| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 86, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.8559, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.7071, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7514, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9992837548163315e-05, | |
| "loss": 0.7098, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9971360452796523e-05, | |
| "loss": 0.6787, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.993559947963185e-05, | |
| "loss": 0.6556, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9885605855918887e-05, | |
| "loss": 0.69, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9821451197042028e-05, | |
| "loss": 0.753, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9743227403932135e-05, | |
| "loss": 0.7316, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9651046531419335e-05, | |
| "loss": 0.7426, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9545040627715554e-05, | |
| "loss": 0.7588, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.942536154525673e-05, | |
| "loss": 0.7027, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9292180723175656e-05, | |
| "loss": 0.6916, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9145688941717074e-05, | |
| "loss": 0.7475, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8986096048946826e-05, | |
| "loss": 0.7039, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.881363066014649e-05, | |
| "loss": 0.7715, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.862853983032423e-05, | |
| "loss": 0.6958, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8431088700310846e-05, | |
| "loss": 0.7214, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8221560116948103e-05, | |
| "loss": 0.6808, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8000254227913346e-05, | |
| "loss": 0.653, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7767488051760858e-05, | |
| "loss": 0.645, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7523595023795814e-05, | |
| "loss": 0.7061, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7268924518431437e-05, | |
| "loss": 0.7223, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.700384134871351e-05, | |
| "loss": 0.6489, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.672872524372919e-05, | |
| "loss": 0.6721, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.644397030464877e-05, | |
| "loss": 0.6771, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.614998444017954e-05, | |
| "loss": 0.1781, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5847188782240473e-05, | |
| "loss": 0.6716, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5536017082694846e-05, | |
| "loss": 0.7402, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5216915092004847e-05, | |
| "loss": 0.6907, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4890339920698334e-05, | |
| "loss": 0.746, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4556759384562418e-05, | |
| "loss": 0.7715, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.421665133450184e-05, | |
| "loss": 0.6999, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3870502972022175e-05, | |
| "loss": 0.6374, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.351881015131833e-05, | |
| "loss": 0.6886, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.316207666896824e-05, | |
| "loss": 0.6923, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2800813542249073e-05, | |
| "loss": 0.648, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2435538277109919e-05, | |
| "loss": 0.6889, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.206677412684953e-05, | |
| "loss": 0.7149, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1695049342560969e-05, | |
| "loss": 0.7076, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1320896416417026e-05, | |
| "loss": 0.7184, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0944851318880314e-05, | |
| "loss": 0.7535, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0567452730930743e-05, | |
| "loss": 0.7158, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0189241272410191e-05, | |
| "loss": 0.7169, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.810758727589814e-06, | |
| "loss": 0.631, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.43254726906926e-06, | |
| "loss": 0.6797, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.055148681119688e-06, | |
| "loss": 0.7235, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.67910358358298e-06, | |
| "loss": 0.7368, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.304950657439034e-06, | |
| "loss": 0.6947, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.93322587315047e-06, | |
| "loss": 0.6969, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.564461722890082e-06, | |
| "loss": 0.7425, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.199186457750931e-06, | |
| "loss": 0.641, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.837923331031761e-06, | |
| "loss": 0.1685, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.48118984868167e-06, | |
| "loss": 0.684, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.129497027977829e-06, | |
| "loss": 0.1894, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.78334866549816e-06, | |
| "loss": 0.6861, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.443240615437586e-06, | |
| "loss": 0.6837, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.109660079301668e-06, | |
| "loss": 0.7266, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.783084907995156e-06, | |
| "loss": 0.665, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.463982917305155e-06, | |
| "loss": 0.655, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.152811217759529e-06, | |
| "loss": 0.6949, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.850015559820465e-06, | |
| "loss": 0.6858, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.5560296953512296e-06, | |
| "loss": 0.6224, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.2712747562708115e-06, | |
| "loss": 0.6903, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.9961586512864947e-06, | |
| "loss": 0.7373, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.7310754815685627e-06, | |
| "loss": 0.6265, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.4764049762041874e-06, | |
| "loss": 0.7343, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2325119482391466e-06, | |
| "loss": 0.6524, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9997457720866554e-06, | |
| "loss": 0.7026, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7784398830519002e-06, | |
| "loss": 0.6995, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.5689112996891576e-06, | |
| "loss": 0.6672, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3714601696757713e-06, | |
| "loss": 0.7158, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1863693398535115e-06, | |
| "loss": 0.6767, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.01390395105318e-06, | |
| "loss": 0.7041, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.543110582829272e-07, | |
| "loss": 0.6582, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.078192768243486e-07, | |
| "loss": 0.7415, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.746384547432738e-07, | |
| "loss": 0.7245, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.549593722844492e-07, | |
| "loss": 0.1694, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4895346858066723e-07, | |
| "loss": 0.6716, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.5677259606786686e-07, | |
| "loss": 0.7095, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7854880295797406e-07, | |
| "loss": 0.6823, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1439414408111471e-07, | |
| "loss": 0.7316, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.440052036815081e-08, | |
| "loss": 0.6417, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.86395472034795e-08, | |
| "loss": 0.6386, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.162451836685291e-09, | |
| "loss": 0.6541, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.1686, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 86, | |
| "total_flos": 204478356652032.0, | |
| "train_loss": 0.6681410723993945, | |
| "train_runtime": 2105.2364, | |
| "train_samples_per_second": 5.225, | |
| "train_steps_per_second": 0.041 | |
| } | |
| ], | |
| "max_steps": 86, | |
| "num_train_epochs": 1, | |
| "total_flos": 204478356652032.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |