| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 49.994350282485875, | |
| "global_step": 4400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 3.5742, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 3.4353, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9e-07, | |
| "loss": 3.2438, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 3.0976, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.8575, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8e-06, | |
| "loss": 2.6829, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 2.4912, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 2.4056, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.7e-06, | |
| "loss": 2.4176, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3e-06, | |
| "loss": 2.2921, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.3e-06, | |
| "loss": 2.2252, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.6e-06, | |
| "loss": 2.2169, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.9e-06, | |
| "loss": 2.156, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 2.194, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.5e-06, | |
| "loss": 2.1136, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 2.1497, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.1e-06, | |
| "loss": 2.1208, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.4e-06, | |
| "loss": 2.2234, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.7000000000000005e-06, | |
| "loss": 2.0122, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6e-06, | |
| "loss": 2.0509, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 6.3e-06, | |
| "loss": 2.0077, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.6e-06, | |
| "loss": 2.0297, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.900000000000001e-06, | |
| "loss": 2.0226, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 7.2e-06, | |
| "loss": 2.0274, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.5e-06, | |
| "loss": 2.0505, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.8e-06, | |
| "loss": 2.0163, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 8.1e-06, | |
| "loss": 2.0764, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.9411, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 8.7e-06, | |
| "loss": 1.9308, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 9e-06, | |
| "loss": 1.9507, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 9.3e-06, | |
| "loss": 1.9036, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.9728, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.9105, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 1.02e-05, | |
| "loss": 1.957, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.05e-05, | |
| "loss": 1.9615, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.08e-05, | |
| "loss": 1.9444, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.11e-05, | |
| "loss": 1.9014, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 1.8752, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 1.1700000000000001e-05, | |
| "loss": 1.8787, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.8929, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 1.2299999999999999e-05, | |
| "loss": 1.9223, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.26e-05, | |
| "loss": 1.8559, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 1.29e-05, | |
| "loss": 1.8533, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 1.32e-05, | |
| "loss": 1.8577, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 1.9091, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 1.8096, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 1.4099999999999999e-05, | |
| "loss": 1.8333, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 1.44e-05, | |
| "loss": 1.8166, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 1.47e-05, | |
| "loss": 1.8077, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.8163, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 1.53e-05, | |
| "loss": 1.8269, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 1.56e-05, | |
| "loss": 1.8201, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.59e-05, | |
| "loss": 1.936, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 1.62e-05, | |
| "loss": 1.7812, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 1.65e-05, | |
| "loss": 1.7872, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 1.782, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 1.71e-05, | |
| "loss": 1.783, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.74e-05, | |
| "loss": 1.7638, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.77e-05, | |
| "loss": 1.7805, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.7768, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 1.83e-05, | |
| "loss": 1.7864, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.86e-05, | |
| "loss": 1.8437, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 1.718, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 1.7334, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 1.95e-05, | |
| "loss": 1.7445, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 1.98e-05, | |
| "loss": 1.7553, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 2.01e-05, | |
| "loss": 1.7567, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 2.04e-05, | |
| "loss": 1.7498, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 2.07e-05, | |
| "loss": 1.7273, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.7588, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 2.13e-05, | |
| "loss": 1.8182, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 2.16e-05, | |
| "loss": 1.7256, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 2.19e-05, | |
| "loss": 1.7198, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 2.22e-05, | |
| "loss": 1.6916, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.7189, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 1.7172, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 2.3100000000000002e-05, | |
| "loss": 1.7219, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 1.7425, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 2.37e-05, | |
| "loss": 1.7384, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.7836, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 2.43e-05, | |
| "loss": 1.6577, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 1.6644, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 2.49e-05, | |
| "loss": 1.6861, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 2.52e-05, | |
| "loss": 1.6939, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 2.55e-05, | |
| "loss": 1.7055, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 2.58e-05, | |
| "loss": 1.7117, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 2.61e-05, | |
| "loss": 1.6968, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "learning_rate": 2.64e-05, | |
| "loss": 1.7166, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 2.6700000000000002e-05, | |
| "loss": 1.7732, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.6664, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 2.7300000000000003e-05, | |
| "loss": 1.6757, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 2.7600000000000003e-05, | |
| "loss": 1.6687, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 2.79e-05, | |
| "loss": 1.6693, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 1.6738, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 10.79, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 1.6807, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 2.88e-05, | |
| "loss": 1.6875, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 2.91e-05, | |
| "loss": 1.7664, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 2.94e-05, | |
| "loss": 1.6674, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 2.97e-05, | |
| "loss": 1.6369, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "learning_rate": 3e-05, | |
| "loss": 1.6675, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "eval_denotation_accuracy": 0.18014835747085836, | |
| "eval_loss": 2.312751293182373, | |
| "eval_runtime": 1826.7928, | |
| "eval_samples_per_second": 1.55, | |
| "eval_steps_per_second": 0.097, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 2.9911764705882352e-05, | |
| "loss": 1.6501, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 2.9823529411764707e-05, | |
| "loss": 1.6657, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 2.973529411764706e-05, | |
| "loss": 1.6592, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 2.9647058823529414e-05, | |
| "loss": 1.6575, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 2.9558823529411766e-05, | |
| "loss": 1.6638, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 12.05, | |
| "learning_rate": 2.9470588235294117e-05, | |
| "loss": 1.7401, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 2.938235294117647e-05, | |
| "loss": 1.6335, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "learning_rate": 2.9294117647058824e-05, | |
| "loss": 1.6446, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 2.9205882352941175e-05, | |
| "loss": 1.6233, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 2.911764705882353e-05, | |
| "loss": 1.6528, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 2.9029411764705882e-05, | |
| "loss": 1.6488, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 12.72, | |
| "learning_rate": 2.8941176470588237e-05, | |
| "loss": 1.644, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "learning_rate": 2.885294117647059e-05, | |
| "loss": 1.6455, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 12.95, | |
| "learning_rate": 2.8764705882352943e-05, | |
| "loss": 1.6705, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 2.8676470588235295e-05, | |
| "loss": 1.7129, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 13.18, | |
| "learning_rate": 2.8588235294117647e-05, | |
| "loss": 1.6363, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 1.6237, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 2.8411764705882353e-05, | |
| "loss": 1.6011, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 13.52, | |
| "learning_rate": 2.8323529411764705e-05, | |
| "loss": 1.6302, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 2.823529411764706e-05, | |
| "loss": 1.6277, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "learning_rate": 2.814705882352941e-05, | |
| "loss": 1.6095, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 13.86, | |
| "learning_rate": 2.8058823529411766e-05, | |
| "loss": 1.6188, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "learning_rate": 2.7970588235294118e-05, | |
| "loss": 1.6085, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 2.7882352941176473e-05, | |
| "loss": 1.6951, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 2.7794117647058824e-05, | |
| "loss": 1.6059, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 2.770588235294118e-05, | |
| "loss": 1.6191, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 2.7617647058823528e-05, | |
| "loss": 1.6178, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 2.7529411764705883e-05, | |
| "loss": 1.6226, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 2.7441176470588234e-05, | |
| "loss": 1.6099, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 2.735294117647059e-05, | |
| "loss": 1.5953, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 2.726470588235294e-05, | |
| "loss": 1.6067, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 14.99, | |
| "learning_rate": 2.7176470588235296e-05, | |
| "loss": 1.6062, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 2.7088235294117647e-05, | |
| "loss": 1.6788, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.5883, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 15.34, | |
| "learning_rate": 2.6911764705882354e-05, | |
| "loss": 1.582, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 2.682352941176471e-05, | |
| "loss": 1.5865, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 15.56, | |
| "learning_rate": 2.673529411764706e-05, | |
| "loss": 1.5797, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 15.68, | |
| "learning_rate": 2.6647058823529412e-05, | |
| "loss": 1.5991, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 2.6558823529411764e-05, | |
| "loss": 1.604, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 15.9, | |
| "learning_rate": 2.647058823529412e-05, | |
| "loss": 1.5845, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.638235294117647e-05, | |
| "loss": 1.661, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "learning_rate": 2.6294117647058825e-05, | |
| "loss": 1.5842, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 16.25, | |
| "learning_rate": 2.6205882352941177e-05, | |
| "loss": 1.576, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 2.6117647058823532e-05, | |
| "loss": 1.5771, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 16.47, | |
| "learning_rate": 2.6029411764705883e-05, | |
| "loss": 1.5764, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 2.594117647058824e-05, | |
| "loss": 1.586, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 2.585294117647059e-05, | |
| "loss": 1.5773, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 2.576470588235294e-05, | |
| "loss": 1.5814, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 16.93, | |
| "learning_rate": 2.5676470588235293e-05, | |
| "loss": 1.5812, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 2.5588235294117648e-05, | |
| "loss": 1.6478, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 17.16, | |
| "learning_rate": 2.55e-05, | |
| "loss": 1.5619, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 17.27, | |
| "learning_rate": 2.5411764705882355e-05, | |
| "loss": 1.577, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 17.38, | |
| "learning_rate": 2.5323529411764706e-05, | |
| "loss": 1.5569, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "learning_rate": 2.523529411764706e-05, | |
| "loss": 1.5831, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 2.5147058823529413e-05, | |
| "loss": 1.5613, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 17.72, | |
| "learning_rate": 2.5058823529411768e-05, | |
| "loss": 1.5594, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 2.497058823529412e-05, | |
| "loss": 1.5723, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 17.95, | |
| "learning_rate": 2.488235294117647e-05, | |
| "loss": 1.576, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 2.4794117647058823e-05, | |
| "loss": 1.6366, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "learning_rate": 2.4705882352941174e-05, | |
| "loss": 1.5548, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 18.29, | |
| "learning_rate": 2.461764705882353e-05, | |
| "loss": 1.5511, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 18.41, | |
| "learning_rate": 2.452941176470588e-05, | |
| "loss": 1.5625, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 2.4441176470588236e-05, | |
| "loss": 1.5676, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 18.63, | |
| "learning_rate": 2.4352941176470587e-05, | |
| "loss": 1.5635, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 2.4264705882352942e-05, | |
| "loss": 1.5552, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 2.4176470588235294e-05, | |
| "loss": 1.5499, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 2.408823529411765e-05, | |
| "loss": 1.5592, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.6249, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 19.2, | |
| "learning_rate": 2.3911764705882356e-05, | |
| "loss": 1.5408, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 2.3823529411764704e-05, | |
| "loss": 1.5567, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 19.43, | |
| "learning_rate": 2.373529411764706e-05, | |
| "loss": 1.5508, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 19.54, | |
| "learning_rate": 2.364705882352941e-05, | |
| "loss": 1.5482, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 19.66, | |
| "learning_rate": 2.3558823529411765e-05, | |
| "loss": 1.541, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 19.77, | |
| "learning_rate": 2.3470588235294117e-05, | |
| "loss": 1.551, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 19.88, | |
| "learning_rate": 2.3382352941176472e-05, | |
| "loss": 1.5503, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 19.99, | |
| "learning_rate": 2.3294117647058824e-05, | |
| "loss": 1.5499, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 20.11, | |
| "learning_rate": 2.320588235294118e-05, | |
| "loss": 1.6213, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 20.23, | |
| "learning_rate": 2.311764705882353e-05, | |
| "loss": 1.5295, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 20.34, | |
| "learning_rate": 2.3029411764705885e-05, | |
| "loss": 1.5382, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 20.45, | |
| "learning_rate": 2.2941176470588233e-05, | |
| "loss": 1.5405, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 20.56, | |
| "learning_rate": 2.2852941176470588e-05, | |
| "loss": 1.5342, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 20.68, | |
| "learning_rate": 2.276470588235294e-05, | |
| "loss": 1.5444, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 20.79, | |
| "learning_rate": 2.2676470588235295e-05, | |
| "loss": 1.5444, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 20.9, | |
| "learning_rate": 2.2588235294117646e-05, | |
| "loss": 1.5363, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.6267, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 21.14, | |
| "learning_rate": 2.2411764705882353e-05, | |
| "loss": 1.5449, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 21.25, | |
| "learning_rate": 2.2323529411764708e-05, | |
| "loss": 1.5241, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 21.36, | |
| "learning_rate": 2.223529411764706e-05, | |
| "loss": 1.5301, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 21.47, | |
| "learning_rate": 2.2147058823529415e-05, | |
| "loss": 1.5385, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 21.59, | |
| "learning_rate": 2.2058823529411766e-05, | |
| "loss": 1.5206, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 2.1970588235294118e-05, | |
| "loss": 1.5306, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 21.81, | |
| "learning_rate": 2.188235294117647e-05, | |
| "loss": 1.5302, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 21.93, | |
| "learning_rate": 2.1794117647058824e-05, | |
| "loss": 1.5401, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 22.05, | |
| "learning_rate": 2.1705882352941176e-05, | |
| "loss": 1.6063, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 22.16, | |
| "learning_rate": 2.161764705882353e-05, | |
| "loss": 1.5251, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 22.27, | |
| "learning_rate": 2.1529411764705882e-05, | |
| "loss": 1.5197, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 22.38, | |
| "learning_rate": 2.1441176470588237e-05, | |
| "loss": 1.524, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 22.5, | |
| "learning_rate": 2.135294117647059e-05, | |
| "loss": 1.5233, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 22.61, | |
| "learning_rate": 2.1264705882352944e-05, | |
| "loss": 1.5256, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "learning_rate": 2.1176470588235296e-05, | |
| "loss": 1.5264, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "eval_denotation_accuracy": 0.48675379724478984, | |
| "eval_loss": 2.4037857055664062, | |
| "eval_runtime": 1745.3919, | |
| "eval_samples_per_second": 1.622, | |
| "eval_steps_per_second": 0.101, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 22.84, | |
| "learning_rate": 2.1088235294117647e-05, | |
| "loss": 1.527, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 22.95, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.5229, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 23.07, | |
| "learning_rate": 2.0911764705882354e-05, | |
| "loss": 1.6007, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 23.18, | |
| "learning_rate": 2.0823529411764705e-05, | |
| "loss": 1.5107, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 23.29, | |
| "learning_rate": 2.073529411764706e-05, | |
| "loss": 1.5148, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 23.41, | |
| "learning_rate": 2.0647058823529412e-05, | |
| "loss": 1.5174, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 23.52, | |
| "learning_rate": 2.0558823529411767e-05, | |
| "loss": 1.5141, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 23.63, | |
| "learning_rate": 2.047058823529412e-05, | |
| "loss": 1.5166, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 23.75, | |
| "learning_rate": 2.0382352941176474e-05, | |
| "loss": 1.5197, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 23.86, | |
| "learning_rate": 2.0294117647058825e-05, | |
| "loss": 1.5154, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 23.97, | |
| "learning_rate": 2.020588235294118e-05, | |
| "loss": 1.5155, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 24.09, | |
| "learning_rate": 2.011764705882353e-05, | |
| "loss": 1.5843, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 24.2, | |
| "learning_rate": 2.0029411764705883e-05, | |
| "loss": 1.5217, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 24.32, | |
| "learning_rate": 1.9941176470588235e-05, | |
| "loss": 1.5099, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 24.43, | |
| "learning_rate": 1.9852941176470586e-05, | |
| "loss": 1.5118, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 24.54, | |
| "learning_rate": 1.976470588235294e-05, | |
| "loss": 1.5216, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 24.66, | |
| "learning_rate": 1.9676470588235293e-05, | |
| "loss": 1.5088, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 24.77, | |
| "learning_rate": 1.9588235294117648e-05, | |
| "loss": 1.5125, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 24.88, | |
| "learning_rate": 1.95e-05, | |
| "loss": 1.517, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 24.99, | |
| "learning_rate": 1.9411764705882355e-05, | |
| "loss": 1.5136, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 25.11, | |
| "learning_rate": 1.9323529411764706e-05, | |
| "loss": 1.5812, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 25.23, | |
| "learning_rate": 1.923529411764706e-05, | |
| "loss": 1.513, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 25.34, | |
| "learning_rate": 1.914705882352941e-05, | |
| "loss": 1.5135, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 25.45, | |
| "learning_rate": 1.9058823529411764e-05, | |
| "loss": 1.5066, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 25.56, | |
| "learning_rate": 1.8970588235294116e-05, | |
| "loss": 1.5087, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 25.68, | |
| "learning_rate": 1.888235294117647e-05, | |
| "loss": 1.5013, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 25.79, | |
| "learning_rate": 1.8794117647058823e-05, | |
| "loss": 1.5059, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 25.9, | |
| "learning_rate": 1.8705882352941178e-05, | |
| "loss": 1.5057, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 1.861764705882353e-05, | |
| "loss": 1.579, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 26.14, | |
| "learning_rate": 1.8529411764705884e-05, | |
| "loss": 1.5009, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 26.25, | |
| "learning_rate": 1.8441176470588236e-05, | |
| "loss": 1.4947, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 26.36, | |
| "learning_rate": 1.835294117647059e-05, | |
| "loss": 1.502, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 26.47, | |
| "learning_rate": 1.826470588235294e-05, | |
| "loss": 1.5006, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 26.59, | |
| "learning_rate": 1.8176470588235294e-05, | |
| "loss": 1.4958, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 26.7, | |
| "learning_rate": 1.8088235294117645e-05, | |
| "loss": 1.5004, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 26.81, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.497, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 26.93, | |
| "learning_rate": 1.7911764705882352e-05, | |
| "loss": 1.5019, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 27.05, | |
| "learning_rate": 1.7823529411764707e-05, | |
| "loss": 1.5746, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 27.16, | |
| "learning_rate": 1.773529411764706e-05, | |
| "loss": 1.4954, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 27.27, | |
| "learning_rate": 1.7647058823529414e-05, | |
| "loss": 1.4968, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 27.38, | |
| "learning_rate": 1.7558823529411765e-05, | |
| "loss": 1.4959, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 27.5, | |
| "learning_rate": 1.747058823529412e-05, | |
| "loss": 1.5, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 27.61, | |
| "learning_rate": 1.7382352941176472e-05, | |
| "loss": 1.4945, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 27.72, | |
| "learning_rate": 1.7294117647058823e-05, | |
| "loss": 1.493, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 27.84, | |
| "learning_rate": 1.7205882352941175e-05, | |
| "loss": 1.5007, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 27.95, | |
| "learning_rate": 1.711764705882353e-05, | |
| "loss": 1.5059, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "learning_rate": 1.702941176470588e-05, | |
| "loss": 1.5709, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 28.18, | |
| "learning_rate": 1.6941176470588237e-05, | |
| "loss": 1.4905, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 28.29, | |
| "learning_rate": 1.6852941176470588e-05, | |
| "loss": 1.4894, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 28.41, | |
| "learning_rate": 1.6764705882352943e-05, | |
| "loss": 1.5007, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 28.52, | |
| "learning_rate": 1.6676470588235295e-05, | |
| "loss": 1.4908, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 28.63, | |
| "learning_rate": 1.658823529411765e-05, | |
| "loss": 1.4912, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 28.75, | |
| "learning_rate": 1.65e-05, | |
| "loss": 1.493, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 28.86, | |
| "learning_rate": 1.6411764705882356e-05, | |
| "loss": 1.4957, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 28.97, | |
| "learning_rate": 1.6323529411764704e-05, | |
| "loss": 1.4919, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 29.09, | |
| "learning_rate": 1.623529411764706e-05, | |
| "loss": 1.5667, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 29.2, | |
| "learning_rate": 1.614705882352941e-05, | |
| "loss": 1.4908, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 29.32, | |
| "learning_rate": 1.6058823529411766e-05, | |
| "loss": 1.4893, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 29.43, | |
| "learning_rate": 1.5970588235294118e-05, | |
| "loss": 1.4898, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 29.54, | |
| "learning_rate": 1.5882352941176473e-05, | |
| "loss": 1.4929, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 29.66, | |
| "learning_rate": 1.5794117647058824e-05, | |
| "loss": 1.4858, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 29.77, | |
| "learning_rate": 1.570588235294118e-05, | |
| "loss": 1.4946, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 29.88, | |
| "learning_rate": 1.561764705882353e-05, | |
| "loss": 1.491, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 29.99, | |
| "learning_rate": 1.5529411764705886e-05, | |
| "loss": 1.4873, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 30.11, | |
| "learning_rate": 1.5441176470588234e-05, | |
| "loss": 1.5672, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 30.23, | |
| "learning_rate": 1.535294117647059e-05, | |
| "loss": 1.4865, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 30.34, | |
| "learning_rate": 1.526470588235294e-05, | |
| "loss": 1.4895, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 30.45, | |
| "learning_rate": 1.5176470588235294e-05, | |
| "loss": 1.485, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 30.56, | |
| "learning_rate": 1.5088235294117647e-05, | |
| "loss": 1.4892, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 30.68, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.4854, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 30.79, | |
| "learning_rate": 1.4911764705882354e-05, | |
| "loss": 1.4862, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 30.9, | |
| "learning_rate": 1.4823529411764707e-05, | |
| "loss": 1.4886, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 1.4735294117647059e-05, | |
| "loss": 1.5558, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 31.14, | |
| "learning_rate": 1.4647058823529412e-05, | |
| "loss": 1.4916, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 1.4558823529411765e-05, | |
| "loss": 1.4892, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 31.36, | |
| "learning_rate": 1.4470588235294118e-05, | |
| "loss": 1.4861, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 31.47, | |
| "learning_rate": 1.4382352941176472e-05, | |
| "loss": 1.4841, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 31.59, | |
| "learning_rate": 1.4294117647058823e-05, | |
| "loss": 1.4834, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 31.7, | |
| "learning_rate": 1.4205882352941177e-05, | |
| "loss": 1.4842, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 31.81, | |
| "learning_rate": 1.411764705882353e-05, | |
| "loss": 1.4834, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 31.93, | |
| "learning_rate": 1.4029411764705883e-05, | |
| "loss": 1.483, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 32.05, | |
| "learning_rate": 1.3941176470588236e-05, | |
| "loss": 1.5567, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 32.16, | |
| "learning_rate": 1.385294117647059e-05, | |
| "loss": 1.4841, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 32.27, | |
| "learning_rate": 1.3764705882352941e-05, | |
| "loss": 1.4858, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 32.38, | |
| "learning_rate": 1.3676470588235295e-05, | |
| "loss": 1.483, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 32.5, | |
| "learning_rate": 1.3588235294117648e-05, | |
| "loss": 1.4781, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 32.61, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 1.4818, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 32.72, | |
| "learning_rate": 1.3411764705882354e-05, | |
| "loss": 1.4802, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 32.84, | |
| "learning_rate": 1.3323529411764706e-05, | |
| "loss": 1.4835, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 32.95, | |
| "learning_rate": 1.323529411764706e-05, | |
| "loss": 1.486, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 33.07, | |
| "learning_rate": 1.3147058823529413e-05, | |
| "loss": 1.5547, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 33.18, | |
| "learning_rate": 1.3058823529411766e-05, | |
| "loss": 1.4813, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 33.29, | |
| "learning_rate": 1.297058823529412e-05, | |
| "loss": 1.4857, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 33.41, | |
| "learning_rate": 1.288235294117647e-05, | |
| "loss": 1.4853, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 33.52, | |
| "learning_rate": 1.2794117647058824e-05, | |
| "loss": 1.4772, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 33.63, | |
| "learning_rate": 1.2705882352941177e-05, | |
| "loss": 1.4826, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 33.75, | |
| "learning_rate": 1.261764705882353e-05, | |
| "loss": 1.484, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 33.86, | |
| "learning_rate": 1.2529411764705884e-05, | |
| "loss": 1.4864, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 33.97, | |
| "learning_rate": 1.2441176470588236e-05, | |
| "loss": 1.4808, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 34.09, | |
| "learning_rate": 1.2352941176470587e-05, | |
| "loss": 1.5574, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 34.09, | |
| "eval_denotation_accuracy": 0.42352525609325326, | |
| "eval_loss": 2.4281280040740967, | |
| "eval_runtime": 1744.8445, | |
| "eval_samples_per_second": 1.622, | |
| "eval_steps_per_second": 0.101, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 34.2, | |
| "learning_rate": 1.226470588235294e-05, | |
| "loss": 1.4832, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 34.32, | |
| "learning_rate": 1.2176470588235294e-05, | |
| "loss": 1.4815, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 34.43, | |
| "learning_rate": 1.2088235294117647e-05, | |
| "loss": 1.4813, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 34.54, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.4826, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 34.66, | |
| "learning_rate": 1.1911764705882352e-05, | |
| "loss": 1.4826, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 34.77, | |
| "learning_rate": 1.1823529411764705e-05, | |
| "loss": 1.4812, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 34.88, | |
| "learning_rate": 1.1735294117647058e-05, | |
| "loss": 1.48, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 34.99, | |
| "learning_rate": 1.1647058823529412e-05, | |
| "loss": 1.4818, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 35.11, | |
| "learning_rate": 1.1558823529411765e-05, | |
| "loss": 1.5577, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 35.23, | |
| "learning_rate": 1.1470588235294117e-05, | |
| "loss": 1.4789, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 35.34, | |
| "learning_rate": 1.138235294117647e-05, | |
| "loss": 1.477, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 35.45, | |
| "learning_rate": 1.1294117647058823e-05, | |
| "loss": 1.4778, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 35.56, | |
| "learning_rate": 1.1205882352941177e-05, | |
| "loss": 1.476, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 35.68, | |
| "learning_rate": 1.111764705882353e-05, | |
| "loss": 1.4799, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 35.79, | |
| "learning_rate": 1.1029411764705883e-05, | |
| "loss": 1.4747, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 35.9, | |
| "learning_rate": 1.0941176470588235e-05, | |
| "loss": 1.4748, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 36.02, | |
| "learning_rate": 1.0852941176470588e-05, | |
| "loss": 1.5561, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 36.14, | |
| "learning_rate": 1.0764705882352941e-05, | |
| "loss": 1.4779, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 36.25, | |
| "learning_rate": 1.0676470588235295e-05, | |
| "loss": 1.4774, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 36.36, | |
| "learning_rate": 1.0588235294117648e-05, | |
| "loss": 1.4743, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 36.47, | |
| "learning_rate": 1.05e-05, | |
| "loss": 1.4765, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 36.59, | |
| "learning_rate": 1.0411764705882353e-05, | |
| "loss": 1.4755, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 36.7, | |
| "learning_rate": 1.0323529411764706e-05, | |
| "loss": 1.4796, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 36.81, | |
| "learning_rate": 1.023529411764706e-05, | |
| "loss": 1.4723, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 36.93, | |
| "learning_rate": 1.0147058823529413e-05, | |
| "loss": 1.4781, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 37.05, | |
| "learning_rate": 1.0058823529411764e-05, | |
| "loss": 1.5533, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 37.16, | |
| "learning_rate": 9.970588235294117e-06, | |
| "loss": 1.4733, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 37.27, | |
| "learning_rate": 9.88235294117647e-06, | |
| "loss": 1.4761, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 37.38, | |
| "learning_rate": 9.794117647058824e-06, | |
| "loss": 1.4738, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 9.705882352941177e-06, | |
| "loss": 1.4731, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 37.61, | |
| "learning_rate": 9.61764705882353e-06, | |
| "loss": 1.4733, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 37.72, | |
| "learning_rate": 9.529411764705882e-06, | |
| "loss": 1.4738, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 37.84, | |
| "learning_rate": 9.441176470588235e-06, | |
| "loss": 1.4742, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 37.95, | |
| "learning_rate": 9.352941176470589e-06, | |
| "loss": 1.4721, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 38.07, | |
| "learning_rate": 9.264705882352942e-06, | |
| "loss": 1.5433, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 38.18, | |
| "learning_rate": 9.176470588235295e-06, | |
| "loss": 1.4746, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 38.29, | |
| "learning_rate": 9.088235294117647e-06, | |
| "loss": 1.4754, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 38.41, | |
| "learning_rate": 9e-06, | |
| "loss": 1.4758, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 38.52, | |
| "learning_rate": 8.911764705882354e-06, | |
| "loss": 1.4705, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 38.63, | |
| "learning_rate": 8.823529411764707e-06, | |
| "loss": 1.4709, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 38.75, | |
| "learning_rate": 8.73529411764706e-06, | |
| "loss": 1.4717, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 38.86, | |
| "learning_rate": 8.647058823529412e-06, | |
| "loss": 1.4741, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 38.97, | |
| "learning_rate": 8.558823529411765e-06, | |
| "loss": 1.47, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 39.09, | |
| "learning_rate": 8.470588235294118e-06, | |
| "loss": 1.543, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 39.2, | |
| "learning_rate": 8.382352941176472e-06, | |
| "loss": 1.4719, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 39.32, | |
| "learning_rate": 8.294117647058825e-06, | |
| "loss": 1.4715, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 39.43, | |
| "learning_rate": 8.205882352941178e-06, | |
| "loss": 1.4681, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 39.54, | |
| "learning_rate": 8.11764705882353e-06, | |
| "loss": 1.4733, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 39.66, | |
| "learning_rate": 8.029411764705883e-06, | |
| "loss": 1.4716, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 39.77, | |
| "learning_rate": 7.941176470588236e-06, | |
| "loss": 1.4694, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 39.88, | |
| "learning_rate": 7.85294117647059e-06, | |
| "loss": 1.4726, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 39.99, | |
| "learning_rate": 7.764705882352943e-06, | |
| "loss": 1.4738, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 40.11, | |
| "learning_rate": 7.676470588235294e-06, | |
| "loss": 1.5443, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 40.23, | |
| "learning_rate": 7.588235294117647e-06, | |
| "loss": 1.4713, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 40.34, | |
| "learning_rate": 7.5e-06, | |
| "loss": 1.4673, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 40.45, | |
| "learning_rate": 7.4117647058823535e-06, | |
| "loss": 1.4705, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 40.56, | |
| "learning_rate": 7.323529411764706e-06, | |
| "loss": 1.4695, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 40.68, | |
| "learning_rate": 7.235294117647059e-06, | |
| "loss": 1.4678, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 40.79, | |
| "learning_rate": 7.147058823529412e-06, | |
| "loss": 1.4679, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 40.9, | |
| "learning_rate": 7.058823529411765e-06, | |
| "loss": 1.4693, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 41.02, | |
| "learning_rate": 6.970588235294118e-06, | |
| "loss": 1.5428, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 41.14, | |
| "learning_rate": 6.882352941176471e-06, | |
| "loss": 1.4741, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 41.25, | |
| "learning_rate": 6.794117647058824e-06, | |
| "loss": 1.4677, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 41.36, | |
| "learning_rate": 6.705882352941177e-06, | |
| "loss": 1.4695, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 41.47, | |
| "learning_rate": 6.61764705882353e-06, | |
| "loss": 1.4659, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 41.59, | |
| "learning_rate": 6.529411764705883e-06, | |
| "loss": 1.4693, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 41.7, | |
| "learning_rate": 6.441176470588235e-06, | |
| "loss": 1.469, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 41.81, | |
| "learning_rate": 6.352941176470589e-06, | |
| "loss": 1.466, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 41.93, | |
| "learning_rate": 6.264705882352942e-06, | |
| "loss": 1.4688, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 42.05, | |
| "learning_rate": 6.176470588235294e-06, | |
| "loss": 1.5432, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 42.16, | |
| "learning_rate": 6.088235294117647e-06, | |
| "loss": 1.4697, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 42.27, | |
| "learning_rate": 6e-06, | |
| "loss": 1.4696, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 42.38, | |
| "learning_rate": 5.911764705882353e-06, | |
| "loss": 1.4661, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 42.5, | |
| "learning_rate": 5.823529411764706e-06, | |
| "loss": 1.4701, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 42.61, | |
| "learning_rate": 5.735294117647058e-06, | |
| "loss": 1.4664, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 42.72, | |
| "learning_rate": 5.647058823529412e-06, | |
| "loss": 1.4661, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 42.84, | |
| "learning_rate": 5.558823529411765e-06, | |
| "loss": 1.4668, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 42.95, | |
| "learning_rate": 5.470588235294117e-06, | |
| "loss": 1.4688, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 43.07, | |
| "learning_rate": 5.382352941176471e-06, | |
| "loss": 1.5398, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 43.18, | |
| "learning_rate": 5.294117647058824e-06, | |
| "loss": 1.4654, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 43.29, | |
| "learning_rate": 5.205882352941176e-06, | |
| "loss": 1.4637, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 43.41, | |
| "learning_rate": 5.11764705882353e-06, | |
| "loss": 1.4706, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 43.52, | |
| "learning_rate": 5.029411764705882e-06, | |
| "loss": 1.4638, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 43.63, | |
| "learning_rate": 4.941176470588235e-06, | |
| "loss": 1.4671, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 43.75, | |
| "learning_rate": 4.852941176470589e-06, | |
| "loss": 1.4681, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 43.86, | |
| "learning_rate": 4.764705882352941e-06, | |
| "loss": 1.467, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 43.97, | |
| "learning_rate": 4.676470588235294e-06, | |
| "loss": 1.4666, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 44.09, | |
| "learning_rate": 4.588235294117648e-06, | |
| "loss": 1.5401, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 44.2, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.4642, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 44.32, | |
| "learning_rate": 4.411764705882353e-06, | |
| "loss": 1.466, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 44.43, | |
| "learning_rate": 4.323529411764706e-06, | |
| "loss": 1.4647, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 44.54, | |
| "learning_rate": 4.235294117647059e-06, | |
| "loss": 1.4689, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 44.66, | |
| "learning_rate": 4.147058823529412e-06, | |
| "loss": 1.4642, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 44.77, | |
| "learning_rate": 4.058823529411765e-06, | |
| "loss": 1.4696, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 44.88, | |
| "learning_rate": 3.970588235294118e-06, | |
| "loss": 1.4666, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 44.99, | |
| "learning_rate": 3.8823529411764714e-06, | |
| "loss": 1.4678, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 45.11, | |
| "learning_rate": 3.7941176470588235e-06, | |
| "loss": 1.5383, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 45.23, | |
| "learning_rate": 3.7058823529411767e-06, | |
| "loss": 1.4664, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 45.34, | |
| "learning_rate": 3.6176470588235296e-06, | |
| "loss": 1.4645, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 45.45, | |
| "learning_rate": 3.5294117647058825e-06, | |
| "loss": 1.4656, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 45.45, | |
| "eval_denotation_accuracy": 0.4959378311550689, | |
| "eval_loss": 2.4232473373413086, | |
| "eval_runtime": 1762.2635, | |
| "eval_samples_per_second": 1.606, | |
| "eval_steps_per_second": 0.1, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 45.56, | |
| "learning_rate": 3.4411764705882353e-06, | |
| "loss": 1.4646, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 45.68, | |
| "learning_rate": 3.3529411764705886e-06, | |
| "loss": 1.4656, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 45.79, | |
| "learning_rate": 3.2647058823529415e-06, | |
| "loss": 1.4653, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 45.9, | |
| "learning_rate": 3.1764705882352943e-06, | |
| "loss": 1.4659, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 46.02, | |
| "learning_rate": 3.088235294117647e-06, | |
| "loss": 1.5391, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 46.14, | |
| "learning_rate": 3e-06, | |
| "loss": 1.4667, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 46.25, | |
| "learning_rate": 2.911764705882353e-06, | |
| "loss": 1.4653, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 46.36, | |
| "learning_rate": 2.823529411764706e-06, | |
| "loss": 1.4646, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 46.47, | |
| "learning_rate": 2.7352941176470587e-06, | |
| "loss": 1.4662, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 46.59, | |
| "learning_rate": 2.647058823529412e-06, | |
| "loss": 1.4678, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 46.7, | |
| "learning_rate": 2.558823529411765e-06, | |
| "loss": 1.4651, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 46.81, | |
| "learning_rate": 2.4705882352941177e-06, | |
| "loss": 1.4628, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 46.93, | |
| "learning_rate": 2.3823529411764705e-06, | |
| "loss": 1.4637, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 47.05, | |
| "learning_rate": 2.294117647058824e-06, | |
| "loss": 1.5399, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 47.16, | |
| "learning_rate": 2.2058823529411767e-06, | |
| "loss": 1.4615, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 47.27, | |
| "learning_rate": 2.1176470588235296e-06, | |
| "loss": 1.4641, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 47.38, | |
| "learning_rate": 2.0294117647058824e-06, | |
| "loss": 1.4653, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 47.5, | |
| "learning_rate": 1.9411764705882357e-06, | |
| "loss": 1.4626, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 47.61, | |
| "learning_rate": 1.8529411764705884e-06, | |
| "loss": 1.4606, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 47.72, | |
| "learning_rate": 1.7647058823529412e-06, | |
| "loss": 1.4667, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 47.84, | |
| "learning_rate": 1.6764705882352943e-06, | |
| "loss": 1.4606, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 47.95, | |
| "learning_rate": 1.5882352941176472e-06, | |
| "loss": 1.4652, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 48.07, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.5363, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 48.18, | |
| "learning_rate": 1.411764705882353e-06, | |
| "loss": 1.4629, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 48.29, | |
| "learning_rate": 1.323529411764706e-06, | |
| "loss": 1.4609, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 48.41, | |
| "learning_rate": 1.2352941176470588e-06, | |
| "loss": 1.4615, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 48.52, | |
| "learning_rate": 1.147058823529412e-06, | |
| "loss": 1.4631, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 48.63, | |
| "learning_rate": 1.0588235294117648e-06, | |
| "loss": 1.4639, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 48.75, | |
| "learning_rate": 9.705882352941179e-07, | |
| "loss": 1.4647, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 48.86, | |
| "learning_rate": 8.823529411764706e-07, | |
| "loss": 1.4614, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 48.97, | |
| "learning_rate": 7.941176470588236e-07, | |
| "loss": 1.4641, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 49.09, | |
| "learning_rate": 7.058823529411765e-07, | |
| "loss": 1.537, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 49.2, | |
| "learning_rate": 6.176470588235294e-07, | |
| "loss": 1.4627, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 49.32, | |
| "learning_rate": 5.294117647058824e-07, | |
| "loss": 1.4662, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 49.43, | |
| "learning_rate": 4.411764705882353e-07, | |
| "loss": 1.4648, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 49.54, | |
| "learning_rate": 3.529411764705882e-07, | |
| "loss": 1.4633, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 49.66, | |
| "learning_rate": 2.647058823529412e-07, | |
| "loss": 1.4631, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 49.77, | |
| "learning_rate": 1.764705882352941e-07, | |
| "loss": 1.4604, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 49.88, | |
| "learning_rate": 8.823529411764706e-08, | |
| "loss": 1.4621, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 49.99, | |
| "learning_rate": 0.0, | |
| "loss": 1.4644, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 49.99, | |
| "step": 4400, | |
| "total_flos": 1.218502107627307e+18, | |
| "train_loss": 1.6152976189960133, | |
| "train_runtime": 26687.1063, | |
| "train_samples_per_second": 21.211, | |
| "train_steps_per_second": 0.165 | |
| } | |
| ], | |
| "max_steps": 4400, | |
| "num_train_epochs": 50, | |
| "total_flos": 1.218502107627307e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |