7.168458781362007e-07, | |
| "loss": 0.1238, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 25.98, | |
| "eval_accuracy": 0.9275, | |
| "eval_f1": 0.9275714852684839, | |
| "eval_loss": 0.21046194434165955, | |
| "eval_precision": 0.9287568114219772, | |
| "eval_recall": 0.9275, | |
| "eval_runtime": 17.3304, | |
| "eval_samples_per_second": 115.404, | |
| "eval_steps_per_second": 1.846, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 26.24, | |
| "learning_rate": 6.571087216248508e-07, | |
| "loss": 0.1275, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 26.56, | |
| "learning_rate": 5.973715651135007e-07, | |
| "loss": 0.1417, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 26.88, | |
| "learning_rate": 5.376344086021506e-07, | |
| "loss": 0.1299, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 26.98, | |
| "eval_accuracy": 0.9325, | |
| "eval_f1": 0.932381459503027, | |
| "eval_loss": 0.20224767923355103, | |
| "eval_precision": 0.9326921008972977, | |
| "eval_recall": 0.9325, | |
| "eval_runtime": 17.2957, | |
| "eval_samples_per_second": 115.636, | |
| "eval_steps_per_second": 1.85, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 27.2, | |
| "learning_rate": 4.778972520908006e-07, | |
| "loss": 0.1393, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 27.52, | |
| "learning_rate": 4.1816009557945043e-07, | |
| "loss": 0.1208, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 27.84, | |
| "learning_rate": 3.5842293906810036e-07, | |
| "loss": 0.1387, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.9335, | |
| "eval_f1": 0.9335696977145349, | |
| "eval_loss": 0.20111316442489624, | |
| "eval_precision": 0.9336932043728637, | |
| "eval_recall": 0.9335, | |
| "eval_runtime": 16.8699, | |
| "eval_samples_per_second": 118.555, | |
| "eval_steps_per_second": 1.897, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 28.16, | |
| "learning_rate": 2.9868578255675034e-07, | |
| "loss": 0.1239, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 28.48, | |
| "learning_rate": 2.389486260454003e-07, | |
| "loss": 0.1357, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 28.8, | |
| "learning_rate": 1.7921146953405018e-07, | |
| "loss": 0.1279, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 28.99, | |
| "eval_accuracy": 0.931, | |
| "eval_f1": 0.9310154849548091, | |
| "eval_loss": 0.20053213834762573, | |
| "eval_precision": 0.931047750983443, | |
| "eval_recall": 0.931, | |
| "eval_runtime": 17.1937, | |
| "eval_samples_per_second": 116.322, | |
| "eval_steps_per_second": 1.861, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 29.12, | |
| "learning_rate": 1.1947431302270014e-07, | |
| "loss": 0.1302, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 29.44, | |
| "learning_rate": 5.973715651135007e-08, | |
| "loss": 0.1233, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 29.76, | |
| "learning_rate": 0.0, | |
| "loss": 0.1256, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 29.76, | |
| "eval_accuracy": 0.931, | |
| "eval_f1": 0.9310052979866259, | |
| "eval_loss": 0.2003646343946457, | |
| "eval_precision": 0.931012460623901, | |
| "eval_recall": 0.931, | |
| "eval_runtime": 17.2043, | |
| "eval_samples_per_second": 116.25, | |
| "eval_steps_per_second": 1.86, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 29.76, | |
| "step": 930, | |
| "total_flos": 2.4317906065576428e+19, | |
| "train_loss": 0.2331973833422507, | |
| "train_runtime": 4457.686, | |
| "train_samples_per_second": 53.84, | |
| "train_steps_per_second": 0.209 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 930, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 500, | |
| "total_flos": 2.4317906065576428e+19, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |