|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 56.65559872836454, |
|
"global_step": 20000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 3.0000000000000004e-07, |
|
"loss": 14.1422, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 6.000000000000001e-07, |
|
"loss": 9.7115, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9e-07, |
|
"loss": 6.8111, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.2000000000000002e-06, |
|
"loss": 5.2187, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.5e-06, |
|
"loss": 4.6434, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.8e-06, |
|
"loss": 4.2415, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 2.1000000000000002e-06, |
|
"loss": 3.7669, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 3.6312, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.7e-06, |
|
"loss": 3.374, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 3e-06, |
|
"loss": 3.1393, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 3.3e-06, |
|
"loss": 2.9762, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 3.6e-06, |
|
"loss": 2.9393, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 3.9e-06, |
|
"loss": 2.814, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.2000000000000004e-06, |
|
"loss": 2.7919, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 4.5e-06, |
|
"loss": 2.7482, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 2.7322, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 5.1e-06, |
|
"loss": 2.6329, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 5.4e-06, |
|
"loss": 2.5376, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 5.7000000000000005e-06, |
|
"loss": 2.5884, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 6e-06, |
|
"loss": 2.499, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 6.3e-06, |
|
"loss": 2.4366, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 6.6e-06, |
|
"loss": 2.575, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.900000000000001e-06, |
|
"loss": 2.4937, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 7.2e-06, |
|
"loss": 2.4204, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 7.5e-06, |
|
"loss": 2.3175, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 7.8e-06, |
|
"loss": 2.4137, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 8.1e-06, |
|
"loss": 2.3216, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 2.2545, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 8.7e-06, |
|
"loss": 2.2574, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 9e-06, |
|
"loss": 2.2952, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 9.3e-06, |
|
"loss": 2.3082, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 9.600000000000001e-06, |
|
"loss": 2.3056, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 9.9e-06, |
|
"loss": 2.2697, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.02e-05, |
|
"loss": 2.2188, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.05e-05, |
|
"loss": 2.2091, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 1.08e-05, |
|
"loss": 2.3702, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 1.11e-05, |
|
"loss": 2.2407, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 1.1400000000000001e-05, |
|
"loss": 2.1957, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 1.1700000000000001e-05, |
|
"loss": 2.2061, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 1.2e-05, |
|
"loss": 2.1418, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 1.2299999999999999e-05, |
|
"loss": 2.1444, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 1.26e-05, |
|
"loss": 2.092, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 1.29e-05, |
|
"loss": 2.1546, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 1.32e-05, |
|
"loss": 2.2028, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 1.3500000000000001e-05, |
|
"loss": 2.1954, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 1.3800000000000002e-05, |
|
"loss": 2.0972, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 1.4099999999999999e-05, |
|
"loss": 2.1261, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 1.44e-05, |
|
"loss": 2.2221, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 1.47e-05, |
|
"loss": 2.164, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 1.5e-05, |
|
"loss": 2.1335, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 1.53e-05, |
|
"loss": 2.0772, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 1.56e-05, |
|
"loss": 2.0248, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 1.59e-05, |
|
"loss": 2.0828, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 1.62e-05, |
|
"loss": 2.1405, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 1.65e-05, |
|
"loss": 2.1052, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 1.6800000000000002e-05, |
|
"loss": 2.0947, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 1.71e-05, |
|
"loss": 2.1731, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 1.74e-05, |
|
"loss": 2.0155, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 1.77e-05, |
|
"loss": 2.0695, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 1.8e-05, |
|
"loss": 2.1026, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 1.83e-05, |
|
"loss": 2.1209, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 1.86e-05, |
|
"loss": 2.0839, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 1.8900000000000002e-05, |
|
"loss": 2.1547, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 1.9200000000000003e-05, |
|
"loss": 2.0353, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 1.95e-05, |
|
"loss": 2.091, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 1.98e-05, |
|
"loss": 2.0464, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 2.01e-05, |
|
"loss": 1.9735, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 2.04e-05, |
|
"loss": 2.0142, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 2.07e-05, |
|
"loss": 2.0526, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 2.1e-05, |
|
"loss": 2.1198, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 2.13e-05, |
|
"loss": 2.1764, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 2.16e-05, |
|
"loss": 1.9925, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 2.19e-05, |
|
"loss": 1.9689, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 2.22e-05, |
|
"loss": 1.8489, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 2.25e-05, |
|
"loss": 1.9314, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 2.2800000000000002e-05, |
|
"loss": 1.8814, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.3100000000000002e-05, |
|
"loss": 1.9512, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 2.3400000000000003e-05, |
|
"loss": 1.9251, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 2.37e-05, |
|
"loss": 1.9665, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 2.4e-05, |
|
"loss": 2.0072, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 2.43e-05, |
|
"loss": 2.0209, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 2.4599999999999998e-05, |
|
"loss": 1.9674, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 2.49e-05, |
|
"loss": 1.895, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 2.52e-05, |
|
"loss": 1.9893, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 2.55e-05, |
|
"loss": 2.0037, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 2.58e-05, |
|
"loss": 2.0346, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 2.61e-05, |
|
"loss": 2.0051, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 2.64e-05, |
|
"loss": 2.0522, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 2.52, |
|
"learning_rate": 2.6700000000000002e-05, |
|
"loss": 1.984, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 2.7000000000000002e-05, |
|
"loss": 1.9341, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 2.7300000000000003e-05, |
|
"loss": 2.052, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 2.7600000000000003e-05, |
|
"loss": 2.0514, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 2.79e-05, |
|
"loss": 2.0879, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 2.8199999999999998e-05, |
|
"loss": 2.0271, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 2.8499999999999998e-05, |
|
"loss": 2.022, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 2.88e-05, |
|
"loss": 1.9329, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 2.91e-05, |
|
"loss": 2.0669, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 2.94e-05, |
|
"loss": 1.9011, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 2.97e-05, |
|
"loss": 1.9225, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 3e-05, |
|
"loss": 1.9818, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"eval_denotation_accuracy": 0.5192511480042388, |
|
"eval_loss": 2.01488995552063, |
|
"eval_runtime": 563.2088, |
|
"eval_samples_per_second": 5.027, |
|
"eval_steps_per_second": 1.257, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 2.998421052631579e-05, |
|
"loss": 1.9532, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 2.9968421052631578e-05, |
|
"loss": 1.9891, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 2.92, |
|
"learning_rate": 2.9952631578947372e-05, |
|
"loss": 1.9292, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 2.993684210526316e-05, |
|
"loss": 2.0779, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 2.992105263157895e-05, |
|
"loss": 1.9649, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 2.9905263157894736e-05, |
|
"loss": 2.1161, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 2.988947368421053e-05, |
|
"loss": 1.8847, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 2.9873684210526317e-05, |
|
"loss": 1.8701, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 2.9857894736842107e-05, |
|
"loss": 1.9537, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 2.9842105263157894e-05, |
|
"loss": 1.8578, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"learning_rate": 2.9826315789473684e-05, |
|
"loss": 1.8608, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"learning_rate": 2.9810526315789475e-05, |
|
"loss": 1.8986, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 3.2, |
|
"learning_rate": 2.9794736842105265e-05, |
|
"loss": 1.8624, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 3.23, |
|
"learning_rate": 2.9778947368421052e-05, |
|
"loss": 1.9263, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 3.26, |
|
"learning_rate": 2.9763157894736842e-05, |
|
"loss": 1.8791, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 3.29, |
|
"learning_rate": 2.974736842105263e-05, |
|
"loss": 1.8547, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 3.31, |
|
"learning_rate": 2.9731578947368423e-05, |
|
"loss": 1.8435, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 3.34, |
|
"learning_rate": 2.9715789473684213e-05, |
|
"loss": 1.9111, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 3.37, |
|
"learning_rate": 2.97e-05, |
|
"loss": 1.8597, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 3.4, |
|
"learning_rate": 2.968421052631579e-05, |
|
"loss": 1.872, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 3.43, |
|
"learning_rate": 2.966842105263158e-05, |
|
"loss": 1.861, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 3.45, |
|
"learning_rate": 2.965263157894737e-05, |
|
"loss": 1.8506, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 3.48, |
|
"learning_rate": 2.9636842105263158e-05, |
|
"loss": 1.9002, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 3.51, |
|
"learning_rate": 2.962105263157895e-05, |
|
"loss": 1.7997, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 3.54, |
|
"learning_rate": 2.9605263157894735e-05, |
|
"loss": 1.846, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 3.57, |
|
"learning_rate": 2.958947368421053e-05, |
|
"loss": 1.9717, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 3.6, |
|
"learning_rate": 2.9573684210526316e-05, |
|
"loss": 1.8678, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 3.62, |
|
"learning_rate": 2.9557894736842107e-05, |
|
"loss": 1.8516, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 3.65, |
|
"learning_rate": 2.9542105263157893e-05, |
|
"loss": 1.8293, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 3.68, |
|
"learning_rate": 2.9526315789473684e-05, |
|
"loss": 1.9229, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 3.71, |
|
"learning_rate": 2.9510526315789474e-05, |
|
"loss": 1.9239, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 3.74, |
|
"learning_rate": 2.9494736842105264e-05, |
|
"loss": 1.8524, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 3.77, |
|
"learning_rate": 2.947894736842105e-05, |
|
"loss": 1.9316, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 3.79, |
|
"learning_rate": 2.9463157894736842e-05, |
|
"loss": 1.8367, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 3.82, |
|
"learning_rate": 2.9447368421052635e-05, |
|
"loss": 1.8685, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 3.85, |
|
"learning_rate": 2.9431578947368422e-05, |
|
"loss": 1.8507, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 3.88, |
|
"learning_rate": 2.9415789473684213e-05, |
|
"loss": 1.8836, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 3.91, |
|
"learning_rate": 2.94e-05, |
|
"loss": 1.9116, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 3.94, |
|
"learning_rate": 2.938421052631579e-05, |
|
"loss": 1.8836, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 3.96, |
|
"learning_rate": 2.936842105263158e-05, |
|
"loss": 1.8195, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 3.99, |
|
"learning_rate": 2.935263157894737e-05, |
|
"loss": 1.9163, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 2.9336842105263158e-05, |
|
"loss": 1.912, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 4.05, |
|
"learning_rate": 2.9321052631578948e-05, |
|
"loss": 1.7437, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 2.9305263157894735e-05, |
|
"loss": 1.7872, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 4.11, |
|
"learning_rate": 2.928947368421053e-05, |
|
"loss": 1.7588, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 4.14, |
|
"learning_rate": 2.9273684210526316e-05, |
|
"loss": 1.7983, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 2.9257894736842106e-05, |
|
"loss": 1.8219, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 4.19, |
|
"learning_rate": 2.9242105263157893e-05, |
|
"loss": 1.8267, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 4.22, |
|
"learning_rate": 2.9226315789473687e-05, |
|
"loss": 1.8059, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 4.25, |
|
"learning_rate": 2.9210526315789474e-05, |
|
"loss": 1.7878, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 2.9194736842105264e-05, |
|
"loss": 1.8023, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 4.31, |
|
"learning_rate": 2.9178947368421054e-05, |
|
"loss": 1.7739, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 4.33, |
|
"learning_rate": 2.916315789473684e-05, |
|
"loss": 1.7929, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 4.36, |
|
"learning_rate": 2.9147368421052635e-05, |
|
"loss": 1.8113, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 4.39, |
|
"learning_rate": 2.9131578947368422e-05, |
|
"loss": 1.811, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 4.42, |
|
"learning_rate": 2.9115789473684212e-05, |
|
"loss": 1.7617, |
|
"step": 1560 |
|
}, |
|
{ |
|
"epoch": 4.45, |
|
"learning_rate": 2.91e-05, |
|
"loss": 1.837, |
|
"step": 1570 |
|
}, |
|
{ |
|
"epoch": 4.47, |
|
"learning_rate": 2.908421052631579e-05, |
|
"loss": 1.8516, |
|
"step": 1580 |
|
}, |
|
{ |
|
"epoch": 4.5, |
|
"learning_rate": 2.906842105263158e-05, |
|
"loss": 1.8067, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 4.53, |
|
"learning_rate": 2.905263157894737e-05, |
|
"loss": 1.7737, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 4.56, |
|
"learning_rate": 2.9036842105263157e-05, |
|
"loss": 1.789, |
|
"step": 1610 |
|
}, |
|
{ |
|
"epoch": 4.59, |
|
"learning_rate": 2.9021052631578948e-05, |
|
"loss": 1.8488, |
|
"step": 1620 |
|
}, |
|
{ |
|
"epoch": 4.62, |
|
"learning_rate": 2.9005263157894738e-05, |
|
"loss": 1.8083, |
|
"step": 1630 |
|
}, |
|
{ |
|
"epoch": 4.64, |
|
"learning_rate": 2.8989473684210528e-05, |
|
"loss": 1.8181, |
|
"step": 1640 |
|
}, |
|
{ |
|
"epoch": 4.67, |
|
"learning_rate": 2.8973684210526315e-05, |
|
"loss": 1.7633, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 4.7, |
|
"learning_rate": 2.8957894736842105e-05, |
|
"loss": 1.7521, |
|
"step": 1660 |
|
}, |
|
{ |
|
"epoch": 4.73, |
|
"learning_rate": 2.8942105263157896e-05, |
|
"loss": 1.7865, |
|
"step": 1670 |
|
}, |
|
{ |
|
"epoch": 4.76, |
|
"learning_rate": 2.8926315789473686e-05, |
|
"loss": 1.7808, |
|
"step": 1680 |
|
}, |
|
{ |
|
"epoch": 4.79, |
|
"learning_rate": 2.8910526315789476e-05, |
|
"loss": 1.9234, |
|
"step": 1690 |
|
}, |
|
{ |
|
"epoch": 4.81, |
|
"learning_rate": 2.8894736842105263e-05, |
|
"loss": 1.7825, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 4.84, |
|
"learning_rate": 2.8878947368421054e-05, |
|
"loss": 1.7962, |
|
"step": 1710 |
|
}, |
|
{ |
|
"epoch": 4.87, |
|
"learning_rate": 2.886315789473684e-05, |
|
"loss": 1.7204, |
|
"step": 1720 |
|
}, |
|
{ |
|
"epoch": 4.9, |
|
"learning_rate": 2.8847368421052634e-05, |
|
"loss": 1.7531, |
|
"step": 1730 |
|
}, |
|
{ |
|
"epoch": 4.93, |
|
"learning_rate": 2.883157894736842e-05, |
|
"loss": 1.772, |
|
"step": 1740 |
|
}, |
|
{ |
|
"epoch": 4.96, |
|
"learning_rate": 2.8815789473684212e-05, |
|
"loss": 1.8145, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 4.98, |
|
"learning_rate": 2.88e-05, |
|
"loss": 1.8345, |
|
"step": 1760 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 2.8784210526315792e-05, |
|
"loss": 1.8886, |
|
"step": 1770 |
|
}, |
|
{ |
|
"epoch": 5.04, |
|
"learning_rate": 2.876842105263158e-05, |
|
"loss": 1.7124, |
|
"step": 1780 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 2.875263157894737e-05, |
|
"loss": 1.7388, |
|
"step": 1790 |
|
}, |
|
{ |
|
"epoch": 5.1, |
|
"learning_rate": 2.8736842105263157e-05, |
|
"loss": 1.7866, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 5.13, |
|
"learning_rate": 2.8721052631578947e-05, |
|
"loss": 1.7238, |
|
"step": 1810 |
|
}, |
|
{ |
|
"epoch": 5.16, |
|
"learning_rate": 2.8705263157894737e-05, |
|
"loss": 1.687, |
|
"step": 1820 |
|
}, |
|
{ |
|
"epoch": 5.18, |
|
"learning_rate": 2.8689473684210528e-05, |
|
"loss": 1.7261, |
|
"step": 1830 |
|
}, |
|
{ |
|
"epoch": 5.21, |
|
"learning_rate": 2.8673684210526318e-05, |
|
"loss": 1.704, |
|
"step": 1840 |
|
}, |
|
{ |
|
"epoch": 5.24, |
|
"learning_rate": 2.8657894736842105e-05, |
|
"loss": 1.7616, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 5.27, |
|
"learning_rate": 2.8642105263157895e-05, |
|
"loss": 1.7323, |
|
"step": 1860 |
|
}, |
|
{ |
|
"epoch": 5.3, |
|
"learning_rate": 2.8626315789473686e-05, |
|
"loss": 1.8417, |
|
"step": 1870 |
|
}, |
|
{ |
|
"epoch": 5.32, |
|
"learning_rate": 2.8610526315789476e-05, |
|
"loss": 1.7533, |
|
"step": 1880 |
|
}, |
|
{ |
|
"epoch": 5.35, |
|
"learning_rate": 2.8594736842105263e-05, |
|
"loss": 1.7302, |
|
"step": 1890 |
|
}, |
|
{ |
|
"epoch": 5.38, |
|
"learning_rate": 2.8578947368421053e-05, |
|
"loss": 1.7701, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 5.41, |
|
"learning_rate": 2.8563157894736844e-05, |
|
"loss": 1.8397, |
|
"step": 1910 |
|
}, |
|
{ |
|
"epoch": 5.44, |
|
"learning_rate": 2.8547368421052634e-05, |
|
"loss": 1.7497, |
|
"step": 1920 |
|
}, |
|
{ |
|
"epoch": 5.47, |
|
"learning_rate": 2.853157894736842e-05, |
|
"loss": 1.7129, |
|
"step": 1930 |
|
}, |
|
{ |
|
"epoch": 5.49, |
|
"learning_rate": 2.851578947368421e-05, |
|
"loss": 1.7424, |
|
"step": 1940 |
|
}, |
|
{ |
|
"epoch": 5.52, |
|
"learning_rate": 2.8499999999999998e-05, |
|
"loss": 1.7278, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 5.55, |
|
"learning_rate": 2.8484210526315792e-05, |
|
"loss": 1.7791, |
|
"step": 1960 |
|
}, |
|
{ |
|
"epoch": 5.58, |
|
"learning_rate": 2.846842105263158e-05, |
|
"loss": 1.7271, |
|
"step": 1970 |
|
}, |
|
{ |
|
"epoch": 5.61, |
|
"learning_rate": 2.845263157894737e-05, |
|
"loss": 1.7126, |
|
"step": 1980 |
|
}, |
|
{ |
|
"epoch": 5.64, |
|
"learning_rate": 2.8436842105263156e-05, |
|
"loss": 1.6985, |
|
"step": 1990 |
|
}, |
|
{ |
|
"epoch": 5.66, |
|
"learning_rate": 2.8421052631578946e-05, |
|
"loss": 1.7464, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 5.66, |
|
"eval_denotation_accuracy": 0.5524549629106322, |
|
"eval_loss": 1.9903446435928345, |
|
"eval_runtime": 409.056, |
|
"eval_samples_per_second": 6.921, |
|
"eval_steps_per_second": 1.731, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 5.69, |
|
"learning_rate": 2.840526315789474e-05, |
|
"loss": 1.8094, |
|
"step": 2010 |
|
}, |
|
{ |
|
"epoch": 5.72, |
|
"learning_rate": 2.8389473684210527e-05, |
|
"loss": 1.6919, |
|
"step": 2020 |
|
}, |
|
{ |
|
"epoch": 5.75, |
|
"learning_rate": 2.8373684210526317e-05, |
|
"loss": 1.7333, |
|
"step": 2030 |
|
}, |
|
{ |
|
"epoch": 5.78, |
|
"learning_rate": 2.8357894736842104e-05, |
|
"loss": 1.7074, |
|
"step": 2040 |
|
}, |
|
{ |
|
"epoch": 5.81, |
|
"learning_rate": 2.8342105263157898e-05, |
|
"loss": 1.6483, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 5.83, |
|
"learning_rate": 2.8326315789473685e-05, |
|
"loss": 1.69, |
|
"step": 2060 |
|
}, |
|
{ |
|
"epoch": 5.86, |
|
"learning_rate": 2.8310526315789475e-05, |
|
"loss": 1.7012, |
|
"step": 2070 |
|
}, |
|
{ |
|
"epoch": 5.89, |
|
"learning_rate": 2.8294736842105262e-05, |
|
"loss": 1.7322, |
|
"step": 2080 |
|
}, |
|
{ |
|
"epoch": 5.92, |
|
"learning_rate": 2.8278947368421053e-05, |
|
"loss": 1.7568, |
|
"step": 2090 |
|
}, |
|
{ |
|
"epoch": 5.95, |
|
"learning_rate": 2.8263157894736843e-05, |
|
"loss": 1.7373, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 5.97, |
|
"learning_rate": 2.8247368421052633e-05, |
|
"loss": 1.7581, |
|
"step": 2110 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"learning_rate": 2.823157894736842e-05, |
|
"loss": 1.869, |
|
"step": 2120 |
|
}, |
|
{ |
|
"epoch": 6.03, |
|
"learning_rate": 2.821578947368421e-05, |
|
"loss": 1.7298, |
|
"step": 2130 |
|
}, |
|
{ |
|
"epoch": 6.06, |
|
"learning_rate": 2.8199999999999998e-05, |
|
"loss": 1.6599, |
|
"step": 2140 |
|
}, |
|
{ |
|
"epoch": 6.09, |
|
"learning_rate": 2.818421052631579e-05, |
|
"loss": 1.7136, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 6.12, |
|
"learning_rate": 2.816842105263158e-05, |
|
"loss": 1.6803, |
|
"step": 2160 |
|
}, |
|
{ |
|
"epoch": 6.15, |
|
"learning_rate": 2.815263157894737e-05, |
|
"loss": 1.6567, |
|
"step": 2170 |
|
}, |
|
{ |
|
"epoch": 6.18, |
|
"learning_rate": 2.813684210526316e-05, |
|
"loss": 1.6761, |
|
"step": 2180 |
|
}, |
|
{ |
|
"epoch": 6.2, |
|
"learning_rate": 2.812105263157895e-05, |
|
"loss": 1.6549, |
|
"step": 2190 |
|
}, |
|
{ |
|
"epoch": 6.23, |
|
"learning_rate": 2.810526315789474e-05, |
|
"loss": 1.6628, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 6.26, |
|
"learning_rate": 2.8089473684210527e-05, |
|
"loss": 1.7003, |
|
"step": 2210 |
|
}, |
|
{ |
|
"epoch": 6.29, |
|
"learning_rate": 2.8073684210526317e-05, |
|
"loss": 1.7405, |
|
"step": 2220 |
|
}, |
|
{ |
|
"epoch": 6.32, |
|
"learning_rate": 2.8057894736842104e-05, |
|
"loss": 1.741, |
|
"step": 2230 |
|
}, |
|
{ |
|
"epoch": 6.34, |
|
"learning_rate": 2.8042105263157898e-05, |
|
"loss": 1.6976, |
|
"step": 2240 |
|
}, |
|
{ |
|
"epoch": 6.37, |
|
"learning_rate": 2.8026315789473685e-05, |
|
"loss": 1.6678, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 6.4, |
|
"learning_rate": 2.8010526315789475e-05, |
|
"loss": 1.688, |
|
"step": 2260 |
|
}, |
|
{ |
|
"epoch": 6.43, |
|
"learning_rate": 2.7994736842105262e-05, |
|
"loss": 1.6566, |
|
"step": 2270 |
|
}, |
|
{ |
|
"epoch": 6.46, |
|
"learning_rate": 2.7978947368421052e-05, |
|
"loss": 1.6851, |
|
"step": 2280 |
|
}, |
|
{ |
|
"epoch": 6.49, |
|
"learning_rate": 2.7963157894736843e-05, |
|
"loss": 1.6712, |
|
"step": 2290 |
|
}, |
|
{ |
|
"epoch": 6.51, |
|
"learning_rate": 2.7947368421052633e-05, |
|
"loss": 1.6816, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 6.54, |
|
"learning_rate": 2.793157894736842e-05, |
|
"loss": 1.6864, |
|
"step": 2310 |
|
}, |
|
{ |
|
"epoch": 6.57, |
|
"learning_rate": 2.791578947368421e-05, |
|
"loss": 1.7062, |
|
"step": 2320 |
|
}, |
|
{ |
|
"epoch": 6.6, |
|
"learning_rate": 2.79e-05, |
|
"loss": 1.6764, |
|
"step": 2330 |
|
}, |
|
{ |
|
"epoch": 6.63, |
|
"learning_rate": 2.788421052631579e-05, |
|
"loss": 1.6874, |
|
"step": 2340 |
|
}, |
|
{ |
|
"epoch": 6.66, |
|
"learning_rate": 2.786842105263158e-05, |
|
"loss": 1.6939, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 6.68, |
|
"learning_rate": 2.7852631578947368e-05, |
|
"loss": 1.6551, |
|
"step": 2360 |
|
}, |
|
{ |
|
"epoch": 6.71, |
|
"learning_rate": 2.783684210526316e-05, |
|
"loss": 1.6876, |
|
"step": 2370 |
|
}, |
|
{ |
|
"epoch": 6.74, |
|
"learning_rate": 2.782105263157895e-05, |
|
"loss": 1.6794, |
|
"step": 2380 |
|
}, |
|
{ |
|
"epoch": 6.77, |
|
"learning_rate": 2.780526315789474e-05, |
|
"loss": 1.6693, |
|
"step": 2390 |
|
}, |
|
{ |
|
"epoch": 6.8, |
|
"learning_rate": 2.7789473684210526e-05, |
|
"loss": 1.7284, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 6.83, |
|
"learning_rate": 2.7773684210526316e-05, |
|
"loss": 1.7203, |
|
"step": 2410 |
|
}, |
|
{ |
|
"epoch": 6.85, |
|
"learning_rate": 2.7757894736842103e-05, |
|
"loss": 1.7067, |
|
"step": 2420 |
|
}, |
|
{ |
|
"epoch": 6.88, |
|
"learning_rate": 2.7742105263157897e-05, |
|
"loss": 1.6756, |
|
"step": 2430 |
|
}, |
|
{ |
|
"epoch": 6.91, |
|
"learning_rate": 2.7726315789473684e-05, |
|
"loss": 1.7367, |
|
"step": 2440 |
|
}, |
|
{ |
|
"epoch": 6.94, |
|
"learning_rate": 2.7710526315789474e-05, |
|
"loss": 1.6639, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 6.97, |
|
"learning_rate": 2.769473684210526e-05, |
|
"loss": 1.7266, |
|
"step": 2460 |
|
}, |
|
{ |
|
"epoch": 6.99, |
|
"learning_rate": 2.7678947368421055e-05, |
|
"loss": 1.696, |
|
"step": 2470 |
|
}, |
|
{ |
|
"epoch": 7.03, |
|
"learning_rate": 2.7663157894736842e-05, |
|
"loss": 1.7717, |
|
"step": 2480 |
|
}, |
|
{ |
|
"epoch": 7.05, |
|
"learning_rate": 2.7647368421052632e-05, |
|
"loss": 1.6731, |
|
"step": 2490 |
|
}, |
|
{ |
|
"epoch": 7.08, |
|
"learning_rate": 2.7631578947368423e-05, |
|
"loss": 1.6213, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 7.11, |
|
"learning_rate": 2.761578947368421e-05, |
|
"loss": 1.6132, |
|
"step": 2510 |
|
}, |
|
{ |
|
"epoch": 7.14, |
|
"learning_rate": 2.7600000000000003e-05, |
|
"loss": 1.6349, |
|
"step": 2520 |
|
}, |
|
{ |
|
"epoch": 7.17, |
|
"learning_rate": 2.758421052631579e-05, |
|
"loss": 1.6066, |
|
"step": 2530 |
|
}, |
|
{ |
|
"epoch": 7.19, |
|
"learning_rate": 2.756842105263158e-05, |
|
"loss": 1.5981, |
|
"step": 2540 |
|
}, |
|
{ |
|
"epoch": 7.22, |
|
"learning_rate": 2.7552631578947368e-05, |
|
"loss": 1.6642, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 7.25, |
|
"learning_rate": 2.7536842105263158e-05, |
|
"loss": 1.6323, |
|
"step": 2560 |
|
}, |
|
{ |
|
"epoch": 7.28, |
|
"learning_rate": 2.752105263157895e-05, |
|
"loss": 1.689, |
|
"step": 2570 |
|
}, |
|
{ |
|
"epoch": 7.31, |
|
"learning_rate": 2.750526315789474e-05, |
|
"loss": 1.6511, |
|
"step": 2580 |
|
}, |
|
{ |
|
"epoch": 7.34, |
|
"learning_rate": 2.7489473684210526e-05, |
|
"loss": 1.6616, |
|
"step": 2590 |
|
}, |
|
{ |
|
"epoch": 7.36, |
|
"learning_rate": 2.7473684210526316e-05, |
|
"loss": 1.6584, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 7.39, |
|
"learning_rate": 2.7457894736842106e-05, |
|
"loss": 1.6503, |
|
"step": 2610 |
|
}, |
|
{ |
|
"epoch": 7.42, |
|
"learning_rate": 2.7442105263157897e-05, |
|
"loss": 1.6598, |
|
"step": 2620 |
|
}, |
|
{ |
|
"epoch": 7.45, |
|
"learning_rate": 2.7426315789473684e-05, |
|
"loss": 1.6312, |
|
"step": 2630 |
|
}, |
|
{ |
|
"epoch": 7.48, |
|
"learning_rate": 2.7410526315789474e-05, |
|
"loss": 1.6501, |
|
"step": 2640 |
|
}, |
|
{ |
|
"epoch": 7.51, |
|
"learning_rate": 2.739473684210526e-05, |
|
"loss": 1.708, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 7.53, |
|
"learning_rate": 2.7378947368421055e-05, |
|
"loss": 1.6449, |
|
"step": 2660 |
|
}, |
|
{ |
|
"epoch": 7.56, |
|
"learning_rate": 2.7363157894736845e-05, |
|
"loss": 1.7256, |
|
"step": 2670 |
|
}, |
|
{ |
|
"epoch": 7.59, |
|
"learning_rate": 2.7347368421052632e-05, |
|
"loss": 1.6688, |
|
"step": 2680 |
|
}, |
|
{ |
|
"epoch": 7.62, |
|
"learning_rate": 2.7331578947368422e-05, |
|
"loss": 1.6438, |
|
"step": 2690 |
|
}, |
|
{ |
|
"epoch": 7.65, |
|
"learning_rate": 2.7315789473684213e-05, |
|
"loss": 1.6481, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 7.68, |
|
"learning_rate": 2.7300000000000003e-05, |
|
"loss": 1.6991, |
|
"step": 2710 |
|
}, |
|
{ |
|
"epoch": 7.7, |
|
"learning_rate": 2.728421052631579e-05, |
|
"loss": 1.6602, |
|
"step": 2720 |
|
}, |
|
{ |
|
"epoch": 7.73, |
|
"learning_rate": 2.726842105263158e-05, |
|
"loss": 1.6537, |
|
"step": 2730 |
|
}, |
|
{ |
|
"epoch": 7.76, |
|
"learning_rate": 2.7252631578947367e-05, |
|
"loss": 1.6564, |
|
"step": 2740 |
|
}, |
|
{ |
|
"epoch": 7.79, |
|
"learning_rate": 2.723684210526316e-05, |
|
"loss": 1.6607, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 7.82, |
|
"learning_rate": 2.7221052631578948e-05, |
|
"loss": 1.6228, |
|
"step": 2760 |
|
}, |
|
{ |
|
"epoch": 7.84, |
|
"learning_rate": 2.7205263157894738e-05, |
|
"loss": 1.6732, |
|
"step": 2770 |
|
}, |
|
{ |
|
"epoch": 7.87, |
|
"learning_rate": 2.7189473684210525e-05, |
|
"loss": 1.7004, |
|
"step": 2780 |
|
}, |
|
{ |
|
"epoch": 7.9, |
|
"learning_rate": 2.7173684210526315e-05, |
|
"loss": 1.678, |
|
"step": 2790 |
|
}, |
|
{ |
|
"epoch": 7.93, |
|
"learning_rate": 2.7157894736842106e-05, |
|
"loss": 1.6379, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 7.96, |
|
"learning_rate": 2.7142105263157896e-05, |
|
"loss": 1.6475, |
|
"step": 2810 |
|
}, |
|
{ |
|
"epoch": 7.99, |
|
"learning_rate": 2.7126315789473683e-05, |
|
"loss": 1.6697, |
|
"step": 2820 |
|
}, |
|
{ |
|
"epoch": 8.02, |
|
"learning_rate": 2.7110526315789473e-05, |
|
"loss": 1.7948, |
|
"step": 2830 |
|
}, |
|
{ |
|
"epoch": 8.05, |
|
"learning_rate": 2.7094736842105267e-05, |
|
"loss": 1.6428, |
|
"step": 2840 |
|
}, |
|
{ |
|
"epoch": 8.07, |
|
"learning_rate": 2.7078947368421054e-05, |
|
"loss": 1.6928, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 8.1, |
|
"learning_rate": 2.7063157894736844e-05, |
|
"loss": 1.6174, |
|
"step": 2860 |
|
}, |
|
{ |
|
"epoch": 8.13, |
|
"learning_rate": 2.704736842105263e-05, |
|
"loss": 1.6166, |
|
"step": 2870 |
|
}, |
|
{ |
|
"epoch": 8.16, |
|
"learning_rate": 2.7031578947368422e-05, |
|
"loss": 1.6341, |
|
"step": 2880 |
|
}, |
|
{ |
|
"epoch": 8.19, |
|
"learning_rate": 2.7015789473684212e-05, |
|
"loss": 1.6341, |
|
"step": 2890 |
|
}, |
|
{ |
|
"epoch": 8.21, |
|
"learning_rate": 2.7000000000000002e-05, |
|
"loss": 1.6289, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 8.24, |
|
"learning_rate": 2.698421052631579e-05, |
|
"loss": 1.6076, |
|
"step": 2910 |
|
}, |
|
{ |
|
"epoch": 8.27, |
|
"learning_rate": 2.696842105263158e-05, |
|
"loss": 1.6466, |
|
"step": 2920 |
|
}, |
|
{ |
|
"epoch": 8.3, |
|
"learning_rate": 2.6952631578947367e-05, |
|
"loss": 1.6026, |
|
"step": 2930 |
|
}, |
|
{ |
|
"epoch": 8.33, |
|
"learning_rate": 2.693684210526316e-05, |
|
"loss": 1.5972, |
|
"step": 2940 |
|
}, |
|
{ |
|
"epoch": 8.36, |
|
"learning_rate": 2.6921052631578947e-05, |
|
"loss": 1.6079, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 8.38, |
|
"learning_rate": 2.6905263157894738e-05, |
|
"loss": 1.6655, |
|
"step": 2960 |
|
}, |
|
{ |
|
"epoch": 8.41, |
|
"learning_rate": 2.6889473684210525e-05, |
|
"loss": 1.6134, |
|
"step": 2970 |
|
}, |
|
{ |
|
"epoch": 8.44, |
|
"learning_rate": 2.687368421052632e-05, |
|
"loss": 1.6399, |
|
"step": 2980 |
|
}, |
|
{ |
|
"epoch": 8.47, |
|
"learning_rate": 2.6857894736842105e-05, |
|
"loss": 1.6358, |
|
"step": 2990 |
|
}, |
|
{ |
|
"epoch": 8.5, |
|
"learning_rate": 2.6842105263157896e-05, |
|
"loss": 1.6447, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 8.5, |
|
"eval_denotation_accuracy": 0.5637583892617449, |
|
"eval_loss": 2.023669958114624, |
|
"eval_runtime": 415.5806, |
|
"eval_samples_per_second": 6.812, |
|
"eval_steps_per_second": 1.704, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 8.53, |
|
"learning_rate": 2.6826315789473686e-05, |
|
"loss": 1.6034, |
|
"step": 3010 |
|
}, |
|
{ |
|
"epoch": 8.55, |
|
"learning_rate": 2.6810526315789473e-05, |
|
"loss": 1.5992, |
|
"step": 3020 |
|
}, |
|
{ |
|
"epoch": 8.58, |
|
"learning_rate": 2.6794736842105267e-05, |
|
"loss": 1.6025, |
|
"step": 3030 |
|
}, |
|
{ |
|
"epoch": 8.61, |
|
"learning_rate": 2.6778947368421054e-05, |
|
"loss": 1.6258, |
|
"step": 3040 |
|
}, |
|
{ |
|
"epoch": 8.64, |
|
"learning_rate": 2.6763157894736844e-05, |
|
"loss": 1.6088, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 8.67, |
|
"learning_rate": 2.674736842105263e-05, |
|
"loss": 1.6179, |
|
"step": 3060 |
|
}, |
|
{ |
|
"epoch": 8.7, |
|
"learning_rate": 2.673157894736842e-05, |
|
"loss": 1.6292, |
|
"step": 3070 |
|
}, |
|
{ |
|
"epoch": 8.72, |
|
"learning_rate": 2.671578947368421e-05, |
|
"loss": 1.6844, |
|
"step": 3080 |
|
}, |
|
{ |
|
"epoch": 8.75, |
|
"learning_rate": 2.6700000000000002e-05, |
|
"loss": 1.6472, |
|
"step": 3090 |
|
}, |
|
{ |
|
"epoch": 8.78, |
|
"learning_rate": 2.668421052631579e-05, |
|
"loss": 1.6238, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 8.81, |
|
"learning_rate": 2.666842105263158e-05, |
|
"loss": 1.6499, |
|
"step": 3110 |
|
}, |
|
{ |
|
"epoch": 8.84, |
|
"learning_rate": 2.665263157894737e-05, |
|
"loss": 1.6397, |
|
"step": 3120 |
|
}, |
|
{ |
|
"epoch": 8.86, |
|
"learning_rate": 2.663684210526316e-05, |
|
"loss": 1.6594, |
|
"step": 3130 |
|
}, |
|
{ |
|
"epoch": 8.89, |
|
"learning_rate": 2.6621052631578947e-05, |
|
"loss": 1.6612, |
|
"step": 3140 |
|
}, |
|
{ |
|
"epoch": 8.92, |
|
"learning_rate": 2.6605263157894737e-05, |
|
"loss": 1.6247, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 8.95, |
|
"learning_rate": 2.6589473684210524e-05, |
|
"loss": 1.633, |
|
"step": 3160 |
|
}, |
|
{ |
|
"epoch": 8.98, |
|
"learning_rate": 2.6573684210526318e-05, |
|
"loss": 1.6364, |
|
"step": 3170 |
|
}, |
|
{ |
|
"epoch": 9.01, |
|
"learning_rate": 2.6557894736842108e-05, |
|
"loss": 1.7977, |
|
"step": 3180 |
|
}, |
|
{ |
|
"epoch": 9.04, |
|
"learning_rate": 2.6542105263157895e-05, |
|
"loss": 1.5909, |
|
"step": 3190 |
|
}, |
|
{ |
|
"epoch": 9.06, |
|
"learning_rate": 2.6526315789473685e-05, |
|
"loss": 1.5751, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 9.09, |
|
"learning_rate": 2.6510526315789472e-05, |
|
"loss": 1.5807, |
|
"step": 3210 |
|
}, |
|
{ |
|
"epoch": 9.12, |
|
"learning_rate": 2.6494736842105266e-05, |
|
"loss": 1.5911, |
|
"step": 3220 |
|
}, |
|
{ |
|
"epoch": 9.15, |
|
"learning_rate": 2.6478947368421053e-05, |
|
"loss": 1.6176, |
|
"step": 3230 |
|
}, |
|
{ |
|
"epoch": 9.18, |
|
"learning_rate": 2.6463157894736843e-05, |
|
"loss": 1.5872, |
|
"step": 3240 |
|
}, |
|
{ |
|
"epoch": 9.21, |
|
"learning_rate": 2.644736842105263e-05, |
|
"loss": 1.5697, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 9.23, |
|
"learning_rate": 2.6431578947368424e-05, |
|
"loss": 1.5871, |
|
"step": 3260 |
|
}, |
|
{ |
|
"epoch": 9.26, |
|
"learning_rate": 2.641578947368421e-05, |
|
"loss": 1.6368, |
|
"step": 3270 |
|
}, |
|
{ |
|
"epoch": 9.29, |
|
"learning_rate": 2.64e-05, |
|
"loss": 1.593, |
|
"step": 3280 |
|
}, |
|
{ |
|
"epoch": 9.32, |
|
"learning_rate": 2.6384210526315788e-05, |
|
"loss": 1.577, |
|
"step": 3290 |
|
}, |
|
{ |
|
"epoch": 9.35, |
|
"learning_rate": 2.636842105263158e-05, |
|
"loss": 1.6126, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 9.38, |
|
"learning_rate": 2.635263157894737e-05, |
|
"loss": 1.6005, |
|
"step": 3310 |
|
}, |
|
{ |
|
"epoch": 9.4, |
|
"learning_rate": 2.633684210526316e-05, |
|
"loss": 1.6161, |
|
"step": 3320 |
|
}, |
|
{ |
|
"epoch": 9.43, |
|
"learning_rate": 2.6321052631578946e-05, |
|
"loss": 1.6188, |
|
"step": 3330 |
|
}, |
|
{ |
|
"epoch": 9.46, |
|
"learning_rate": 2.6305263157894737e-05, |
|
"loss": 1.6316, |
|
"step": 3340 |
|
}, |
|
{ |
|
"epoch": 9.49, |
|
"learning_rate": 2.6289473684210527e-05, |
|
"loss": 1.6132, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 9.52, |
|
"learning_rate": 2.6273684210526317e-05, |
|
"loss": 1.6491, |
|
"step": 3360 |
|
}, |
|
{ |
|
"epoch": 9.55, |
|
"learning_rate": 2.6257894736842108e-05, |
|
"loss": 1.6004, |
|
"step": 3370 |
|
}, |
|
{ |
|
"epoch": 9.57, |
|
"learning_rate": 2.6242105263157895e-05, |
|
"loss": 1.6161, |
|
"step": 3380 |
|
}, |
|
{ |
|
"epoch": 9.6, |
|
"learning_rate": 2.6226315789473685e-05, |
|
"loss": 1.607, |
|
"step": 3390 |
|
}, |
|
{ |
|
"epoch": 9.63, |
|
"learning_rate": 2.6210526315789475e-05, |
|
"loss": 1.6207, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 9.66, |
|
"learning_rate": 2.6194736842105266e-05, |
|
"loss": 1.6111, |
|
"step": 3410 |
|
}, |
|
{ |
|
"epoch": 9.69, |
|
"learning_rate": 2.6178947368421053e-05, |
|
"loss": 1.5797, |
|
"step": 3420 |
|
}, |
|
{ |
|
"epoch": 9.71, |
|
"learning_rate": 2.6163157894736843e-05, |
|
"loss": 1.5955, |
|
"step": 3430 |
|
}, |
|
{ |
|
"epoch": 9.74, |
|
"learning_rate": 2.614736842105263e-05, |
|
"loss": 1.6303, |
|
"step": 3440 |
|
}, |
|
{ |
|
"epoch": 9.77, |
|
"learning_rate": 2.6131578947368424e-05, |
|
"loss": 1.6379, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 9.8, |
|
"learning_rate": 2.611578947368421e-05, |
|
"loss": 1.6016, |
|
"step": 3460 |
|
}, |
|
{ |
|
"epoch": 9.83, |
|
"learning_rate": 2.61e-05, |
|
"loss": 1.6299, |
|
"step": 3470 |
|
}, |
|
{ |
|
"epoch": 9.86, |
|
"learning_rate": 2.6084210526315788e-05, |
|
"loss": 1.5963, |
|
"step": 3480 |
|
}, |
|
{ |
|
"epoch": 9.88, |
|
"learning_rate": 2.6068421052631578e-05, |
|
"loss": 1.5872, |
|
"step": 3490 |
|
}, |
|
{ |
|
"epoch": 9.91, |
|
"learning_rate": 2.605263157894737e-05, |
|
"loss": 1.626, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 9.94, |
|
"learning_rate": 2.603684210526316e-05, |
|
"loss": 1.5876, |
|
"step": 3510 |
|
}, |
|
{ |
|
"epoch": 9.97, |
|
"learning_rate": 2.602105263157895e-05, |
|
"loss": 1.6061, |
|
"step": 3520 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"learning_rate": 2.6005263157894736e-05, |
|
"loss": 1.5925, |
|
"step": 3530 |
|
}, |
|
{ |
|
"epoch": 10.03, |
|
"learning_rate": 2.598947368421053e-05, |
|
"loss": 1.7235, |
|
"step": 3540 |
|
}, |
|
{ |
|
"epoch": 10.06, |
|
"learning_rate": 2.5973684210526317e-05, |
|
"loss": 1.5714, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 10.08, |
|
"learning_rate": 2.5957894736842107e-05, |
|
"loss": 1.5669, |
|
"step": 3560 |
|
}, |
|
{ |
|
"epoch": 10.11, |
|
"learning_rate": 2.5942105263157894e-05, |
|
"loss": 1.573, |
|
"step": 3570 |
|
}, |
|
{ |
|
"epoch": 10.14, |
|
"learning_rate": 2.5926315789473684e-05, |
|
"loss": 1.5635, |
|
"step": 3580 |
|
}, |
|
{ |
|
"epoch": 10.17, |
|
"learning_rate": 2.5910526315789475e-05, |
|
"loss": 1.5614, |
|
"step": 3590 |
|
}, |
|
{ |
|
"epoch": 10.2, |
|
"learning_rate": 2.5894736842105265e-05, |
|
"loss": 1.5664, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 10.23, |
|
"learning_rate": 2.5878947368421052e-05, |
|
"loss": 1.5666, |
|
"step": 3610 |
|
}, |
|
{ |
|
"epoch": 10.25, |
|
"learning_rate": 2.5863157894736842e-05, |
|
"loss": 1.5893, |
|
"step": 3620 |
|
}, |
|
{ |
|
"epoch": 10.28, |
|
"learning_rate": 2.584736842105263e-05, |
|
"loss": 1.5843, |
|
"step": 3630 |
|
}, |
|
{ |
|
"epoch": 10.31, |
|
"learning_rate": 2.5831578947368423e-05, |
|
"loss": 1.5478, |
|
"step": 3640 |
|
}, |
|
{ |
|
"epoch": 10.34, |
|
"learning_rate": 2.581578947368421e-05, |
|
"loss": 1.5738, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 10.37, |
|
"learning_rate": 2.58e-05, |
|
"loss": 1.5638, |
|
"step": 3660 |
|
}, |
|
{ |
|
"epoch": 10.4, |
|
"learning_rate": 2.578421052631579e-05, |
|
"loss": 1.5695, |
|
"step": 3670 |
|
}, |
|
{ |
|
"epoch": 10.42, |
|
"learning_rate": 2.576842105263158e-05, |
|
"loss": 1.5776, |
|
"step": 3680 |
|
}, |
|
{ |
|
"epoch": 10.45, |
|
"learning_rate": 2.575263157894737e-05, |
|
"loss": 1.5758, |
|
"step": 3690 |
|
}, |
|
{ |
|
"epoch": 10.48, |
|
"learning_rate": 2.5736842105263158e-05, |
|
"loss": 1.5606, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 10.51, |
|
"learning_rate": 2.572105263157895e-05, |
|
"loss": 1.6042, |
|
"step": 3710 |
|
}, |
|
{ |
|
"epoch": 10.54, |
|
"learning_rate": 2.5705263157894736e-05, |
|
"loss": 1.563, |
|
"step": 3720 |
|
}, |
|
{ |
|
"epoch": 10.57, |
|
"learning_rate": 2.568947368421053e-05, |
|
"loss": 1.5776, |
|
"step": 3730 |
|
}, |
|
{ |
|
"epoch": 10.59, |
|
"learning_rate": 2.5673684210526316e-05, |
|
"loss": 1.6218, |
|
"step": 3740 |
|
}, |
|
{ |
|
"epoch": 10.62, |
|
"learning_rate": 2.5657894736842107e-05, |
|
"loss": 1.5918, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 10.65, |
|
"learning_rate": 2.5642105263157894e-05, |
|
"loss": 1.5627, |
|
"step": 3760 |
|
}, |
|
{ |
|
"epoch": 10.68, |
|
"learning_rate": 2.5626315789473684e-05, |
|
"loss": 1.5722, |
|
"step": 3770 |
|
}, |
|
{ |
|
"epoch": 10.71, |
|
"learning_rate": 2.5610526315789474e-05, |
|
"loss": 1.5993, |
|
"step": 3780 |
|
}, |
|
{ |
|
"epoch": 10.73, |
|
"learning_rate": 2.5594736842105265e-05, |
|
"loss": 1.5965, |
|
"step": 3790 |
|
}, |
|
{ |
|
"epoch": 10.76, |
|
"learning_rate": 2.557894736842105e-05, |
|
"loss": 1.5654, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 10.79, |
|
"learning_rate": 2.5563157894736842e-05, |
|
"loss": 1.5903, |
|
"step": 3810 |
|
}, |
|
{ |
|
"epoch": 10.82, |
|
"learning_rate": 2.5547368421052632e-05, |
|
"loss": 1.5632, |
|
"step": 3820 |
|
}, |
|
{ |
|
"epoch": 10.85, |
|
"learning_rate": 2.5531578947368423e-05, |
|
"loss": 1.5944, |
|
"step": 3830 |
|
}, |
|
{ |
|
"epoch": 10.88, |
|
"learning_rate": 2.5515789473684213e-05, |
|
"loss": 1.5708, |
|
"step": 3840 |
|
}, |
|
{ |
|
"epoch": 10.9, |
|
"learning_rate": 2.55e-05, |
|
"loss": 1.5838, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 10.93, |
|
"learning_rate": 2.548421052631579e-05, |
|
"loss": 1.6397, |
|
"step": 3860 |
|
}, |
|
{ |
|
"epoch": 10.96, |
|
"learning_rate": 2.546842105263158e-05, |
|
"loss": 1.6055, |
|
"step": 3870 |
|
}, |
|
{ |
|
"epoch": 10.99, |
|
"learning_rate": 2.545263157894737e-05, |
|
"loss": 1.6096, |
|
"step": 3880 |
|
}, |
|
{ |
|
"epoch": 11.02, |
|
"learning_rate": 2.5436842105263158e-05, |
|
"loss": 1.73, |
|
"step": 3890 |
|
}, |
|
{ |
|
"epoch": 11.05, |
|
"learning_rate": 2.5421052631578948e-05, |
|
"loss": 1.558, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 11.08, |
|
"learning_rate": 2.5405263157894735e-05, |
|
"loss": 1.6073, |
|
"step": 3910 |
|
}, |
|
{ |
|
"epoch": 11.1, |
|
"learning_rate": 2.538947368421053e-05, |
|
"loss": 1.5865, |
|
"step": 3920 |
|
}, |
|
{ |
|
"epoch": 11.13, |
|
"learning_rate": 2.5373684210526316e-05, |
|
"loss": 1.5615, |
|
"step": 3930 |
|
}, |
|
{ |
|
"epoch": 11.16, |
|
"learning_rate": 2.5357894736842106e-05, |
|
"loss": 1.5534, |
|
"step": 3940 |
|
}, |
|
{ |
|
"epoch": 11.19, |
|
"learning_rate": 2.5342105263157893e-05, |
|
"loss": 1.5786, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 11.22, |
|
"learning_rate": 2.5326315789473687e-05, |
|
"loss": 1.5597, |
|
"step": 3960 |
|
}, |
|
{ |
|
"epoch": 11.25, |
|
"learning_rate": 2.5310526315789474e-05, |
|
"loss": 1.5256, |
|
"step": 3970 |
|
}, |
|
{ |
|
"epoch": 11.27, |
|
"learning_rate": 2.5294736842105264e-05, |
|
"loss": 1.5662, |
|
"step": 3980 |
|
}, |
|
{ |
|
"epoch": 11.3, |
|
"learning_rate": 2.527894736842105e-05, |
|
"loss": 1.5891, |
|
"step": 3990 |
|
}, |
|
{ |
|
"epoch": 11.33, |
|
"learning_rate": 2.526315789473684e-05, |
|
"loss": 1.5694, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 11.33, |
|
"eval_denotation_accuracy": 0.555987283645355, |
|
"eval_loss": 2.0672969818115234, |
|
"eval_runtime": 545.5849, |
|
"eval_samples_per_second": 5.189, |
|
"eval_steps_per_second": 1.298, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 11.36, |
|
"learning_rate": 2.5247368421052635e-05, |
|
"loss": 1.5932, |
|
"step": 4010 |
|
}, |
|
{ |
|
"epoch": 11.39, |
|
"learning_rate": 2.5231578947368422e-05, |
|
"loss": 1.5728, |
|
"step": 4020 |
|
}, |
|
{ |
|
"epoch": 11.42, |
|
"learning_rate": 2.5215789473684212e-05, |
|
"loss": 1.5914, |
|
"step": 4030 |
|
}, |
|
{ |
|
"epoch": 11.44, |
|
"learning_rate": 2.52e-05, |
|
"loss": 1.579, |
|
"step": 4040 |
|
}, |
|
{ |
|
"epoch": 11.47, |
|
"learning_rate": 2.518421052631579e-05, |
|
"loss": 1.5675, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 11.5, |
|
"learning_rate": 2.516842105263158e-05, |
|
"loss": 1.5588, |
|
"step": 4060 |
|
}, |
|
{ |
|
"epoch": 11.53, |
|
"learning_rate": 2.515263157894737e-05, |
|
"loss": 1.5447, |
|
"step": 4070 |
|
}, |
|
{ |
|
"epoch": 11.56, |
|
"learning_rate": 2.5136842105263157e-05, |
|
"loss": 1.5659, |
|
"step": 4080 |
|
}, |
|
{ |
|
"epoch": 11.58, |
|
"learning_rate": 2.5121052631578948e-05, |
|
"loss": 1.5522, |
|
"step": 4090 |
|
}, |
|
{ |
|
"epoch": 11.61, |
|
"learning_rate": 2.5105263157894738e-05, |
|
"loss": 1.5465, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 11.64, |
|
"learning_rate": 2.5089473684210528e-05, |
|
"loss": 1.5612, |
|
"step": 4110 |
|
}, |
|
{ |
|
"epoch": 11.67, |
|
"learning_rate": 2.5073684210526315e-05, |
|
"loss": 1.5501, |
|
"step": 4120 |
|
}, |
|
{ |
|
"epoch": 11.7, |
|
"learning_rate": 2.5057894736842106e-05, |
|
"loss": 1.5516, |
|
"step": 4130 |
|
}, |
|
{ |
|
"epoch": 11.73, |
|
"learning_rate": 2.5042105263157893e-05, |
|
"loss": 1.5716, |
|
"step": 4140 |
|
}, |
|
{ |
|
"epoch": 11.75, |
|
"learning_rate": 2.5026315789473686e-05, |
|
"loss": 1.5688, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 11.78, |
|
"learning_rate": 2.5010526315789473e-05, |
|
"loss": 1.5453, |
|
"step": 4160 |
|
}, |
|
{ |
|
"epoch": 11.81, |
|
"learning_rate": 2.4994736842105264e-05, |
|
"loss": 1.5601, |
|
"step": 4170 |
|
}, |
|
{ |
|
"epoch": 11.84, |
|
"learning_rate": 2.4978947368421054e-05, |
|
"loss": 1.5772, |
|
"step": 4180 |
|
}, |
|
{ |
|
"epoch": 11.87, |
|
"learning_rate": 2.4963157894736844e-05, |
|
"loss": 1.573, |
|
"step": 4190 |
|
}, |
|
{ |
|
"epoch": 11.9, |
|
"learning_rate": 2.4947368421052635e-05, |
|
"loss": 1.5693, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 11.92, |
|
"learning_rate": 2.493157894736842e-05, |
|
"loss": 1.5634, |
|
"step": 4210 |
|
}, |
|
{ |
|
"epoch": 11.95, |
|
"learning_rate": 2.4915789473684212e-05, |
|
"loss": 1.5836, |
|
"step": 4220 |
|
}, |
|
{ |
|
"epoch": 11.98, |
|
"learning_rate": 2.49e-05, |
|
"loss": 1.5607, |
|
"step": 4230 |
|
}, |
|
{ |
|
"epoch": 12.01, |
|
"learning_rate": 2.4884210526315792e-05, |
|
"loss": 1.6897, |
|
"step": 4240 |
|
}, |
|
{ |
|
"epoch": 12.04, |
|
"learning_rate": 2.486842105263158e-05, |
|
"loss": 1.5572, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 12.07, |
|
"learning_rate": 2.485263157894737e-05, |
|
"loss": 1.5313, |
|
"step": 4260 |
|
}, |
|
{ |
|
"epoch": 12.1, |
|
"learning_rate": 2.4836842105263157e-05, |
|
"loss": 1.5374, |
|
"step": 4270 |
|
}, |
|
{ |
|
"epoch": 12.12, |
|
"learning_rate": 2.4821052631578947e-05, |
|
"loss": 1.5246, |
|
"step": 4280 |
|
}, |
|
{ |
|
"epoch": 12.15, |
|
"learning_rate": 2.4805263157894737e-05, |
|
"loss": 1.5319, |
|
"step": 4290 |
|
}, |
|
{ |
|
"epoch": 12.18, |
|
"learning_rate": 2.4789473684210528e-05, |
|
"loss": 1.5379, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 12.21, |
|
"learning_rate": 2.4773684210526315e-05, |
|
"loss": 1.531, |
|
"step": 4310 |
|
}, |
|
{ |
|
"epoch": 12.24, |
|
"learning_rate": 2.4757894736842105e-05, |
|
"loss": 1.5313, |
|
"step": 4320 |
|
}, |
|
{ |
|
"epoch": 12.27, |
|
"learning_rate": 2.4742105263157895e-05, |
|
"loss": 1.538, |
|
"step": 4330 |
|
}, |
|
{ |
|
"epoch": 12.29, |
|
"learning_rate": 2.4726315789473686e-05, |
|
"loss": 1.5354, |
|
"step": 4340 |
|
}, |
|
{ |
|
"epoch": 12.32, |
|
"learning_rate": 2.4710526315789476e-05, |
|
"loss": 1.5123, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 12.35, |
|
"learning_rate": 2.4694736842105263e-05, |
|
"loss": 1.5324, |
|
"step": 4360 |
|
}, |
|
{ |
|
"epoch": 12.38, |
|
"learning_rate": 2.4678947368421053e-05, |
|
"loss": 1.5454, |
|
"step": 4370 |
|
}, |
|
{ |
|
"epoch": 12.41, |
|
"learning_rate": 2.4663157894736844e-05, |
|
"loss": 1.5264, |
|
"step": 4380 |
|
}, |
|
{ |
|
"epoch": 12.44, |
|
"learning_rate": 2.4647368421052634e-05, |
|
"loss": 1.5431, |
|
"step": 4390 |
|
}, |
|
{ |
|
"epoch": 12.46, |
|
"learning_rate": 2.463157894736842e-05, |
|
"loss": 1.561, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 12.49, |
|
"learning_rate": 2.461578947368421e-05, |
|
"loss": 1.5431, |
|
"step": 4410 |
|
}, |
|
{ |
|
"epoch": 12.52, |
|
"learning_rate": 2.4599999999999998e-05, |
|
"loss": 1.5622, |
|
"step": 4420 |
|
}, |
|
{ |
|
"epoch": 12.55, |
|
"learning_rate": 2.4584210526315792e-05, |
|
"loss": 1.5437, |
|
"step": 4430 |
|
}, |
|
{ |
|
"epoch": 12.58, |
|
"learning_rate": 2.456842105263158e-05, |
|
"loss": 1.5525, |
|
"step": 4440 |
|
}, |
|
{ |
|
"epoch": 12.6, |
|
"learning_rate": 2.455263157894737e-05, |
|
"loss": 1.5432, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 12.63, |
|
"learning_rate": 2.4536842105263156e-05, |
|
"loss": 1.5883, |
|
"step": 4460 |
|
}, |
|
{ |
|
"epoch": 12.66, |
|
"learning_rate": 2.452105263157895e-05, |
|
"loss": 1.5411, |
|
"step": 4470 |
|
}, |
|
{ |
|
"epoch": 12.69, |
|
"learning_rate": 2.4505263157894737e-05, |
|
"loss": 1.5631, |
|
"step": 4480 |
|
}, |
|
{ |
|
"epoch": 12.72, |
|
"learning_rate": 2.4489473684210527e-05, |
|
"loss": 1.5579, |
|
"step": 4490 |
|
}, |
|
{ |
|
"epoch": 12.75, |
|
"learning_rate": 2.4473684210526318e-05, |
|
"loss": 1.5554, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 12.77, |
|
"learning_rate": 2.4457894736842105e-05, |
|
"loss": 1.5325, |
|
"step": 4510 |
|
}, |
|
{ |
|
"epoch": 12.8, |
|
"learning_rate": 2.4442105263157898e-05, |
|
"loss": 1.5589, |
|
"step": 4520 |
|
}, |
|
{ |
|
"epoch": 12.83, |
|
"learning_rate": 2.4426315789473685e-05, |
|
"loss": 1.5583, |
|
"step": 4530 |
|
}, |
|
{ |
|
"epoch": 12.86, |
|
"learning_rate": 2.4410526315789476e-05, |
|
"loss": 1.5552, |
|
"step": 4540 |
|
}, |
|
{ |
|
"epoch": 12.89, |
|
"learning_rate": 2.4394736842105262e-05, |
|
"loss": 1.559, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 12.92, |
|
"learning_rate": 2.4378947368421053e-05, |
|
"loss": 1.5267, |
|
"step": 4560 |
|
}, |
|
{ |
|
"epoch": 12.94, |
|
"learning_rate": 2.4363157894736843e-05, |
|
"loss": 1.5576, |
|
"step": 4570 |
|
}, |
|
{ |
|
"epoch": 12.97, |
|
"learning_rate": 2.4347368421052633e-05, |
|
"loss": 1.5553, |
|
"step": 4580 |
|
}, |
|
{ |
|
"epoch": 13.0, |
|
"learning_rate": 2.433157894736842e-05, |
|
"loss": 1.6593, |
|
"step": 4590 |
|
}, |
|
{ |
|
"epoch": 13.03, |
|
"learning_rate": 2.431578947368421e-05, |
|
"loss": 1.5177, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 13.06, |
|
"learning_rate": 2.43e-05, |
|
"loss": 1.5141, |
|
"step": 4610 |
|
}, |
|
{ |
|
"epoch": 13.09, |
|
"learning_rate": 2.428421052631579e-05, |
|
"loss": 1.5269, |
|
"step": 4620 |
|
}, |
|
{ |
|
"epoch": 13.12, |
|
"learning_rate": 2.426842105263158e-05, |
|
"loss": 1.5197, |
|
"step": 4630 |
|
}, |
|
{ |
|
"epoch": 13.14, |
|
"learning_rate": 2.425263157894737e-05, |
|
"loss": 1.5164, |
|
"step": 4640 |
|
}, |
|
{ |
|
"epoch": 13.17, |
|
"learning_rate": 2.4236842105263156e-05, |
|
"loss": 1.5322, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 13.2, |
|
"learning_rate": 2.422105263157895e-05, |
|
"loss": 1.5146, |
|
"step": 4660 |
|
}, |
|
{ |
|
"epoch": 13.23, |
|
"learning_rate": 2.420526315789474e-05, |
|
"loss": 1.5173, |
|
"step": 4670 |
|
}, |
|
{ |
|
"epoch": 13.26, |
|
"learning_rate": 2.4189473684210527e-05, |
|
"loss": 1.548, |
|
"step": 4680 |
|
}, |
|
{ |
|
"epoch": 13.29, |
|
"learning_rate": 2.4173684210526317e-05, |
|
"loss": 1.521, |
|
"step": 4690 |
|
}, |
|
{ |
|
"epoch": 13.31, |
|
"learning_rate": 2.4157894736842104e-05, |
|
"loss": 1.5316, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 13.34, |
|
"learning_rate": 2.4142105263157898e-05, |
|
"loss": 1.508, |
|
"step": 4710 |
|
}, |
|
{ |
|
"epoch": 13.37, |
|
"learning_rate": 2.4126315789473685e-05, |
|
"loss": 1.5215, |
|
"step": 4720 |
|
}, |
|
{ |
|
"epoch": 13.4, |
|
"learning_rate": 2.4110526315789475e-05, |
|
"loss": 1.515, |
|
"step": 4730 |
|
}, |
|
{ |
|
"epoch": 13.43, |
|
"learning_rate": 2.4094736842105262e-05, |
|
"loss": 1.5275, |
|
"step": 4740 |
|
}, |
|
{ |
|
"epoch": 13.45, |
|
"learning_rate": 2.4078947368421056e-05, |
|
"loss": 1.5286, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 13.48, |
|
"learning_rate": 2.4063157894736843e-05, |
|
"loss": 1.5315, |
|
"step": 4760 |
|
}, |
|
{ |
|
"epoch": 13.51, |
|
"learning_rate": 2.4047368421052633e-05, |
|
"loss": 1.5211, |
|
"step": 4770 |
|
}, |
|
{ |
|
"epoch": 13.54, |
|
"learning_rate": 2.403157894736842e-05, |
|
"loss": 1.5282, |
|
"step": 4780 |
|
}, |
|
{ |
|
"epoch": 13.57, |
|
"learning_rate": 2.401578947368421e-05, |
|
"loss": 1.5341, |
|
"step": 4790 |
|
}, |
|
{ |
|
"epoch": 13.6, |
|
"learning_rate": 2.4e-05, |
|
"loss": 1.5497, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 13.62, |
|
"learning_rate": 2.398421052631579e-05, |
|
"loss": 1.5217, |
|
"step": 4810 |
|
}, |
|
{ |
|
"epoch": 13.65, |
|
"learning_rate": 2.3968421052631578e-05, |
|
"loss": 1.5514, |
|
"step": 4820 |
|
}, |
|
{ |
|
"epoch": 13.68, |
|
"learning_rate": 2.3952631578947368e-05, |
|
"loss": 1.5398, |
|
"step": 4830 |
|
}, |
|
{ |
|
"epoch": 13.71, |
|
"learning_rate": 2.393684210526316e-05, |
|
"loss": 1.5226, |
|
"step": 4840 |
|
}, |
|
{ |
|
"epoch": 13.74, |
|
"learning_rate": 2.392105263157895e-05, |
|
"loss": 1.5465, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 13.77, |
|
"learning_rate": 2.390526315789474e-05, |
|
"loss": 1.5332, |
|
"step": 4860 |
|
}, |
|
{ |
|
"epoch": 13.79, |
|
"learning_rate": 2.3889473684210526e-05, |
|
"loss": 1.5446, |
|
"step": 4870 |
|
}, |
|
{ |
|
"epoch": 13.82, |
|
"learning_rate": 2.3873684210526317e-05, |
|
"loss": 1.5414, |
|
"step": 4880 |
|
}, |
|
{ |
|
"epoch": 13.85, |
|
"learning_rate": 2.3857894736842107e-05, |
|
"loss": 1.5127, |
|
"step": 4890 |
|
}, |
|
{ |
|
"epoch": 13.88, |
|
"learning_rate": 2.3842105263157897e-05, |
|
"loss": 1.5044, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 13.91, |
|
"learning_rate": 2.3826315789473684e-05, |
|
"loss": 1.5433, |
|
"step": 4910 |
|
}, |
|
{ |
|
"epoch": 13.94, |
|
"learning_rate": 2.3810526315789475e-05, |
|
"loss": 1.5447, |
|
"step": 4920 |
|
}, |
|
{ |
|
"epoch": 13.96, |
|
"learning_rate": 2.379473684210526e-05, |
|
"loss": 1.585, |
|
"step": 4930 |
|
}, |
|
{ |
|
"epoch": 13.99, |
|
"learning_rate": 2.3778947368421055e-05, |
|
"loss": 1.5463, |
|
"step": 4940 |
|
}, |
|
{ |
|
"epoch": 14.02, |
|
"learning_rate": 2.3763157894736842e-05, |
|
"loss": 1.6538, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 14.05, |
|
"learning_rate": 2.3747368421052632e-05, |
|
"loss": 1.5099, |
|
"step": 4960 |
|
}, |
|
{ |
|
"epoch": 14.08, |
|
"learning_rate": 2.373157894736842e-05, |
|
"loss": 1.5006, |
|
"step": 4970 |
|
}, |
|
{ |
|
"epoch": 14.11, |
|
"learning_rate": 2.371578947368421e-05, |
|
"loss": 1.5441, |
|
"step": 4980 |
|
}, |
|
{ |
|
"epoch": 14.14, |
|
"learning_rate": 2.37e-05, |
|
"loss": 1.5238, |
|
"step": 4990 |
|
}, |
|
{ |
|
"epoch": 14.16, |
|
"learning_rate": 2.368421052631579e-05, |
|
"loss": 1.5332, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 14.16, |
|
"eval_denotation_accuracy": 0.5588131402331332, |
|
"eval_loss": 2.120304584503174, |
|
"eval_runtime": 363.847, |
|
"eval_samples_per_second": 7.781, |
|
"eval_steps_per_second": 1.946, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 14.19, |
|
"learning_rate": 2.366842105263158e-05, |
|
"loss": 1.5345, |
|
"step": 5010 |
|
}, |
|
{ |
|
"epoch": 14.22, |
|
"learning_rate": 2.3652631578947368e-05, |
|
"loss": 1.5463, |
|
"step": 5020 |
|
}, |
|
{ |
|
"epoch": 14.25, |
|
"learning_rate": 2.363684210526316e-05, |
|
"loss": 1.5225, |
|
"step": 5030 |
|
}, |
|
{ |
|
"epoch": 14.28, |
|
"learning_rate": 2.362105263157895e-05, |
|
"loss": 1.5379, |
|
"step": 5040 |
|
}, |
|
{ |
|
"epoch": 14.31, |
|
"learning_rate": 2.360526315789474e-05, |
|
"loss": 1.5255, |
|
"step": 5050 |
|
}, |
|
{ |
|
"epoch": 14.33, |
|
"learning_rate": 2.3589473684210526e-05, |
|
"loss": 1.524, |
|
"step": 5060 |
|
}, |
|
{ |
|
"epoch": 14.36, |
|
"learning_rate": 2.3573684210526316e-05, |
|
"loss": 1.5206, |
|
"step": 5070 |
|
}, |
|
{ |
|
"epoch": 14.39, |
|
"learning_rate": 2.3557894736842106e-05, |
|
"loss": 1.5192, |
|
"step": 5080 |
|
}, |
|
{ |
|
"epoch": 14.42, |
|
"learning_rate": 2.3542105263157897e-05, |
|
"loss": 1.5225, |
|
"step": 5090 |
|
}, |
|
{ |
|
"epoch": 14.45, |
|
"learning_rate": 2.3526315789473684e-05, |
|
"loss": 1.5119, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 14.47, |
|
"learning_rate": 2.3510526315789474e-05, |
|
"loss": 1.5378, |
|
"step": 5110 |
|
}, |
|
{ |
|
"epoch": 14.5, |
|
"learning_rate": 2.349473684210526e-05, |
|
"loss": 1.5179, |
|
"step": 5120 |
|
}, |
|
{ |
|
"epoch": 14.53, |
|
"learning_rate": 2.3478947368421055e-05, |
|
"loss": 1.5317, |
|
"step": 5130 |
|
}, |
|
{ |
|
"epoch": 14.56, |
|
"learning_rate": 2.346315789473684e-05, |
|
"loss": 1.514, |
|
"step": 5140 |
|
}, |
|
{ |
|
"epoch": 14.59, |
|
"learning_rate": 2.3447368421052632e-05, |
|
"loss": 1.5168, |
|
"step": 5150 |
|
}, |
|
{ |
|
"epoch": 14.62, |
|
"learning_rate": 2.343157894736842e-05, |
|
"loss": 1.4988, |
|
"step": 5160 |
|
}, |
|
{ |
|
"epoch": 14.64, |
|
"learning_rate": 2.3415789473684213e-05, |
|
"loss": 1.5204, |
|
"step": 5170 |
|
}, |
|
{ |
|
"epoch": 14.67, |
|
"learning_rate": 2.3400000000000003e-05, |
|
"loss": 1.5065, |
|
"step": 5180 |
|
}, |
|
{ |
|
"epoch": 14.7, |
|
"learning_rate": 2.338421052631579e-05, |
|
"loss": 1.509, |
|
"step": 5190 |
|
}, |
|
{ |
|
"epoch": 14.73, |
|
"learning_rate": 2.336842105263158e-05, |
|
"loss": 1.5171, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 14.76, |
|
"learning_rate": 2.3352631578947367e-05, |
|
"loss": 1.5005, |
|
"step": 5210 |
|
}, |
|
{ |
|
"epoch": 14.79, |
|
"learning_rate": 2.333684210526316e-05, |
|
"loss": 1.5134, |
|
"step": 5220 |
|
}, |
|
{ |
|
"epoch": 14.81, |
|
"learning_rate": 2.3321052631578948e-05, |
|
"loss": 1.4947, |
|
"step": 5230 |
|
}, |
|
{ |
|
"epoch": 14.84, |
|
"learning_rate": 2.3305263157894738e-05, |
|
"loss": 1.5001, |
|
"step": 5240 |
|
}, |
|
{ |
|
"epoch": 14.87, |
|
"learning_rate": 2.3289473684210525e-05, |
|
"loss": 1.5067, |
|
"step": 5250 |
|
}, |
|
{ |
|
"epoch": 14.9, |
|
"learning_rate": 2.3273684210526316e-05, |
|
"loss": 1.4931, |
|
"step": 5260 |
|
}, |
|
{ |
|
"epoch": 14.93, |
|
"learning_rate": 2.3257894736842106e-05, |
|
"loss": 1.4966, |
|
"step": 5270 |
|
}, |
|
{ |
|
"epoch": 14.96, |
|
"learning_rate": 2.3242105263157896e-05, |
|
"loss": 1.5149, |
|
"step": 5280 |
|
}, |
|
{ |
|
"epoch": 14.98, |
|
"learning_rate": 2.3226315789473683e-05, |
|
"loss": 1.5046, |
|
"step": 5290 |
|
}, |
|
{ |
|
"epoch": 15.01, |
|
"learning_rate": 2.3210526315789473e-05, |
|
"loss": 1.6353, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 15.04, |
|
"learning_rate": 2.3194736842105264e-05, |
|
"loss": 1.505, |
|
"step": 5310 |
|
}, |
|
{ |
|
"epoch": 15.07, |
|
"learning_rate": 2.3178947368421054e-05, |
|
"loss": 1.5005, |
|
"step": 5320 |
|
}, |
|
{ |
|
"epoch": 15.1, |
|
"learning_rate": 2.316315789473684e-05, |
|
"loss": 1.5189, |
|
"step": 5330 |
|
}, |
|
{ |
|
"epoch": 15.13, |
|
"learning_rate": 2.314736842105263e-05, |
|
"loss": 1.4973, |
|
"step": 5340 |
|
}, |
|
{ |
|
"epoch": 15.16, |
|
"learning_rate": 2.3131578947368422e-05, |
|
"loss": 1.5069, |
|
"step": 5350 |
|
}, |
|
{ |
|
"epoch": 15.18, |
|
"learning_rate": 2.3115789473684212e-05, |
|
"loss": 1.5069, |
|
"step": 5360 |
|
}, |
|
{ |
|
"epoch": 15.21, |
|
"learning_rate": 2.3100000000000002e-05, |
|
"loss": 1.5092, |
|
"step": 5370 |
|
}, |
|
{ |
|
"epoch": 15.24, |
|
"learning_rate": 2.308421052631579e-05, |
|
"loss": 1.5012, |
|
"step": 5380 |
|
}, |
|
{ |
|
"epoch": 15.27, |
|
"learning_rate": 2.306842105263158e-05, |
|
"loss": 1.5132, |
|
"step": 5390 |
|
}, |
|
{ |
|
"epoch": 15.3, |
|
"learning_rate": 2.3052631578947367e-05, |
|
"loss": 1.5186, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 15.32, |
|
"learning_rate": 2.303684210526316e-05, |
|
"loss": 1.5145, |
|
"step": 5410 |
|
}, |
|
{ |
|
"epoch": 15.35, |
|
"learning_rate": 2.3021052631578947e-05, |
|
"loss": 1.5089, |
|
"step": 5420 |
|
}, |
|
{ |
|
"epoch": 15.38, |
|
"learning_rate": 2.3005263157894738e-05, |
|
"loss": 1.4856, |
|
"step": 5430 |
|
}, |
|
{ |
|
"epoch": 15.41, |
|
"learning_rate": 2.2989473684210525e-05, |
|
"loss": 1.5063, |
|
"step": 5440 |
|
}, |
|
{ |
|
"epoch": 15.44, |
|
"learning_rate": 2.297368421052632e-05, |
|
"loss": 1.5018, |
|
"step": 5450 |
|
}, |
|
{ |
|
"epoch": 15.47, |
|
"learning_rate": 2.2957894736842105e-05, |
|
"loss": 1.5089, |
|
"step": 5460 |
|
}, |
|
{ |
|
"epoch": 15.49, |
|
"learning_rate": 2.2942105263157896e-05, |
|
"loss": 1.5036, |
|
"step": 5470 |
|
}, |
|
{ |
|
"epoch": 15.52, |
|
"learning_rate": 2.2926315789473683e-05, |
|
"loss": 1.517, |
|
"step": 5480 |
|
}, |
|
{ |
|
"epoch": 15.55, |
|
"learning_rate": 2.2910526315789473e-05, |
|
"loss": 1.5061, |
|
"step": 5490 |
|
}, |
|
{ |
|
"epoch": 15.58, |
|
"learning_rate": 2.2894736842105263e-05, |
|
"loss": 1.4994, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 15.61, |
|
"learning_rate": 2.2878947368421054e-05, |
|
"loss": 1.5256, |
|
"step": 5510 |
|
}, |
|
{ |
|
"epoch": 15.64, |
|
"learning_rate": 2.2863157894736844e-05, |
|
"loss": 1.5177, |
|
"step": 5520 |
|
}, |
|
{ |
|
"epoch": 15.66, |
|
"learning_rate": 2.284736842105263e-05, |
|
"loss": 1.5047, |
|
"step": 5530 |
|
}, |
|
{ |
|
"epoch": 15.69, |
|
"learning_rate": 2.283157894736842e-05, |
|
"loss": 1.5303, |
|
"step": 5540 |
|
}, |
|
{ |
|
"epoch": 15.72, |
|
"learning_rate": 2.281578947368421e-05, |
|
"loss": 1.5366, |
|
"step": 5550 |
|
}, |
|
{ |
|
"epoch": 15.75, |
|
"learning_rate": 2.2800000000000002e-05, |
|
"loss": 1.5397, |
|
"step": 5560 |
|
}, |
|
{ |
|
"epoch": 15.78, |
|
"learning_rate": 2.278421052631579e-05, |
|
"loss": 1.5016, |
|
"step": 5570 |
|
}, |
|
{ |
|
"epoch": 15.81, |
|
"learning_rate": 2.276842105263158e-05, |
|
"loss": 1.51, |
|
"step": 5580 |
|
}, |
|
{ |
|
"epoch": 15.83, |
|
"learning_rate": 2.275263157894737e-05, |
|
"loss": 1.5091, |
|
"step": 5590 |
|
}, |
|
{ |
|
"epoch": 15.86, |
|
"learning_rate": 2.273684210526316e-05, |
|
"loss": 1.5116, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 15.89, |
|
"learning_rate": 2.2721052631578947e-05, |
|
"loss": 1.503, |
|
"step": 5610 |
|
}, |
|
{ |
|
"epoch": 15.92, |
|
"learning_rate": 2.2705263157894737e-05, |
|
"loss": 1.5021, |
|
"step": 5620 |
|
}, |
|
{ |
|
"epoch": 15.95, |
|
"learning_rate": 2.2689473684210524e-05, |
|
"loss": 1.5252, |
|
"step": 5630 |
|
}, |
|
{ |
|
"epoch": 15.97, |
|
"learning_rate": 2.2673684210526318e-05, |
|
"loss": 1.5087, |
|
"step": 5640 |
|
}, |
|
{ |
|
"epoch": 16.01, |
|
"learning_rate": 2.2657894736842105e-05, |
|
"loss": 1.6352, |
|
"step": 5650 |
|
}, |
|
{ |
|
"epoch": 16.03, |
|
"learning_rate": 2.2642105263157895e-05, |
|
"loss": 1.4857, |
|
"step": 5660 |
|
}, |
|
{ |
|
"epoch": 16.06, |
|
"learning_rate": 2.2626315789473685e-05, |
|
"loss": 1.5041, |
|
"step": 5670 |
|
}, |
|
{ |
|
"epoch": 16.09, |
|
"learning_rate": 2.2610526315789472e-05, |
|
"loss": 1.544, |
|
"step": 5680 |
|
}, |
|
{ |
|
"epoch": 16.12, |
|
"learning_rate": 2.2594736842105266e-05, |
|
"loss": 1.5226, |
|
"step": 5690 |
|
}, |
|
{ |
|
"epoch": 16.15, |
|
"learning_rate": 2.2578947368421053e-05, |
|
"loss": 1.5118, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 16.18, |
|
"learning_rate": 2.2563157894736843e-05, |
|
"loss": 1.5095, |
|
"step": 5710 |
|
}, |
|
{ |
|
"epoch": 16.2, |
|
"learning_rate": 2.254736842105263e-05, |
|
"loss": 1.5032, |
|
"step": 5720 |
|
}, |
|
{ |
|
"epoch": 16.23, |
|
"learning_rate": 2.2531578947368424e-05, |
|
"loss": 1.4881, |
|
"step": 5730 |
|
}, |
|
{ |
|
"epoch": 16.26, |
|
"learning_rate": 2.251578947368421e-05, |
|
"loss": 1.4976, |
|
"step": 5740 |
|
}, |
|
{ |
|
"epoch": 16.29, |
|
"learning_rate": 2.25e-05, |
|
"loss": 1.5119, |
|
"step": 5750 |
|
}, |
|
{ |
|
"epoch": 16.32, |
|
"learning_rate": 2.248421052631579e-05, |
|
"loss": 1.5084, |
|
"step": 5760 |
|
}, |
|
{ |
|
"epoch": 16.34, |
|
"learning_rate": 2.246842105263158e-05, |
|
"loss": 1.5138, |
|
"step": 5770 |
|
}, |
|
{ |
|
"epoch": 16.37, |
|
"learning_rate": 2.245263157894737e-05, |
|
"loss": 1.4983, |
|
"step": 5780 |
|
}, |
|
{ |
|
"epoch": 16.4, |
|
"learning_rate": 2.243684210526316e-05, |
|
"loss": 1.4821, |
|
"step": 5790 |
|
}, |
|
{ |
|
"epoch": 16.43, |
|
"learning_rate": 2.2421052631578946e-05, |
|
"loss": 1.5076, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 16.46, |
|
"learning_rate": 2.2405263157894737e-05, |
|
"loss": 1.4939, |
|
"step": 5810 |
|
}, |
|
{ |
|
"epoch": 16.49, |
|
"learning_rate": 2.2389473684210527e-05, |
|
"loss": 1.4924, |
|
"step": 5820 |
|
}, |
|
{ |
|
"epoch": 16.51, |
|
"learning_rate": 2.2373684210526317e-05, |
|
"loss": 1.4835, |
|
"step": 5830 |
|
}, |
|
{ |
|
"epoch": 16.54, |
|
"learning_rate": 2.2357894736842108e-05, |
|
"loss": 1.4857, |
|
"step": 5840 |
|
}, |
|
{ |
|
"epoch": 16.57, |
|
"learning_rate": 2.2342105263157895e-05, |
|
"loss": 1.5067, |
|
"step": 5850 |
|
}, |
|
{ |
|
"epoch": 16.6, |
|
"learning_rate": 2.2326315789473685e-05, |
|
"loss": 1.5082, |
|
"step": 5860 |
|
}, |
|
{ |
|
"epoch": 16.63, |
|
"learning_rate": 2.2310526315789475e-05, |
|
"loss": 1.5156, |
|
"step": 5870 |
|
}, |
|
{ |
|
"epoch": 16.66, |
|
"learning_rate": 2.2294736842105266e-05, |
|
"loss": 1.5161, |
|
"step": 5880 |
|
}, |
|
{ |
|
"epoch": 16.68, |
|
"learning_rate": 2.2278947368421053e-05, |
|
"loss": 1.5042, |
|
"step": 5890 |
|
}, |
|
{ |
|
"epoch": 16.71, |
|
"learning_rate": 2.2263157894736843e-05, |
|
"loss": 1.5196, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 16.74, |
|
"learning_rate": 2.224736842105263e-05, |
|
"loss": 1.5198, |
|
"step": 5910 |
|
}, |
|
{ |
|
"epoch": 16.77, |
|
"learning_rate": 2.2231578947368424e-05, |
|
"loss": 1.5017, |
|
"step": 5920 |
|
}, |
|
{ |
|
"epoch": 16.8, |
|
"learning_rate": 2.221578947368421e-05, |
|
"loss": 1.5016, |
|
"step": 5930 |
|
}, |
|
{ |
|
"epoch": 16.83, |
|
"learning_rate": 2.22e-05, |
|
"loss": 1.487, |
|
"step": 5940 |
|
}, |
|
{ |
|
"epoch": 16.85, |
|
"learning_rate": 2.2184210526315788e-05, |
|
"loss": 1.5135, |
|
"step": 5950 |
|
}, |
|
{ |
|
"epoch": 16.88, |
|
"learning_rate": 2.216842105263158e-05, |
|
"loss": 1.4875, |
|
"step": 5960 |
|
}, |
|
{ |
|
"epoch": 16.91, |
|
"learning_rate": 2.215263157894737e-05, |
|
"loss": 1.5154, |
|
"step": 5970 |
|
}, |
|
{ |
|
"epoch": 16.94, |
|
"learning_rate": 2.213684210526316e-05, |
|
"loss": 1.5004, |
|
"step": 5980 |
|
}, |
|
{ |
|
"epoch": 16.97, |
|
"learning_rate": 2.2121052631578946e-05, |
|
"loss": 1.4995, |
|
"step": 5990 |
|
}, |
|
{ |
|
"epoch": 16.99, |
|
"learning_rate": 2.2105263157894736e-05, |
|
"loss": 1.4987, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 16.99, |
|
"eval_denotation_accuracy": 0.5732956552454963, |
|
"eval_loss": 2.120565891265869, |
|
"eval_runtime": 374.0788, |
|
"eval_samples_per_second": 7.568, |
|
"eval_steps_per_second": 1.893, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 17.03, |
|
"learning_rate": 2.208947368421053e-05, |
|
"loss": 1.6182, |
|
"step": 6010 |
|
}, |
|
{ |
|
"epoch": 17.05, |
|
"learning_rate": 2.2073684210526317e-05, |
|
"loss": 1.494, |
|
"step": 6020 |
|
}, |
|
{ |
|
"epoch": 17.08, |
|
"learning_rate": 2.2057894736842107e-05, |
|
"loss": 1.4837, |
|
"step": 6030 |
|
}, |
|
{ |
|
"epoch": 17.11, |
|
"learning_rate": 2.2042105263157894e-05, |
|
"loss": 1.4926, |
|
"step": 6040 |
|
}, |
|
{ |
|
"epoch": 17.14, |
|
"learning_rate": 2.2026315789473684e-05, |
|
"loss": 1.494, |
|
"step": 6050 |
|
}, |
|
{ |
|
"epoch": 17.17, |
|
"learning_rate": 2.2010526315789475e-05, |
|
"loss": 1.4765, |
|
"step": 6060 |
|
}, |
|
{ |
|
"epoch": 17.19, |
|
"learning_rate": 2.1994736842105265e-05, |
|
"loss": 1.4961, |
|
"step": 6070 |
|
}, |
|
{ |
|
"epoch": 17.22, |
|
"learning_rate": 2.1978947368421052e-05, |
|
"loss": 1.4857, |
|
"step": 6080 |
|
}, |
|
{ |
|
"epoch": 17.25, |
|
"learning_rate": 2.1963157894736842e-05, |
|
"loss": 1.5008, |
|
"step": 6090 |
|
}, |
|
{ |
|
"epoch": 17.28, |
|
"learning_rate": 2.1947368421052633e-05, |
|
"loss": 1.4987, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 17.31, |
|
"learning_rate": 2.1931578947368423e-05, |
|
"loss": 1.4967, |
|
"step": 6110 |
|
}, |
|
{ |
|
"epoch": 17.34, |
|
"learning_rate": 2.191578947368421e-05, |
|
"loss": 1.4939, |
|
"step": 6120 |
|
}, |
|
{ |
|
"epoch": 17.36, |
|
"learning_rate": 2.19e-05, |
|
"loss": 1.4838, |
|
"step": 6130 |
|
}, |
|
{ |
|
"epoch": 17.39, |
|
"learning_rate": 2.1884210526315787e-05, |
|
"loss": 1.5048, |
|
"step": 6140 |
|
}, |
|
{ |
|
"epoch": 17.42, |
|
"learning_rate": 2.186842105263158e-05, |
|
"loss": 1.494, |
|
"step": 6150 |
|
}, |
|
{ |
|
"epoch": 17.45, |
|
"learning_rate": 2.1852631578947368e-05, |
|
"loss": 1.5121, |
|
"step": 6160 |
|
}, |
|
{ |
|
"epoch": 17.48, |
|
"learning_rate": 2.183684210526316e-05, |
|
"loss": 1.5, |
|
"step": 6170 |
|
}, |
|
{ |
|
"epoch": 17.51, |
|
"learning_rate": 2.182105263157895e-05, |
|
"loss": 1.4775, |
|
"step": 6180 |
|
}, |
|
{ |
|
"epoch": 17.53, |
|
"learning_rate": 2.1805263157894736e-05, |
|
"loss": 1.4803, |
|
"step": 6190 |
|
}, |
|
{ |
|
"epoch": 17.56, |
|
"learning_rate": 2.178947368421053e-05, |
|
"loss": 1.4727, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 17.59, |
|
"learning_rate": 2.1773684210526316e-05, |
|
"loss": 1.4691, |
|
"step": 6210 |
|
}, |
|
{ |
|
"epoch": 17.62, |
|
"learning_rate": 2.1757894736842107e-05, |
|
"loss": 1.4769, |
|
"step": 6220 |
|
}, |
|
{ |
|
"epoch": 17.65, |
|
"learning_rate": 2.1742105263157894e-05, |
|
"loss": 1.4741, |
|
"step": 6230 |
|
}, |
|
{ |
|
"epoch": 17.68, |
|
"learning_rate": 2.1726315789473687e-05, |
|
"loss": 1.4794, |
|
"step": 6240 |
|
}, |
|
{ |
|
"epoch": 17.7, |
|
"learning_rate": 2.1710526315789474e-05, |
|
"loss": 1.4919, |
|
"step": 6250 |
|
}, |
|
{ |
|
"epoch": 17.73, |
|
"learning_rate": 2.1694736842105265e-05, |
|
"loss": 1.5191, |
|
"step": 6260 |
|
}, |
|
{ |
|
"epoch": 17.76, |
|
"learning_rate": 2.167894736842105e-05, |
|
"loss": 1.4839, |
|
"step": 6270 |
|
}, |
|
{ |
|
"epoch": 17.79, |
|
"learning_rate": 2.1663157894736842e-05, |
|
"loss": 1.5003, |
|
"step": 6280 |
|
}, |
|
{ |
|
"epoch": 17.82, |
|
"learning_rate": 2.1647368421052632e-05, |
|
"loss": 1.4856, |
|
"step": 6290 |
|
}, |
|
{ |
|
"epoch": 17.84, |
|
"learning_rate": 2.1631578947368423e-05, |
|
"loss": 1.5092, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 17.87, |
|
"learning_rate": 2.161578947368421e-05, |
|
"loss": 1.4915, |
|
"step": 6310 |
|
}, |
|
{ |
|
"epoch": 17.9, |
|
"learning_rate": 2.16e-05, |
|
"loss": 1.5019, |
|
"step": 6320 |
|
}, |
|
{ |
|
"epoch": 17.93, |
|
"learning_rate": 2.1584210526315787e-05, |
|
"loss": 1.4821, |
|
"step": 6330 |
|
}, |
|
{ |
|
"epoch": 17.96, |
|
"learning_rate": 2.156842105263158e-05, |
|
"loss": 1.4982, |
|
"step": 6340 |
|
}, |
|
{ |
|
"epoch": 17.99, |
|
"learning_rate": 2.155263157894737e-05, |
|
"loss": 1.485, |
|
"step": 6350 |
|
}, |
|
{ |
|
"epoch": 18.02, |
|
"learning_rate": 2.1536842105263158e-05, |
|
"loss": 1.6133, |
|
"step": 6360 |
|
}, |
|
{ |
|
"epoch": 18.05, |
|
"learning_rate": 2.1521052631578948e-05, |
|
"loss": 1.484, |
|
"step": 6370 |
|
}, |
|
{ |
|
"epoch": 18.07, |
|
"learning_rate": 2.150526315789474e-05, |
|
"loss": 1.4784, |
|
"step": 6380 |
|
}, |
|
{ |
|
"epoch": 18.1, |
|
"learning_rate": 2.148947368421053e-05, |
|
"loss": 1.4662, |
|
"step": 6390 |
|
}, |
|
{ |
|
"epoch": 18.13, |
|
"learning_rate": 2.1473684210526316e-05, |
|
"loss": 1.4904, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 18.16, |
|
"learning_rate": 2.1457894736842106e-05, |
|
"loss": 1.4876, |
|
"step": 6410 |
|
}, |
|
{ |
|
"epoch": 18.19, |
|
"learning_rate": 2.1442105263157893e-05, |
|
"loss": 1.5111, |
|
"step": 6420 |
|
}, |
|
{ |
|
"epoch": 18.21, |
|
"learning_rate": 2.1426315789473687e-05, |
|
"loss": 1.4777, |
|
"step": 6430 |
|
}, |
|
{ |
|
"epoch": 18.24, |
|
"learning_rate": 2.1410526315789474e-05, |
|
"loss": 1.4912, |
|
"step": 6440 |
|
}, |
|
{ |
|
"epoch": 18.27, |
|
"learning_rate": 2.1394736842105264e-05, |
|
"loss": 1.4765, |
|
"step": 6450 |
|
}, |
|
{ |
|
"epoch": 18.3, |
|
"learning_rate": 2.137894736842105e-05, |
|
"loss": 1.4948, |
|
"step": 6460 |
|
}, |
|
{ |
|
"epoch": 18.33, |
|
"learning_rate": 2.136315789473684e-05, |
|
"loss": 1.475, |
|
"step": 6470 |
|
}, |
|
{ |
|
"epoch": 18.36, |
|
"learning_rate": 2.1347368421052632e-05, |
|
"loss": 1.4844, |
|
"step": 6480 |
|
}, |
|
{ |
|
"epoch": 18.38, |
|
"learning_rate": 2.1331578947368422e-05, |
|
"loss": 1.4933, |
|
"step": 6490 |
|
}, |
|
{ |
|
"epoch": 18.41, |
|
"learning_rate": 2.1315789473684212e-05, |
|
"loss": 1.4813, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 18.44, |
|
"learning_rate": 2.13e-05, |
|
"loss": 1.4783, |
|
"step": 6510 |
|
}, |
|
{ |
|
"epoch": 18.47, |
|
"learning_rate": 2.1284210526315793e-05, |
|
"loss": 1.4781, |
|
"step": 6520 |
|
}, |
|
{ |
|
"epoch": 18.5, |
|
"learning_rate": 2.126842105263158e-05, |
|
"loss": 1.4926, |
|
"step": 6530 |
|
}, |
|
{ |
|
"epoch": 18.53, |
|
"learning_rate": 2.125263157894737e-05, |
|
"loss": 1.4912, |
|
"step": 6540 |
|
}, |
|
{ |
|
"epoch": 18.55, |
|
"learning_rate": 2.1236842105263157e-05, |
|
"loss": 1.4714, |
|
"step": 6550 |
|
}, |
|
{ |
|
"epoch": 18.58, |
|
"learning_rate": 2.1221052631578948e-05, |
|
"loss": 1.4949, |
|
"step": 6560 |
|
}, |
|
{ |
|
"epoch": 18.61, |
|
"learning_rate": 2.1205263157894738e-05, |
|
"loss": 1.4994, |
|
"step": 6570 |
|
}, |
|
{ |
|
"epoch": 18.64, |
|
"learning_rate": 2.118947368421053e-05, |
|
"loss": 1.4855, |
|
"step": 6580 |
|
}, |
|
{ |
|
"epoch": 18.67, |
|
"learning_rate": 2.1173684210526315e-05, |
|
"loss": 1.4913, |
|
"step": 6590 |
|
}, |
|
{ |
|
"epoch": 18.7, |
|
"learning_rate": 2.1157894736842106e-05, |
|
"loss": 1.4812, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 18.72, |
|
"learning_rate": 2.1142105263157893e-05, |
|
"loss": 1.4969, |
|
"step": 6610 |
|
}, |
|
{ |
|
"epoch": 18.75, |
|
"learning_rate": 2.1126315789473686e-05, |
|
"loss": 1.4852, |
|
"step": 6620 |
|
}, |
|
{ |
|
"epoch": 18.78, |
|
"learning_rate": 2.1110526315789473e-05, |
|
"loss": 1.4857, |
|
"step": 6630 |
|
}, |
|
{ |
|
"epoch": 18.81, |
|
"learning_rate": 2.1094736842105264e-05, |
|
"loss": 1.4952, |
|
"step": 6640 |
|
}, |
|
{ |
|
"epoch": 18.84, |
|
"learning_rate": 2.107894736842105e-05, |
|
"loss": 1.4932, |
|
"step": 6650 |
|
}, |
|
{ |
|
"epoch": 18.86, |
|
"learning_rate": 2.1063157894736844e-05, |
|
"loss": 1.4755, |
|
"step": 6660 |
|
}, |
|
{ |
|
"epoch": 18.89, |
|
"learning_rate": 2.1047368421052635e-05, |
|
"loss": 1.4817, |
|
"step": 6670 |
|
}, |
|
{ |
|
"epoch": 18.92, |
|
"learning_rate": 2.103157894736842e-05, |
|
"loss": 1.4837, |
|
"step": 6680 |
|
}, |
|
{ |
|
"epoch": 18.95, |
|
"learning_rate": 2.1015789473684212e-05, |
|
"loss": 1.4852, |
|
"step": 6690 |
|
}, |
|
{ |
|
"epoch": 18.98, |
|
"learning_rate": 2.1e-05, |
|
"loss": 1.4877, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 19.01, |
|
"learning_rate": 2.0984210526315793e-05, |
|
"loss": 1.6275, |
|
"step": 6710 |
|
}, |
|
{ |
|
"epoch": 19.04, |
|
"learning_rate": 2.096842105263158e-05, |
|
"loss": 1.4863, |
|
"step": 6720 |
|
}, |
|
{ |
|
"epoch": 19.06, |
|
"learning_rate": 2.095263157894737e-05, |
|
"loss": 1.4741, |
|
"step": 6730 |
|
}, |
|
{ |
|
"epoch": 19.09, |
|
"learning_rate": 2.0936842105263157e-05, |
|
"loss": 1.4807, |
|
"step": 6740 |
|
}, |
|
{ |
|
"epoch": 19.12, |
|
"learning_rate": 2.0921052631578947e-05, |
|
"loss": 1.4751, |
|
"step": 6750 |
|
}, |
|
{ |
|
"epoch": 19.15, |
|
"learning_rate": 2.0905263157894737e-05, |
|
"loss": 1.4796, |
|
"step": 6760 |
|
}, |
|
{ |
|
"epoch": 19.18, |
|
"learning_rate": 2.0889473684210528e-05, |
|
"loss": 1.4745, |
|
"step": 6770 |
|
}, |
|
{ |
|
"epoch": 19.21, |
|
"learning_rate": 2.0873684210526315e-05, |
|
"loss": 1.4767, |
|
"step": 6780 |
|
}, |
|
{ |
|
"epoch": 19.23, |
|
"learning_rate": 2.0857894736842105e-05, |
|
"loss": 1.4831, |
|
"step": 6790 |
|
}, |
|
{ |
|
"epoch": 19.26, |
|
"learning_rate": 2.0842105263157895e-05, |
|
"loss": 1.4649, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 19.29, |
|
"learning_rate": 2.0826315789473686e-05, |
|
"loss": 1.4696, |
|
"step": 6810 |
|
}, |
|
{ |
|
"epoch": 19.32, |
|
"learning_rate": 2.0810526315789473e-05, |
|
"loss": 1.472, |
|
"step": 6820 |
|
}, |
|
{ |
|
"epoch": 19.35, |
|
"learning_rate": 2.0794736842105263e-05, |
|
"loss": 1.4631, |
|
"step": 6830 |
|
}, |
|
{ |
|
"epoch": 19.38, |
|
"learning_rate": 2.0778947368421053e-05, |
|
"loss": 1.4755, |
|
"step": 6840 |
|
}, |
|
{ |
|
"epoch": 19.4, |
|
"learning_rate": 2.0763157894736844e-05, |
|
"loss": 1.483, |
|
"step": 6850 |
|
}, |
|
{ |
|
"epoch": 19.43, |
|
"learning_rate": 2.0747368421052634e-05, |
|
"loss": 1.4855, |
|
"step": 6860 |
|
}, |
|
{ |
|
"epoch": 19.46, |
|
"learning_rate": 2.073157894736842e-05, |
|
"loss": 1.476, |
|
"step": 6870 |
|
}, |
|
{ |
|
"epoch": 19.49, |
|
"learning_rate": 2.071578947368421e-05, |
|
"loss": 1.4683, |
|
"step": 6880 |
|
}, |
|
{ |
|
"epoch": 19.52, |
|
"learning_rate": 2.07e-05, |
|
"loss": 1.4768, |
|
"step": 6890 |
|
}, |
|
{ |
|
"epoch": 19.55, |
|
"learning_rate": 2.0684210526315792e-05, |
|
"loss": 1.4775, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 19.57, |
|
"learning_rate": 2.066842105263158e-05, |
|
"loss": 1.476, |
|
"step": 6910 |
|
}, |
|
{ |
|
"epoch": 19.6, |
|
"learning_rate": 2.065263157894737e-05, |
|
"loss": 1.4981, |
|
"step": 6920 |
|
}, |
|
{ |
|
"epoch": 19.63, |
|
"learning_rate": 2.0636842105263156e-05, |
|
"loss": 1.4777, |
|
"step": 6930 |
|
}, |
|
{ |
|
"epoch": 19.66, |
|
"learning_rate": 2.062105263157895e-05, |
|
"loss": 1.4649, |
|
"step": 6940 |
|
}, |
|
{ |
|
"epoch": 19.69, |
|
"learning_rate": 2.0605263157894737e-05, |
|
"loss": 1.4702, |
|
"step": 6950 |
|
}, |
|
{ |
|
"epoch": 19.71, |
|
"learning_rate": 2.0589473684210527e-05, |
|
"loss": 1.4714, |
|
"step": 6960 |
|
}, |
|
{ |
|
"epoch": 19.74, |
|
"learning_rate": 2.0573684210526314e-05, |
|
"loss": 1.4763, |
|
"step": 6970 |
|
}, |
|
{ |
|
"epoch": 19.77, |
|
"learning_rate": 2.0557894736842105e-05, |
|
"loss": 1.4748, |
|
"step": 6980 |
|
}, |
|
{ |
|
"epoch": 19.8, |
|
"learning_rate": 2.0542105263157895e-05, |
|
"loss": 1.4655, |
|
"step": 6990 |
|
}, |
|
{ |
|
"epoch": 19.83, |
|
"learning_rate": 2.0526315789473685e-05, |
|
"loss": 1.4764, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 19.83, |
|
"eval_denotation_accuracy": 0.5810667608618862, |
|
"eval_loss": 2.136824607849121, |
|
"eval_runtime": 369.0602, |
|
"eval_samples_per_second": 7.671, |
|
"eval_steps_per_second": 1.918, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 19.86, |
|
"learning_rate": 2.0510526315789476e-05, |
|
"loss": 1.4886, |
|
"step": 7010 |
|
}, |
|
{ |
|
"epoch": 19.88, |
|
"learning_rate": 2.0494736842105263e-05, |
|
"loss": 1.4839, |
|
"step": 7020 |
|
}, |
|
{ |
|
"epoch": 19.91, |
|
"learning_rate": 2.0478947368421053e-05, |
|
"loss": 1.4785, |
|
"step": 7030 |
|
}, |
|
{ |
|
"epoch": 19.94, |
|
"learning_rate": 2.0463157894736843e-05, |
|
"loss": 1.485, |
|
"step": 7040 |
|
}, |
|
{ |
|
"epoch": 19.97, |
|
"learning_rate": 2.0447368421052634e-05, |
|
"loss": 1.4685, |
|
"step": 7050 |
|
}, |
|
{ |
|
"epoch": 20.0, |
|
"learning_rate": 2.043157894736842e-05, |
|
"loss": 1.4821, |
|
"step": 7060 |
|
}, |
|
{ |
|
"epoch": 20.03, |
|
"learning_rate": 2.041578947368421e-05, |
|
"loss": 1.6062, |
|
"step": 7070 |
|
}, |
|
{ |
|
"epoch": 20.06, |
|
"learning_rate": 2.04e-05, |
|
"loss": 1.4742, |
|
"step": 7080 |
|
}, |
|
{ |
|
"epoch": 20.08, |
|
"learning_rate": 2.038421052631579e-05, |
|
"loss": 1.4791, |
|
"step": 7090 |
|
}, |
|
{ |
|
"epoch": 20.11, |
|
"learning_rate": 2.036842105263158e-05, |
|
"loss": 1.4936, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 20.14, |
|
"learning_rate": 2.035263157894737e-05, |
|
"loss": 1.4764, |
|
"step": 7110 |
|
}, |
|
{ |
|
"epoch": 20.17, |
|
"learning_rate": 2.0336842105263156e-05, |
|
"loss": 1.4798, |
|
"step": 7120 |
|
}, |
|
{ |
|
"epoch": 20.2, |
|
"learning_rate": 2.032105263157895e-05, |
|
"loss": 1.4762, |
|
"step": 7130 |
|
}, |
|
{ |
|
"epoch": 20.23, |
|
"learning_rate": 2.0305263157894736e-05, |
|
"loss": 1.4768, |
|
"step": 7140 |
|
}, |
|
{ |
|
"epoch": 20.25, |
|
"learning_rate": 2.0289473684210527e-05, |
|
"loss": 1.4949, |
|
"step": 7150 |
|
}, |
|
{ |
|
"epoch": 20.28, |
|
"learning_rate": 2.0273684210526314e-05, |
|
"loss": 1.4789, |
|
"step": 7160 |
|
}, |
|
{ |
|
"epoch": 20.31, |
|
"learning_rate": 2.0257894736842104e-05, |
|
"loss": 1.4794, |
|
"step": 7170 |
|
}, |
|
{ |
|
"epoch": 20.34, |
|
"learning_rate": 2.0242105263157898e-05, |
|
"loss": 1.4721, |
|
"step": 7180 |
|
}, |
|
{ |
|
"epoch": 20.37, |
|
"learning_rate": 2.0226315789473685e-05, |
|
"loss": 1.4656, |
|
"step": 7190 |
|
}, |
|
{ |
|
"epoch": 20.4, |
|
"learning_rate": 2.0210526315789475e-05, |
|
"loss": 1.4754, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 20.42, |
|
"learning_rate": 2.0194736842105262e-05, |
|
"loss": 1.4671, |
|
"step": 7210 |
|
}, |
|
{ |
|
"epoch": 20.45, |
|
"learning_rate": 2.0178947368421056e-05, |
|
"loss": 1.462, |
|
"step": 7220 |
|
}, |
|
{ |
|
"epoch": 20.48, |
|
"learning_rate": 2.0163157894736843e-05, |
|
"loss": 1.4633, |
|
"step": 7230 |
|
}, |
|
{ |
|
"epoch": 20.51, |
|
"learning_rate": 2.0147368421052633e-05, |
|
"loss": 1.4769, |
|
"step": 7240 |
|
}, |
|
{ |
|
"epoch": 20.54, |
|
"learning_rate": 2.013157894736842e-05, |
|
"loss": 1.5024, |
|
"step": 7250 |
|
}, |
|
{ |
|
"epoch": 20.57, |
|
"learning_rate": 2.011578947368421e-05, |
|
"loss": 1.4809, |
|
"step": 7260 |
|
}, |
|
{ |
|
"epoch": 20.59, |
|
"learning_rate": 2.01e-05, |
|
"loss": 1.4737, |
|
"step": 7270 |
|
}, |
|
{ |
|
"epoch": 20.62, |
|
"learning_rate": 2.008421052631579e-05, |
|
"loss": 1.4723, |
|
"step": 7280 |
|
}, |
|
{ |
|
"epoch": 20.65, |
|
"learning_rate": 2.0068421052631578e-05, |
|
"loss": 1.4826, |
|
"step": 7290 |
|
}, |
|
{ |
|
"epoch": 20.68, |
|
"learning_rate": 2.0052631578947368e-05, |
|
"loss": 1.4702, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 20.71, |
|
"learning_rate": 2.0036842105263155e-05, |
|
"loss": 1.4806, |
|
"step": 7310 |
|
}, |
|
{ |
|
"epoch": 20.73, |
|
"learning_rate": 2.002105263157895e-05, |
|
"loss": 1.4684, |
|
"step": 7320 |
|
}, |
|
{ |
|
"epoch": 20.76, |
|
"learning_rate": 2.0005263157894736e-05, |
|
"loss": 1.4692, |
|
"step": 7330 |
|
}, |
|
{ |
|
"epoch": 20.79, |
|
"learning_rate": 1.9989473684210526e-05, |
|
"loss": 1.4731, |
|
"step": 7340 |
|
}, |
|
{ |
|
"epoch": 20.82, |
|
"learning_rate": 1.9973684210526317e-05, |
|
"loss": 1.4706, |
|
"step": 7350 |
|
}, |
|
{ |
|
"epoch": 20.85, |
|
"learning_rate": 1.9957894736842107e-05, |
|
"loss": 1.4721, |
|
"step": 7360 |
|
}, |
|
{ |
|
"epoch": 20.88, |
|
"learning_rate": 1.9942105263157897e-05, |
|
"loss": 1.4863, |
|
"step": 7370 |
|
}, |
|
{ |
|
"epoch": 20.9, |
|
"learning_rate": 1.9926315789473684e-05, |
|
"loss": 1.4697, |
|
"step": 7380 |
|
}, |
|
{ |
|
"epoch": 20.93, |
|
"learning_rate": 1.9910526315789475e-05, |
|
"loss": 1.4793, |
|
"step": 7390 |
|
}, |
|
{ |
|
"epoch": 20.96, |
|
"learning_rate": 1.989473684210526e-05, |
|
"loss": 1.4908, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 20.99, |
|
"learning_rate": 1.9878947368421055e-05, |
|
"loss": 1.4859, |
|
"step": 7410 |
|
}, |
|
{ |
|
"epoch": 21.02, |
|
"learning_rate": 1.9863157894736842e-05, |
|
"loss": 1.5996, |
|
"step": 7420 |
|
}, |
|
{ |
|
"epoch": 21.05, |
|
"learning_rate": 1.9847368421052633e-05, |
|
"loss": 1.4811, |
|
"step": 7430 |
|
}, |
|
{ |
|
"epoch": 21.08, |
|
"learning_rate": 1.983157894736842e-05, |
|
"loss": 1.4806, |
|
"step": 7440 |
|
}, |
|
{ |
|
"epoch": 21.1, |
|
"learning_rate": 1.9815789473684213e-05, |
|
"loss": 1.4688, |
|
"step": 7450 |
|
}, |
|
{ |
|
"epoch": 21.13, |
|
"learning_rate": 1.98e-05, |
|
"loss": 1.484, |
|
"step": 7460 |
|
}, |
|
{ |
|
"epoch": 21.16, |
|
"learning_rate": 1.978421052631579e-05, |
|
"loss": 1.4733, |
|
"step": 7470 |
|
}, |
|
{ |
|
"epoch": 21.19, |
|
"learning_rate": 1.9768421052631577e-05, |
|
"loss": 1.4781, |
|
"step": 7480 |
|
}, |
|
{ |
|
"epoch": 21.22, |
|
"learning_rate": 1.9752631578947368e-05, |
|
"loss": 1.4683, |
|
"step": 7490 |
|
}, |
|
{ |
|
"epoch": 21.25, |
|
"learning_rate": 1.9736842105263158e-05, |
|
"loss": 1.481, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 21.27, |
|
"learning_rate": 1.972105263157895e-05, |
|
"loss": 1.4658, |
|
"step": 7510 |
|
}, |
|
{ |
|
"epoch": 21.3, |
|
"learning_rate": 1.970526315789474e-05, |
|
"loss": 1.4639, |
|
"step": 7520 |
|
}, |
|
{ |
|
"epoch": 21.33, |
|
"learning_rate": 1.9689473684210526e-05, |
|
"loss": 1.4731, |
|
"step": 7530 |
|
}, |
|
{ |
|
"epoch": 21.36, |
|
"learning_rate": 1.9673684210526316e-05, |
|
"loss": 1.4683, |
|
"step": 7540 |
|
}, |
|
{ |
|
"epoch": 21.39, |
|
"learning_rate": 1.9657894736842106e-05, |
|
"loss": 1.4687, |
|
"step": 7550 |
|
}, |
|
{ |
|
"epoch": 21.42, |
|
"learning_rate": 1.9642105263157897e-05, |
|
"loss": 1.4824, |
|
"step": 7560 |
|
}, |
|
{ |
|
"epoch": 21.44, |
|
"learning_rate": 1.9626315789473684e-05, |
|
"loss": 1.47, |
|
"step": 7570 |
|
}, |
|
{ |
|
"epoch": 21.47, |
|
"learning_rate": 1.9610526315789474e-05, |
|
"loss": 1.4842, |
|
"step": 7580 |
|
}, |
|
{ |
|
"epoch": 21.5, |
|
"learning_rate": 1.9594736842105264e-05, |
|
"loss": 1.4659, |
|
"step": 7590 |
|
}, |
|
{ |
|
"epoch": 21.53, |
|
"learning_rate": 1.9578947368421055e-05, |
|
"loss": 1.4776, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 21.56, |
|
"learning_rate": 1.956315789473684e-05, |
|
"loss": 1.4647, |
|
"step": 7610 |
|
}, |
|
{ |
|
"epoch": 21.58, |
|
"learning_rate": 1.9547368421052632e-05, |
|
"loss": 1.5014, |
|
"step": 7620 |
|
}, |
|
{ |
|
"epoch": 21.61, |
|
"learning_rate": 1.953157894736842e-05, |
|
"loss": 1.4737, |
|
"step": 7630 |
|
}, |
|
{ |
|
"epoch": 21.64, |
|
"learning_rate": 1.9515789473684213e-05, |
|
"loss": 1.4659, |
|
"step": 7640 |
|
}, |
|
{ |
|
"epoch": 21.67, |
|
"learning_rate": 1.95e-05, |
|
"loss": 1.4784, |
|
"step": 7650 |
|
}, |
|
{ |
|
"epoch": 21.7, |
|
"learning_rate": 1.948421052631579e-05, |
|
"loss": 1.4617, |
|
"step": 7660 |
|
}, |
|
{ |
|
"epoch": 21.73, |
|
"learning_rate": 1.946842105263158e-05, |
|
"loss": 1.4775, |
|
"step": 7670 |
|
}, |
|
{ |
|
"epoch": 21.75, |
|
"learning_rate": 1.9452631578947367e-05, |
|
"loss": 1.4701, |
|
"step": 7680 |
|
}, |
|
{ |
|
"epoch": 21.78, |
|
"learning_rate": 1.943684210526316e-05, |
|
"loss": 1.4719, |
|
"step": 7690 |
|
}, |
|
{ |
|
"epoch": 21.81, |
|
"learning_rate": 1.9421052631578948e-05, |
|
"loss": 1.4584, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 21.84, |
|
"learning_rate": 1.9405263157894738e-05, |
|
"loss": 1.479, |
|
"step": 7710 |
|
}, |
|
{ |
|
"epoch": 21.87, |
|
"learning_rate": 1.9389473684210525e-05, |
|
"loss": 1.4675, |
|
"step": 7720 |
|
}, |
|
{ |
|
"epoch": 21.9, |
|
"learning_rate": 1.937368421052632e-05, |
|
"loss": 1.4719, |
|
"step": 7730 |
|
}, |
|
{ |
|
"epoch": 21.92, |
|
"learning_rate": 1.9357894736842106e-05, |
|
"loss": 1.4812, |
|
"step": 7740 |
|
}, |
|
{ |
|
"epoch": 21.95, |
|
"learning_rate": 1.9342105263157896e-05, |
|
"loss": 1.484, |
|
"step": 7750 |
|
}, |
|
{ |
|
"epoch": 21.98, |
|
"learning_rate": 1.9326315789473683e-05, |
|
"loss": 1.4798, |
|
"step": 7760 |
|
}, |
|
{ |
|
"epoch": 22.01, |
|
"learning_rate": 1.9310526315789474e-05, |
|
"loss": 1.5941, |
|
"step": 7770 |
|
}, |
|
{ |
|
"epoch": 22.04, |
|
"learning_rate": 1.9294736842105264e-05, |
|
"loss": 1.4656, |
|
"step": 7780 |
|
}, |
|
{ |
|
"epoch": 22.07, |
|
"learning_rate": 1.9278947368421054e-05, |
|
"loss": 1.4704, |
|
"step": 7790 |
|
}, |
|
{ |
|
"epoch": 22.1, |
|
"learning_rate": 1.926315789473684e-05, |
|
"loss": 1.468, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 22.12, |
|
"learning_rate": 1.924736842105263e-05, |
|
"loss": 1.4671, |
|
"step": 7810 |
|
}, |
|
{ |
|
"epoch": 22.15, |
|
"learning_rate": 1.923157894736842e-05, |
|
"loss": 1.4543, |
|
"step": 7820 |
|
}, |
|
{ |
|
"epoch": 22.18, |
|
"learning_rate": 1.9215789473684212e-05, |
|
"loss": 1.4557, |
|
"step": 7830 |
|
}, |
|
{ |
|
"epoch": 22.21, |
|
"learning_rate": 1.9200000000000003e-05, |
|
"loss": 1.4602, |
|
"step": 7840 |
|
}, |
|
{ |
|
"epoch": 22.24, |
|
"learning_rate": 1.918421052631579e-05, |
|
"loss": 1.4582, |
|
"step": 7850 |
|
}, |
|
{ |
|
"epoch": 22.27, |
|
"learning_rate": 1.916842105263158e-05, |
|
"loss": 1.4604, |
|
"step": 7860 |
|
}, |
|
{ |
|
"epoch": 22.29, |
|
"learning_rate": 1.915263157894737e-05, |
|
"loss": 1.4581, |
|
"step": 7870 |
|
}, |
|
{ |
|
"epoch": 22.32, |
|
"learning_rate": 1.913684210526316e-05, |
|
"loss": 1.4616, |
|
"step": 7880 |
|
}, |
|
{ |
|
"epoch": 22.35, |
|
"learning_rate": 1.9121052631578947e-05, |
|
"loss": 1.4587, |
|
"step": 7890 |
|
}, |
|
{ |
|
"epoch": 22.38, |
|
"learning_rate": 1.9105263157894738e-05, |
|
"loss": 1.4643, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 22.41, |
|
"learning_rate": 1.9089473684210525e-05, |
|
"loss": 1.4661, |
|
"step": 7910 |
|
}, |
|
{ |
|
"epoch": 22.44, |
|
"learning_rate": 1.907368421052632e-05, |
|
"loss": 1.4728, |
|
"step": 7920 |
|
}, |
|
{ |
|
"epoch": 22.46, |
|
"learning_rate": 1.9057894736842105e-05, |
|
"loss": 1.4609, |
|
"step": 7930 |
|
}, |
|
{ |
|
"epoch": 22.49, |
|
"learning_rate": 1.9042105263157896e-05, |
|
"loss": 1.4672, |
|
"step": 7940 |
|
}, |
|
{ |
|
"epoch": 22.52, |
|
"learning_rate": 1.9026315789473683e-05, |
|
"loss": 1.4643, |
|
"step": 7950 |
|
}, |
|
{ |
|
"epoch": 22.55, |
|
"learning_rate": 1.9010526315789473e-05, |
|
"loss": 1.4592, |
|
"step": 7960 |
|
}, |
|
{ |
|
"epoch": 22.58, |
|
"learning_rate": 1.8994736842105263e-05, |
|
"loss": 1.4795, |
|
"step": 7970 |
|
}, |
|
{ |
|
"epoch": 22.6, |
|
"learning_rate": 1.8978947368421054e-05, |
|
"loss": 1.4669, |
|
"step": 7980 |
|
}, |
|
{ |
|
"epoch": 22.63, |
|
"learning_rate": 1.896315789473684e-05, |
|
"loss": 1.4639, |
|
"step": 7990 |
|
}, |
|
{ |
|
"epoch": 22.66, |
|
"learning_rate": 1.894736842105263e-05, |
|
"loss": 1.4649, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 22.66, |
|
"eval_denotation_accuracy": 0.5694101024373013, |
|
"eval_loss": 2.1465563774108887, |
|
"eval_runtime": 367.8906, |
|
"eval_samples_per_second": 7.695, |
|
"eval_steps_per_second": 1.924, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 22.69, |
|
"learning_rate": 1.8931578947368425e-05, |
|
"loss": 1.4677, |
|
"step": 8010 |
|
}, |
|
{ |
|
"epoch": 22.72, |
|
"learning_rate": 1.891578947368421e-05, |
|
"loss": 1.4773, |
|
"step": 8020 |
|
}, |
|
{ |
|
"epoch": 22.75, |
|
"learning_rate": 1.8900000000000002e-05, |
|
"loss": 1.4968, |
|
"step": 8030 |
|
}, |
|
{ |
|
"epoch": 22.77, |
|
"learning_rate": 1.888421052631579e-05, |
|
"loss": 1.4555, |
|
"step": 8040 |
|
}, |
|
{ |
|
"epoch": 22.8, |
|
"learning_rate": 1.886842105263158e-05, |
|
"loss": 1.4749, |
|
"step": 8050 |
|
}, |
|
{ |
|
"epoch": 22.83, |
|
"learning_rate": 1.885263157894737e-05, |
|
"loss": 1.4622, |
|
"step": 8060 |
|
}, |
|
{ |
|
"epoch": 22.86, |
|
"learning_rate": 1.883684210526316e-05, |
|
"loss": 1.4731, |
|
"step": 8070 |
|
}, |
|
{ |
|
"epoch": 22.89, |
|
"learning_rate": 1.8821052631578947e-05, |
|
"loss": 1.4608, |
|
"step": 8080 |
|
}, |
|
{ |
|
"epoch": 22.92, |
|
"learning_rate": 1.8805263157894737e-05, |
|
"loss": 1.4608, |
|
"step": 8090 |
|
}, |
|
{ |
|
"epoch": 22.94, |
|
"learning_rate": 1.8789473684210524e-05, |
|
"loss": 1.4561, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 22.97, |
|
"learning_rate": 1.8773684210526318e-05, |
|
"loss": 1.4714, |
|
"step": 8110 |
|
}, |
|
{ |
|
"epoch": 23.0, |
|
"learning_rate": 1.8757894736842105e-05, |
|
"loss": 1.5981, |
|
"step": 8120 |
|
}, |
|
{ |
|
"epoch": 23.03, |
|
"learning_rate": 1.8742105263157895e-05, |
|
"loss": 1.4634, |
|
"step": 8130 |
|
}, |
|
{ |
|
"epoch": 23.06, |
|
"learning_rate": 1.8726315789473682e-05, |
|
"loss": 1.4691, |
|
"step": 8140 |
|
}, |
|
{ |
|
"epoch": 23.09, |
|
"learning_rate": 1.8710526315789476e-05, |
|
"loss": 1.4606, |
|
"step": 8150 |
|
}, |
|
{ |
|
"epoch": 23.12, |
|
"learning_rate": 1.8694736842105263e-05, |
|
"loss": 1.4549, |
|
"step": 8160 |
|
}, |
|
{ |
|
"epoch": 23.14, |
|
"learning_rate": 1.8678947368421053e-05, |
|
"loss": 1.4595, |
|
"step": 8170 |
|
}, |
|
{ |
|
"epoch": 23.17, |
|
"learning_rate": 1.8663157894736844e-05, |
|
"loss": 1.4512, |
|
"step": 8180 |
|
}, |
|
{ |
|
"epoch": 23.2, |
|
"learning_rate": 1.864736842105263e-05, |
|
"loss": 1.4609, |
|
"step": 8190 |
|
}, |
|
{ |
|
"epoch": 23.23, |
|
"learning_rate": 1.8631578947368424e-05, |
|
"loss": 1.4571, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 23.26, |
|
"learning_rate": 1.861578947368421e-05, |
|
"loss": 1.455, |
|
"step": 8210 |
|
}, |
|
{ |
|
"epoch": 23.29, |
|
"learning_rate": 1.86e-05, |
|
"loss": 1.4581, |
|
"step": 8220 |
|
}, |
|
{ |
|
"epoch": 23.31, |
|
"learning_rate": 1.858421052631579e-05, |
|
"loss": 1.4637, |
|
"step": 8230 |
|
}, |
|
{ |
|
"epoch": 23.34, |
|
"learning_rate": 1.856842105263158e-05, |
|
"loss": 1.4472, |
|
"step": 8240 |
|
}, |
|
{ |
|
"epoch": 23.37, |
|
"learning_rate": 1.855263157894737e-05, |
|
"loss": 1.4485, |
|
"step": 8250 |
|
}, |
|
{ |
|
"epoch": 23.4, |
|
"learning_rate": 1.853684210526316e-05, |
|
"loss": 1.4692, |
|
"step": 8260 |
|
}, |
|
{ |
|
"epoch": 23.43, |
|
"learning_rate": 1.8521052631578946e-05, |
|
"loss": 1.4647, |
|
"step": 8270 |
|
}, |
|
{ |
|
"epoch": 23.45, |
|
"learning_rate": 1.8505263157894737e-05, |
|
"loss": 1.4576, |
|
"step": 8280 |
|
}, |
|
{ |
|
"epoch": 23.48, |
|
"learning_rate": 1.8489473684210527e-05, |
|
"loss": 1.4552, |
|
"step": 8290 |
|
}, |
|
{ |
|
"epoch": 23.51, |
|
"learning_rate": 1.8473684210526317e-05, |
|
"loss": 1.4576, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 23.54, |
|
"learning_rate": 1.8457894736842104e-05, |
|
"loss": 1.4523, |
|
"step": 8310 |
|
}, |
|
{ |
|
"epoch": 23.57, |
|
"learning_rate": 1.8442105263157895e-05, |
|
"loss": 1.4652, |
|
"step": 8320 |
|
}, |
|
{ |
|
"epoch": 23.6, |
|
"learning_rate": 1.842631578947368e-05, |
|
"loss": 1.4659, |
|
"step": 8330 |
|
}, |
|
{ |
|
"epoch": 23.62, |
|
"learning_rate": 1.8410526315789475e-05, |
|
"loss": 1.4585, |
|
"step": 8340 |
|
}, |
|
{ |
|
"epoch": 23.65, |
|
"learning_rate": 1.8394736842105266e-05, |
|
"loss": 1.4553, |
|
"step": 8350 |
|
}, |
|
{ |
|
"epoch": 23.68, |
|
"learning_rate": 1.8378947368421053e-05, |
|
"loss": 1.4472, |
|
"step": 8360 |
|
}, |
|
{ |
|
"epoch": 23.71, |
|
"learning_rate": 1.8363157894736843e-05, |
|
"loss": 1.4547, |
|
"step": 8370 |
|
}, |
|
{ |
|
"epoch": 23.74, |
|
"learning_rate": 1.834736842105263e-05, |
|
"loss": 1.4645, |
|
"step": 8380 |
|
}, |
|
{ |
|
"epoch": 23.77, |
|
"learning_rate": 1.8331578947368424e-05, |
|
"loss": 1.4682, |
|
"step": 8390 |
|
}, |
|
{ |
|
"epoch": 23.79, |
|
"learning_rate": 1.831578947368421e-05, |
|
"loss": 1.4623, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 23.82, |
|
"learning_rate": 1.83e-05, |
|
"loss": 1.4597, |
|
"step": 8410 |
|
}, |
|
{ |
|
"epoch": 23.85, |
|
"learning_rate": 1.8284210526315788e-05, |
|
"loss": 1.4626, |
|
"step": 8420 |
|
}, |
|
{ |
|
"epoch": 23.88, |
|
"learning_rate": 1.826842105263158e-05, |
|
"loss": 1.4676, |
|
"step": 8430 |
|
}, |
|
{ |
|
"epoch": 23.91, |
|
"learning_rate": 1.825263157894737e-05, |
|
"loss": 1.464, |
|
"step": 8440 |
|
}, |
|
{ |
|
"epoch": 23.94, |
|
"learning_rate": 1.823684210526316e-05, |
|
"loss": 1.4632, |
|
"step": 8450 |
|
}, |
|
{ |
|
"epoch": 23.96, |
|
"learning_rate": 1.8221052631578946e-05, |
|
"loss": 1.4781, |
|
"step": 8460 |
|
}, |
|
{ |
|
"epoch": 23.99, |
|
"learning_rate": 1.8205263157894736e-05, |
|
"loss": 1.4641, |
|
"step": 8470 |
|
}, |
|
{ |
|
"epoch": 24.02, |
|
"learning_rate": 1.8189473684210527e-05, |
|
"loss": 1.5895, |
|
"step": 8480 |
|
}, |
|
{ |
|
"epoch": 24.05, |
|
"learning_rate": 1.8173684210526317e-05, |
|
"loss": 1.4559, |
|
"step": 8490 |
|
}, |
|
{ |
|
"epoch": 24.08, |
|
"learning_rate": 1.8157894736842107e-05, |
|
"loss": 1.455, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 24.11, |
|
"learning_rate": 1.8142105263157894e-05, |
|
"loss": 1.4503, |
|
"step": 8510 |
|
}, |
|
{ |
|
"epoch": 24.14, |
|
"learning_rate": 1.8126315789473685e-05, |
|
"loss": 1.4536, |
|
"step": 8520 |
|
}, |
|
{ |
|
"epoch": 24.16, |
|
"learning_rate": 1.8110526315789475e-05, |
|
"loss": 1.4639, |
|
"step": 8530 |
|
}, |
|
{ |
|
"epoch": 24.19, |
|
"learning_rate": 1.8094736842105265e-05, |
|
"loss": 1.4554, |
|
"step": 8540 |
|
}, |
|
{ |
|
"epoch": 24.22, |
|
"learning_rate": 1.8078947368421052e-05, |
|
"loss": 1.4527, |
|
"step": 8550 |
|
}, |
|
{ |
|
"epoch": 24.25, |
|
"learning_rate": 1.8063157894736842e-05, |
|
"loss": 1.4564, |
|
"step": 8560 |
|
}, |
|
{ |
|
"epoch": 24.28, |
|
"learning_rate": 1.8047368421052633e-05, |
|
"loss": 1.4552, |
|
"step": 8570 |
|
}, |
|
{ |
|
"epoch": 24.31, |
|
"learning_rate": 1.8031578947368423e-05, |
|
"loss": 1.4553, |
|
"step": 8580 |
|
}, |
|
{ |
|
"epoch": 24.33, |
|
"learning_rate": 1.801578947368421e-05, |
|
"loss": 1.4624, |
|
"step": 8590 |
|
}, |
|
{ |
|
"epoch": 24.36, |
|
"learning_rate": 1.8e-05, |
|
"loss": 1.4666, |
|
"step": 8600 |
|
}, |
|
{ |
|
"epoch": 24.39, |
|
"learning_rate": 1.7984210526315787e-05, |
|
"loss": 1.483, |
|
"step": 8610 |
|
}, |
|
{ |
|
"epoch": 24.42, |
|
"learning_rate": 1.796842105263158e-05, |
|
"loss": 1.4497, |
|
"step": 8620 |
|
}, |
|
{ |
|
"epoch": 24.45, |
|
"learning_rate": 1.7952631578947368e-05, |
|
"loss": 1.4499, |
|
"step": 8630 |
|
}, |
|
{ |
|
"epoch": 24.47, |
|
"learning_rate": 1.793684210526316e-05, |
|
"loss": 1.4621, |
|
"step": 8640 |
|
}, |
|
{ |
|
"epoch": 24.5, |
|
"learning_rate": 1.7921052631578945e-05, |
|
"loss": 1.4563, |
|
"step": 8650 |
|
}, |
|
{ |
|
"epoch": 24.53, |
|
"learning_rate": 1.7905263157894736e-05, |
|
"loss": 1.4604, |
|
"step": 8660 |
|
}, |
|
{ |
|
"epoch": 24.56, |
|
"learning_rate": 1.788947368421053e-05, |
|
"loss": 1.459, |
|
"step": 8670 |
|
}, |
|
{ |
|
"epoch": 24.59, |
|
"learning_rate": 1.7873684210526316e-05, |
|
"loss": 1.4535, |
|
"step": 8680 |
|
}, |
|
{ |
|
"epoch": 24.62, |
|
"learning_rate": 1.7857894736842107e-05, |
|
"loss": 1.4607, |
|
"step": 8690 |
|
}, |
|
{ |
|
"epoch": 24.64, |
|
"learning_rate": 1.7842105263157894e-05, |
|
"loss": 1.4631, |
|
"step": 8700 |
|
}, |
|
{ |
|
"epoch": 24.67, |
|
"learning_rate": 1.7826315789473687e-05, |
|
"loss": 1.4546, |
|
"step": 8710 |
|
}, |
|
{ |
|
"epoch": 24.7, |
|
"learning_rate": 1.7810526315789474e-05, |
|
"loss": 1.4585, |
|
"step": 8720 |
|
}, |
|
{ |
|
"epoch": 24.73, |
|
"learning_rate": 1.7794736842105265e-05, |
|
"loss": 1.4513, |
|
"step": 8730 |
|
}, |
|
{ |
|
"epoch": 24.76, |
|
"learning_rate": 1.777894736842105e-05, |
|
"loss": 1.4654, |
|
"step": 8740 |
|
}, |
|
{ |
|
"epoch": 24.79, |
|
"learning_rate": 1.7763157894736842e-05, |
|
"loss": 1.469, |
|
"step": 8750 |
|
}, |
|
{ |
|
"epoch": 24.81, |
|
"learning_rate": 1.7747368421052632e-05, |
|
"loss": 1.4568, |
|
"step": 8760 |
|
}, |
|
{ |
|
"epoch": 24.84, |
|
"learning_rate": 1.7731578947368423e-05, |
|
"loss": 1.4563, |
|
"step": 8770 |
|
}, |
|
{ |
|
"epoch": 24.87, |
|
"learning_rate": 1.771578947368421e-05, |
|
"loss": 1.4524, |
|
"step": 8780 |
|
}, |
|
{ |
|
"epoch": 24.9, |
|
"learning_rate": 1.77e-05, |
|
"loss": 1.4634, |
|
"step": 8790 |
|
}, |
|
{ |
|
"epoch": 24.93, |
|
"learning_rate": 1.7684210526315787e-05, |
|
"loss": 1.4664, |
|
"step": 8800 |
|
}, |
|
{ |
|
"epoch": 24.96, |
|
"learning_rate": 1.766842105263158e-05, |
|
"loss": 1.4586, |
|
"step": 8810 |
|
}, |
|
{ |
|
"epoch": 24.98, |
|
"learning_rate": 1.7652631578947368e-05, |
|
"loss": 1.4597, |
|
"step": 8820 |
|
}, |
|
{ |
|
"epoch": 25.01, |
|
"learning_rate": 1.7636842105263158e-05, |
|
"loss": 1.5807, |
|
"step": 8830 |
|
}, |
|
{ |
|
"epoch": 25.04, |
|
"learning_rate": 1.7621052631578948e-05, |
|
"loss": 1.4547, |
|
"step": 8840 |
|
}, |
|
{ |
|
"epoch": 25.07, |
|
"learning_rate": 1.760526315789474e-05, |
|
"loss": 1.457, |
|
"step": 8850 |
|
}, |
|
{ |
|
"epoch": 25.1, |
|
"learning_rate": 1.758947368421053e-05, |
|
"loss": 1.4622, |
|
"step": 8860 |
|
}, |
|
{ |
|
"epoch": 25.13, |
|
"learning_rate": 1.7573684210526316e-05, |
|
"loss": 1.4571, |
|
"step": 8870 |
|
}, |
|
{ |
|
"epoch": 25.16, |
|
"learning_rate": 1.7557894736842106e-05, |
|
"loss": 1.4649, |
|
"step": 8880 |
|
}, |
|
{ |
|
"epoch": 25.18, |
|
"learning_rate": 1.7542105263157893e-05, |
|
"loss": 1.4504, |
|
"step": 8890 |
|
}, |
|
{ |
|
"epoch": 25.21, |
|
"learning_rate": 1.7526315789473687e-05, |
|
"loss": 1.4478, |
|
"step": 8900 |
|
}, |
|
{ |
|
"epoch": 25.24, |
|
"learning_rate": 1.7510526315789474e-05, |
|
"loss": 1.4472, |
|
"step": 8910 |
|
}, |
|
{ |
|
"epoch": 25.27, |
|
"learning_rate": 1.7494736842105264e-05, |
|
"loss": 1.4518, |
|
"step": 8920 |
|
}, |
|
{ |
|
"epoch": 25.3, |
|
"learning_rate": 1.747894736842105e-05, |
|
"loss": 1.4583, |
|
"step": 8930 |
|
}, |
|
{ |
|
"epoch": 25.32, |
|
"learning_rate": 1.7463157894736845e-05, |
|
"loss": 1.4494, |
|
"step": 8940 |
|
}, |
|
{ |
|
"epoch": 25.35, |
|
"learning_rate": 1.7447368421052632e-05, |
|
"loss": 1.4479, |
|
"step": 8950 |
|
}, |
|
{ |
|
"epoch": 25.38, |
|
"learning_rate": 1.7431578947368422e-05, |
|
"loss": 1.4529, |
|
"step": 8960 |
|
}, |
|
{ |
|
"epoch": 25.41, |
|
"learning_rate": 1.741578947368421e-05, |
|
"loss": 1.4584, |
|
"step": 8970 |
|
}, |
|
{ |
|
"epoch": 25.44, |
|
"learning_rate": 1.74e-05, |
|
"loss": 1.4568, |
|
"step": 8980 |
|
}, |
|
{ |
|
"epoch": 25.47, |
|
"learning_rate": 1.738421052631579e-05, |
|
"loss": 1.4552, |
|
"step": 8990 |
|
}, |
|
{ |
|
"epoch": 25.49, |
|
"learning_rate": 1.736842105263158e-05, |
|
"loss": 1.4578, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 25.49, |
|
"eval_denotation_accuracy": 0.5655245496291064, |
|
"eval_loss": 2.1795523166656494, |
|
"eval_runtime": 384.2494, |
|
"eval_samples_per_second": 7.368, |
|
"eval_steps_per_second": 1.843, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 25.52, |
|
"learning_rate": 1.735263157894737e-05, |
|
"loss": 1.4569, |
|
"step": 9010 |
|
}, |
|
{ |
|
"epoch": 25.55, |
|
"learning_rate": 1.7336842105263157e-05, |
|
"loss": 1.4494, |
|
"step": 9020 |
|
}, |
|
{ |
|
"epoch": 25.58, |
|
"learning_rate": 1.7321052631578948e-05, |
|
"loss": 1.4565, |
|
"step": 9030 |
|
}, |
|
{ |
|
"epoch": 25.61, |
|
"learning_rate": 1.7305263157894738e-05, |
|
"loss": 1.4538, |
|
"step": 9040 |
|
}, |
|
{ |
|
"epoch": 25.64, |
|
"learning_rate": 1.728947368421053e-05, |
|
"loss": 1.4593, |
|
"step": 9050 |
|
}, |
|
{ |
|
"epoch": 25.66, |
|
"learning_rate": 1.7273684210526315e-05, |
|
"loss": 1.4494, |
|
"step": 9060 |
|
}, |
|
{ |
|
"epoch": 25.69, |
|
"learning_rate": 1.7257894736842106e-05, |
|
"loss": 1.4544, |
|
"step": 9070 |
|
}, |
|
{ |
|
"epoch": 25.72, |
|
"learning_rate": 1.7242105263157896e-05, |
|
"loss": 1.4545, |
|
"step": 9080 |
|
}, |
|
{ |
|
"epoch": 25.75, |
|
"learning_rate": 1.7226315789473686e-05, |
|
"loss": 1.4551, |
|
"step": 9090 |
|
}, |
|
{ |
|
"epoch": 25.78, |
|
"learning_rate": 1.7210526315789473e-05, |
|
"loss": 1.4566, |
|
"step": 9100 |
|
}, |
|
{ |
|
"epoch": 25.81, |
|
"learning_rate": 1.7194736842105264e-05, |
|
"loss": 1.4476, |
|
"step": 9110 |
|
}, |
|
{ |
|
"epoch": 25.83, |
|
"learning_rate": 1.717894736842105e-05, |
|
"loss": 1.4564, |
|
"step": 9120 |
|
}, |
|
{ |
|
"epoch": 25.86, |
|
"learning_rate": 1.7163157894736844e-05, |
|
"loss": 1.4504, |
|
"step": 9130 |
|
}, |
|
{ |
|
"epoch": 25.89, |
|
"learning_rate": 1.714736842105263e-05, |
|
"loss": 1.4464, |
|
"step": 9140 |
|
}, |
|
{ |
|
"epoch": 25.92, |
|
"learning_rate": 1.713157894736842e-05, |
|
"loss": 1.4613, |
|
"step": 9150 |
|
}, |
|
{ |
|
"epoch": 25.95, |
|
"learning_rate": 1.711578947368421e-05, |
|
"loss": 1.452, |
|
"step": 9160 |
|
}, |
|
{ |
|
"epoch": 25.97, |
|
"learning_rate": 1.71e-05, |
|
"loss": 1.4672, |
|
"step": 9170 |
|
}, |
|
{ |
|
"epoch": 26.01, |
|
"learning_rate": 1.7084210526315793e-05, |
|
"loss": 1.5794, |
|
"step": 9180 |
|
}, |
|
{ |
|
"epoch": 26.03, |
|
"learning_rate": 1.706842105263158e-05, |
|
"loss": 1.4436, |
|
"step": 9190 |
|
}, |
|
{ |
|
"epoch": 26.06, |
|
"learning_rate": 1.705263157894737e-05, |
|
"loss": 1.4512, |
|
"step": 9200 |
|
}, |
|
{ |
|
"epoch": 26.09, |
|
"learning_rate": 1.7036842105263157e-05, |
|
"loss": 1.4582, |
|
"step": 9210 |
|
}, |
|
{ |
|
"epoch": 26.12, |
|
"learning_rate": 1.702105263157895e-05, |
|
"loss": 1.4483, |
|
"step": 9220 |
|
}, |
|
{ |
|
"epoch": 26.15, |
|
"learning_rate": 1.7005263157894738e-05, |
|
"loss": 1.4564, |
|
"step": 9230 |
|
}, |
|
{ |
|
"epoch": 26.18, |
|
"learning_rate": 1.6989473684210528e-05, |
|
"loss": 1.4485, |
|
"step": 9240 |
|
}, |
|
{ |
|
"epoch": 26.2, |
|
"learning_rate": 1.6973684210526315e-05, |
|
"loss": 1.4552, |
|
"step": 9250 |
|
}, |
|
{ |
|
"epoch": 26.23, |
|
"learning_rate": 1.6957894736842105e-05, |
|
"loss": 1.4425, |
|
"step": 9260 |
|
}, |
|
{ |
|
"epoch": 26.26, |
|
"learning_rate": 1.6942105263157896e-05, |
|
"loss": 1.4505, |
|
"step": 9270 |
|
}, |
|
{ |
|
"epoch": 26.29, |
|
"learning_rate": 1.6926315789473686e-05, |
|
"loss": 1.4553, |
|
"step": 9280 |
|
}, |
|
{ |
|
"epoch": 26.32, |
|
"learning_rate": 1.6910526315789473e-05, |
|
"loss": 1.4532, |
|
"step": 9290 |
|
}, |
|
{ |
|
"epoch": 26.34, |
|
"learning_rate": 1.6894736842105263e-05, |
|
"loss": 1.4607, |
|
"step": 9300 |
|
}, |
|
{ |
|
"epoch": 26.37, |
|
"learning_rate": 1.687894736842105e-05, |
|
"loss": 1.4574, |
|
"step": 9310 |
|
}, |
|
{ |
|
"epoch": 26.4, |
|
"learning_rate": 1.6863157894736844e-05, |
|
"loss": 1.4539, |
|
"step": 9320 |
|
}, |
|
{ |
|
"epoch": 26.43, |
|
"learning_rate": 1.684736842105263e-05, |
|
"loss": 1.4613, |
|
"step": 9330 |
|
}, |
|
{ |
|
"epoch": 26.46, |
|
"learning_rate": 1.683157894736842e-05, |
|
"loss": 1.4561, |
|
"step": 9340 |
|
}, |
|
{ |
|
"epoch": 26.49, |
|
"learning_rate": 1.681578947368421e-05, |
|
"loss": 1.4589, |
|
"step": 9350 |
|
}, |
|
{ |
|
"epoch": 26.51, |
|
"learning_rate": 1.6800000000000002e-05, |
|
"loss": 1.4499, |
|
"step": 9360 |
|
}, |
|
{ |
|
"epoch": 26.54, |
|
"learning_rate": 1.6784210526315792e-05, |
|
"loss": 1.4518, |
|
"step": 9370 |
|
}, |
|
{ |
|
"epoch": 26.57, |
|
"learning_rate": 1.676842105263158e-05, |
|
"loss": 1.4559, |
|
"step": 9380 |
|
}, |
|
{ |
|
"epoch": 26.6, |
|
"learning_rate": 1.675263157894737e-05, |
|
"loss": 1.4573, |
|
"step": 9390 |
|
}, |
|
{ |
|
"epoch": 26.63, |
|
"learning_rate": 1.6736842105263156e-05, |
|
"loss": 1.4822, |
|
"step": 9400 |
|
}, |
|
{ |
|
"epoch": 26.66, |
|
"learning_rate": 1.672105263157895e-05, |
|
"loss": 1.4559, |
|
"step": 9410 |
|
}, |
|
{ |
|
"epoch": 26.68, |
|
"learning_rate": 1.6705263157894737e-05, |
|
"loss": 1.4518, |
|
"step": 9420 |
|
}, |
|
{ |
|
"epoch": 26.71, |
|
"learning_rate": 1.6689473684210527e-05, |
|
"loss": 1.4506, |
|
"step": 9430 |
|
}, |
|
{ |
|
"epoch": 26.74, |
|
"learning_rate": 1.6673684210526314e-05, |
|
"loss": 1.4456, |
|
"step": 9440 |
|
}, |
|
{ |
|
"epoch": 26.77, |
|
"learning_rate": 1.6657894736842105e-05, |
|
"loss": 1.4601, |
|
"step": 9450 |
|
}, |
|
{ |
|
"epoch": 26.8, |
|
"learning_rate": 1.6642105263157895e-05, |
|
"loss": 1.453, |
|
"step": 9460 |
|
}, |
|
{ |
|
"epoch": 26.83, |
|
"learning_rate": 1.6626315789473685e-05, |
|
"loss": 1.4559, |
|
"step": 9470 |
|
}, |
|
{ |
|
"epoch": 26.85, |
|
"learning_rate": 1.6610526315789472e-05, |
|
"loss": 1.4578, |
|
"step": 9480 |
|
}, |
|
{ |
|
"epoch": 26.88, |
|
"learning_rate": 1.6594736842105263e-05, |
|
"loss": 1.4607, |
|
"step": 9490 |
|
}, |
|
{ |
|
"epoch": 26.91, |
|
"learning_rate": 1.6578947368421053e-05, |
|
"loss": 1.4576, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 26.94, |
|
"learning_rate": 1.6563157894736843e-05, |
|
"loss": 1.4611, |
|
"step": 9510 |
|
}, |
|
{ |
|
"epoch": 26.97, |
|
"learning_rate": 1.6547368421052634e-05, |
|
"loss": 1.4605, |
|
"step": 9520 |
|
}, |
|
{ |
|
"epoch": 26.99, |
|
"learning_rate": 1.653157894736842e-05, |
|
"loss": 1.4587, |
|
"step": 9530 |
|
}, |
|
{ |
|
"epoch": 27.03, |
|
"learning_rate": 1.651578947368421e-05, |
|
"loss": 1.587, |
|
"step": 9540 |
|
}, |
|
{ |
|
"epoch": 27.05, |
|
"learning_rate": 1.65e-05, |
|
"loss": 1.4476, |
|
"step": 9550 |
|
}, |
|
{ |
|
"epoch": 27.08, |
|
"learning_rate": 1.648421052631579e-05, |
|
"loss": 1.4579, |
|
"step": 9560 |
|
}, |
|
{ |
|
"epoch": 27.11, |
|
"learning_rate": 1.646842105263158e-05, |
|
"loss": 1.4519, |
|
"step": 9570 |
|
}, |
|
{ |
|
"epoch": 27.14, |
|
"learning_rate": 1.645263157894737e-05, |
|
"loss": 1.4559, |
|
"step": 9580 |
|
}, |
|
{ |
|
"epoch": 27.17, |
|
"learning_rate": 1.6436842105263156e-05, |
|
"loss": 1.4634, |
|
"step": 9590 |
|
}, |
|
{ |
|
"epoch": 27.19, |
|
"learning_rate": 1.642105263157895e-05, |
|
"loss": 1.4613, |
|
"step": 9600 |
|
}, |
|
{ |
|
"epoch": 27.22, |
|
"learning_rate": 1.6405263157894737e-05, |
|
"loss": 1.4512, |
|
"step": 9610 |
|
}, |
|
{ |
|
"epoch": 27.25, |
|
"learning_rate": 1.6389473684210527e-05, |
|
"loss": 1.4537, |
|
"step": 9620 |
|
}, |
|
{ |
|
"epoch": 27.28, |
|
"learning_rate": 1.6373684210526314e-05, |
|
"loss": 1.4579, |
|
"step": 9630 |
|
}, |
|
{ |
|
"epoch": 27.31, |
|
"learning_rate": 1.6357894736842108e-05, |
|
"loss": 1.4478, |
|
"step": 9640 |
|
}, |
|
{ |
|
"epoch": 27.34, |
|
"learning_rate": 1.6342105263157894e-05, |
|
"loss": 1.468, |
|
"step": 9650 |
|
}, |
|
{ |
|
"epoch": 27.36, |
|
"learning_rate": 1.6326315789473685e-05, |
|
"loss": 1.4684, |
|
"step": 9660 |
|
}, |
|
{ |
|
"epoch": 27.39, |
|
"learning_rate": 1.6310526315789475e-05, |
|
"loss": 1.4503, |
|
"step": 9670 |
|
}, |
|
{ |
|
"epoch": 27.42, |
|
"learning_rate": 1.6294736842105262e-05, |
|
"loss": 1.4523, |
|
"step": 9680 |
|
}, |
|
{ |
|
"epoch": 27.45, |
|
"learning_rate": 1.6278947368421056e-05, |
|
"loss": 1.4587, |
|
"step": 9690 |
|
}, |
|
{ |
|
"epoch": 27.48, |
|
"learning_rate": 1.6263157894736843e-05, |
|
"loss": 1.4599, |
|
"step": 9700 |
|
}, |
|
{ |
|
"epoch": 27.51, |
|
"learning_rate": 1.6247368421052633e-05, |
|
"loss": 1.4459, |
|
"step": 9710 |
|
}, |
|
{ |
|
"epoch": 27.53, |
|
"learning_rate": 1.623157894736842e-05, |
|
"loss": 1.4492, |
|
"step": 9720 |
|
}, |
|
{ |
|
"epoch": 27.56, |
|
"learning_rate": 1.621578947368421e-05, |
|
"loss": 1.4537, |
|
"step": 9730 |
|
}, |
|
{ |
|
"epoch": 27.59, |
|
"learning_rate": 1.62e-05, |
|
"loss": 1.4507, |
|
"step": 9740 |
|
}, |
|
{ |
|
"epoch": 27.62, |
|
"learning_rate": 1.618421052631579e-05, |
|
"loss": 1.4546, |
|
"step": 9750 |
|
}, |
|
{ |
|
"epoch": 27.65, |
|
"learning_rate": 1.6168421052631578e-05, |
|
"loss": 1.4561, |
|
"step": 9760 |
|
}, |
|
{ |
|
"epoch": 27.68, |
|
"learning_rate": 1.615263157894737e-05, |
|
"loss": 1.4472, |
|
"step": 9770 |
|
}, |
|
{ |
|
"epoch": 27.7, |
|
"learning_rate": 1.613684210526316e-05, |
|
"loss": 1.4475, |
|
"step": 9780 |
|
}, |
|
{ |
|
"epoch": 27.73, |
|
"learning_rate": 1.612105263157895e-05, |
|
"loss": 1.4499, |
|
"step": 9790 |
|
}, |
|
{ |
|
"epoch": 27.76, |
|
"learning_rate": 1.6105263157894736e-05, |
|
"loss": 1.4549, |
|
"step": 9800 |
|
}, |
|
{ |
|
"epoch": 27.79, |
|
"learning_rate": 1.6089473684210526e-05, |
|
"loss": 1.4532, |
|
"step": 9810 |
|
}, |
|
{ |
|
"epoch": 27.82, |
|
"learning_rate": 1.6073684210526313e-05, |
|
"loss": 1.4622, |
|
"step": 9820 |
|
}, |
|
{ |
|
"epoch": 27.84, |
|
"learning_rate": 1.6057894736842107e-05, |
|
"loss": 1.4459, |
|
"step": 9830 |
|
}, |
|
{ |
|
"epoch": 27.87, |
|
"learning_rate": 1.6042105263157897e-05, |
|
"loss": 1.4534, |
|
"step": 9840 |
|
}, |
|
{ |
|
"epoch": 27.9, |
|
"learning_rate": 1.6026315789473684e-05, |
|
"loss": 1.4538, |
|
"step": 9850 |
|
}, |
|
{ |
|
"epoch": 27.93, |
|
"learning_rate": 1.6010526315789475e-05, |
|
"loss": 1.4614, |
|
"step": 9860 |
|
}, |
|
{ |
|
"epoch": 27.96, |
|
"learning_rate": 1.599473684210526e-05, |
|
"loss": 1.4608, |
|
"step": 9870 |
|
}, |
|
{ |
|
"epoch": 27.99, |
|
"learning_rate": 1.5978947368421055e-05, |
|
"loss": 1.4524, |
|
"step": 9880 |
|
}, |
|
{ |
|
"epoch": 28.02, |
|
"learning_rate": 1.5963157894736842e-05, |
|
"loss": 1.5717, |
|
"step": 9890 |
|
}, |
|
{ |
|
"epoch": 28.05, |
|
"learning_rate": 1.5947368421052633e-05, |
|
"loss": 1.4465, |
|
"step": 9900 |
|
}, |
|
{ |
|
"epoch": 28.07, |
|
"learning_rate": 1.593157894736842e-05, |
|
"loss": 1.4488, |
|
"step": 9910 |
|
}, |
|
{ |
|
"epoch": 28.1, |
|
"learning_rate": 1.5915789473684213e-05, |
|
"loss": 1.4412, |
|
"step": 9920 |
|
}, |
|
{ |
|
"epoch": 28.13, |
|
"learning_rate": 1.59e-05, |
|
"loss": 1.4391, |
|
"step": 9930 |
|
}, |
|
{ |
|
"epoch": 28.16, |
|
"learning_rate": 1.588421052631579e-05, |
|
"loss": 1.4502, |
|
"step": 9940 |
|
}, |
|
{ |
|
"epoch": 28.19, |
|
"learning_rate": 1.5868421052631578e-05, |
|
"loss": 1.4522, |
|
"step": 9950 |
|
}, |
|
{ |
|
"epoch": 28.21, |
|
"learning_rate": 1.5852631578947368e-05, |
|
"loss": 1.446, |
|
"step": 9960 |
|
}, |
|
{ |
|
"epoch": 28.24, |
|
"learning_rate": 1.5836842105263158e-05, |
|
"loss": 1.4494, |
|
"step": 9970 |
|
}, |
|
{ |
|
"epoch": 28.27, |
|
"learning_rate": 1.582105263157895e-05, |
|
"loss": 1.4492, |
|
"step": 9980 |
|
}, |
|
{ |
|
"epoch": 28.3, |
|
"learning_rate": 1.5805263157894735e-05, |
|
"loss": 1.4441, |
|
"step": 9990 |
|
}, |
|
{ |
|
"epoch": 28.33, |
|
"learning_rate": 1.5789473684210526e-05, |
|
"loss": 1.45, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 28.33, |
|
"eval_denotation_accuracy": 0.5655245496291064, |
|
"eval_loss": 2.1741864681243896, |
|
"eval_runtime": 343.9405, |
|
"eval_samples_per_second": 8.231, |
|
"eval_steps_per_second": 2.058, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 28.36, |
|
"learning_rate": 1.5773684210526316e-05, |
|
"loss": 1.4492, |
|
"step": 10010 |
|
}, |
|
{ |
|
"epoch": 28.38, |
|
"learning_rate": 1.5757894736842107e-05, |
|
"loss": 1.4592, |
|
"step": 10020 |
|
}, |
|
{ |
|
"epoch": 28.41, |
|
"learning_rate": 1.5742105263157897e-05, |
|
"loss": 1.4474, |
|
"step": 10030 |
|
}, |
|
{ |
|
"epoch": 28.44, |
|
"learning_rate": 1.5726315789473684e-05, |
|
"loss": 1.448, |
|
"step": 10040 |
|
}, |
|
{ |
|
"epoch": 28.47, |
|
"learning_rate": 1.5710526315789474e-05, |
|
"loss": 1.4454, |
|
"step": 10050 |
|
}, |
|
{ |
|
"epoch": 28.5, |
|
"learning_rate": 1.5694736842105264e-05, |
|
"loss": 1.4551, |
|
"step": 10060 |
|
}, |
|
{ |
|
"epoch": 28.53, |
|
"learning_rate": 1.5678947368421055e-05, |
|
"loss": 1.444, |
|
"step": 10070 |
|
}, |
|
{ |
|
"epoch": 28.55, |
|
"learning_rate": 1.5663157894736842e-05, |
|
"loss": 1.444, |
|
"step": 10080 |
|
}, |
|
{ |
|
"epoch": 28.58, |
|
"learning_rate": 1.5647368421052632e-05, |
|
"loss": 1.4529, |
|
"step": 10090 |
|
}, |
|
{ |
|
"epoch": 28.61, |
|
"learning_rate": 1.563157894736842e-05, |
|
"loss": 1.4558, |
|
"step": 10100 |
|
}, |
|
{ |
|
"epoch": 28.64, |
|
"learning_rate": 1.5615789473684213e-05, |
|
"loss": 1.4422, |
|
"step": 10110 |
|
}, |
|
{ |
|
"epoch": 28.67, |
|
"learning_rate": 1.56e-05, |
|
"loss": 1.4477, |
|
"step": 10120 |
|
}, |
|
{ |
|
"epoch": 28.7, |
|
"learning_rate": 1.558421052631579e-05, |
|
"loss": 1.446, |
|
"step": 10130 |
|
}, |
|
{ |
|
"epoch": 28.72, |
|
"learning_rate": 1.5568421052631577e-05, |
|
"loss": 1.4423, |
|
"step": 10140 |
|
}, |
|
{ |
|
"epoch": 28.75, |
|
"learning_rate": 1.5552631578947367e-05, |
|
"loss": 1.4442, |
|
"step": 10150 |
|
}, |
|
{ |
|
"epoch": 28.78, |
|
"learning_rate": 1.5536842105263158e-05, |
|
"loss": 1.4479, |
|
"step": 10160 |
|
}, |
|
{ |
|
"epoch": 28.81, |
|
"learning_rate": 1.5521052631578948e-05, |
|
"loss": 1.4556, |
|
"step": 10170 |
|
}, |
|
{ |
|
"epoch": 28.84, |
|
"learning_rate": 1.550526315789474e-05, |
|
"loss": 1.4418, |
|
"step": 10180 |
|
}, |
|
{ |
|
"epoch": 28.86, |
|
"learning_rate": 1.5489473684210525e-05, |
|
"loss": 1.451, |
|
"step": 10190 |
|
}, |
|
{ |
|
"epoch": 28.89, |
|
"learning_rate": 1.547368421052632e-05, |
|
"loss": 1.4399, |
|
"step": 10200 |
|
}, |
|
{ |
|
"epoch": 28.92, |
|
"learning_rate": 1.5457894736842106e-05, |
|
"loss": 1.4433, |
|
"step": 10210 |
|
}, |
|
{ |
|
"epoch": 28.95, |
|
"learning_rate": 1.5442105263157896e-05, |
|
"loss": 1.4507, |
|
"step": 10220 |
|
}, |
|
{ |
|
"epoch": 28.98, |
|
"learning_rate": 1.5426315789473683e-05, |
|
"loss": 1.4389, |
|
"step": 10230 |
|
}, |
|
{ |
|
"epoch": 29.01, |
|
"learning_rate": 1.5410526315789474e-05, |
|
"loss": 1.568, |
|
"step": 10240 |
|
}, |
|
{ |
|
"epoch": 29.04, |
|
"learning_rate": 1.5394736842105264e-05, |
|
"loss": 1.4508, |
|
"step": 10250 |
|
}, |
|
{ |
|
"epoch": 29.06, |
|
"learning_rate": 1.5378947368421054e-05, |
|
"loss": 1.4691, |
|
"step": 10260 |
|
}, |
|
{ |
|
"epoch": 29.09, |
|
"learning_rate": 1.536315789473684e-05, |
|
"loss": 1.4477, |
|
"step": 10270 |
|
}, |
|
{ |
|
"epoch": 29.12, |
|
"learning_rate": 1.534736842105263e-05, |
|
"loss": 1.4421, |
|
"step": 10280 |
|
}, |
|
{ |
|
"epoch": 29.15, |
|
"learning_rate": 1.533157894736842e-05, |
|
"loss": 1.4447, |
|
"step": 10290 |
|
}, |
|
{ |
|
"epoch": 29.18, |
|
"learning_rate": 1.5315789473684212e-05, |
|
"loss": 1.4369, |
|
"step": 10300 |
|
}, |
|
{ |
|
"epoch": 29.21, |
|
"learning_rate": 1.53e-05, |
|
"loss": 1.447, |
|
"step": 10310 |
|
}, |
|
{ |
|
"epoch": 29.23, |
|
"learning_rate": 1.528421052631579e-05, |
|
"loss": 1.4429, |
|
"step": 10320 |
|
}, |
|
{ |
|
"epoch": 29.26, |
|
"learning_rate": 1.5268421052631576e-05, |
|
"loss": 1.445, |
|
"step": 10330 |
|
}, |
|
{ |
|
"epoch": 29.29, |
|
"learning_rate": 1.525263157894737e-05, |
|
"loss": 1.4455, |
|
"step": 10340 |
|
}, |
|
{ |
|
"epoch": 29.32, |
|
"learning_rate": 1.5236842105263159e-05, |
|
"loss": 1.4524, |
|
"step": 10350 |
|
}, |
|
{ |
|
"epoch": 29.35, |
|
"learning_rate": 1.5221052631578948e-05, |
|
"loss": 1.4439, |
|
"step": 10360 |
|
}, |
|
{ |
|
"epoch": 29.38, |
|
"learning_rate": 1.5205263157894736e-05, |
|
"loss": 1.4389, |
|
"step": 10370 |
|
}, |
|
{ |
|
"epoch": 29.4, |
|
"learning_rate": 1.5189473684210525e-05, |
|
"loss": 1.4435, |
|
"step": 10380 |
|
}, |
|
{ |
|
"epoch": 29.43, |
|
"learning_rate": 1.5173684210526317e-05, |
|
"loss": 1.4401, |
|
"step": 10390 |
|
}, |
|
{ |
|
"epoch": 29.46, |
|
"learning_rate": 1.5157894736842105e-05, |
|
"loss": 1.4424, |
|
"step": 10400 |
|
}, |
|
{ |
|
"epoch": 29.49, |
|
"learning_rate": 1.5142105263157894e-05, |
|
"loss": 1.4546, |
|
"step": 10410 |
|
}, |
|
{ |
|
"epoch": 29.52, |
|
"learning_rate": 1.5126315789473684e-05, |
|
"loss": 1.4429, |
|
"step": 10420 |
|
}, |
|
{ |
|
"epoch": 29.55, |
|
"learning_rate": 1.5110526315789473e-05, |
|
"loss": 1.4474, |
|
"step": 10430 |
|
}, |
|
{ |
|
"epoch": 29.57, |
|
"learning_rate": 1.5094736842105265e-05, |
|
"loss": 1.4457, |
|
"step": 10440 |
|
}, |
|
{ |
|
"epoch": 29.6, |
|
"learning_rate": 1.5078947368421054e-05, |
|
"loss": 1.4507, |
|
"step": 10450 |
|
}, |
|
{ |
|
"epoch": 29.63, |
|
"learning_rate": 1.5063157894736842e-05, |
|
"loss": 1.4445, |
|
"step": 10460 |
|
}, |
|
{ |
|
"epoch": 29.66, |
|
"learning_rate": 1.5047368421052631e-05, |
|
"loss": 1.4514, |
|
"step": 10470 |
|
}, |
|
{ |
|
"epoch": 29.69, |
|
"learning_rate": 1.5031578947368423e-05, |
|
"loss": 1.443, |
|
"step": 10480 |
|
}, |
|
{ |
|
"epoch": 29.71, |
|
"learning_rate": 1.5015789473684212e-05, |
|
"loss": 1.4455, |
|
"step": 10490 |
|
}, |
|
{ |
|
"epoch": 29.74, |
|
"learning_rate": 1.5e-05, |
|
"loss": 1.4482, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 29.77, |
|
"learning_rate": 1.4984210526315789e-05, |
|
"loss": 1.4545, |
|
"step": 10510 |
|
}, |
|
{ |
|
"epoch": 29.8, |
|
"learning_rate": 1.496842105263158e-05, |
|
"loss": 1.4469, |
|
"step": 10520 |
|
}, |
|
{ |
|
"epoch": 29.83, |
|
"learning_rate": 1.4952631578947368e-05, |
|
"loss": 1.4595, |
|
"step": 10530 |
|
}, |
|
{ |
|
"epoch": 29.86, |
|
"learning_rate": 1.4936842105263158e-05, |
|
"loss": 1.4502, |
|
"step": 10540 |
|
}, |
|
{ |
|
"epoch": 29.88, |
|
"learning_rate": 1.4921052631578947e-05, |
|
"loss": 1.4431, |
|
"step": 10550 |
|
}, |
|
{ |
|
"epoch": 29.91, |
|
"learning_rate": 1.4905263157894737e-05, |
|
"loss": 1.4417, |
|
"step": 10560 |
|
}, |
|
{ |
|
"epoch": 29.94, |
|
"learning_rate": 1.4889473684210526e-05, |
|
"loss": 1.4488, |
|
"step": 10570 |
|
}, |
|
{ |
|
"epoch": 29.97, |
|
"learning_rate": 1.4873684210526315e-05, |
|
"loss": 1.438, |
|
"step": 10580 |
|
}, |
|
{ |
|
"epoch": 30.0, |
|
"learning_rate": 1.4857894736842107e-05, |
|
"loss": 1.4436, |
|
"step": 10590 |
|
}, |
|
{ |
|
"epoch": 30.03, |
|
"learning_rate": 1.4842105263157895e-05, |
|
"loss": 1.5746, |
|
"step": 10600 |
|
}, |
|
{ |
|
"epoch": 30.06, |
|
"learning_rate": 1.4826315789473686e-05, |
|
"loss": 1.4384, |
|
"step": 10610 |
|
}, |
|
{ |
|
"epoch": 30.08, |
|
"learning_rate": 1.4810526315789474e-05, |
|
"loss": 1.4467, |
|
"step": 10620 |
|
}, |
|
{ |
|
"epoch": 30.11, |
|
"learning_rate": 1.4794736842105265e-05, |
|
"loss": 1.4427, |
|
"step": 10630 |
|
}, |
|
{ |
|
"epoch": 30.14, |
|
"learning_rate": 1.4778947368421053e-05, |
|
"loss": 1.442, |
|
"step": 10640 |
|
}, |
|
{ |
|
"epoch": 30.17, |
|
"learning_rate": 1.4763157894736842e-05, |
|
"loss": 1.4459, |
|
"step": 10650 |
|
}, |
|
{ |
|
"epoch": 30.2, |
|
"learning_rate": 1.4747368421052632e-05, |
|
"loss": 1.4418, |
|
"step": 10660 |
|
}, |
|
{ |
|
"epoch": 30.23, |
|
"learning_rate": 1.4731578947368421e-05, |
|
"loss": 1.4442, |
|
"step": 10670 |
|
}, |
|
{ |
|
"epoch": 30.25, |
|
"learning_rate": 1.4715789473684211e-05, |
|
"loss": 1.441, |
|
"step": 10680 |
|
}, |
|
{ |
|
"epoch": 30.28, |
|
"learning_rate": 1.47e-05, |
|
"loss": 1.4451, |
|
"step": 10690 |
|
}, |
|
{ |
|
"epoch": 30.31, |
|
"learning_rate": 1.468421052631579e-05, |
|
"loss": 1.441, |
|
"step": 10700 |
|
}, |
|
{ |
|
"epoch": 30.34, |
|
"learning_rate": 1.4668421052631579e-05, |
|
"loss": 1.4487, |
|
"step": 10710 |
|
}, |
|
{ |
|
"epoch": 30.37, |
|
"learning_rate": 1.4652631578947367e-05, |
|
"loss": 1.4508, |
|
"step": 10720 |
|
}, |
|
{ |
|
"epoch": 30.4, |
|
"learning_rate": 1.4636842105263158e-05, |
|
"loss": 1.4441, |
|
"step": 10730 |
|
}, |
|
{ |
|
"epoch": 30.42, |
|
"learning_rate": 1.4621052631578946e-05, |
|
"loss": 1.4405, |
|
"step": 10740 |
|
}, |
|
{ |
|
"epoch": 30.45, |
|
"learning_rate": 1.4605263157894737e-05, |
|
"loss": 1.4432, |
|
"step": 10750 |
|
}, |
|
{ |
|
"epoch": 30.48, |
|
"learning_rate": 1.4589473684210527e-05, |
|
"loss": 1.4416, |
|
"step": 10760 |
|
}, |
|
{ |
|
"epoch": 30.51, |
|
"learning_rate": 1.4573684210526317e-05, |
|
"loss": 1.4374, |
|
"step": 10770 |
|
}, |
|
{ |
|
"epoch": 30.54, |
|
"learning_rate": 1.4557894736842106e-05, |
|
"loss": 1.4431, |
|
"step": 10780 |
|
}, |
|
{ |
|
"epoch": 30.57, |
|
"learning_rate": 1.4542105263157895e-05, |
|
"loss": 1.4433, |
|
"step": 10790 |
|
}, |
|
{ |
|
"epoch": 30.59, |
|
"learning_rate": 1.4526315789473685e-05, |
|
"loss": 1.4387, |
|
"step": 10800 |
|
}, |
|
{ |
|
"epoch": 30.62, |
|
"learning_rate": 1.4510526315789474e-05, |
|
"loss": 1.4442, |
|
"step": 10810 |
|
}, |
|
{ |
|
"epoch": 30.65, |
|
"learning_rate": 1.4494736842105264e-05, |
|
"loss": 1.4588, |
|
"step": 10820 |
|
}, |
|
{ |
|
"epoch": 30.68, |
|
"learning_rate": 1.4478947368421053e-05, |
|
"loss": 1.4447, |
|
"step": 10830 |
|
}, |
|
{ |
|
"epoch": 30.71, |
|
"learning_rate": 1.4463157894736843e-05, |
|
"loss": 1.4411, |
|
"step": 10840 |
|
}, |
|
{ |
|
"epoch": 30.73, |
|
"learning_rate": 1.4447368421052632e-05, |
|
"loss": 1.4487, |
|
"step": 10850 |
|
}, |
|
{ |
|
"epoch": 30.76, |
|
"learning_rate": 1.443157894736842e-05, |
|
"loss": 1.4416, |
|
"step": 10860 |
|
}, |
|
{ |
|
"epoch": 30.79, |
|
"learning_rate": 1.441578947368421e-05, |
|
"loss": 1.4445, |
|
"step": 10870 |
|
}, |
|
{ |
|
"epoch": 30.82, |
|
"learning_rate": 1.44e-05, |
|
"loss": 1.448, |
|
"step": 10880 |
|
}, |
|
{ |
|
"epoch": 30.85, |
|
"learning_rate": 1.438421052631579e-05, |
|
"loss": 1.4417, |
|
"step": 10890 |
|
}, |
|
{ |
|
"epoch": 30.88, |
|
"learning_rate": 1.4368421052631578e-05, |
|
"loss": 1.4531, |
|
"step": 10900 |
|
}, |
|
{ |
|
"epoch": 30.9, |
|
"learning_rate": 1.4352631578947369e-05, |
|
"loss": 1.442, |
|
"step": 10910 |
|
}, |
|
{ |
|
"epoch": 30.93, |
|
"learning_rate": 1.4336842105263159e-05, |
|
"loss": 1.4445, |
|
"step": 10920 |
|
}, |
|
{ |
|
"epoch": 30.96, |
|
"learning_rate": 1.4321052631578948e-05, |
|
"loss": 1.4446, |
|
"step": 10930 |
|
}, |
|
{ |
|
"epoch": 30.99, |
|
"learning_rate": 1.4305263157894738e-05, |
|
"loss": 1.4354, |
|
"step": 10940 |
|
}, |
|
{ |
|
"epoch": 31.02, |
|
"learning_rate": 1.4289473684210527e-05, |
|
"loss": 1.5688, |
|
"step": 10950 |
|
}, |
|
{ |
|
"epoch": 31.05, |
|
"learning_rate": 1.4273684210526317e-05, |
|
"loss": 1.4468, |
|
"step": 10960 |
|
}, |
|
{ |
|
"epoch": 31.08, |
|
"learning_rate": 1.4257894736842106e-05, |
|
"loss": 1.4484, |
|
"step": 10970 |
|
}, |
|
{ |
|
"epoch": 31.1, |
|
"learning_rate": 1.4242105263157896e-05, |
|
"loss": 1.4418, |
|
"step": 10980 |
|
}, |
|
{ |
|
"epoch": 31.13, |
|
"learning_rate": 1.4226315789473685e-05, |
|
"loss": 1.4436, |
|
"step": 10990 |
|
}, |
|
{ |
|
"epoch": 31.16, |
|
"learning_rate": 1.4210526315789473e-05, |
|
"loss": 1.4387, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 31.16, |
|
"eval_denotation_accuracy": 0.5754150476863299, |
|
"eval_loss": 2.172001600265503, |
|
"eval_runtime": 356.2446, |
|
"eval_samples_per_second": 7.947, |
|
"eval_steps_per_second": 1.987, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 31.19, |
|
"learning_rate": 1.4194736842105264e-05, |
|
"loss": 1.4366, |
|
"step": 11010 |
|
}, |
|
{ |
|
"epoch": 31.22, |
|
"learning_rate": 1.4178947368421052e-05, |
|
"loss": 1.4424, |
|
"step": 11020 |
|
}, |
|
{ |
|
"epoch": 31.25, |
|
"learning_rate": 1.4163157894736843e-05, |
|
"loss": 1.4401, |
|
"step": 11030 |
|
}, |
|
{ |
|
"epoch": 31.27, |
|
"learning_rate": 1.4147368421052631e-05, |
|
"loss": 1.4421, |
|
"step": 11040 |
|
}, |
|
{ |
|
"epoch": 31.3, |
|
"learning_rate": 1.4131578947368422e-05, |
|
"loss": 1.4503, |
|
"step": 11050 |
|
}, |
|
{ |
|
"epoch": 31.33, |
|
"learning_rate": 1.411578947368421e-05, |
|
"loss": 1.44, |
|
"step": 11060 |
|
}, |
|
{ |
|
"epoch": 31.36, |
|
"learning_rate": 1.4099999999999999e-05, |
|
"loss": 1.4547, |
|
"step": 11070 |
|
}, |
|
{ |
|
"epoch": 31.39, |
|
"learning_rate": 1.408421052631579e-05, |
|
"loss": 1.4453, |
|
"step": 11080 |
|
}, |
|
{ |
|
"epoch": 31.42, |
|
"learning_rate": 1.406842105263158e-05, |
|
"loss": 1.4466, |
|
"step": 11090 |
|
}, |
|
{ |
|
"epoch": 31.44, |
|
"learning_rate": 1.405263157894737e-05, |
|
"loss": 1.4418, |
|
"step": 11100 |
|
}, |
|
{ |
|
"epoch": 31.47, |
|
"learning_rate": 1.4036842105263158e-05, |
|
"loss": 1.4453, |
|
"step": 11110 |
|
}, |
|
{ |
|
"epoch": 31.5, |
|
"learning_rate": 1.4021052631578949e-05, |
|
"loss": 1.4463, |
|
"step": 11120 |
|
}, |
|
{ |
|
"epoch": 31.53, |
|
"learning_rate": 1.4005263157894737e-05, |
|
"loss": 1.4414, |
|
"step": 11130 |
|
}, |
|
{ |
|
"epoch": 31.56, |
|
"learning_rate": 1.3989473684210526e-05, |
|
"loss": 1.4417, |
|
"step": 11140 |
|
}, |
|
{ |
|
"epoch": 31.58, |
|
"learning_rate": 1.3973684210526316e-05, |
|
"loss": 1.4382, |
|
"step": 11150 |
|
}, |
|
{ |
|
"epoch": 31.61, |
|
"learning_rate": 1.3957894736842105e-05, |
|
"loss": 1.4421, |
|
"step": 11160 |
|
}, |
|
{ |
|
"epoch": 31.64, |
|
"learning_rate": 1.3942105263157895e-05, |
|
"loss": 1.44, |
|
"step": 11170 |
|
}, |
|
{ |
|
"epoch": 31.67, |
|
"learning_rate": 1.3926315789473684e-05, |
|
"loss": 1.4404, |
|
"step": 11180 |
|
}, |
|
{ |
|
"epoch": 31.7, |
|
"learning_rate": 1.3910526315789474e-05, |
|
"loss": 1.4501, |
|
"step": 11190 |
|
}, |
|
{ |
|
"epoch": 31.73, |
|
"learning_rate": 1.3894736842105263e-05, |
|
"loss": 1.4476, |
|
"step": 11200 |
|
}, |
|
{ |
|
"epoch": 31.75, |
|
"learning_rate": 1.3878947368421052e-05, |
|
"loss": 1.4348, |
|
"step": 11210 |
|
}, |
|
{ |
|
"epoch": 31.78, |
|
"learning_rate": 1.3863157894736842e-05, |
|
"loss": 1.4393, |
|
"step": 11220 |
|
}, |
|
{ |
|
"epoch": 31.81, |
|
"learning_rate": 1.384736842105263e-05, |
|
"loss": 1.441, |
|
"step": 11230 |
|
}, |
|
{ |
|
"epoch": 31.84, |
|
"learning_rate": 1.3831578947368421e-05, |
|
"loss": 1.4393, |
|
"step": 11240 |
|
}, |
|
{ |
|
"epoch": 31.87, |
|
"learning_rate": 1.3815789473684211e-05, |
|
"loss": 1.4382, |
|
"step": 11250 |
|
}, |
|
{ |
|
"epoch": 31.9, |
|
"learning_rate": 1.3800000000000002e-05, |
|
"loss": 1.4465, |
|
"step": 11260 |
|
}, |
|
{ |
|
"epoch": 31.92, |
|
"learning_rate": 1.378421052631579e-05, |
|
"loss": 1.4413, |
|
"step": 11270 |
|
}, |
|
{ |
|
"epoch": 31.95, |
|
"learning_rate": 1.3768421052631579e-05, |
|
"loss": 1.4435, |
|
"step": 11280 |
|
}, |
|
{ |
|
"epoch": 31.98, |
|
"learning_rate": 1.375263157894737e-05, |
|
"loss": 1.4449, |
|
"step": 11290 |
|
}, |
|
{ |
|
"epoch": 32.01, |
|
"learning_rate": 1.3736842105263158e-05, |
|
"loss": 1.5643, |
|
"step": 11300 |
|
}, |
|
{ |
|
"epoch": 32.04, |
|
"learning_rate": 1.3721052631578948e-05, |
|
"loss": 1.4375, |
|
"step": 11310 |
|
}, |
|
{ |
|
"epoch": 32.07, |
|
"learning_rate": 1.3705263157894737e-05, |
|
"loss": 1.4359, |
|
"step": 11320 |
|
}, |
|
{ |
|
"epoch": 32.1, |
|
"learning_rate": 1.3689473684210527e-05, |
|
"loss": 1.4309, |
|
"step": 11330 |
|
}, |
|
{ |
|
"epoch": 32.12, |
|
"learning_rate": 1.3673684210526316e-05, |
|
"loss": 1.4376, |
|
"step": 11340 |
|
}, |
|
{ |
|
"epoch": 32.15, |
|
"learning_rate": 1.3657894736842106e-05, |
|
"loss": 1.436, |
|
"step": 11350 |
|
}, |
|
{ |
|
"epoch": 32.18, |
|
"learning_rate": 1.3642105263157895e-05, |
|
"loss": 1.4455, |
|
"step": 11360 |
|
}, |
|
{ |
|
"epoch": 32.21, |
|
"learning_rate": 1.3626315789473684e-05, |
|
"loss": 1.4416, |
|
"step": 11370 |
|
}, |
|
{ |
|
"epoch": 32.24, |
|
"learning_rate": 1.3610526315789474e-05, |
|
"loss": 1.4381, |
|
"step": 11380 |
|
}, |
|
{ |
|
"epoch": 32.27, |
|
"learning_rate": 1.3594736842105263e-05, |
|
"loss": 1.438, |
|
"step": 11390 |
|
}, |
|
{ |
|
"epoch": 32.29, |
|
"learning_rate": 1.3578947368421053e-05, |
|
"loss": 1.4397, |
|
"step": 11400 |
|
}, |
|
{ |
|
"epoch": 32.32, |
|
"learning_rate": 1.3563157894736842e-05, |
|
"loss": 1.4342, |
|
"step": 11410 |
|
}, |
|
{ |
|
"epoch": 32.35, |
|
"learning_rate": 1.3547368421052634e-05, |
|
"loss": 1.4357, |
|
"step": 11420 |
|
}, |
|
{ |
|
"epoch": 32.38, |
|
"learning_rate": 1.3531578947368422e-05, |
|
"loss": 1.4391, |
|
"step": 11430 |
|
}, |
|
{ |
|
"epoch": 32.41, |
|
"learning_rate": 1.3515789473684211e-05, |
|
"loss": 1.4376, |
|
"step": 11440 |
|
}, |
|
{ |
|
"epoch": 32.44, |
|
"learning_rate": 1.3500000000000001e-05, |
|
"loss": 1.4378, |
|
"step": 11450 |
|
}, |
|
{ |
|
"epoch": 32.46, |
|
"learning_rate": 1.348421052631579e-05, |
|
"loss": 1.4369, |
|
"step": 11460 |
|
}, |
|
{ |
|
"epoch": 32.49, |
|
"learning_rate": 1.346842105263158e-05, |
|
"loss": 1.4441, |
|
"step": 11470 |
|
}, |
|
{ |
|
"epoch": 32.52, |
|
"learning_rate": 1.3452631578947369e-05, |
|
"loss": 1.448, |
|
"step": 11480 |
|
}, |
|
{ |
|
"epoch": 32.55, |
|
"learning_rate": 1.343684210526316e-05, |
|
"loss": 1.4375, |
|
"step": 11490 |
|
}, |
|
{ |
|
"epoch": 32.58, |
|
"learning_rate": 1.3421052631578948e-05, |
|
"loss": 1.4345, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 32.6, |
|
"learning_rate": 1.3405263157894736e-05, |
|
"loss": 1.4377, |
|
"step": 11510 |
|
}, |
|
{ |
|
"epoch": 32.63, |
|
"learning_rate": 1.3389473684210527e-05, |
|
"loss": 1.4494, |
|
"step": 11520 |
|
}, |
|
{ |
|
"epoch": 32.66, |
|
"learning_rate": 1.3373684210526315e-05, |
|
"loss": 1.4433, |
|
"step": 11530 |
|
}, |
|
{ |
|
"epoch": 32.69, |
|
"learning_rate": 1.3357894736842106e-05, |
|
"loss": 1.4489, |
|
"step": 11540 |
|
}, |
|
{ |
|
"epoch": 32.72, |
|
"learning_rate": 1.3342105263157894e-05, |
|
"loss": 1.4412, |
|
"step": 11550 |
|
}, |
|
{ |
|
"epoch": 32.75, |
|
"learning_rate": 1.3326315789473685e-05, |
|
"loss": 1.4376, |
|
"step": 11560 |
|
}, |
|
{ |
|
"epoch": 32.77, |
|
"learning_rate": 1.3310526315789473e-05, |
|
"loss": 1.4435, |
|
"step": 11570 |
|
}, |
|
{ |
|
"epoch": 32.8, |
|
"learning_rate": 1.3294736842105262e-05, |
|
"loss": 1.4378, |
|
"step": 11580 |
|
}, |
|
{ |
|
"epoch": 32.83, |
|
"learning_rate": 1.3278947368421054e-05, |
|
"loss": 1.4386, |
|
"step": 11590 |
|
}, |
|
{ |
|
"epoch": 32.86, |
|
"learning_rate": 1.3263157894736843e-05, |
|
"loss": 1.4428, |
|
"step": 11600 |
|
}, |
|
{ |
|
"epoch": 32.89, |
|
"learning_rate": 1.3247368421052633e-05, |
|
"loss": 1.4409, |
|
"step": 11610 |
|
}, |
|
{ |
|
"epoch": 32.92, |
|
"learning_rate": 1.3231578947368422e-05, |
|
"loss": 1.441, |
|
"step": 11620 |
|
}, |
|
{ |
|
"epoch": 32.94, |
|
"learning_rate": 1.3215789473684212e-05, |
|
"loss": 1.4356, |
|
"step": 11630 |
|
}, |
|
{ |
|
"epoch": 32.97, |
|
"learning_rate": 1.32e-05, |
|
"loss": 1.4436, |
|
"step": 11640 |
|
}, |
|
{ |
|
"epoch": 33.0, |
|
"learning_rate": 1.318421052631579e-05, |
|
"loss": 1.5665, |
|
"step": 11650 |
|
}, |
|
{ |
|
"epoch": 33.03, |
|
"learning_rate": 1.316842105263158e-05, |
|
"loss": 1.434, |
|
"step": 11660 |
|
}, |
|
{ |
|
"epoch": 33.06, |
|
"learning_rate": 1.3152631578947368e-05, |
|
"loss": 1.4356, |
|
"step": 11670 |
|
}, |
|
{ |
|
"epoch": 33.09, |
|
"learning_rate": 1.3136842105263159e-05, |
|
"loss": 1.4409, |
|
"step": 11680 |
|
}, |
|
{ |
|
"epoch": 33.12, |
|
"learning_rate": 1.3121052631578947e-05, |
|
"loss": 1.4436, |
|
"step": 11690 |
|
}, |
|
{ |
|
"epoch": 33.14, |
|
"learning_rate": 1.3105263157894738e-05, |
|
"loss": 1.438, |
|
"step": 11700 |
|
}, |
|
{ |
|
"epoch": 33.17, |
|
"learning_rate": 1.3089473684210526e-05, |
|
"loss": 1.4404, |
|
"step": 11710 |
|
}, |
|
{ |
|
"epoch": 33.2, |
|
"learning_rate": 1.3073684210526315e-05, |
|
"loss": 1.4388, |
|
"step": 11720 |
|
}, |
|
{ |
|
"epoch": 33.23, |
|
"learning_rate": 1.3057894736842105e-05, |
|
"loss": 1.4436, |
|
"step": 11730 |
|
}, |
|
{ |
|
"epoch": 33.26, |
|
"learning_rate": 1.3042105263157894e-05, |
|
"loss": 1.4393, |
|
"step": 11740 |
|
}, |
|
{ |
|
"epoch": 33.29, |
|
"learning_rate": 1.3026315789473684e-05, |
|
"loss": 1.4385, |
|
"step": 11750 |
|
}, |
|
{ |
|
"epoch": 33.31, |
|
"learning_rate": 1.3010526315789475e-05, |
|
"loss": 1.4362, |
|
"step": 11760 |
|
}, |
|
{ |
|
"epoch": 33.34, |
|
"learning_rate": 1.2994736842105265e-05, |
|
"loss": 1.4369, |
|
"step": 11770 |
|
}, |
|
{ |
|
"epoch": 33.37, |
|
"learning_rate": 1.2978947368421054e-05, |
|
"loss": 1.435, |
|
"step": 11780 |
|
}, |
|
{ |
|
"epoch": 33.4, |
|
"learning_rate": 1.2963157894736842e-05, |
|
"loss": 1.4342, |
|
"step": 11790 |
|
}, |
|
{ |
|
"epoch": 33.43, |
|
"learning_rate": 1.2947368421052633e-05, |
|
"loss": 1.4362, |
|
"step": 11800 |
|
}, |
|
{ |
|
"epoch": 33.45, |
|
"learning_rate": 1.2931578947368421e-05, |
|
"loss": 1.4349, |
|
"step": 11810 |
|
}, |
|
{ |
|
"epoch": 33.48, |
|
"learning_rate": 1.2915789473684212e-05, |
|
"loss": 1.436, |
|
"step": 11820 |
|
}, |
|
{ |
|
"epoch": 33.51, |
|
"learning_rate": 1.29e-05, |
|
"loss": 1.4361, |
|
"step": 11830 |
|
}, |
|
{ |
|
"epoch": 33.54, |
|
"learning_rate": 1.288421052631579e-05, |
|
"loss": 1.4409, |
|
"step": 11840 |
|
}, |
|
{ |
|
"epoch": 33.57, |
|
"learning_rate": 1.2868421052631579e-05, |
|
"loss": 1.4433, |
|
"step": 11850 |
|
}, |
|
{ |
|
"epoch": 33.6, |
|
"learning_rate": 1.2852631578947368e-05, |
|
"loss": 1.4505, |
|
"step": 11860 |
|
}, |
|
{ |
|
"epoch": 33.62, |
|
"learning_rate": 1.2836842105263158e-05, |
|
"loss": 1.4358, |
|
"step": 11870 |
|
}, |
|
{ |
|
"epoch": 33.65, |
|
"learning_rate": 1.2821052631578947e-05, |
|
"loss": 1.442, |
|
"step": 11880 |
|
}, |
|
{ |
|
"epoch": 33.68, |
|
"learning_rate": 1.2805263157894737e-05, |
|
"loss": 1.4447, |
|
"step": 11890 |
|
}, |
|
{ |
|
"epoch": 33.71, |
|
"learning_rate": 1.2789473684210526e-05, |
|
"loss": 1.4462, |
|
"step": 11900 |
|
}, |
|
{ |
|
"epoch": 33.74, |
|
"learning_rate": 1.2773684210526316e-05, |
|
"loss": 1.4398, |
|
"step": 11910 |
|
}, |
|
{ |
|
"epoch": 33.77, |
|
"learning_rate": 1.2757894736842106e-05, |
|
"loss": 1.4486, |
|
"step": 11920 |
|
}, |
|
{ |
|
"epoch": 33.79, |
|
"learning_rate": 1.2742105263157895e-05, |
|
"loss": 1.4388, |
|
"step": 11930 |
|
}, |
|
{ |
|
"epoch": 33.82, |
|
"learning_rate": 1.2726315789473685e-05, |
|
"loss": 1.4432, |
|
"step": 11940 |
|
}, |
|
{ |
|
"epoch": 33.85, |
|
"learning_rate": 1.2710526315789474e-05, |
|
"loss": 1.4357, |
|
"step": 11950 |
|
}, |
|
{ |
|
"epoch": 33.88, |
|
"learning_rate": 1.2694736842105264e-05, |
|
"loss": 1.4354, |
|
"step": 11960 |
|
}, |
|
{ |
|
"epoch": 33.91, |
|
"learning_rate": 1.2678947368421053e-05, |
|
"loss": 1.4359, |
|
"step": 11970 |
|
}, |
|
{ |
|
"epoch": 33.94, |
|
"learning_rate": 1.2663157894736843e-05, |
|
"loss": 1.4386, |
|
"step": 11980 |
|
}, |
|
{ |
|
"epoch": 33.96, |
|
"learning_rate": 1.2647368421052632e-05, |
|
"loss": 1.4405, |
|
"step": 11990 |
|
}, |
|
{ |
|
"epoch": 33.99, |
|
"learning_rate": 1.263157894736842e-05, |
|
"loss": 1.4414, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 33.99, |
|
"eval_denotation_accuracy": 0.5683504062168845, |
|
"eval_loss": 2.178483009338379, |
|
"eval_runtime": 358.5197, |
|
"eval_samples_per_second": 7.896, |
|
"eval_steps_per_second": 1.975, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 34.02, |
|
"learning_rate": 1.2615789473684211e-05, |
|
"loss": 1.563, |
|
"step": 12010 |
|
}, |
|
{ |
|
"epoch": 34.05, |
|
"learning_rate": 1.26e-05, |
|
"loss": 1.4362, |
|
"step": 12020 |
|
}, |
|
{ |
|
"epoch": 34.08, |
|
"learning_rate": 1.258421052631579e-05, |
|
"loss": 1.4417, |
|
"step": 12030 |
|
}, |
|
{ |
|
"epoch": 34.11, |
|
"learning_rate": 1.2568421052631579e-05, |
|
"loss": 1.4401, |
|
"step": 12040 |
|
}, |
|
{ |
|
"epoch": 34.14, |
|
"learning_rate": 1.2552631578947369e-05, |
|
"loss": 1.4399, |
|
"step": 12050 |
|
}, |
|
{ |
|
"epoch": 34.16, |
|
"learning_rate": 1.2536842105263158e-05, |
|
"loss": 1.4323, |
|
"step": 12060 |
|
}, |
|
{ |
|
"epoch": 34.19, |
|
"learning_rate": 1.2521052631578946e-05, |
|
"loss": 1.4393, |
|
"step": 12070 |
|
}, |
|
{ |
|
"epoch": 34.22, |
|
"learning_rate": 1.2505263157894737e-05, |
|
"loss": 1.4359, |
|
"step": 12080 |
|
}, |
|
{ |
|
"epoch": 34.25, |
|
"learning_rate": 1.2489473684210527e-05, |
|
"loss": 1.4371, |
|
"step": 12090 |
|
}, |
|
{ |
|
"epoch": 34.28, |
|
"learning_rate": 1.2473684210526317e-05, |
|
"loss": 1.4342, |
|
"step": 12100 |
|
}, |
|
{ |
|
"epoch": 34.31, |
|
"learning_rate": 1.2457894736842106e-05, |
|
"loss": 1.437, |
|
"step": 12110 |
|
}, |
|
{ |
|
"epoch": 34.33, |
|
"learning_rate": 1.2442105263157896e-05, |
|
"loss": 1.4368, |
|
"step": 12120 |
|
}, |
|
{ |
|
"epoch": 34.36, |
|
"learning_rate": 1.2426315789473685e-05, |
|
"loss": 1.4359, |
|
"step": 12130 |
|
}, |
|
{ |
|
"epoch": 34.39, |
|
"learning_rate": 1.2410526315789474e-05, |
|
"loss": 1.4348, |
|
"step": 12140 |
|
}, |
|
{ |
|
"epoch": 34.42, |
|
"learning_rate": 1.2394736842105264e-05, |
|
"loss": 1.4334, |
|
"step": 12150 |
|
}, |
|
{ |
|
"epoch": 34.45, |
|
"learning_rate": 1.2378947368421053e-05, |
|
"loss": 1.4343, |
|
"step": 12160 |
|
}, |
|
{ |
|
"epoch": 34.47, |
|
"learning_rate": 1.2363157894736843e-05, |
|
"loss": 1.4329, |
|
"step": 12170 |
|
}, |
|
{ |
|
"epoch": 34.5, |
|
"learning_rate": 1.2347368421052631e-05, |
|
"loss": 1.434, |
|
"step": 12180 |
|
}, |
|
{ |
|
"epoch": 34.53, |
|
"learning_rate": 1.2331578947368422e-05, |
|
"loss": 1.4382, |
|
"step": 12190 |
|
}, |
|
{ |
|
"epoch": 34.56, |
|
"learning_rate": 1.231578947368421e-05, |
|
"loss": 1.433, |
|
"step": 12200 |
|
}, |
|
{ |
|
"epoch": 34.59, |
|
"learning_rate": 1.2299999999999999e-05, |
|
"loss": 1.4331, |
|
"step": 12210 |
|
}, |
|
{ |
|
"epoch": 34.62, |
|
"learning_rate": 1.228421052631579e-05, |
|
"loss": 1.4368, |
|
"step": 12220 |
|
}, |
|
{ |
|
"epoch": 34.64, |
|
"learning_rate": 1.2268421052631578e-05, |
|
"loss": 1.4369, |
|
"step": 12230 |
|
}, |
|
{ |
|
"epoch": 34.67, |
|
"learning_rate": 1.2252631578947368e-05, |
|
"loss": 1.4364, |
|
"step": 12240 |
|
}, |
|
{ |
|
"epoch": 34.7, |
|
"learning_rate": 1.2236842105263159e-05, |
|
"loss": 1.4365, |
|
"step": 12250 |
|
}, |
|
{ |
|
"epoch": 34.73, |
|
"learning_rate": 1.2221052631578949e-05, |
|
"loss": 1.4349, |
|
"step": 12260 |
|
}, |
|
{ |
|
"epoch": 34.76, |
|
"learning_rate": 1.2205263157894738e-05, |
|
"loss": 1.4343, |
|
"step": 12270 |
|
}, |
|
{ |
|
"epoch": 34.79, |
|
"learning_rate": 1.2189473684210526e-05, |
|
"loss": 1.4377, |
|
"step": 12280 |
|
}, |
|
{ |
|
"epoch": 34.81, |
|
"learning_rate": 1.2173684210526317e-05, |
|
"loss": 1.4331, |
|
"step": 12290 |
|
}, |
|
{ |
|
"epoch": 34.84, |
|
"learning_rate": 1.2157894736842105e-05, |
|
"loss": 1.4334, |
|
"step": 12300 |
|
}, |
|
{ |
|
"epoch": 34.87, |
|
"learning_rate": 1.2142105263157896e-05, |
|
"loss": 1.4392, |
|
"step": 12310 |
|
}, |
|
{ |
|
"epoch": 34.9, |
|
"learning_rate": 1.2126315789473684e-05, |
|
"loss": 1.4378, |
|
"step": 12320 |
|
}, |
|
{ |
|
"epoch": 34.93, |
|
"learning_rate": 1.2110526315789475e-05, |
|
"loss": 1.4321, |
|
"step": 12330 |
|
}, |
|
{ |
|
"epoch": 34.96, |
|
"learning_rate": 1.2094736842105263e-05, |
|
"loss": 1.4379, |
|
"step": 12340 |
|
}, |
|
{ |
|
"epoch": 34.98, |
|
"learning_rate": 1.2078947368421052e-05, |
|
"loss": 1.4351, |
|
"step": 12350 |
|
}, |
|
{ |
|
"epoch": 35.01, |
|
"learning_rate": 1.2063157894736842e-05, |
|
"loss": 1.5578, |
|
"step": 12360 |
|
}, |
|
{ |
|
"epoch": 35.04, |
|
"learning_rate": 1.2047368421052631e-05, |
|
"loss": 1.4329, |
|
"step": 12370 |
|
}, |
|
{ |
|
"epoch": 35.07, |
|
"learning_rate": 1.2031578947368421e-05, |
|
"loss": 1.4325, |
|
"step": 12380 |
|
}, |
|
{ |
|
"epoch": 35.1, |
|
"learning_rate": 1.201578947368421e-05, |
|
"loss": 1.4342, |
|
"step": 12390 |
|
}, |
|
{ |
|
"epoch": 35.13, |
|
"learning_rate": 1.2e-05, |
|
"loss": 1.4406, |
|
"step": 12400 |
|
}, |
|
{ |
|
"epoch": 35.16, |
|
"learning_rate": 1.1984210526315789e-05, |
|
"loss": 1.4337, |
|
"step": 12410 |
|
}, |
|
{ |
|
"epoch": 35.18, |
|
"learning_rate": 1.196842105263158e-05, |
|
"loss": 1.4424, |
|
"step": 12420 |
|
}, |
|
{ |
|
"epoch": 35.21, |
|
"learning_rate": 1.195263157894737e-05, |
|
"loss": 1.4368, |
|
"step": 12430 |
|
}, |
|
{ |
|
"epoch": 35.24, |
|
"learning_rate": 1.1936842105263158e-05, |
|
"loss": 1.4357, |
|
"step": 12440 |
|
}, |
|
{ |
|
"epoch": 35.27, |
|
"learning_rate": 1.1921052631578949e-05, |
|
"loss": 1.4333, |
|
"step": 12450 |
|
}, |
|
{ |
|
"epoch": 35.3, |
|
"learning_rate": 1.1905263157894737e-05, |
|
"loss": 1.4336, |
|
"step": 12460 |
|
}, |
|
{ |
|
"epoch": 35.32, |
|
"learning_rate": 1.1889473684210528e-05, |
|
"loss": 1.4349, |
|
"step": 12470 |
|
}, |
|
{ |
|
"epoch": 35.35, |
|
"learning_rate": 1.1873684210526316e-05, |
|
"loss": 1.4329, |
|
"step": 12480 |
|
}, |
|
{ |
|
"epoch": 35.38, |
|
"learning_rate": 1.1857894736842105e-05, |
|
"loss": 1.4317, |
|
"step": 12490 |
|
}, |
|
{ |
|
"epoch": 35.41, |
|
"learning_rate": 1.1842105263157895e-05, |
|
"loss": 1.4354, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 35.44, |
|
"learning_rate": 1.1826315789473684e-05, |
|
"loss": 1.4361, |
|
"step": 12510 |
|
}, |
|
{ |
|
"epoch": 35.47, |
|
"learning_rate": 1.1810526315789474e-05, |
|
"loss": 1.4375, |
|
"step": 12520 |
|
}, |
|
{ |
|
"epoch": 35.49, |
|
"learning_rate": 1.1794736842105263e-05, |
|
"loss": 1.4348, |
|
"step": 12530 |
|
}, |
|
{ |
|
"epoch": 35.52, |
|
"learning_rate": 1.1778947368421053e-05, |
|
"loss": 1.4358, |
|
"step": 12540 |
|
}, |
|
{ |
|
"epoch": 35.55, |
|
"learning_rate": 1.1763157894736842e-05, |
|
"loss": 1.4317, |
|
"step": 12550 |
|
}, |
|
{ |
|
"epoch": 35.58, |
|
"learning_rate": 1.174736842105263e-05, |
|
"loss": 1.4334, |
|
"step": 12560 |
|
}, |
|
{ |
|
"epoch": 35.61, |
|
"learning_rate": 1.173157894736842e-05, |
|
"loss": 1.4378, |
|
"step": 12570 |
|
}, |
|
{ |
|
"epoch": 35.64, |
|
"learning_rate": 1.171578947368421e-05, |
|
"loss": 1.4345, |
|
"step": 12580 |
|
}, |
|
{ |
|
"epoch": 35.66, |
|
"learning_rate": 1.1700000000000001e-05, |
|
"loss": 1.4424, |
|
"step": 12590 |
|
}, |
|
{ |
|
"epoch": 35.69, |
|
"learning_rate": 1.168421052631579e-05, |
|
"loss": 1.437, |
|
"step": 12600 |
|
}, |
|
{ |
|
"epoch": 35.72, |
|
"learning_rate": 1.166842105263158e-05, |
|
"loss": 1.4333, |
|
"step": 12610 |
|
}, |
|
{ |
|
"epoch": 35.75, |
|
"learning_rate": 1.1652631578947369e-05, |
|
"loss": 1.4348, |
|
"step": 12620 |
|
}, |
|
{ |
|
"epoch": 35.78, |
|
"learning_rate": 1.1636842105263158e-05, |
|
"loss": 1.4365, |
|
"step": 12630 |
|
}, |
|
{ |
|
"epoch": 35.81, |
|
"learning_rate": 1.1621052631578948e-05, |
|
"loss": 1.4356, |
|
"step": 12640 |
|
}, |
|
{ |
|
"epoch": 35.83, |
|
"learning_rate": 1.1605263157894737e-05, |
|
"loss": 1.4344, |
|
"step": 12650 |
|
}, |
|
{ |
|
"epoch": 35.86, |
|
"learning_rate": 1.1589473684210527e-05, |
|
"loss": 1.4359, |
|
"step": 12660 |
|
}, |
|
{ |
|
"epoch": 35.89, |
|
"learning_rate": 1.1573684210526316e-05, |
|
"loss": 1.4332, |
|
"step": 12670 |
|
}, |
|
{ |
|
"epoch": 35.92, |
|
"learning_rate": 1.1557894736842106e-05, |
|
"loss": 1.437, |
|
"step": 12680 |
|
}, |
|
{ |
|
"epoch": 35.95, |
|
"learning_rate": 1.1542105263157895e-05, |
|
"loss": 1.4394, |
|
"step": 12690 |
|
}, |
|
{ |
|
"epoch": 35.97, |
|
"learning_rate": 1.1526315789473683e-05, |
|
"loss": 1.4346, |
|
"step": 12700 |
|
}, |
|
{ |
|
"epoch": 36.01, |
|
"learning_rate": 1.1510526315789474e-05, |
|
"loss": 1.5572, |
|
"step": 12710 |
|
}, |
|
{ |
|
"epoch": 36.03, |
|
"learning_rate": 1.1494736842105262e-05, |
|
"loss": 1.433, |
|
"step": 12720 |
|
}, |
|
{ |
|
"epoch": 36.06, |
|
"learning_rate": 1.1478947368421053e-05, |
|
"loss": 1.4303, |
|
"step": 12730 |
|
}, |
|
{ |
|
"epoch": 36.09, |
|
"learning_rate": 1.1463157894736841e-05, |
|
"loss": 1.4337, |
|
"step": 12740 |
|
}, |
|
{ |
|
"epoch": 36.12, |
|
"learning_rate": 1.1447368421052632e-05, |
|
"loss": 1.4343, |
|
"step": 12750 |
|
}, |
|
{ |
|
"epoch": 36.15, |
|
"learning_rate": 1.1431578947368422e-05, |
|
"loss": 1.4361, |
|
"step": 12760 |
|
}, |
|
{ |
|
"epoch": 36.18, |
|
"learning_rate": 1.141578947368421e-05, |
|
"loss": 1.4425, |
|
"step": 12770 |
|
}, |
|
{ |
|
"epoch": 36.2, |
|
"learning_rate": 1.1400000000000001e-05, |
|
"loss": 1.4347, |
|
"step": 12780 |
|
}, |
|
{ |
|
"epoch": 36.23, |
|
"learning_rate": 1.138421052631579e-05, |
|
"loss": 1.4311, |
|
"step": 12790 |
|
}, |
|
{ |
|
"epoch": 36.26, |
|
"learning_rate": 1.136842105263158e-05, |
|
"loss": 1.4336, |
|
"step": 12800 |
|
}, |
|
{ |
|
"epoch": 36.29, |
|
"learning_rate": 1.1352631578947369e-05, |
|
"loss": 1.4356, |
|
"step": 12810 |
|
}, |
|
{ |
|
"epoch": 36.32, |
|
"learning_rate": 1.1336842105263159e-05, |
|
"loss": 1.4348, |
|
"step": 12820 |
|
}, |
|
{ |
|
"epoch": 36.34, |
|
"learning_rate": 1.1321052631578948e-05, |
|
"loss": 1.4379, |
|
"step": 12830 |
|
}, |
|
{ |
|
"epoch": 36.37, |
|
"learning_rate": 1.1305263157894736e-05, |
|
"loss": 1.438, |
|
"step": 12840 |
|
}, |
|
{ |
|
"epoch": 36.4, |
|
"learning_rate": 1.1289473684210527e-05, |
|
"loss": 1.4303, |
|
"step": 12850 |
|
}, |
|
{ |
|
"epoch": 36.43, |
|
"learning_rate": 1.1273684210526315e-05, |
|
"loss": 1.4343, |
|
"step": 12860 |
|
}, |
|
{ |
|
"epoch": 36.46, |
|
"learning_rate": 1.1257894736842106e-05, |
|
"loss": 1.4314, |
|
"step": 12870 |
|
}, |
|
{ |
|
"epoch": 36.49, |
|
"learning_rate": 1.1242105263157894e-05, |
|
"loss": 1.4331, |
|
"step": 12880 |
|
}, |
|
{ |
|
"epoch": 36.51, |
|
"learning_rate": 1.1226315789473685e-05, |
|
"loss": 1.4341, |
|
"step": 12890 |
|
}, |
|
{ |
|
"epoch": 36.54, |
|
"learning_rate": 1.1210526315789473e-05, |
|
"loss": 1.4332, |
|
"step": 12900 |
|
}, |
|
{ |
|
"epoch": 36.57, |
|
"learning_rate": 1.1194736842105264e-05, |
|
"loss": 1.4327, |
|
"step": 12910 |
|
}, |
|
{ |
|
"epoch": 36.6, |
|
"learning_rate": 1.1178947368421054e-05, |
|
"loss": 1.4338, |
|
"step": 12920 |
|
}, |
|
{ |
|
"epoch": 36.63, |
|
"learning_rate": 1.1163157894736842e-05, |
|
"loss": 1.4302, |
|
"step": 12930 |
|
}, |
|
{ |
|
"epoch": 36.66, |
|
"learning_rate": 1.1147368421052633e-05, |
|
"loss": 1.4345, |
|
"step": 12940 |
|
}, |
|
{ |
|
"epoch": 36.68, |
|
"learning_rate": 1.1131578947368421e-05, |
|
"loss": 1.4332, |
|
"step": 12950 |
|
}, |
|
{ |
|
"epoch": 36.71, |
|
"learning_rate": 1.1115789473684212e-05, |
|
"loss": 1.4319, |
|
"step": 12960 |
|
}, |
|
{ |
|
"epoch": 36.74, |
|
"learning_rate": 1.11e-05, |
|
"loss": 1.4384, |
|
"step": 12970 |
|
}, |
|
{ |
|
"epoch": 36.77, |
|
"learning_rate": 1.108421052631579e-05, |
|
"loss": 1.4363, |
|
"step": 12980 |
|
}, |
|
{ |
|
"epoch": 36.8, |
|
"learning_rate": 1.106842105263158e-05, |
|
"loss": 1.4329, |
|
"step": 12990 |
|
}, |
|
{ |
|
"epoch": 36.83, |
|
"learning_rate": 1.1052631578947368e-05, |
|
"loss": 1.431, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 36.83, |
|
"eval_denotation_accuracy": 0.571529494878135, |
|
"eval_loss": 2.1800432205200195, |
|
"eval_runtime": 373.5447, |
|
"eval_samples_per_second": 7.579, |
|
"eval_steps_per_second": 1.895, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 36.85, |
|
"learning_rate": 1.1036842105263158e-05, |
|
"loss": 1.433, |
|
"step": 13010 |
|
}, |
|
{ |
|
"epoch": 36.88, |
|
"learning_rate": 1.1021052631578947e-05, |
|
"loss": 1.4333, |
|
"step": 13020 |
|
}, |
|
{ |
|
"epoch": 36.91, |
|
"learning_rate": 1.1005263157894737e-05, |
|
"loss": 1.4356, |
|
"step": 13030 |
|
}, |
|
{ |
|
"epoch": 36.94, |
|
"learning_rate": 1.0989473684210526e-05, |
|
"loss": 1.434, |
|
"step": 13040 |
|
}, |
|
{ |
|
"epoch": 36.97, |
|
"learning_rate": 1.0973684210526316e-05, |
|
"loss": 1.432, |
|
"step": 13050 |
|
}, |
|
{ |
|
"epoch": 36.99, |
|
"learning_rate": 1.0957894736842105e-05, |
|
"loss": 1.4389, |
|
"step": 13060 |
|
}, |
|
{ |
|
"epoch": 37.03, |
|
"learning_rate": 1.0942105263157894e-05, |
|
"loss": 1.5555, |
|
"step": 13070 |
|
}, |
|
{ |
|
"epoch": 37.05, |
|
"learning_rate": 1.0926315789473684e-05, |
|
"loss": 1.4337, |
|
"step": 13080 |
|
}, |
|
{ |
|
"epoch": 37.08, |
|
"learning_rate": 1.0910526315789474e-05, |
|
"loss": 1.4387, |
|
"step": 13090 |
|
}, |
|
{ |
|
"epoch": 37.11, |
|
"learning_rate": 1.0894736842105265e-05, |
|
"loss": 1.4395, |
|
"step": 13100 |
|
}, |
|
{ |
|
"epoch": 37.14, |
|
"learning_rate": 1.0878947368421053e-05, |
|
"loss": 1.442, |
|
"step": 13110 |
|
}, |
|
{ |
|
"epoch": 37.17, |
|
"learning_rate": 1.0863157894736844e-05, |
|
"loss": 1.4446, |
|
"step": 13120 |
|
}, |
|
{ |
|
"epoch": 37.19, |
|
"learning_rate": 1.0847368421052632e-05, |
|
"loss": 1.4369, |
|
"step": 13130 |
|
}, |
|
{ |
|
"epoch": 37.22, |
|
"learning_rate": 1.0831578947368421e-05, |
|
"loss": 1.4341, |
|
"step": 13140 |
|
}, |
|
{ |
|
"epoch": 37.25, |
|
"learning_rate": 1.0815789473684211e-05, |
|
"loss": 1.4337, |
|
"step": 13150 |
|
}, |
|
{ |
|
"epoch": 37.28, |
|
"learning_rate": 1.08e-05, |
|
"loss": 1.4353, |
|
"step": 13160 |
|
}, |
|
{ |
|
"epoch": 37.31, |
|
"learning_rate": 1.078421052631579e-05, |
|
"loss": 1.4331, |
|
"step": 13170 |
|
}, |
|
{ |
|
"epoch": 37.34, |
|
"learning_rate": 1.0768421052631579e-05, |
|
"loss": 1.4359, |
|
"step": 13180 |
|
}, |
|
{ |
|
"epoch": 37.36, |
|
"learning_rate": 1.075263157894737e-05, |
|
"loss": 1.4319, |
|
"step": 13190 |
|
}, |
|
{ |
|
"epoch": 37.39, |
|
"learning_rate": 1.0736842105263158e-05, |
|
"loss": 1.4307, |
|
"step": 13200 |
|
}, |
|
{ |
|
"epoch": 37.42, |
|
"learning_rate": 1.0721052631578947e-05, |
|
"loss": 1.4306, |
|
"step": 13210 |
|
}, |
|
{ |
|
"epoch": 37.45, |
|
"learning_rate": 1.0705263157894737e-05, |
|
"loss": 1.4331, |
|
"step": 13220 |
|
}, |
|
{ |
|
"epoch": 37.48, |
|
"learning_rate": 1.0689473684210526e-05, |
|
"loss": 1.4386, |
|
"step": 13230 |
|
}, |
|
{ |
|
"epoch": 37.51, |
|
"learning_rate": 1.0673684210526316e-05, |
|
"loss": 1.4362, |
|
"step": 13240 |
|
}, |
|
{ |
|
"epoch": 37.53, |
|
"learning_rate": 1.0657894736842106e-05, |
|
"loss": 1.4335, |
|
"step": 13250 |
|
}, |
|
{ |
|
"epoch": 37.56, |
|
"learning_rate": 1.0642105263157897e-05, |
|
"loss": 1.4323, |
|
"step": 13260 |
|
}, |
|
{ |
|
"epoch": 37.59, |
|
"learning_rate": 1.0626315789473685e-05, |
|
"loss": 1.4331, |
|
"step": 13270 |
|
}, |
|
{ |
|
"epoch": 37.62, |
|
"learning_rate": 1.0610526315789474e-05, |
|
"loss": 1.431, |
|
"step": 13280 |
|
}, |
|
{ |
|
"epoch": 37.65, |
|
"learning_rate": 1.0594736842105264e-05, |
|
"loss": 1.4342, |
|
"step": 13290 |
|
}, |
|
{ |
|
"epoch": 37.68, |
|
"learning_rate": 1.0578947368421053e-05, |
|
"loss": 1.4299, |
|
"step": 13300 |
|
}, |
|
{ |
|
"epoch": 37.7, |
|
"learning_rate": 1.0563157894736843e-05, |
|
"loss": 1.4336, |
|
"step": 13310 |
|
}, |
|
{ |
|
"epoch": 37.73, |
|
"learning_rate": 1.0547368421052632e-05, |
|
"loss": 1.4364, |
|
"step": 13320 |
|
}, |
|
{ |
|
"epoch": 37.76, |
|
"learning_rate": 1.0531578947368422e-05, |
|
"loss": 1.4352, |
|
"step": 13330 |
|
}, |
|
{ |
|
"epoch": 37.79, |
|
"learning_rate": 1.051578947368421e-05, |
|
"loss": 1.435, |
|
"step": 13340 |
|
}, |
|
{ |
|
"epoch": 37.82, |
|
"learning_rate": 1.05e-05, |
|
"loss": 1.4351, |
|
"step": 13350 |
|
}, |
|
{ |
|
"epoch": 37.84, |
|
"learning_rate": 1.048421052631579e-05, |
|
"loss": 1.4322, |
|
"step": 13360 |
|
}, |
|
{ |
|
"epoch": 37.87, |
|
"learning_rate": 1.0468421052631578e-05, |
|
"loss": 1.4318, |
|
"step": 13370 |
|
}, |
|
{ |
|
"epoch": 37.9, |
|
"learning_rate": 1.0452631578947369e-05, |
|
"loss": 1.4326, |
|
"step": 13380 |
|
}, |
|
{ |
|
"epoch": 37.93, |
|
"learning_rate": 1.0436842105263157e-05, |
|
"loss": 1.4351, |
|
"step": 13390 |
|
}, |
|
{ |
|
"epoch": 37.96, |
|
"learning_rate": 1.0421052631578948e-05, |
|
"loss": 1.4349, |
|
"step": 13400 |
|
}, |
|
{ |
|
"epoch": 37.99, |
|
"learning_rate": 1.0405263157894736e-05, |
|
"loss": 1.4353, |
|
"step": 13410 |
|
}, |
|
{ |
|
"epoch": 38.02, |
|
"learning_rate": 1.0389473684210527e-05, |
|
"loss": 1.557, |
|
"step": 13420 |
|
}, |
|
{ |
|
"epoch": 38.05, |
|
"learning_rate": 1.0373684210526317e-05, |
|
"loss": 1.4375, |
|
"step": 13430 |
|
}, |
|
{ |
|
"epoch": 38.07, |
|
"learning_rate": 1.0357894736842106e-05, |
|
"loss": 1.4359, |
|
"step": 13440 |
|
}, |
|
{ |
|
"epoch": 38.1, |
|
"learning_rate": 1.0342105263157896e-05, |
|
"loss": 1.4353, |
|
"step": 13450 |
|
}, |
|
{ |
|
"epoch": 38.13, |
|
"learning_rate": 1.0326315789473685e-05, |
|
"loss": 1.4326, |
|
"step": 13460 |
|
}, |
|
{ |
|
"epoch": 38.16, |
|
"learning_rate": 1.0310526315789475e-05, |
|
"loss": 1.4343, |
|
"step": 13470 |
|
}, |
|
{ |
|
"epoch": 38.19, |
|
"learning_rate": 1.0294736842105264e-05, |
|
"loss": 1.4347, |
|
"step": 13480 |
|
}, |
|
{ |
|
"epoch": 38.21, |
|
"learning_rate": 1.0278947368421052e-05, |
|
"loss": 1.4322, |
|
"step": 13490 |
|
}, |
|
{ |
|
"epoch": 38.24, |
|
"learning_rate": 1.0263157894736843e-05, |
|
"loss": 1.4324, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 38.27, |
|
"learning_rate": 1.0247368421052631e-05, |
|
"loss": 1.4313, |
|
"step": 13510 |
|
}, |
|
{ |
|
"epoch": 38.3, |
|
"learning_rate": 1.0231578947368422e-05, |
|
"loss": 1.4285, |
|
"step": 13520 |
|
}, |
|
{ |
|
"epoch": 38.33, |
|
"learning_rate": 1.021578947368421e-05, |
|
"loss": 1.4301, |
|
"step": 13530 |
|
}, |
|
{ |
|
"epoch": 38.36, |
|
"learning_rate": 1.02e-05, |
|
"loss": 1.4311, |
|
"step": 13540 |
|
}, |
|
{ |
|
"epoch": 38.38, |
|
"learning_rate": 1.018421052631579e-05, |
|
"loss": 1.4426, |
|
"step": 13550 |
|
}, |
|
{ |
|
"epoch": 38.41, |
|
"learning_rate": 1.0168421052631578e-05, |
|
"loss": 1.433, |
|
"step": 13560 |
|
}, |
|
{ |
|
"epoch": 38.44, |
|
"learning_rate": 1.0152631578947368e-05, |
|
"loss": 1.4326, |
|
"step": 13570 |
|
}, |
|
{ |
|
"epoch": 38.47, |
|
"learning_rate": 1.0136842105263157e-05, |
|
"loss": 1.4286, |
|
"step": 13580 |
|
}, |
|
{ |
|
"epoch": 38.5, |
|
"learning_rate": 1.0121052631578949e-05, |
|
"loss": 1.4321, |
|
"step": 13590 |
|
}, |
|
{ |
|
"epoch": 38.53, |
|
"learning_rate": 1.0105263157894738e-05, |
|
"loss": 1.4361, |
|
"step": 13600 |
|
}, |
|
{ |
|
"epoch": 38.55, |
|
"learning_rate": 1.0089473684210528e-05, |
|
"loss": 1.4319, |
|
"step": 13610 |
|
}, |
|
{ |
|
"epoch": 38.58, |
|
"learning_rate": 1.0073684210526317e-05, |
|
"loss": 1.431, |
|
"step": 13620 |
|
}, |
|
{ |
|
"epoch": 38.61, |
|
"learning_rate": 1.0057894736842105e-05, |
|
"loss": 1.4281, |
|
"step": 13630 |
|
}, |
|
{ |
|
"epoch": 38.64, |
|
"learning_rate": 1.0042105263157896e-05, |
|
"loss": 1.4335, |
|
"step": 13640 |
|
}, |
|
{ |
|
"epoch": 38.67, |
|
"learning_rate": 1.0026315789473684e-05, |
|
"loss": 1.4363, |
|
"step": 13650 |
|
}, |
|
{ |
|
"epoch": 38.7, |
|
"learning_rate": 1.0010526315789474e-05, |
|
"loss": 1.4297, |
|
"step": 13660 |
|
}, |
|
{ |
|
"epoch": 38.72, |
|
"learning_rate": 9.994736842105263e-06, |
|
"loss": 1.4328, |
|
"step": 13670 |
|
}, |
|
{ |
|
"epoch": 38.75, |
|
"learning_rate": 9.978947368421053e-06, |
|
"loss": 1.4329, |
|
"step": 13680 |
|
}, |
|
{ |
|
"epoch": 38.78, |
|
"learning_rate": 9.963157894736842e-06, |
|
"loss": 1.4306, |
|
"step": 13690 |
|
}, |
|
{ |
|
"epoch": 38.81, |
|
"learning_rate": 9.94736842105263e-06, |
|
"loss": 1.4342, |
|
"step": 13700 |
|
}, |
|
{ |
|
"epoch": 38.84, |
|
"learning_rate": 9.931578947368421e-06, |
|
"loss": 1.4362, |
|
"step": 13710 |
|
}, |
|
{ |
|
"epoch": 38.86, |
|
"learning_rate": 9.91578947368421e-06, |
|
"loss": 1.4336, |
|
"step": 13720 |
|
}, |
|
{ |
|
"epoch": 38.89, |
|
"learning_rate": 9.9e-06, |
|
"loss": 1.439, |
|
"step": 13730 |
|
}, |
|
{ |
|
"epoch": 38.92, |
|
"learning_rate": 9.884210526315789e-06, |
|
"loss": 1.4341, |
|
"step": 13740 |
|
}, |
|
{ |
|
"epoch": 38.95, |
|
"learning_rate": 9.868421052631579e-06, |
|
"loss": 1.4328, |
|
"step": 13750 |
|
}, |
|
{ |
|
"epoch": 38.98, |
|
"learning_rate": 9.85263157894737e-06, |
|
"loss": 1.4378, |
|
"step": 13760 |
|
}, |
|
{ |
|
"epoch": 39.01, |
|
"learning_rate": 9.836842105263158e-06, |
|
"loss": 1.5636, |
|
"step": 13770 |
|
}, |
|
{ |
|
"epoch": 39.04, |
|
"learning_rate": 9.821052631578948e-06, |
|
"loss": 1.4318, |
|
"step": 13780 |
|
}, |
|
{ |
|
"epoch": 39.06, |
|
"learning_rate": 9.805263157894737e-06, |
|
"loss": 1.4361, |
|
"step": 13790 |
|
}, |
|
{ |
|
"epoch": 39.09, |
|
"learning_rate": 9.789473684210527e-06, |
|
"loss": 1.4359, |
|
"step": 13800 |
|
}, |
|
{ |
|
"epoch": 39.12, |
|
"learning_rate": 9.773684210526316e-06, |
|
"loss": 1.4312, |
|
"step": 13810 |
|
}, |
|
{ |
|
"epoch": 39.15, |
|
"learning_rate": 9.757894736842106e-06, |
|
"loss": 1.4304, |
|
"step": 13820 |
|
}, |
|
{ |
|
"epoch": 39.18, |
|
"learning_rate": 9.742105263157895e-06, |
|
"loss": 1.4334, |
|
"step": 13830 |
|
}, |
|
{ |
|
"epoch": 39.21, |
|
"learning_rate": 9.726315789473684e-06, |
|
"loss": 1.4335, |
|
"step": 13840 |
|
}, |
|
{ |
|
"epoch": 39.23, |
|
"learning_rate": 9.710526315789474e-06, |
|
"loss": 1.4316, |
|
"step": 13850 |
|
}, |
|
{ |
|
"epoch": 39.26, |
|
"learning_rate": 9.694736842105263e-06, |
|
"loss": 1.4287, |
|
"step": 13860 |
|
}, |
|
{ |
|
"epoch": 39.29, |
|
"learning_rate": 9.678947368421053e-06, |
|
"loss": 1.4313, |
|
"step": 13870 |
|
}, |
|
{ |
|
"epoch": 39.32, |
|
"learning_rate": 9.663157894736842e-06, |
|
"loss": 1.4315, |
|
"step": 13880 |
|
}, |
|
{ |
|
"epoch": 39.35, |
|
"learning_rate": 9.647368421052632e-06, |
|
"loss": 1.4299, |
|
"step": 13890 |
|
}, |
|
{ |
|
"epoch": 39.38, |
|
"learning_rate": 9.63157894736842e-06, |
|
"loss": 1.4298, |
|
"step": 13900 |
|
}, |
|
{ |
|
"epoch": 39.4, |
|
"learning_rate": 9.61578947368421e-06, |
|
"loss": 1.4294, |
|
"step": 13910 |
|
}, |
|
{ |
|
"epoch": 39.43, |
|
"learning_rate": 9.600000000000001e-06, |
|
"loss": 1.4292, |
|
"step": 13920 |
|
}, |
|
{ |
|
"epoch": 39.46, |
|
"learning_rate": 9.58421052631579e-06, |
|
"loss": 1.4318, |
|
"step": 13930 |
|
}, |
|
{ |
|
"epoch": 39.49, |
|
"learning_rate": 9.56842105263158e-06, |
|
"loss": 1.4333, |
|
"step": 13940 |
|
}, |
|
{ |
|
"epoch": 39.52, |
|
"learning_rate": 9.552631578947369e-06, |
|
"loss": 1.4315, |
|
"step": 13950 |
|
}, |
|
{ |
|
"epoch": 39.55, |
|
"learning_rate": 9.53684210526316e-06, |
|
"loss": 1.4308, |
|
"step": 13960 |
|
}, |
|
{ |
|
"epoch": 39.57, |
|
"learning_rate": 9.521052631578948e-06, |
|
"loss": 1.4315, |
|
"step": 13970 |
|
}, |
|
{ |
|
"epoch": 39.6, |
|
"learning_rate": 9.505263157894737e-06, |
|
"loss": 1.4305, |
|
"step": 13980 |
|
}, |
|
{ |
|
"epoch": 39.63, |
|
"learning_rate": 9.489473684210527e-06, |
|
"loss": 1.4323, |
|
"step": 13990 |
|
}, |
|
{ |
|
"epoch": 39.66, |
|
"learning_rate": 9.473684210526315e-06, |
|
"loss": 1.4319, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 39.66, |
|
"eval_denotation_accuracy": 0.5778876722006359, |
|
"eval_loss": 2.177828311920166, |
|
"eval_runtime": 341.1576, |
|
"eval_samples_per_second": 8.298, |
|
"eval_steps_per_second": 2.075, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 39.69, |
|
"learning_rate": 9.457894736842106e-06, |
|
"loss": 1.4318, |
|
"step": 14010 |
|
}, |
|
{ |
|
"epoch": 39.71, |
|
"learning_rate": 9.442105263157894e-06, |
|
"loss": 1.4328, |
|
"step": 14020 |
|
}, |
|
{ |
|
"epoch": 39.74, |
|
"learning_rate": 9.426315789473685e-06, |
|
"loss": 1.4336, |
|
"step": 14030 |
|
}, |
|
{ |
|
"epoch": 39.77, |
|
"learning_rate": 9.410526315789473e-06, |
|
"loss": 1.4325, |
|
"step": 14040 |
|
}, |
|
{ |
|
"epoch": 39.8, |
|
"learning_rate": 9.394736842105262e-06, |
|
"loss": 1.4311, |
|
"step": 14050 |
|
}, |
|
{ |
|
"epoch": 39.83, |
|
"learning_rate": 9.378947368421052e-06, |
|
"loss": 1.4327, |
|
"step": 14060 |
|
}, |
|
{ |
|
"epoch": 39.86, |
|
"learning_rate": 9.363157894736841e-06, |
|
"loss": 1.4325, |
|
"step": 14070 |
|
}, |
|
{ |
|
"epoch": 39.88, |
|
"learning_rate": 9.347368421052631e-06, |
|
"loss": 1.4332, |
|
"step": 14080 |
|
}, |
|
{ |
|
"epoch": 39.91, |
|
"learning_rate": 9.331578947368422e-06, |
|
"loss": 1.4337, |
|
"step": 14090 |
|
}, |
|
{ |
|
"epoch": 39.94, |
|
"learning_rate": 9.315789473684212e-06, |
|
"loss": 1.4315, |
|
"step": 14100 |
|
}, |
|
{ |
|
"epoch": 39.97, |
|
"learning_rate": 9.3e-06, |
|
"loss": 1.4296, |
|
"step": 14110 |
|
}, |
|
{ |
|
"epoch": 40.0, |
|
"learning_rate": 9.28421052631579e-06, |
|
"loss": 1.4475, |
|
"step": 14120 |
|
}, |
|
{ |
|
"epoch": 40.03, |
|
"learning_rate": 9.26842105263158e-06, |
|
"loss": 1.5554, |
|
"step": 14130 |
|
}, |
|
{ |
|
"epoch": 40.06, |
|
"learning_rate": 9.252631578947368e-06, |
|
"loss": 1.4305, |
|
"step": 14140 |
|
}, |
|
{ |
|
"epoch": 40.08, |
|
"learning_rate": 9.236842105263159e-06, |
|
"loss": 1.4331, |
|
"step": 14150 |
|
}, |
|
{ |
|
"epoch": 40.11, |
|
"learning_rate": 9.221052631578947e-06, |
|
"loss": 1.4349, |
|
"step": 14160 |
|
}, |
|
{ |
|
"epoch": 40.14, |
|
"learning_rate": 9.205263157894738e-06, |
|
"loss": 1.4333, |
|
"step": 14170 |
|
}, |
|
{ |
|
"epoch": 40.17, |
|
"learning_rate": 9.189473684210526e-06, |
|
"loss": 1.4289, |
|
"step": 14180 |
|
}, |
|
{ |
|
"epoch": 40.2, |
|
"learning_rate": 9.173684210526315e-06, |
|
"loss": 1.4321, |
|
"step": 14190 |
|
}, |
|
{ |
|
"epoch": 40.23, |
|
"learning_rate": 9.157894736842105e-06, |
|
"loss": 1.435, |
|
"step": 14200 |
|
}, |
|
{ |
|
"epoch": 40.25, |
|
"learning_rate": 9.142105263157894e-06, |
|
"loss": 1.4276, |
|
"step": 14210 |
|
}, |
|
{ |
|
"epoch": 40.28, |
|
"learning_rate": 9.126315789473684e-06, |
|
"loss": 1.4337, |
|
"step": 14220 |
|
}, |
|
{ |
|
"epoch": 40.31, |
|
"learning_rate": 9.110526315789473e-06, |
|
"loss": 1.4326, |
|
"step": 14230 |
|
}, |
|
{ |
|
"epoch": 40.34, |
|
"learning_rate": 9.094736842105263e-06, |
|
"loss": 1.4431, |
|
"step": 14240 |
|
}, |
|
{ |
|
"epoch": 40.37, |
|
"learning_rate": 9.078947368421054e-06, |
|
"loss": 1.4355, |
|
"step": 14250 |
|
}, |
|
{ |
|
"epoch": 40.4, |
|
"learning_rate": 9.063157894736842e-06, |
|
"loss": 1.43, |
|
"step": 14260 |
|
}, |
|
{ |
|
"epoch": 40.42, |
|
"learning_rate": 9.047368421052633e-06, |
|
"loss": 1.4316, |
|
"step": 14270 |
|
}, |
|
{ |
|
"epoch": 40.45, |
|
"learning_rate": 9.031578947368421e-06, |
|
"loss": 1.4322, |
|
"step": 14280 |
|
}, |
|
{ |
|
"epoch": 40.48, |
|
"learning_rate": 9.015789473684212e-06, |
|
"loss": 1.4332, |
|
"step": 14290 |
|
}, |
|
{ |
|
"epoch": 40.51, |
|
"learning_rate": 9e-06, |
|
"loss": 1.4298, |
|
"step": 14300 |
|
}, |
|
{ |
|
"epoch": 40.54, |
|
"learning_rate": 8.98421052631579e-06, |
|
"loss": 1.4304, |
|
"step": 14310 |
|
}, |
|
{ |
|
"epoch": 40.57, |
|
"learning_rate": 8.96842105263158e-06, |
|
"loss": 1.433, |
|
"step": 14320 |
|
}, |
|
{ |
|
"epoch": 40.59, |
|
"learning_rate": 8.952631578947368e-06, |
|
"loss": 1.4295, |
|
"step": 14330 |
|
}, |
|
{ |
|
"epoch": 40.62, |
|
"learning_rate": 8.936842105263158e-06, |
|
"loss": 1.4269, |
|
"step": 14340 |
|
}, |
|
{ |
|
"epoch": 40.65, |
|
"learning_rate": 8.921052631578947e-06, |
|
"loss": 1.4351, |
|
"step": 14350 |
|
}, |
|
{ |
|
"epoch": 40.68, |
|
"learning_rate": 8.905263157894737e-06, |
|
"loss": 1.4289, |
|
"step": 14360 |
|
}, |
|
{ |
|
"epoch": 40.71, |
|
"learning_rate": 8.889473684210526e-06, |
|
"loss": 1.4306, |
|
"step": 14370 |
|
}, |
|
{ |
|
"epoch": 40.73, |
|
"learning_rate": 8.873684210526316e-06, |
|
"loss": 1.4338, |
|
"step": 14380 |
|
}, |
|
{ |
|
"epoch": 40.76, |
|
"learning_rate": 8.857894736842105e-06, |
|
"loss": 1.4284, |
|
"step": 14390 |
|
}, |
|
{ |
|
"epoch": 40.79, |
|
"learning_rate": 8.842105263157893e-06, |
|
"loss": 1.4361, |
|
"step": 14400 |
|
}, |
|
{ |
|
"epoch": 40.82, |
|
"learning_rate": 8.826315789473684e-06, |
|
"loss": 1.4295, |
|
"step": 14410 |
|
}, |
|
{ |
|
"epoch": 40.85, |
|
"learning_rate": 8.810526315789474e-06, |
|
"loss": 1.4303, |
|
"step": 14420 |
|
}, |
|
{ |
|
"epoch": 40.88, |
|
"learning_rate": 8.794736842105264e-06, |
|
"loss": 1.4279, |
|
"step": 14430 |
|
}, |
|
{ |
|
"epoch": 40.9, |
|
"learning_rate": 8.778947368421053e-06, |
|
"loss": 1.4304, |
|
"step": 14440 |
|
}, |
|
{ |
|
"epoch": 40.93, |
|
"learning_rate": 8.763157894736843e-06, |
|
"loss": 1.4327, |
|
"step": 14450 |
|
}, |
|
{ |
|
"epoch": 40.96, |
|
"learning_rate": 8.747368421052632e-06, |
|
"loss": 1.4279, |
|
"step": 14460 |
|
}, |
|
{ |
|
"epoch": 40.99, |
|
"learning_rate": 8.731578947368422e-06, |
|
"loss": 1.4316, |
|
"step": 14470 |
|
}, |
|
{ |
|
"epoch": 41.02, |
|
"learning_rate": 8.715789473684211e-06, |
|
"loss": 1.5542, |
|
"step": 14480 |
|
}, |
|
{ |
|
"epoch": 41.05, |
|
"learning_rate": 8.7e-06, |
|
"loss": 1.4297, |
|
"step": 14490 |
|
}, |
|
{ |
|
"epoch": 41.08, |
|
"learning_rate": 8.68421052631579e-06, |
|
"loss": 1.4296, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 41.1, |
|
"learning_rate": 8.668421052631579e-06, |
|
"loss": 1.4287, |
|
"step": 14510 |
|
}, |
|
{ |
|
"epoch": 41.13, |
|
"learning_rate": 8.652631578947369e-06, |
|
"loss": 1.4305, |
|
"step": 14520 |
|
}, |
|
{ |
|
"epoch": 41.16, |
|
"learning_rate": 8.636842105263158e-06, |
|
"loss": 1.4293, |
|
"step": 14530 |
|
}, |
|
{ |
|
"epoch": 41.19, |
|
"learning_rate": 8.621052631578948e-06, |
|
"loss": 1.4329, |
|
"step": 14540 |
|
}, |
|
{ |
|
"epoch": 41.22, |
|
"learning_rate": 8.605263157894737e-06, |
|
"loss": 1.4309, |
|
"step": 14550 |
|
}, |
|
{ |
|
"epoch": 41.25, |
|
"learning_rate": 8.589473684210525e-06, |
|
"loss": 1.4331, |
|
"step": 14560 |
|
}, |
|
{ |
|
"epoch": 41.27, |
|
"learning_rate": 8.573684210526316e-06, |
|
"loss": 1.4307, |
|
"step": 14570 |
|
}, |
|
{ |
|
"epoch": 41.3, |
|
"learning_rate": 8.557894736842104e-06, |
|
"loss": 1.432, |
|
"step": 14580 |
|
}, |
|
{ |
|
"epoch": 41.33, |
|
"learning_rate": 8.542105263157896e-06, |
|
"loss": 1.4288, |
|
"step": 14590 |
|
}, |
|
{ |
|
"epoch": 41.36, |
|
"learning_rate": 8.526315789473685e-06, |
|
"loss": 1.4297, |
|
"step": 14600 |
|
}, |
|
{ |
|
"epoch": 41.39, |
|
"learning_rate": 8.510526315789475e-06, |
|
"loss": 1.433, |
|
"step": 14610 |
|
}, |
|
{ |
|
"epoch": 41.42, |
|
"learning_rate": 8.494736842105264e-06, |
|
"loss": 1.4293, |
|
"step": 14620 |
|
}, |
|
{ |
|
"epoch": 41.44, |
|
"learning_rate": 8.478947368421053e-06, |
|
"loss": 1.4293, |
|
"step": 14630 |
|
}, |
|
{ |
|
"epoch": 41.47, |
|
"learning_rate": 8.463157894736843e-06, |
|
"loss": 1.435, |
|
"step": 14640 |
|
}, |
|
{ |
|
"epoch": 41.5, |
|
"learning_rate": 8.447368421052632e-06, |
|
"loss": 1.4266, |
|
"step": 14650 |
|
}, |
|
{ |
|
"epoch": 41.53, |
|
"learning_rate": 8.431578947368422e-06, |
|
"loss": 1.4298, |
|
"step": 14660 |
|
}, |
|
{ |
|
"epoch": 41.56, |
|
"learning_rate": 8.41578947368421e-06, |
|
"loss": 1.4384, |
|
"step": 14670 |
|
}, |
|
{ |
|
"epoch": 41.58, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 1.4358, |
|
"step": 14680 |
|
}, |
|
{ |
|
"epoch": 41.61, |
|
"learning_rate": 8.38421052631579e-06, |
|
"loss": 1.43, |
|
"step": 14690 |
|
}, |
|
{ |
|
"epoch": 41.64, |
|
"learning_rate": 8.368421052631578e-06, |
|
"loss": 1.4327, |
|
"step": 14700 |
|
}, |
|
{ |
|
"epoch": 41.67, |
|
"learning_rate": 8.352631578947369e-06, |
|
"loss": 1.4299, |
|
"step": 14710 |
|
}, |
|
{ |
|
"epoch": 41.7, |
|
"learning_rate": 8.336842105263157e-06, |
|
"loss": 1.4313, |
|
"step": 14720 |
|
}, |
|
{ |
|
"epoch": 41.73, |
|
"learning_rate": 8.321052631578947e-06, |
|
"loss": 1.4278, |
|
"step": 14730 |
|
}, |
|
{ |
|
"epoch": 41.75, |
|
"learning_rate": 8.305263157894736e-06, |
|
"loss": 1.428, |
|
"step": 14740 |
|
}, |
|
{ |
|
"epoch": 41.78, |
|
"learning_rate": 8.289473684210526e-06, |
|
"loss": 1.429, |
|
"step": 14750 |
|
}, |
|
{ |
|
"epoch": 41.81, |
|
"learning_rate": 8.273684210526317e-06, |
|
"loss": 1.4287, |
|
"step": 14760 |
|
}, |
|
{ |
|
"epoch": 41.84, |
|
"learning_rate": 8.257894736842105e-06, |
|
"loss": 1.43, |
|
"step": 14770 |
|
}, |
|
{ |
|
"epoch": 41.87, |
|
"learning_rate": 8.242105263157896e-06, |
|
"loss": 1.4293, |
|
"step": 14780 |
|
}, |
|
{ |
|
"epoch": 41.9, |
|
"learning_rate": 8.226315789473684e-06, |
|
"loss": 1.4286, |
|
"step": 14790 |
|
}, |
|
{ |
|
"epoch": 41.92, |
|
"learning_rate": 8.210526315789475e-06, |
|
"loss": 1.434, |
|
"step": 14800 |
|
}, |
|
{ |
|
"epoch": 41.95, |
|
"learning_rate": 8.194736842105263e-06, |
|
"loss": 1.4282, |
|
"step": 14810 |
|
}, |
|
{ |
|
"epoch": 41.98, |
|
"learning_rate": 8.178947368421054e-06, |
|
"loss": 1.4293, |
|
"step": 14820 |
|
}, |
|
{ |
|
"epoch": 42.01, |
|
"learning_rate": 8.163157894736842e-06, |
|
"loss": 1.555, |
|
"step": 14830 |
|
}, |
|
{ |
|
"epoch": 42.04, |
|
"learning_rate": 8.147368421052631e-06, |
|
"loss": 1.4281, |
|
"step": 14840 |
|
}, |
|
{ |
|
"epoch": 42.07, |
|
"learning_rate": 8.131578947368421e-06, |
|
"loss": 1.4272, |
|
"step": 14850 |
|
}, |
|
{ |
|
"epoch": 42.1, |
|
"learning_rate": 8.11578947368421e-06, |
|
"loss": 1.4316, |
|
"step": 14860 |
|
}, |
|
{ |
|
"epoch": 42.12, |
|
"learning_rate": 8.1e-06, |
|
"loss": 1.4275, |
|
"step": 14870 |
|
}, |
|
{ |
|
"epoch": 42.15, |
|
"learning_rate": 8.084210526315789e-06, |
|
"loss": 1.4261, |
|
"step": 14880 |
|
}, |
|
{ |
|
"epoch": 42.18, |
|
"learning_rate": 8.06842105263158e-06, |
|
"loss": 1.4293, |
|
"step": 14890 |
|
}, |
|
{ |
|
"epoch": 42.21, |
|
"learning_rate": 8.052631578947368e-06, |
|
"loss": 1.4279, |
|
"step": 14900 |
|
}, |
|
{ |
|
"epoch": 42.24, |
|
"learning_rate": 8.036842105263157e-06, |
|
"loss": 1.4273, |
|
"step": 14910 |
|
}, |
|
{ |
|
"epoch": 42.27, |
|
"learning_rate": 8.021052631578949e-06, |
|
"loss": 1.43, |
|
"step": 14920 |
|
}, |
|
{ |
|
"epoch": 42.29, |
|
"learning_rate": 8.005263157894737e-06, |
|
"loss": 1.4286, |
|
"step": 14930 |
|
}, |
|
{ |
|
"epoch": 42.32, |
|
"learning_rate": 7.989473684210528e-06, |
|
"loss": 1.4282, |
|
"step": 14940 |
|
}, |
|
{ |
|
"epoch": 42.35, |
|
"learning_rate": 7.973684210526316e-06, |
|
"loss": 1.4303, |
|
"step": 14950 |
|
}, |
|
{ |
|
"epoch": 42.38, |
|
"learning_rate": 7.957894736842107e-06, |
|
"loss": 1.4268, |
|
"step": 14960 |
|
}, |
|
{ |
|
"epoch": 42.41, |
|
"learning_rate": 7.942105263157895e-06, |
|
"loss": 1.426, |
|
"step": 14970 |
|
}, |
|
{ |
|
"epoch": 42.44, |
|
"learning_rate": 7.926315789473684e-06, |
|
"loss": 1.4271, |
|
"step": 14980 |
|
}, |
|
{ |
|
"epoch": 42.46, |
|
"learning_rate": 7.910526315789474e-06, |
|
"loss": 1.4281, |
|
"step": 14990 |
|
}, |
|
{ |
|
"epoch": 42.49, |
|
"learning_rate": 7.894736842105263e-06, |
|
"loss": 1.4318, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 42.49, |
|
"eval_denotation_accuracy": 0.5708230307311903, |
|
"eval_loss": 2.1977450847625732, |
|
"eval_runtime": 342.8816, |
|
"eval_samples_per_second": 8.256, |
|
"eval_steps_per_second": 2.065, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 42.52, |
|
"learning_rate": 7.878947368421053e-06, |
|
"loss": 1.4315, |
|
"step": 15010 |
|
}, |
|
{ |
|
"epoch": 42.55, |
|
"learning_rate": 7.863157894736842e-06, |
|
"loss": 1.4316, |
|
"step": 15020 |
|
}, |
|
{ |
|
"epoch": 42.58, |
|
"learning_rate": 7.847368421052632e-06, |
|
"loss": 1.4299, |
|
"step": 15030 |
|
}, |
|
{ |
|
"epoch": 42.6, |
|
"learning_rate": 7.831578947368421e-06, |
|
"loss": 1.4309, |
|
"step": 15040 |
|
}, |
|
{ |
|
"epoch": 42.63, |
|
"learning_rate": 7.81578947368421e-06, |
|
"loss": 1.4283, |
|
"step": 15050 |
|
}, |
|
{ |
|
"epoch": 42.66, |
|
"learning_rate": 7.8e-06, |
|
"loss": 1.4293, |
|
"step": 15060 |
|
}, |
|
{ |
|
"epoch": 42.69, |
|
"learning_rate": 7.784210526315789e-06, |
|
"loss": 1.4278, |
|
"step": 15070 |
|
}, |
|
{ |
|
"epoch": 42.72, |
|
"learning_rate": 7.768421052631579e-06, |
|
"loss": 1.4293, |
|
"step": 15080 |
|
}, |
|
{ |
|
"epoch": 42.75, |
|
"learning_rate": 7.75263157894737e-06, |
|
"loss": 1.4329, |
|
"step": 15090 |
|
}, |
|
{ |
|
"epoch": 42.77, |
|
"learning_rate": 7.73684210526316e-06, |
|
"loss": 1.4268, |
|
"step": 15100 |
|
}, |
|
{ |
|
"epoch": 42.8, |
|
"learning_rate": 7.721052631578948e-06, |
|
"loss": 1.427, |
|
"step": 15110 |
|
}, |
|
{ |
|
"epoch": 42.83, |
|
"learning_rate": 7.705263157894737e-06, |
|
"loss": 1.4275, |
|
"step": 15120 |
|
}, |
|
{ |
|
"epoch": 42.86, |
|
"learning_rate": 7.689473684210527e-06, |
|
"loss": 1.4275, |
|
"step": 15130 |
|
}, |
|
{ |
|
"epoch": 42.89, |
|
"learning_rate": 7.673684210526316e-06, |
|
"loss": 1.4309, |
|
"step": 15140 |
|
}, |
|
{ |
|
"epoch": 42.92, |
|
"learning_rate": 7.657894736842106e-06, |
|
"loss": 1.4273, |
|
"step": 15150 |
|
}, |
|
{ |
|
"epoch": 42.94, |
|
"learning_rate": 7.642105263157895e-06, |
|
"loss": 1.4312, |
|
"step": 15160 |
|
}, |
|
{ |
|
"epoch": 42.97, |
|
"learning_rate": 7.626315789473685e-06, |
|
"loss": 1.4264, |
|
"step": 15170 |
|
}, |
|
{ |
|
"epoch": 43.0, |
|
"learning_rate": 7.610526315789474e-06, |
|
"loss": 1.5505, |
|
"step": 15180 |
|
}, |
|
{ |
|
"epoch": 43.03, |
|
"learning_rate": 7.594736842105262e-06, |
|
"loss": 1.4282, |
|
"step": 15190 |
|
}, |
|
{ |
|
"epoch": 43.06, |
|
"learning_rate": 7.578947368421053e-06, |
|
"loss": 1.427, |
|
"step": 15200 |
|
}, |
|
{ |
|
"epoch": 43.09, |
|
"learning_rate": 7.563157894736842e-06, |
|
"loss": 1.4284, |
|
"step": 15210 |
|
}, |
|
{ |
|
"epoch": 43.12, |
|
"learning_rate": 7.5473684210526326e-06, |
|
"loss": 1.4278, |
|
"step": 15220 |
|
}, |
|
{ |
|
"epoch": 43.14, |
|
"learning_rate": 7.531578947368421e-06, |
|
"loss": 1.4264, |
|
"step": 15230 |
|
}, |
|
{ |
|
"epoch": 43.17, |
|
"learning_rate": 7.5157894736842115e-06, |
|
"loss": 1.4283, |
|
"step": 15240 |
|
}, |
|
{ |
|
"epoch": 43.2, |
|
"learning_rate": 7.5e-06, |
|
"loss": 1.4262, |
|
"step": 15250 |
|
}, |
|
{ |
|
"epoch": 43.23, |
|
"learning_rate": 7.48421052631579e-06, |
|
"loss": 1.4298, |
|
"step": 15260 |
|
}, |
|
{ |
|
"epoch": 43.26, |
|
"learning_rate": 7.468421052631579e-06, |
|
"loss": 1.4263, |
|
"step": 15270 |
|
}, |
|
{ |
|
"epoch": 43.29, |
|
"learning_rate": 7.452631578947369e-06, |
|
"loss": 1.4272, |
|
"step": 15280 |
|
}, |
|
{ |
|
"epoch": 43.31, |
|
"learning_rate": 7.436842105263157e-06, |
|
"loss": 1.4274, |
|
"step": 15290 |
|
}, |
|
{ |
|
"epoch": 43.34, |
|
"learning_rate": 7.421052631578948e-06, |
|
"loss": 1.429, |
|
"step": 15300 |
|
}, |
|
{ |
|
"epoch": 43.37, |
|
"learning_rate": 7.405263157894737e-06, |
|
"loss": 1.4264, |
|
"step": 15310 |
|
}, |
|
{ |
|
"epoch": 43.4, |
|
"learning_rate": 7.389473684210527e-06, |
|
"loss": 1.4284, |
|
"step": 15320 |
|
}, |
|
{ |
|
"epoch": 43.43, |
|
"learning_rate": 7.373684210526316e-06, |
|
"loss": 1.4274, |
|
"step": 15330 |
|
}, |
|
{ |
|
"epoch": 43.45, |
|
"learning_rate": 7.357894736842106e-06, |
|
"loss": 1.4287, |
|
"step": 15340 |
|
}, |
|
{ |
|
"epoch": 43.48, |
|
"learning_rate": 7.342105263157895e-06, |
|
"loss": 1.4279, |
|
"step": 15350 |
|
}, |
|
{ |
|
"epoch": 43.51, |
|
"learning_rate": 7.326315789473684e-06, |
|
"loss": 1.4313, |
|
"step": 15360 |
|
}, |
|
{ |
|
"epoch": 43.54, |
|
"learning_rate": 7.310526315789473e-06, |
|
"loss": 1.4283, |
|
"step": 15370 |
|
}, |
|
{ |
|
"epoch": 43.57, |
|
"learning_rate": 7.2947368421052636e-06, |
|
"loss": 1.4261, |
|
"step": 15380 |
|
}, |
|
{ |
|
"epoch": 43.6, |
|
"learning_rate": 7.278947368421053e-06, |
|
"loss": 1.4273, |
|
"step": 15390 |
|
}, |
|
{ |
|
"epoch": 43.62, |
|
"learning_rate": 7.2631578947368426e-06, |
|
"loss": 1.4275, |
|
"step": 15400 |
|
}, |
|
{ |
|
"epoch": 43.65, |
|
"learning_rate": 7.247368421052632e-06, |
|
"loss": 1.4278, |
|
"step": 15410 |
|
}, |
|
{ |
|
"epoch": 43.68, |
|
"learning_rate": 7.2315789473684215e-06, |
|
"loss": 1.4265, |
|
"step": 15420 |
|
}, |
|
{ |
|
"epoch": 43.71, |
|
"learning_rate": 7.21578947368421e-06, |
|
"loss": 1.4297, |
|
"step": 15430 |
|
}, |
|
{ |
|
"epoch": 43.74, |
|
"learning_rate": 7.2e-06, |
|
"loss": 1.4275, |
|
"step": 15440 |
|
}, |
|
{ |
|
"epoch": 43.77, |
|
"learning_rate": 7.184210526315789e-06, |
|
"loss": 1.4284, |
|
"step": 15450 |
|
}, |
|
{ |
|
"epoch": 43.79, |
|
"learning_rate": 7.1684210526315795e-06, |
|
"loss": 1.4289, |
|
"step": 15460 |
|
}, |
|
{ |
|
"epoch": 43.82, |
|
"learning_rate": 7.152631578947369e-06, |
|
"loss": 1.4292, |
|
"step": 15470 |
|
}, |
|
{ |
|
"epoch": 43.85, |
|
"learning_rate": 7.1368421052631585e-06, |
|
"loss": 1.4262, |
|
"step": 15480 |
|
}, |
|
{ |
|
"epoch": 43.88, |
|
"learning_rate": 7.121052631578948e-06, |
|
"loss": 1.428, |
|
"step": 15490 |
|
}, |
|
{ |
|
"epoch": 43.91, |
|
"learning_rate": 7.105263157894737e-06, |
|
"loss": 1.4269, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 43.94, |
|
"learning_rate": 7.089473684210526e-06, |
|
"loss": 1.4283, |
|
"step": 15510 |
|
}, |
|
{ |
|
"epoch": 43.96, |
|
"learning_rate": 7.073684210526316e-06, |
|
"loss": 1.4288, |
|
"step": 15520 |
|
}, |
|
{ |
|
"epoch": 43.99, |
|
"learning_rate": 7.057894736842105e-06, |
|
"loss": 1.4249, |
|
"step": 15530 |
|
}, |
|
{ |
|
"epoch": 44.02, |
|
"learning_rate": 7.042105263157895e-06, |
|
"loss": 1.5519, |
|
"step": 15540 |
|
}, |
|
{ |
|
"epoch": 44.05, |
|
"learning_rate": 7.026315789473685e-06, |
|
"loss": 1.4271, |
|
"step": 15550 |
|
}, |
|
{ |
|
"epoch": 44.08, |
|
"learning_rate": 7.010526315789474e-06, |
|
"loss": 1.4279, |
|
"step": 15560 |
|
}, |
|
{ |
|
"epoch": 44.11, |
|
"learning_rate": 6.994736842105263e-06, |
|
"loss": 1.4269, |
|
"step": 15570 |
|
}, |
|
{ |
|
"epoch": 44.14, |
|
"learning_rate": 6.9789473684210525e-06, |
|
"loss": 1.4299, |
|
"step": 15580 |
|
}, |
|
{ |
|
"epoch": 44.16, |
|
"learning_rate": 6.963157894736842e-06, |
|
"loss": 1.4262, |
|
"step": 15590 |
|
}, |
|
{ |
|
"epoch": 44.19, |
|
"learning_rate": 6.9473684210526315e-06, |
|
"loss": 1.4258, |
|
"step": 15600 |
|
}, |
|
{ |
|
"epoch": 44.22, |
|
"learning_rate": 6.931578947368421e-06, |
|
"loss": 1.4259, |
|
"step": 15610 |
|
}, |
|
{ |
|
"epoch": 44.25, |
|
"learning_rate": 6.9157894736842105e-06, |
|
"loss": 1.4278, |
|
"step": 15620 |
|
}, |
|
{ |
|
"epoch": 44.28, |
|
"learning_rate": 6.900000000000001e-06, |
|
"loss": 1.4249, |
|
"step": 15630 |
|
}, |
|
{ |
|
"epoch": 44.31, |
|
"learning_rate": 6.8842105263157895e-06, |
|
"loss": 1.4281, |
|
"step": 15640 |
|
}, |
|
{ |
|
"epoch": 44.33, |
|
"learning_rate": 6.868421052631579e-06, |
|
"loss": 1.4258, |
|
"step": 15650 |
|
}, |
|
{ |
|
"epoch": 44.36, |
|
"learning_rate": 6.8526315789473685e-06, |
|
"loss": 1.4272, |
|
"step": 15660 |
|
}, |
|
{ |
|
"epoch": 44.39, |
|
"learning_rate": 6.836842105263158e-06, |
|
"loss": 1.4268, |
|
"step": 15670 |
|
}, |
|
{ |
|
"epoch": 44.42, |
|
"learning_rate": 6.8210526315789475e-06, |
|
"loss": 1.4277, |
|
"step": 15680 |
|
}, |
|
{ |
|
"epoch": 44.45, |
|
"learning_rate": 6.805263157894737e-06, |
|
"loss": 1.4272, |
|
"step": 15690 |
|
}, |
|
{ |
|
"epoch": 44.47, |
|
"learning_rate": 6.7894736842105264e-06, |
|
"loss": 1.4277, |
|
"step": 15700 |
|
}, |
|
{ |
|
"epoch": 44.5, |
|
"learning_rate": 6.773684210526317e-06, |
|
"loss": 1.4272, |
|
"step": 15710 |
|
}, |
|
{ |
|
"epoch": 44.53, |
|
"learning_rate": 6.7578947368421054e-06, |
|
"loss": 1.4265, |
|
"step": 15720 |
|
}, |
|
{ |
|
"epoch": 44.56, |
|
"learning_rate": 6.742105263157895e-06, |
|
"loss": 1.4262, |
|
"step": 15730 |
|
}, |
|
{ |
|
"epoch": 44.59, |
|
"learning_rate": 6.726315789473684e-06, |
|
"loss": 1.4259, |
|
"step": 15740 |
|
}, |
|
{ |
|
"epoch": 44.62, |
|
"learning_rate": 6.710526315789474e-06, |
|
"loss": 1.427, |
|
"step": 15750 |
|
}, |
|
{ |
|
"epoch": 44.64, |
|
"learning_rate": 6.694736842105263e-06, |
|
"loss": 1.4266, |
|
"step": 15760 |
|
}, |
|
{ |
|
"epoch": 44.67, |
|
"learning_rate": 6.678947368421053e-06, |
|
"loss": 1.4265, |
|
"step": 15770 |
|
}, |
|
{ |
|
"epoch": 44.7, |
|
"learning_rate": 6.663157894736842e-06, |
|
"loss": 1.4267, |
|
"step": 15780 |
|
}, |
|
{ |
|
"epoch": 44.73, |
|
"learning_rate": 6.647368421052631e-06, |
|
"loss": 1.4263, |
|
"step": 15790 |
|
}, |
|
{ |
|
"epoch": 44.76, |
|
"learning_rate": 6.631578947368421e-06, |
|
"loss": 1.4324, |
|
"step": 15800 |
|
}, |
|
{ |
|
"epoch": 44.79, |
|
"learning_rate": 6.615789473684211e-06, |
|
"loss": 1.4265, |
|
"step": 15810 |
|
}, |
|
{ |
|
"epoch": 44.81, |
|
"learning_rate": 6.6e-06, |
|
"loss": 1.428, |
|
"step": 15820 |
|
}, |
|
{ |
|
"epoch": 44.84, |
|
"learning_rate": 6.58421052631579e-06, |
|
"loss": 1.4252, |
|
"step": 15830 |
|
}, |
|
{ |
|
"epoch": 44.87, |
|
"learning_rate": 6.568421052631579e-06, |
|
"loss": 1.4256, |
|
"step": 15840 |
|
}, |
|
{ |
|
"epoch": 44.9, |
|
"learning_rate": 6.552631578947369e-06, |
|
"loss": 1.4266, |
|
"step": 15850 |
|
}, |
|
{ |
|
"epoch": 44.93, |
|
"learning_rate": 6.5368421052631575e-06, |
|
"loss": 1.4275, |
|
"step": 15860 |
|
}, |
|
{ |
|
"epoch": 44.96, |
|
"learning_rate": 6.521052631578947e-06, |
|
"loss": 1.4265, |
|
"step": 15870 |
|
}, |
|
{ |
|
"epoch": 44.98, |
|
"learning_rate": 6.505263157894737e-06, |
|
"loss": 1.4257, |
|
"step": 15880 |
|
}, |
|
{ |
|
"epoch": 45.01, |
|
"learning_rate": 6.489473684210527e-06, |
|
"loss": 1.5515, |
|
"step": 15890 |
|
}, |
|
{ |
|
"epoch": 45.04, |
|
"learning_rate": 6.473684210526316e-06, |
|
"loss": 1.4271, |
|
"step": 15900 |
|
}, |
|
{ |
|
"epoch": 45.07, |
|
"learning_rate": 6.457894736842106e-06, |
|
"loss": 1.4269, |
|
"step": 15910 |
|
}, |
|
{ |
|
"epoch": 45.1, |
|
"learning_rate": 6.442105263157895e-06, |
|
"loss": 1.4258, |
|
"step": 15920 |
|
}, |
|
{ |
|
"epoch": 45.13, |
|
"learning_rate": 6.426315789473684e-06, |
|
"loss": 1.4271, |
|
"step": 15930 |
|
}, |
|
{ |
|
"epoch": 45.16, |
|
"learning_rate": 6.410526315789473e-06, |
|
"loss": 1.4269, |
|
"step": 15940 |
|
}, |
|
{ |
|
"epoch": 45.18, |
|
"learning_rate": 6.394736842105263e-06, |
|
"loss": 1.4265, |
|
"step": 15950 |
|
}, |
|
{ |
|
"epoch": 45.21, |
|
"learning_rate": 6.378947368421053e-06, |
|
"loss": 1.4269, |
|
"step": 15960 |
|
}, |
|
{ |
|
"epoch": 45.24, |
|
"learning_rate": 6.363157894736843e-06, |
|
"loss": 1.4267, |
|
"step": 15970 |
|
}, |
|
{ |
|
"epoch": 45.27, |
|
"learning_rate": 6.347368421052632e-06, |
|
"loss": 1.4318, |
|
"step": 15980 |
|
}, |
|
{ |
|
"epoch": 45.3, |
|
"learning_rate": 6.331578947368422e-06, |
|
"loss": 1.4256, |
|
"step": 15990 |
|
}, |
|
{ |
|
"epoch": 45.32, |
|
"learning_rate": 6.31578947368421e-06, |
|
"loss": 1.4257, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 45.32, |
|
"eval_denotation_accuracy": 0.572942423172024, |
|
"eval_loss": 2.181939125061035, |
|
"eval_runtime": 339.4655, |
|
"eval_samples_per_second": 8.34, |
|
"eval_steps_per_second": 2.086, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 45.35, |
|
"learning_rate": 6.3e-06, |
|
"loss": 1.429, |
|
"step": 16010 |
|
}, |
|
{ |
|
"epoch": 45.38, |
|
"learning_rate": 6.284210526315789e-06, |
|
"loss": 1.4262, |
|
"step": 16020 |
|
}, |
|
{ |
|
"epoch": 45.41, |
|
"learning_rate": 6.268421052631579e-06, |
|
"loss": 1.4312, |
|
"step": 16030 |
|
}, |
|
{ |
|
"epoch": 45.44, |
|
"learning_rate": 6.252631578947368e-06, |
|
"loss": 1.4271, |
|
"step": 16040 |
|
}, |
|
{ |
|
"epoch": 45.47, |
|
"learning_rate": 6.236842105263159e-06, |
|
"loss": 1.427, |
|
"step": 16050 |
|
}, |
|
{ |
|
"epoch": 45.49, |
|
"learning_rate": 6.221052631578948e-06, |
|
"loss": 1.4278, |
|
"step": 16060 |
|
}, |
|
{ |
|
"epoch": 45.52, |
|
"learning_rate": 6.205263157894737e-06, |
|
"loss": 1.4258, |
|
"step": 16070 |
|
}, |
|
{ |
|
"epoch": 45.55, |
|
"learning_rate": 6.189473684210526e-06, |
|
"loss": 1.4253, |
|
"step": 16080 |
|
}, |
|
{ |
|
"epoch": 45.58, |
|
"learning_rate": 6.173684210526316e-06, |
|
"loss": 1.4278, |
|
"step": 16090 |
|
}, |
|
{ |
|
"epoch": 45.61, |
|
"learning_rate": 6.157894736842105e-06, |
|
"loss": 1.4276, |
|
"step": 16100 |
|
}, |
|
{ |
|
"epoch": 45.64, |
|
"learning_rate": 6.142105263157895e-06, |
|
"loss": 1.4278, |
|
"step": 16110 |
|
}, |
|
{ |
|
"epoch": 45.66, |
|
"learning_rate": 6.126315789473684e-06, |
|
"loss": 1.4258, |
|
"step": 16120 |
|
}, |
|
{ |
|
"epoch": 45.69, |
|
"learning_rate": 6.1105263157894746e-06, |
|
"loss": 1.4263, |
|
"step": 16130 |
|
}, |
|
{ |
|
"epoch": 45.72, |
|
"learning_rate": 6.094736842105263e-06, |
|
"loss": 1.4297, |
|
"step": 16140 |
|
}, |
|
{ |
|
"epoch": 45.75, |
|
"learning_rate": 6.078947368421053e-06, |
|
"loss": 1.4296, |
|
"step": 16150 |
|
}, |
|
{ |
|
"epoch": 45.78, |
|
"learning_rate": 6.063157894736842e-06, |
|
"loss": 1.4314, |
|
"step": 16160 |
|
}, |
|
{ |
|
"epoch": 45.81, |
|
"learning_rate": 6.047368421052632e-06, |
|
"loss": 1.4254, |
|
"step": 16170 |
|
}, |
|
{ |
|
"epoch": 45.83, |
|
"learning_rate": 6.031578947368421e-06, |
|
"loss": 1.4265, |
|
"step": 16180 |
|
}, |
|
{ |
|
"epoch": 45.86, |
|
"learning_rate": 6.015789473684211e-06, |
|
"loss": 1.4254, |
|
"step": 16190 |
|
}, |
|
{ |
|
"epoch": 45.89, |
|
"learning_rate": 6e-06, |
|
"loss": 1.4242, |
|
"step": 16200 |
|
}, |
|
{ |
|
"epoch": 45.92, |
|
"learning_rate": 5.98421052631579e-06, |
|
"loss": 1.4248, |
|
"step": 16210 |
|
}, |
|
{ |
|
"epoch": 45.95, |
|
"learning_rate": 5.968421052631579e-06, |
|
"loss": 1.4274, |
|
"step": 16220 |
|
}, |
|
{ |
|
"epoch": 45.97, |
|
"learning_rate": 5.952631578947369e-06, |
|
"loss": 1.428, |
|
"step": 16230 |
|
}, |
|
{ |
|
"epoch": 46.01, |
|
"learning_rate": 5.936842105263158e-06, |
|
"loss": 1.5503, |
|
"step": 16240 |
|
}, |
|
{ |
|
"epoch": 46.03, |
|
"learning_rate": 5.921052631578948e-06, |
|
"loss": 1.4249, |
|
"step": 16250 |
|
}, |
|
{ |
|
"epoch": 46.06, |
|
"learning_rate": 5.905263157894737e-06, |
|
"loss": 1.4247, |
|
"step": 16260 |
|
}, |
|
{ |
|
"epoch": 46.09, |
|
"learning_rate": 5.889473684210527e-06, |
|
"loss": 1.4243, |
|
"step": 16270 |
|
}, |
|
{ |
|
"epoch": 46.12, |
|
"learning_rate": 5.873684210526315e-06, |
|
"loss": 1.4257, |
|
"step": 16280 |
|
}, |
|
{ |
|
"epoch": 46.15, |
|
"learning_rate": 5.857894736842105e-06, |
|
"loss": 1.4288, |
|
"step": 16290 |
|
}, |
|
{ |
|
"epoch": 46.18, |
|
"learning_rate": 5.842105263157895e-06, |
|
"loss": 1.4253, |
|
"step": 16300 |
|
}, |
|
{ |
|
"epoch": 46.2, |
|
"learning_rate": 5.8263157894736846e-06, |
|
"loss": 1.4309, |
|
"step": 16310 |
|
}, |
|
{ |
|
"epoch": 46.23, |
|
"learning_rate": 5.810526315789474e-06, |
|
"loss": 1.4274, |
|
"step": 16320 |
|
}, |
|
{ |
|
"epoch": 46.26, |
|
"learning_rate": 5.7947368421052635e-06, |
|
"loss": 1.4261, |
|
"step": 16330 |
|
}, |
|
{ |
|
"epoch": 46.29, |
|
"learning_rate": 5.778947368421053e-06, |
|
"loss": 1.4261, |
|
"step": 16340 |
|
}, |
|
{ |
|
"epoch": 46.32, |
|
"learning_rate": 5.763157894736842e-06, |
|
"loss": 1.4272, |
|
"step": 16350 |
|
}, |
|
{ |
|
"epoch": 46.34, |
|
"learning_rate": 5.747368421052631e-06, |
|
"loss": 1.4262, |
|
"step": 16360 |
|
}, |
|
{ |
|
"epoch": 46.37, |
|
"learning_rate": 5.731578947368421e-06, |
|
"loss": 1.4265, |
|
"step": 16370 |
|
}, |
|
{ |
|
"epoch": 46.4, |
|
"learning_rate": 5.715789473684211e-06, |
|
"loss": 1.4247, |
|
"step": 16380 |
|
}, |
|
{ |
|
"epoch": 46.43, |
|
"learning_rate": 5.7000000000000005e-06, |
|
"loss": 1.4248, |
|
"step": 16390 |
|
}, |
|
{ |
|
"epoch": 46.46, |
|
"learning_rate": 5.68421052631579e-06, |
|
"loss": 1.4259, |
|
"step": 16400 |
|
}, |
|
{ |
|
"epoch": 46.49, |
|
"learning_rate": 5.6684210526315795e-06, |
|
"loss": 1.4254, |
|
"step": 16410 |
|
}, |
|
{ |
|
"epoch": 46.51, |
|
"learning_rate": 5.652631578947368e-06, |
|
"loss": 1.4268, |
|
"step": 16420 |
|
}, |
|
{ |
|
"epoch": 46.54, |
|
"learning_rate": 5.636842105263158e-06, |
|
"loss": 1.427, |
|
"step": 16430 |
|
}, |
|
{ |
|
"epoch": 46.57, |
|
"learning_rate": 5.621052631578947e-06, |
|
"loss": 1.4264, |
|
"step": 16440 |
|
}, |
|
{ |
|
"epoch": 46.6, |
|
"learning_rate": 5.605263157894737e-06, |
|
"loss": 1.4265, |
|
"step": 16450 |
|
}, |
|
{ |
|
"epoch": 46.63, |
|
"learning_rate": 5.589473684210527e-06, |
|
"loss": 1.4251, |
|
"step": 16460 |
|
}, |
|
{ |
|
"epoch": 46.66, |
|
"learning_rate": 5.573684210526316e-06, |
|
"loss": 1.4258, |
|
"step": 16470 |
|
}, |
|
{ |
|
"epoch": 46.68, |
|
"learning_rate": 5.557894736842106e-06, |
|
"loss": 1.4255, |
|
"step": 16480 |
|
}, |
|
{ |
|
"epoch": 46.71, |
|
"learning_rate": 5.542105263157895e-06, |
|
"loss": 1.4271, |
|
"step": 16490 |
|
}, |
|
{ |
|
"epoch": 46.74, |
|
"learning_rate": 5.526315789473684e-06, |
|
"loss": 1.4261, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 46.77, |
|
"learning_rate": 5.5105263157894735e-06, |
|
"loss": 1.4262, |
|
"step": 16510 |
|
}, |
|
{ |
|
"epoch": 46.8, |
|
"learning_rate": 5.494736842105263e-06, |
|
"loss": 1.4251, |
|
"step": 16520 |
|
}, |
|
{ |
|
"epoch": 46.83, |
|
"learning_rate": 5.4789473684210525e-06, |
|
"loss": 1.4256, |
|
"step": 16530 |
|
}, |
|
{ |
|
"epoch": 46.85, |
|
"learning_rate": 5.463157894736842e-06, |
|
"loss": 1.425, |
|
"step": 16540 |
|
}, |
|
{ |
|
"epoch": 46.88, |
|
"learning_rate": 5.447368421052632e-06, |
|
"loss": 1.4269, |
|
"step": 16550 |
|
}, |
|
{ |
|
"epoch": 46.91, |
|
"learning_rate": 5.431578947368422e-06, |
|
"loss": 1.4257, |
|
"step": 16560 |
|
}, |
|
{ |
|
"epoch": 46.94, |
|
"learning_rate": 5.4157894736842105e-06, |
|
"loss": 1.4249, |
|
"step": 16570 |
|
}, |
|
{ |
|
"epoch": 46.97, |
|
"learning_rate": 5.4e-06, |
|
"loss": 1.4279, |
|
"step": 16580 |
|
}, |
|
{ |
|
"epoch": 46.99, |
|
"learning_rate": 5.3842105263157895e-06, |
|
"loss": 1.4262, |
|
"step": 16590 |
|
}, |
|
{ |
|
"epoch": 47.03, |
|
"learning_rate": 5.368421052631579e-06, |
|
"loss": 1.5532, |
|
"step": 16600 |
|
}, |
|
{ |
|
"epoch": 47.05, |
|
"learning_rate": 5.3526315789473684e-06, |
|
"loss": 1.4252, |
|
"step": 16610 |
|
}, |
|
{ |
|
"epoch": 47.08, |
|
"learning_rate": 5.336842105263158e-06, |
|
"loss": 1.4255, |
|
"step": 16620 |
|
}, |
|
{ |
|
"epoch": 47.11, |
|
"learning_rate": 5.321052631578948e-06, |
|
"loss": 1.4251, |
|
"step": 16630 |
|
}, |
|
{ |
|
"epoch": 47.14, |
|
"learning_rate": 5.305263157894737e-06, |
|
"loss": 1.4238, |
|
"step": 16640 |
|
}, |
|
{ |
|
"epoch": 47.17, |
|
"learning_rate": 5.289473684210526e-06, |
|
"loss": 1.4243, |
|
"step": 16650 |
|
}, |
|
{ |
|
"epoch": 47.19, |
|
"learning_rate": 5.273684210526316e-06, |
|
"loss": 1.4262, |
|
"step": 16660 |
|
}, |
|
{ |
|
"epoch": 47.22, |
|
"learning_rate": 5.257894736842105e-06, |
|
"loss": 1.4283, |
|
"step": 16670 |
|
}, |
|
{ |
|
"epoch": 47.25, |
|
"learning_rate": 5.242105263157895e-06, |
|
"loss": 1.425, |
|
"step": 16680 |
|
}, |
|
{ |
|
"epoch": 47.28, |
|
"learning_rate": 5.226315789473684e-06, |
|
"loss": 1.4269, |
|
"step": 16690 |
|
}, |
|
{ |
|
"epoch": 47.31, |
|
"learning_rate": 5.210526315789474e-06, |
|
"loss": 1.4249, |
|
"step": 16700 |
|
}, |
|
{ |
|
"epoch": 47.34, |
|
"learning_rate": 5.194736842105263e-06, |
|
"loss": 1.4254, |
|
"step": 16710 |
|
}, |
|
{ |
|
"epoch": 47.36, |
|
"learning_rate": 5.178947368421053e-06, |
|
"loss": 1.4247, |
|
"step": 16720 |
|
}, |
|
{ |
|
"epoch": 47.39, |
|
"learning_rate": 5.163157894736842e-06, |
|
"loss": 1.4337, |
|
"step": 16730 |
|
}, |
|
{ |
|
"epoch": 47.42, |
|
"learning_rate": 5.147368421052632e-06, |
|
"loss": 1.4268, |
|
"step": 16740 |
|
}, |
|
{ |
|
"epoch": 47.45, |
|
"learning_rate": 5.131578947368421e-06, |
|
"loss": 1.4242, |
|
"step": 16750 |
|
}, |
|
{ |
|
"epoch": 47.48, |
|
"learning_rate": 5.115789473684211e-06, |
|
"loss": 1.4241, |
|
"step": 16760 |
|
}, |
|
{ |
|
"epoch": 47.51, |
|
"learning_rate": 5.1e-06, |
|
"loss": 1.4259, |
|
"step": 16770 |
|
}, |
|
{ |
|
"epoch": 47.53, |
|
"learning_rate": 5.084210526315789e-06, |
|
"loss": 1.4265, |
|
"step": 16780 |
|
}, |
|
{ |
|
"epoch": 47.56, |
|
"learning_rate": 5.0684210526315784e-06, |
|
"loss": 1.4268, |
|
"step": 16790 |
|
}, |
|
{ |
|
"epoch": 47.59, |
|
"learning_rate": 5.052631578947369e-06, |
|
"loss": 1.4248, |
|
"step": 16800 |
|
}, |
|
{ |
|
"epoch": 47.62, |
|
"learning_rate": 5.036842105263158e-06, |
|
"loss": 1.4245, |
|
"step": 16810 |
|
}, |
|
{ |
|
"epoch": 47.65, |
|
"learning_rate": 5.021052631578948e-06, |
|
"loss": 1.4267, |
|
"step": 16820 |
|
}, |
|
{ |
|
"epoch": 47.68, |
|
"learning_rate": 5.005263157894737e-06, |
|
"loss": 1.4252, |
|
"step": 16830 |
|
}, |
|
{ |
|
"epoch": 47.7, |
|
"learning_rate": 4.989473684210527e-06, |
|
"loss": 1.4241, |
|
"step": 16840 |
|
}, |
|
{ |
|
"epoch": 47.73, |
|
"learning_rate": 4.973684210526315e-06, |
|
"loss": 1.4243, |
|
"step": 16850 |
|
}, |
|
{ |
|
"epoch": 47.76, |
|
"learning_rate": 4.957894736842105e-06, |
|
"loss": 1.4271, |
|
"step": 16860 |
|
}, |
|
{ |
|
"epoch": 47.79, |
|
"learning_rate": 4.942105263157894e-06, |
|
"loss": 1.4239, |
|
"step": 16870 |
|
}, |
|
{ |
|
"epoch": 47.82, |
|
"learning_rate": 4.926315789473685e-06, |
|
"loss": 1.4292, |
|
"step": 16880 |
|
}, |
|
{ |
|
"epoch": 47.84, |
|
"learning_rate": 4.910526315789474e-06, |
|
"loss": 1.4252, |
|
"step": 16890 |
|
}, |
|
{ |
|
"epoch": 47.87, |
|
"learning_rate": 4.894736842105264e-06, |
|
"loss": 1.4254, |
|
"step": 16900 |
|
}, |
|
{ |
|
"epoch": 47.9, |
|
"learning_rate": 4.878947368421053e-06, |
|
"loss": 1.4261, |
|
"step": 16910 |
|
}, |
|
{ |
|
"epoch": 47.93, |
|
"learning_rate": 4.863157894736842e-06, |
|
"loss": 1.4252, |
|
"step": 16920 |
|
}, |
|
{ |
|
"epoch": 47.96, |
|
"learning_rate": 4.847368421052631e-06, |
|
"loss": 1.4256, |
|
"step": 16930 |
|
}, |
|
{ |
|
"epoch": 47.99, |
|
"learning_rate": 4.831578947368421e-06, |
|
"loss": 1.4258, |
|
"step": 16940 |
|
}, |
|
{ |
|
"epoch": 48.02, |
|
"learning_rate": 4.81578947368421e-06, |
|
"loss": 1.5516, |
|
"step": 16950 |
|
}, |
|
{ |
|
"epoch": 48.05, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 1.4234, |
|
"step": 16960 |
|
}, |
|
{ |
|
"epoch": 48.07, |
|
"learning_rate": 4.78421052631579e-06, |
|
"loss": 1.4246, |
|
"step": 16970 |
|
}, |
|
{ |
|
"epoch": 48.1, |
|
"learning_rate": 4.76842105263158e-06, |
|
"loss": 1.4253, |
|
"step": 16980 |
|
}, |
|
{ |
|
"epoch": 48.13, |
|
"learning_rate": 4.752631578947368e-06, |
|
"loss": 1.4241, |
|
"step": 16990 |
|
}, |
|
{ |
|
"epoch": 48.16, |
|
"learning_rate": 4.736842105263158e-06, |
|
"loss": 1.4254, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 48.16, |
|
"eval_denotation_accuracy": 0.5757682797598022, |
|
"eval_loss": 2.200056314468384, |
|
"eval_runtime": 339.9367, |
|
"eval_samples_per_second": 8.328, |
|
"eval_steps_per_second": 2.083, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 48.19, |
|
"learning_rate": 4.721052631578947e-06, |
|
"loss": 1.4268, |
|
"step": 17010 |
|
}, |
|
{ |
|
"epoch": 48.21, |
|
"learning_rate": 4.705263157894737e-06, |
|
"loss": 1.4245, |
|
"step": 17020 |
|
}, |
|
{ |
|
"epoch": 48.24, |
|
"learning_rate": 4.689473684210526e-06, |
|
"loss": 1.4318, |
|
"step": 17030 |
|
}, |
|
{ |
|
"epoch": 48.27, |
|
"learning_rate": 4.673684210526316e-06, |
|
"loss": 1.4239, |
|
"step": 17040 |
|
}, |
|
{ |
|
"epoch": 48.3, |
|
"learning_rate": 4.657894736842106e-06, |
|
"loss": 1.4245, |
|
"step": 17050 |
|
}, |
|
{ |
|
"epoch": 48.33, |
|
"learning_rate": 4.642105263157895e-06, |
|
"loss": 1.4254, |
|
"step": 17060 |
|
}, |
|
{ |
|
"epoch": 48.36, |
|
"learning_rate": 4.626315789473684e-06, |
|
"loss": 1.4241, |
|
"step": 17070 |
|
}, |
|
{ |
|
"epoch": 48.38, |
|
"learning_rate": 4.610526315789474e-06, |
|
"loss": 1.4248, |
|
"step": 17080 |
|
}, |
|
{ |
|
"epoch": 48.41, |
|
"learning_rate": 4.594736842105263e-06, |
|
"loss": 1.4248, |
|
"step": 17090 |
|
}, |
|
{ |
|
"epoch": 48.44, |
|
"learning_rate": 4.578947368421053e-06, |
|
"loss": 1.4245, |
|
"step": 17100 |
|
}, |
|
{ |
|
"epoch": 48.47, |
|
"learning_rate": 4.563157894736842e-06, |
|
"loss": 1.4241, |
|
"step": 17110 |
|
}, |
|
{ |
|
"epoch": 48.5, |
|
"learning_rate": 4.547368421052632e-06, |
|
"loss": 1.4271, |
|
"step": 17120 |
|
}, |
|
{ |
|
"epoch": 48.53, |
|
"learning_rate": 4.531578947368421e-06, |
|
"loss": 1.4247, |
|
"step": 17130 |
|
}, |
|
{ |
|
"epoch": 48.55, |
|
"learning_rate": 4.515789473684211e-06, |
|
"loss": 1.4268, |
|
"step": 17140 |
|
}, |
|
{ |
|
"epoch": 48.58, |
|
"learning_rate": 4.5e-06, |
|
"loss": 1.4245, |
|
"step": 17150 |
|
}, |
|
{ |
|
"epoch": 48.61, |
|
"learning_rate": 4.48421052631579e-06, |
|
"loss": 1.4276, |
|
"step": 17160 |
|
}, |
|
{ |
|
"epoch": 48.64, |
|
"learning_rate": 4.468421052631579e-06, |
|
"loss": 1.4245, |
|
"step": 17170 |
|
}, |
|
{ |
|
"epoch": 48.67, |
|
"learning_rate": 4.452631578947369e-06, |
|
"loss": 1.4257, |
|
"step": 17180 |
|
}, |
|
{ |
|
"epoch": 48.7, |
|
"learning_rate": 4.436842105263158e-06, |
|
"loss": 1.4254, |
|
"step": 17190 |
|
}, |
|
{ |
|
"epoch": 48.72, |
|
"learning_rate": 4.421052631578947e-06, |
|
"loss": 1.4259, |
|
"step": 17200 |
|
}, |
|
{ |
|
"epoch": 48.75, |
|
"learning_rate": 4.405263157894737e-06, |
|
"loss": 1.4278, |
|
"step": 17210 |
|
}, |
|
{ |
|
"epoch": 48.78, |
|
"learning_rate": 4.3894736842105266e-06, |
|
"loss": 1.4242, |
|
"step": 17220 |
|
}, |
|
{ |
|
"epoch": 48.81, |
|
"learning_rate": 4.373684210526316e-06, |
|
"loss": 1.4246, |
|
"step": 17230 |
|
}, |
|
{ |
|
"epoch": 48.84, |
|
"learning_rate": 4.3578947368421055e-06, |
|
"loss": 1.4262, |
|
"step": 17240 |
|
}, |
|
{ |
|
"epoch": 48.86, |
|
"learning_rate": 4.342105263157895e-06, |
|
"loss": 1.4243, |
|
"step": 17250 |
|
}, |
|
{ |
|
"epoch": 48.89, |
|
"learning_rate": 4.3263157894736845e-06, |
|
"loss": 1.4249, |
|
"step": 17260 |
|
}, |
|
{ |
|
"epoch": 48.92, |
|
"learning_rate": 4.310526315789474e-06, |
|
"loss": 1.4274, |
|
"step": 17270 |
|
}, |
|
{ |
|
"epoch": 48.95, |
|
"learning_rate": 4.294736842105263e-06, |
|
"loss": 1.4233, |
|
"step": 17280 |
|
}, |
|
{ |
|
"epoch": 48.98, |
|
"learning_rate": 4.278947368421052e-06, |
|
"loss": 1.4248, |
|
"step": 17290 |
|
}, |
|
{ |
|
"epoch": 49.01, |
|
"learning_rate": 4.2631578947368425e-06, |
|
"loss": 1.5487, |
|
"step": 17300 |
|
}, |
|
{ |
|
"epoch": 49.04, |
|
"learning_rate": 4.247368421052632e-06, |
|
"loss": 1.4247, |
|
"step": 17310 |
|
}, |
|
{ |
|
"epoch": 49.06, |
|
"learning_rate": 4.2315789473684215e-06, |
|
"loss": 1.4237, |
|
"step": 17320 |
|
}, |
|
{ |
|
"epoch": 49.09, |
|
"learning_rate": 4.215789473684211e-06, |
|
"loss": 1.4244, |
|
"step": 17330 |
|
}, |
|
{ |
|
"epoch": 49.12, |
|
"learning_rate": 4.2000000000000004e-06, |
|
"loss": 1.4264, |
|
"step": 17340 |
|
}, |
|
{ |
|
"epoch": 49.15, |
|
"learning_rate": 4.184210526315789e-06, |
|
"loss": 1.4297, |
|
"step": 17350 |
|
}, |
|
{ |
|
"epoch": 49.18, |
|
"learning_rate": 4.168421052631579e-06, |
|
"loss": 1.4262, |
|
"step": 17360 |
|
}, |
|
{ |
|
"epoch": 49.21, |
|
"learning_rate": 4.152631578947368e-06, |
|
"loss": 1.4243, |
|
"step": 17370 |
|
}, |
|
{ |
|
"epoch": 49.23, |
|
"learning_rate": 4.136842105263158e-06, |
|
"loss": 1.426, |
|
"step": 17380 |
|
}, |
|
{ |
|
"epoch": 49.26, |
|
"learning_rate": 4.121052631578948e-06, |
|
"loss": 1.4252, |
|
"step": 17390 |
|
}, |
|
{ |
|
"epoch": 49.29, |
|
"learning_rate": 4.105263157894737e-06, |
|
"loss": 1.4254, |
|
"step": 17400 |
|
}, |
|
{ |
|
"epoch": 49.32, |
|
"learning_rate": 4.089473684210527e-06, |
|
"loss": 1.4257, |
|
"step": 17410 |
|
}, |
|
{ |
|
"epoch": 49.35, |
|
"learning_rate": 4.0736842105263155e-06, |
|
"loss": 1.424, |
|
"step": 17420 |
|
}, |
|
{ |
|
"epoch": 49.38, |
|
"learning_rate": 4.057894736842105e-06, |
|
"loss": 1.4258, |
|
"step": 17430 |
|
}, |
|
{ |
|
"epoch": 49.4, |
|
"learning_rate": 4.0421052631578945e-06, |
|
"loss": 1.4247, |
|
"step": 17440 |
|
}, |
|
{ |
|
"epoch": 49.43, |
|
"learning_rate": 4.026315789473684e-06, |
|
"loss": 1.424, |
|
"step": 17450 |
|
}, |
|
{ |
|
"epoch": 49.46, |
|
"learning_rate": 4.010526315789474e-06, |
|
"loss": 1.4246, |
|
"step": 17460 |
|
}, |
|
{ |
|
"epoch": 49.49, |
|
"learning_rate": 3.994736842105264e-06, |
|
"loss": 1.4247, |
|
"step": 17470 |
|
}, |
|
{ |
|
"epoch": 49.52, |
|
"learning_rate": 3.978947368421053e-06, |
|
"loss": 1.4225, |
|
"step": 17480 |
|
}, |
|
{ |
|
"epoch": 49.55, |
|
"learning_rate": 3.963157894736842e-06, |
|
"loss": 1.4285, |
|
"step": 17490 |
|
}, |
|
{ |
|
"epoch": 49.57, |
|
"learning_rate": 3.9473684210526315e-06, |
|
"loss": 1.4258, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 49.6, |
|
"learning_rate": 3.931578947368421e-06, |
|
"loss": 1.4241, |
|
"step": 17510 |
|
}, |
|
{ |
|
"epoch": 49.63, |
|
"learning_rate": 3.9157894736842104e-06, |
|
"loss": 1.4243, |
|
"step": 17520 |
|
}, |
|
{ |
|
"epoch": 49.66, |
|
"learning_rate": 3.9e-06, |
|
"loss": 1.4268, |
|
"step": 17530 |
|
}, |
|
{ |
|
"epoch": 49.69, |
|
"learning_rate": 3.884210526315789e-06, |
|
"loss": 1.4265, |
|
"step": 17540 |
|
}, |
|
{ |
|
"epoch": 49.71, |
|
"learning_rate": 3.86842105263158e-06, |
|
"loss": 1.4227, |
|
"step": 17550 |
|
}, |
|
{ |
|
"epoch": 49.74, |
|
"learning_rate": 3.852631578947368e-06, |
|
"loss": 1.426, |
|
"step": 17560 |
|
}, |
|
{ |
|
"epoch": 49.77, |
|
"learning_rate": 3.836842105263158e-06, |
|
"loss": 1.4237, |
|
"step": 17570 |
|
}, |
|
{ |
|
"epoch": 49.8, |
|
"learning_rate": 3.821052631578947e-06, |
|
"loss": 1.4235, |
|
"step": 17580 |
|
}, |
|
{ |
|
"epoch": 49.83, |
|
"learning_rate": 3.805263157894737e-06, |
|
"loss": 1.4249, |
|
"step": 17590 |
|
}, |
|
{ |
|
"epoch": 49.86, |
|
"learning_rate": 3.7894736842105264e-06, |
|
"loss": 1.4235, |
|
"step": 17600 |
|
}, |
|
{ |
|
"epoch": 49.88, |
|
"learning_rate": 3.7736842105263163e-06, |
|
"loss": 1.4244, |
|
"step": 17610 |
|
}, |
|
{ |
|
"epoch": 49.91, |
|
"learning_rate": 3.7578947368421058e-06, |
|
"loss": 1.4285, |
|
"step": 17620 |
|
}, |
|
{ |
|
"epoch": 49.94, |
|
"learning_rate": 3.742105263157895e-06, |
|
"loss": 1.4255, |
|
"step": 17630 |
|
}, |
|
{ |
|
"epoch": 49.97, |
|
"learning_rate": 3.7263157894736843e-06, |
|
"loss": 1.425, |
|
"step": 17640 |
|
}, |
|
{ |
|
"epoch": 50.0, |
|
"learning_rate": 3.710526315789474e-06, |
|
"loss": 1.425, |
|
"step": 17650 |
|
}, |
|
{ |
|
"epoch": 50.03, |
|
"learning_rate": 3.6947368421052633e-06, |
|
"loss": 1.5484, |
|
"step": 17660 |
|
}, |
|
{ |
|
"epoch": 50.06, |
|
"learning_rate": 3.678947368421053e-06, |
|
"loss": 1.4246, |
|
"step": 17670 |
|
}, |
|
{ |
|
"epoch": 50.08, |
|
"learning_rate": 3.663157894736842e-06, |
|
"loss": 1.4261, |
|
"step": 17680 |
|
}, |
|
{ |
|
"epoch": 50.11, |
|
"learning_rate": 3.6473684210526318e-06, |
|
"loss": 1.4246, |
|
"step": 17690 |
|
}, |
|
{ |
|
"epoch": 50.14, |
|
"learning_rate": 3.6315789473684213e-06, |
|
"loss": 1.4268, |
|
"step": 17700 |
|
}, |
|
{ |
|
"epoch": 50.17, |
|
"learning_rate": 3.6157894736842108e-06, |
|
"loss": 1.4249, |
|
"step": 17710 |
|
}, |
|
{ |
|
"epoch": 50.2, |
|
"learning_rate": 3.6e-06, |
|
"loss": 1.4249, |
|
"step": 17720 |
|
}, |
|
{ |
|
"epoch": 50.23, |
|
"learning_rate": 3.5842105263157898e-06, |
|
"loss": 1.4245, |
|
"step": 17730 |
|
}, |
|
{ |
|
"epoch": 50.25, |
|
"learning_rate": 3.5684210526315792e-06, |
|
"loss": 1.4253, |
|
"step": 17740 |
|
}, |
|
{ |
|
"epoch": 50.28, |
|
"learning_rate": 3.5526315789473683e-06, |
|
"loss": 1.426, |
|
"step": 17750 |
|
}, |
|
{ |
|
"epoch": 50.31, |
|
"learning_rate": 3.536842105263158e-06, |
|
"loss": 1.4276, |
|
"step": 17760 |
|
}, |
|
{ |
|
"epoch": 50.34, |
|
"learning_rate": 3.5210526315789473e-06, |
|
"loss": 1.4233, |
|
"step": 17770 |
|
}, |
|
{ |
|
"epoch": 50.37, |
|
"learning_rate": 3.505263157894737e-06, |
|
"loss": 1.4239, |
|
"step": 17780 |
|
}, |
|
{ |
|
"epoch": 50.4, |
|
"learning_rate": 3.4894736842105263e-06, |
|
"loss": 1.4237, |
|
"step": 17790 |
|
}, |
|
{ |
|
"epoch": 50.42, |
|
"learning_rate": 3.4736842105263158e-06, |
|
"loss": 1.4244, |
|
"step": 17800 |
|
}, |
|
{ |
|
"epoch": 50.45, |
|
"learning_rate": 3.4578947368421053e-06, |
|
"loss": 1.4239, |
|
"step": 17810 |
|
}, |
|
{ |
|
"epoch": 50.48, |
|
"learning_rate": 3.4421052631578947e-06, |
|
"loss": 1.4247, |
|
"step": 17820 |
|
}, |
|
{ |
|
"epoch": 50.51, |
|
"learning_rate": 3.4263157894736842e-06, |
|
"loss": 1.4261, |
|
"step": 17830 |
|
}, |
|
{ |
|
"epoch": 50.54, |
|
"learning_rate": 3.4105263157894737e-06, |
|
"loss": 1.4238, |
|
"step": 17840 |
|
}, |
|
{ |
|
"epoch": 50.57, |
|
"learning_rate": 3.3947368421052632e-06, |
|
"loss": 1.4244, |
|
"step": 17850 |
|
}, |
|
{ |
|
"epoch": 50.59, |
|
"learning_rate": 3.3789473684210527e-06, |
|
"loss": 1.4234, |
|
"step": 17860 |
|
}, |
|
{ |
|
"epoch": 50.62, |
|
"learning_rate": 3.363157894736842e-06, |
|
"loss": 1.424, |
|
"step": 17870 |
|
}, |
|
{ |
|
"epoch": 50.65, |
|
"learning_rate": 3.3473684210526317e-06, |
|
"loss": 1.4319, |
|
"step": 17880 |
|
}, |
|
{ |
|
"epoch": 50.68, |
|
"learning_rate": 3.331578947368421e-06, |
|
"loss": 1.4242, |
|
"step": 17890 |
|
}, |
|
{ |
|
"epoch": 50.71, |
|
"learning_rate": 3.3157894736842107e-06, |
|
"loss": 1.4245, |
|
"step": 17900 |
|
}, |
|
{ |
|
"epoch": 50.73, |
|
"learning_rate": 3.3e-06, |
|
"loss": 1.4315, |
|
"step": 17910 |
|
}, |
|
{ |
|
"epoch": 50.76, |
|
"learning_rate": 3.2842105263157897e-06, |
|
"loss": 1.427, |
|
"step": 17920 |
|
}, |
|
{ |
|
"epoch": 50.79, |
|
"learning_rate": 3.2684210526315787e-06, |
|
"loss": 1.4243, |
|
"step": 17930 |
|
}, |
|
{ |
|
"epoch": 50.82, |
|
"learning_rate": 3.2526315789473686e-06, |
|
"loss": 1.4257, |
|
"step": 17940 |
|
}, |
|
{ |
|
"epoch": 50.85, |
|
"learning_rate": 3.236842105263158e-06, |
|
"loss": 1.4236, |
|
"step": 17950 |
|
}, |
|
{ |
|
"epoch": 50.88, |
|
"learning_rate": 3.2210526315789476e-06, |
|
"loss": 1.4241, |
|
"step": 17960 |
|
}, |
|
{ |
|
"epoch": 50.9, |
|
"learning_rate": 3.2052631578947367e-06, |
|
"loss": 1.424, |
|
"step": 17970 |
|
}, |
|
{ |
|
"epoch": 50.93, |
|
"learning_rate": 3.1894736842105266e-06, |
|
"loss": 1.4244, |
|
"step": 17980 |
|
}, |
|
{ |
|
"epoch": 50.96, |
|
"learning_rate": 3.173684210526316e-06, |
|
"loss": 1.4231, |
|
"step": 17990 |
|
}, |
|
{ |
|
"epoch": 50.99, |
|
"learning_rate": 3.157894736842105e-06, |
|
"loss": 1.4248, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 50.99, |
|
"eval_denotation_accuracy": 0.5817732250088308, |
|
"eval_loss": 2.1932122707366943, |
|
"eval_runtime": 338.4738, |
|
"eval_samples_per_second": 8.364, |
|
"eval_steps_per_second": 2.092, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 51.02, |
|
"learning_rate": 3.1421052631578947e-06, |
|
"loss": 1.548, |
|
"step": 18010 |
|
}, |
|
{ |
|
"epoch": 51.05, |
|
"learning_rate": 3.126315789473684e-06, |
|
"loss": 1.4236, |
|
"step": 18020 |
|
}, |
|
{ |
|
"epoch": 51.08, |
|
"learning_rate": 3.110526315789474e-06, |
|
"loss": 1.423, |
|
"step": 18030 |
|
}, |
|
{ |
|
"epoch": 51.1, |
|
"learning_rate": 3.094736842105263e-06, |
|
"loss": 1.4243, |
|
"step": 18040 |
|
}, |
|
{ |
|
"epoch": 51.13, |
|
"learning_rate": 3.0789473684210526e-06, |
|
"loss": 1.432, |
|
"step": 18050 |
|
}, |
|
{ |
|
"epoch": 51.16, |
|
"learning_rate": 3.063157894736842e-06, |
|
"loss": 1.4236, |
|
"step": 18060 |
|
}, |
|
{ |
|
"epoch": 51.19, |
|
"learning_rate": 3.0473684210526316e-06, |
|
"loss": 1.4244, |
|
"step": 18070 |
|
}, |
|
{ |
|
"epoch": 51.22, |
|
"learning_rate": 3.031578947368421e-06, |
|
"loss": 1.4245, |
|
"step": 18080 |
|
}, |
|
{ |
|
"epoch": 51.25, |
|
"learning_rate": 3.0157894736842106e-06, |
|
"loss": 1.4227, |
|
"step": 18090 |
|
}, |
|
{ |
|
"epoch": 51.27, |
|
"learning_rate": 3e-06, |
|
"loss": 1.4227, |
|
"step": 18100 |
|
}, |
|
{ |
|
"epoch": 51.3, |
|
"learning_rate": 2.9842105263157896e-06, |
|
"loss": 1.4233, |
|
"step": 18110 |
|
}, |
|
{ |
|
"epoch": 51.33, |
|
"learning_rate": 2.968421052631579e-06, |
|
"loss": 1.4241, |
|
"step": 18120 |
|
}, |
|
{ |
|
"epoch": 51.36, |
|
"learning_rate": 2.9526315789473685e-06, |
|
"loss": 1.4239, |
|
"step": 18130 |
|
}, |
|
{ |
|
"epoch": 51.39, |
|
"learning_rate": 2.9368421052631576e-06, |
|
"loss": 1.424, |
|
"step": 18140 |
|
}, |
|
{ |
|
"epoch": 51.42, |
|
"learning_rate": 2.9210526315789475e-06, |
|
"loss": 1.4234, |
|
"step": 18150 |
|
}, |
|
{ |
|
"epoch": 51.44, |
|
"learning_rate": 2.905263157894737e-06, |
|
"loss": 1.4247, |
|
"step": 18160 |
|
}, |
|
{ |
|
"epoch": 51.47, |
|
"learning_rate": 2.8894736842105265e-06, |
|
"loss": 1.4223, |
|
"step": 18170 |
|
}, |
|
{ |
|
"epoch": 51.5, |
|
"learning_rate": 2.8736842105263156e-06, |
|
"loss": 1.4246, |
|
"step": 18180 |
|
}, |
|
{ |
|
"epoch": 51.53, |
|
"learning_rate": 2.8578947368421055e-06, |
|
"loss": 1.424, |
|
"step": 18190 |
|
}, |
|
{ |
|
"epoch": 51.56, |
|
"learning_rate": 2.842105263157895e-06, |
|
"loss": 1.4245, |
|
"step": 18200 |
|
}, |
|
{ |
|
"epoch": 51.58, |
|
"learning_rate": 2.826315789473684e-06, |
|
"loss": 1.4237, |
|
"step": 18210 |
|
}, |
|
{ |
|
"epoch": 51.61, |
|
"learning_rate": 2.8105263157894735e-06, |
|
"loss": 1.4234, |
|
"step": 18220 |
|
}, |
|
{ |
|
"epoch": 51.64, |
|
"learning_rate": 2.7947368421052635e-06, |
|
"loss": 1.4231, |
|
"step": 18230 |
|
}, |
|
{ |
|
"epoch": 51.67, |
|
"learning_rate": 2.778947368421053e-06, |
|
"loss": 1.4237, |
|
"step": 18240 |
|
}, |
|
{ |
|
"epoch": 51.7, |
|
"learning_rate": 2.763157894736842e-06, |
|
"loss": 1.425, |
|
"step": 18250 |
|
}, |
|
{ |
|
"epoch": 51.73, |
|
"learning_rate": 2.7473684210526315e-06, |
|
"loss": 1.4223, |
|
"step": 18260 |
|
}, |
|
{ |
|
"epoch": 51.75, |
|
"learning_rate": 2.731578947368421e-06, |
|
"loss": 1.4241, |
|
"step": 18270 |
|
}, |
|
{ |
|
"epoch": 51.78, |
|
"learning_rate": 2.715789473684211e-06, |
|
"loss": 1.4231, |
|
"step": 18280 |
|
}, |
|
{ |
|
"epoch": 51.81, |
|
"learning_rate": 2.7e-06, |
|
"loss": 1.4242, |
|
"step": 18290 |
|
}, |
|
{ |
|
"epoch": 51.84, |
|
"learning_rate": 2.6842105263157895e-06, |
|
"loss": 1.4239, |
|
"step": 18300 |
|
}, |
|
{ |
|
"epoch": 51.87, |
|
"learning_rate": 2.668421052631579e-06, |
|
"loss": 1.4231, |
|
"step": 18310 |
|
}, |
|
{ |
|
"epoch": 51.9, |
|
"learning_rate": 2.6526315789473685e-06, |
|
"loss": 1.4238, |
|
"step": 18320 |
|
}, |
|
{ |
|
"epoch": 51.92, |
|
"learning_rate": 2.636842105263158e-06, |
|
"loss": 1.4236, |
|
"step": 18330 |
|
}, |
|
{ |
|
"epoch": 51.95, |
|
"learning_rate": 2.6210526315789474e-06, |
|
"loss": 1.4251, |
|
"step": 18340 |
|
}, |
|
{ |
|
"epoch": 51.98, |
|
"learning_rate": 2.605263157894737e-06, |
|
"loss": 1.4227, |
|
"step": 18350 |
|
}, |
|
{ |
|
"epoch": 52.01, |
|
"learning_rate": 2.5894736842105264e-06, |
|
"loss": 1.5467, |
|
"step": 18360 |
|
}, |
|
{ |
|
"epoch": 52.04, |
|
"learning_rate": 2.573684210526316e-06, |
|
"loss": 1.4229, |
|
"step": 18370 |
|
}, |
|
{ |
|
"epoch": 52.07, |
|
"learning_rate": 2.5578947368421054e-06, |
|
"loss": 1.4236, |
|
"step": 18380 |
|
}, |
|
{ |
|
"epoch": 52.1, |
|
"learning_rate": 2.5421052631578945e-06, |
|
"loss": 1.4228, |
|
"step": 18390 |
|
}, |
|
{ |
|
"epoch": 52.12, |
|
"learning_rate": 2.5263157894736844e-06, |
|
"loss": 1.4229, |
|
"step": 18400 |
|
}, |
|
{ |
|
"epoch": 52.15, |
|
"learning_rate": 2.510526315789474e-06, |
|
"loss": 1.4234, |
|
"step": 18410 |
|
}, |
|
{ |
|
"epoch": 52.18, |
|
"learning_rate": 2.4947368421052634e-06, |
|
"loss": 1.4245, |
|
"step": 18420 |
|
}, |
|
{ |
|
"epoch": 52.21, |
|
"learning_rate": 2.4789473684210524e-06, |
|
"loss": 1.4238, |
|
"step": 18430 |
|
}, |
|
{ |
|
"epoch": 52.24, |
|
"learning_rate": 2.4631578947368424e-06, |
|
"loss": 1.4229, |
|
"step": 18440 |
|
}, |
|
{ |
|
"epoch": 52.27, |
|
"learning_rate": 2.447368421052632e-06, |
|
"loss": 1.4236, |
|
"step": 18450 |
|
}, |
|
{ |
|
"epoch": 52.29, |
|
"learning_rate": 2.431578947368421e-06, |
|
"loss": 1.4233, |
|
"step": 18460 |
|
}, |
|
{ |
|
"epoch": 52.32, |
|
"learning_rate": 2.4157894736842104e-06, |
|
"loss": 1.4237, |
|
"step": 18470 |
|
}, |
|
{ |
|
"epoch": 52.35, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 1.4244, |
|
"step": 18480 |
|
}, |
|
{ |
|
"epoch": 52.38, |
|
"learning_rate": 2.38421052631579e-06, |
|
"loss": 1.4237, |
|
"step": 18490 |
|
}, |
|
{ |
|
"epoch": 52.41, |
|
"learning_rate": 2.368421052631579e-06, |
|
"loss": 1.4231, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 52.44, |
|
"learning_rate": 2.3526315789473684e-06, |
|
"loss": 1.4243, |
|
"step": 18510 |
|
}, |
|
{ |
|
"epoch": 52.46, |
|
"learning_rate": 2.336842105263158e-06, |
|
"loss": 1.4236, |
|
"step": 18520 |
|
}, |
|
{ |
|
"epoch": 52.49, |
|
"learning_rate": 2.3210526315789473e-06, |
|
"loss": 1.4243, |
|
"step": 18530 |
|
}, |
|
{ |
|
"epoch": 52.52, |
|
"learning_rate": 2.305263157894737e-06, |
|
"loss": 1.4224, |
|
"step": 18540 |
|
}, |
|
{ |
|
"epoch": 52.55, |
|
"learning_rate": 2.2894736842105263e-06, |
|
"loss": 1.4223, |
|
"step": 18550 |
|
}, |
|
{ |
|
"epoch": 52.58, |
|
"learning_rate": 2.273684210526316e-06, |
|
"loss": 1.4245, |
|
"step": 18560 |
|
}, |
|
{ |
|
"epoch": 52.6, |
|
"learning_rate": 2.2578947368421053e-06, |
|
"loss": 1.4243, |
|
"step": 18570 |
|
}, |
|
{ |
|
"epoch": 52.63, |
|
"learning_rate": 2.242105263157895e-06, |
|
"loss": 1.4226, |
|
"step": 18580 |
|
}, |
|
{ |
|
"epoch": 52.66, |
|
"learning_rate": 2.2263157894736843e-06, |
|
"loss": 1.4248, |
|
"step": 18590 |
|
}, |
|
{ |
|
"epoch": 52.69, |
|
"learning_rate": 2.2105263157894734e-06, |
|
"loss": 1.4232, |
|
"step": 18600 |
|
}, |
|
{ |
|
"epoch": 52.72, |
|
"learning_rate": 2.1947368421052633e-06, |
|
"loss": 1.4236, |
|
"step": 18610 |
|
}, |
|
{ |
|
"epoch": 52.75, |
|
"learning_rate": 2.1789473684210528e-06, |
|
"loss": 1.4237, |
|
"step": 18620 |
|
}, |
|
{ |
|
"epoch": 52.77, |
|
"learning_rate": 2.1631578947368423e-06, |
|
"loss": 1.4223, |
|
"step": 18630 |
|
}, |
|
{ |
|
"epoch": 52.8, |
|
"learning_rate": 2.1473684210526313e-06, |
|
"loss": 1.4226, |
|
"step": 18640 |
|
}, |
|
{ |
|
"epoch": 52.83, |
|
"learning_rate": 2.1315789473684212e-06, |
|
"loss": 1.4217, |
|
"step": 18650 |
|
}, |
|
{ |
|
"epoch": 52.86, |
|
"learning_rate": 2.1157894736842107e-06, |
|
"loss": 1.4244, |
|
"step": 18660 |
|
}, |
|
{ |
|
"epoch": 52.89, |
|
"learning_rate": 2.1000000000000002e-06, |
|
"loss": 1.4257, |
|
"step": 18670 |
|
}, |
|
{ |
|
"epoch": 52.92, |
|
"learning_rate": 2.0842105263157893e-06, |
|
"loss": 1.4224, |
|
"step": 18680 |
|
}, |
|
{ |
|
"epoch": 52.94, |
|
"learning_rate": 2.068421052631579e-06, |
|
"loss": 1.4231, |
|
"step": 18690 |
|
}, |
|
{ |
|
"epoch": 52.97, |
|
"learning_rate": 2.0526315789473687e-06, |
|
"loss": 1.423, |
|
"step": 18700 |
|
}, |
|
{ |
|
"epoch": 53.0, |
|
"learning_rate": 2.0368421052631578e-06, |
|
"loss": 1.5492, |
|
"step": 18710 |
|
}, |
|
{ |
|
"epoch": 53.03, |
|
"learning_rate": 2.0210526315789473e-06, |
|
"loss": 1.4238, |
|
"step": 18720 |
|
}, |
|
{ |
|
"epoch": 53.06, |
|
"learning_rate": 2.005263157894737e-06, |
|
"loss": 1.4227, |
|
"step": 18730 |
|
}, |
|
{ |
|
"epoch": 53.09, |
|
"learning_rate": 1.9894736842105267e-06, |
|
"loss": 1.4226, |
|
"step": 18740 |
|
}, |
|
{ |
|
"epoch": 53.12, |
|
"learning_rate": 1.9736842105263157e-06, |
|
"loss": 1.4235, |
|
"step": 18750 |
|
}, |
|
{ |
|
"epoch": 53.14, |
|
"learning_rate": 1.9578947368421052e-06, |
|
"loss": 1.4218, |
|
"step": 18760 |
|
}, |
|
{ |
|
"epoch": 53.17, |
|
"learning_rate": 1.9421052631578947e-06, |
|
"loss": 1.4228, |
|
"step": 18770 |
|
}, |
|
{ |
|
"epoch": 53.2, |
|
"learning_rate": 1.926315789473684e-06, |
|
"loss": 1.4231, |
|
"step": 18780 |
|
}, |
|
{ |
|
"epoch": 53.23, |
|
"learning_rate": 1.9105263157894737e-06, |
|
"loss": 1.422, |
|
"step": 18790 |
|
}, |
|
{ |
|
"epoch": 53.26, |
|
"learning_rate": 1.8947368421052632e-06, |
|
"loss": 1.4231, |
|
"step": 18800 |
|
}, |
|
{ |
|
"epoch": 53.29, |
|
"learning_rate": 1.8789473684210529e-06, |
|
"loss": 1.4239, |
|
"step": 18810 |
|
}, |
|
{ |
|
"epoch": 53.31, |
|
"learning_rate": 1.8631578947368422e-06, |
|
"loss": 1.423, |
|
"step": 18820 |
|
}, |
|
{ |
|
"epoch": 53.34, |
|
"learning_rate": 1.8473684210526317e-06, |
|
"loss": 1.4238, |
|
"step": 18830 |
|
}, |
|
{ |
|
"epoch": 53.37, |
|
"learning_rate": 1.831578947368421e-06, |
|
"loss": 1.4233, |
|
"step": 18840 |
|
}, |
|
{ |
|
"epoch": 53.4, |
|
"learning_rate": 1.8157894736842106e-06, |
|
"loss": 1.424, |
|
"step": 18850 |
|
}, |
|
{ |
|
"epoch": 53.43, |
|
"learning_rate": 1.8e-06, |
|
"loss": 1.4224, |
|
"step": 18860 |
|
}, |
|
{ |
|
"epoch": 53.45, |
|
"learning_rate": 1.7842105263157896e-06, |
|
"loss": 1.4238, |
|
"step": 18870 |
|
}, |
|
{ |
|
"epoch": 53.48, |
|
"learning_rate": 1.768421052631579e-06, |
|
"loss": 1.4226, |
|
"step": 18880 |
|
}, |
|
{ |
|
"epoch": 53.51, |
|
"learning_rate": 1.7526315789473686e-06, |
|
"loss": 1.4226, |
|
"step": 18890 |
|
}, |
|
{ |
|
"epoch": 53.54, |
|
"learning_rate": 1.7368421052631579e-06, |
|
"loss": 1.4234, |
|
"step": 18900 |
|
}, |
|
{ |
|
"epoch": 53.57, |
|
"learning_rate": 1.7210526315789474e-06, |
|
"loss": 1.4231, |
|
"step": 18910 |
|
}, |
|
{ |
|
"epoch": 53.6, |
|
"learning_rate": 1.7052631578947369e-06, |
|
"loss": 1.4234, |
|
"step": 18920 |
|
}, |
|
{ |
|
"epoch": 53.62, |
|
"learning_rate": 1.6894736842105264e-06, |
|
"loss": 1.4227, |
|
"step": 18930 |
|
}, |
|
{ |
|
"epoch": 53.65, |
|
"learning_rate": 1.6736842105263158e-06, |
|
"loss": 1.4225, |
|
"step": 18940 |
|
}, |
|
{ |
|
"epoch": 53.68, |
|
"learning_rate": 1.6578947368421053e-06, |
|
"loss": 1.4225, |
|
"step": 18950 |
|
}, |
|
{ |
|
"epoch": 53.71, |
|
"learning_rate": 1.6421052631578948e-06, |
|
"loss": 1.4233, |
|
"step": 18960 |
|
}, |
|
{ |
|
"epoch": 53.74, |
|
"learning_rate": 1.6263157894736843e-06, |
|
"loss": 1.4247, |
|
"step": 18970 |
|
}, |
|
{ |
|
"epoch": 53.77, |
|
"learning_rate": 1.6105263157894738e-06, |
|
"loss": 1.4239, |
|
"step": 18980 |
|
}, |
|
{ |
|
"epoch": 53.79, |
|
"learning_rate": 1.5947368421052633e-06, |
|
"loss": 1.428, |
|
"step": 18990 |
|
}, |
|
{ |
|
"epoch": 53.82, |
|
"learning_rate": 1.5789473684210526e-06, |
|
"loss": 1.423, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 53.82, |
|
"eval_denotation_accuracy": 0.5782409042741081, |
|
"eval_loss": 2.194387435913086, |
|
"eval_runtime": 320.0346, |
|
"eval_samples_per_second": 8.846, |
|
"eval_steps_per_second": 2.212, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 53.85, |
|
"learning_rate": 1.563157894736842e-06, |
|
"loss": 1.4241, |
|
"step": 19010 |
|
}, |
|
{ |
|
"epoch": 53.88, |
|
"learning_rate": 1.5473684210526316e-06, |
|
"loss": 1.4229, |
|
"step": 19020 |
|
}, |
|
{ |
|
"epoch": 53.91, |
|
"learning_rate": 1.531578947368421e-06, |
|
"loss": 1.4278, |
|
"step": 19030 |
|
}, |
|
{ |
|
"epoch": 53.94, |
|
"learning_rate": 1.5157894736842105e-06, |
|
"loss": 1.4226, |
|
"step": 19040 |
|
}, |
|
{ |
|
"epoch": 53.96, |
|
"learning_rate": 1.5e-06, |
|
"loss": 1.4231, |
|
"step": 19050 |
|
}, |
|
{ |
|
"epoch": 53.99, |
|
"learning_rate": 1.4842105263157895e-06, |
|
"loss": 1.4231, |
|
"step": 19060 |
|
}, |
|
{ |
|
"epoch": 54.02, |
|
"learning_rate": 1.4684210526315788e-06, |
|
"loss": 1.548, |
|
"step": 19070 |
|
}, |
|
{ |
|
"epoch": 54.05, |
|
"learning_rate": 1.4526315789473685e-06, |
|
"loss": 1.4215, |
|
"step": 19080 |
|
}, |
|
{ |
|
"epoch": 54.08, |
|
"learning_rate": 1.4368421052631578e-06, |
|
"loss": 1.4229, |
|
"step": 19090 |
|
}, |
|
{ |
|
"epoch": 54.11, |
|
"learning_rate": 1.4210526315789475e-06, |
|
"loss": 1.4234, |
|
"step": 19100 |
|
}, |
|
{ |
|
"epoch": 54.14, |
|
"learning_rate": 1.4052631578947368e-06, |
|
"loss": 1.4227, |
|
"step": 19110 |
|
}, |
|
{ |
|
"epoch": 54.16, |
|
"learning_rate": 1.3894736842105265e-06, |
|
"loss": 1.423, |
|
"step": 19120 |
|
}, |
|
{ |
|
"epoch": 54.19, |
|
"learning_rate": 1.3736842105263158e-06, |
|
"loss": 1.4233, |
|
"step": 19130 |
|
}, |
|
{ |
|
"epoch": 54.22, |
|
"learning_rate": 1.3578947368421055e-06, |
|
"loss": 1.4235, |
|
"step": 19140 |
|
}, |
|
{ |
|
"epoch": 54.25, |
|
"learning_rate": 1.3421052631578947e-06, |
|
"loss": 1.4223, |
|
"step": 19150 |
|
}, |
|
{ |
|
"epoch": 54.28, |
|
"learning_rate": 1.3263157894736842e-06, |
|
"loss": 1.4233, |
|
"step": 19160 |
|
}, |
|
{ |
|
"epoch": 54.31, |
|
"learning_rate": 1.3105263157894737e-06, |
|
"loss": 1.4229, |
|
"step": 19170 |
|
}, |
|
{ |
|
"epoch": 54.33, |
|
"learning_rate": 1.2947368421052632e-06, |
|
"loss": 1.4233, |
|
"step": 19180 |
|
}, |
|
{ |
|
"epoch": 54.36, |
|
"learning_rate": 1.2789473684210527e-06, |
|
"loss": 1.4234, |
|
"step": 19190 |
|
}, |
|
{ |
|
"epoch": 54.39, |
|
"learning_rate": 1.2631578947368422e-06, |
|
"loss": 1.4231, |
|
"step": 19200 |
|
}, |
|
{ |
|
"epoch": 54.42, |
|
"learning_rate": 1.2473684210526317e-06, |
|
"loss": 1.4236, |
|
"step": 19210 |
|
}, |
|
{ |
|
"epoch": 54.45, |
|
"learning_rate": 1.2315789473684212e-06, |
|
"loss": 1.4234, |
|
"step": 19220 |
|
}, |
|
{ |
|
"epoch": 54.47, |
|
"learning_rate": 1.2157894736842105e-06, |
|
"loss": 1.4228, |
|
"step": 19230 |
|
}, |
|
{ |
|
"epoch": 54.5, |
|
"learning_rate": 1.2000000000000002e-06, |
|
"loss": 1.423, |
|
"step": 19240 |
|
}, |
|
{ |
|
"epoch": 54.53, |
|
"learning_rate": 1.1842105263157894e-06, |
|
"loss": 1.4237, |
|
"step": 19250 |
|
}, |
|
{ |
|
"epoch": 54.56, |
|
"learning_rate": 1.168421052631579e-06, |
|
"loss": 1.4231, |
|
"step": 19260 |
|
}, |
|
{ |
|
"epoch": 54.59, |
|
"learning_rate": 1.1526315789473684e-06, |
|
"loss": 1.4231, |
|
"step": 19270 |
|
}, |
|
{ |
|
"epoch": 54.62, |
|
"learning_rate": 1.136842105263158e-06, |
|
"loss": 1.4223, |
|
"step": 19280 |
|
}, |
|
{ |
|
"epoch": 54.64, |
|
"learning_rate": 1.1210526315789474e-06, |
|
"loss": 1.4228, |
|
"step": 19290 |
|
}, |
|
{ |
|
"epoch": 54.67, |
|
"learning_rate": 1.1052631578947367e-06, |
|
"loss": 1.4224, |
|
"step": 19300 |
|
}, |
|
{ |
|
"epoch": 54.7, |
|
"learning_rate": 1.0894736842105264e-06, |
|
"loss": 1.4242, |
|
"step": 19310 |
|
}, |
|
{ |
|
"epoch": 54.73, |
|
"learning_rate": 1.0736842105263157e-06, |
|
"loss": 1.4224, |
|
"step": 19320 |
|
}, |
|
{ |
|
"epoch": 54.76, |
|
"learning_rate": 1.0578947368421054e-06, |
|
"loss": 1.422, |
|
"step": 19330 |
|
}, |
|
{ |
|
"epoch": 54.79, |
|
"learning_rate": 1.0421052631578946e-06, |
|
"loss": 1.4226, |
|
"step": 19340 |
|
}, |
|
{ |
|
"epoch": 54.81, |
|
"learning_rate": 1.0263157894736843e-06, |
|
"loss": 1.4228, |
|
"step": 19350 |
|
}, |
|
{ |
|
"epoch": 54.84, |
|
"learning_rate": 1.0105263157894736e-06, |
|
"loss": 1.4228, |
|
"step": 19360 |
|
}, |
|
{ |
|
"epoch": 54.87, |
|
"learning_rate": 9.947368421052633e-07, |
|
"loss": 1.4232, |
|
"step": 19370 |
|
}, |
|
{ |
|
"epoch": 54.9, |
|
"learning_rate": 9.789473684210526e-07, |
|
"loss": 1.4241, |
|
"step": 19380 |
|
}, |
|
{ |
|
"epoch": 54.93, |
|
"learning_rate": 9.63157894736842e-07, |
|
"loss": 1.4232, |
|
"step": 19390 |
|
}, |
|
{ |
|
"epoch": 54.96, |
|
"learning_rate": 9.473684210526316e-07, |
|
"loss": 1.4225, |
|
"step": 19400 |
|
}, |
|
{ |
|
"epoch": 54.98, |
|
"learning_rate": 9.315789473684211e-07, |
|
"loss": 1.4237, |
|
"step": 19410 |
|
}, |
|
{ |
|
"epoch": 55.01, |
|
"learning_rate": 9.157894736842105e-07, |
|
"loss": 1.5472, |
|
"step": 19420 |
|
}, |
|
{ |
|
"epoch": 55.04, |
|
"learning_rate": 9e-07, |
|
"loss": 1.4235, |
|
"step": 19430 |
|
}, |
|
{ |
|
"epoch": 55.07, |
|
"learning_rate": 8.842105263157895e-07, |
|
"loss": 1.422, |
|
"step": 19440 |
|
}, |
|
{ |
|
"epoch": 55.1, |
|
"learning_rate": 8.684210526315789e-07, |
|
"loss": 1.4247, |
|
"step": 19450 |
|
}, |
|
{ |
|
"epoch": 55.13, |
|
"learning_rate": 8.526315789473684e-07, |
|
"loss": 1.4235, |
|
"step": 19460 |
|
}, |
|
{ |
|
"epoch": 55.16, |
|
"learning_rate": 8.368421052631579e-07, |
|
"loss": 1.4226, |
|
"step": 19470 |
|
}, |
|
{ |
|
"epoch": 55.18, |
|
"learning_rate": 8.210526315789474e-07, |
|
"loss": 1.4232, |
|
"step": 19480 |
|
}, |
|
{ |
|
"epoch": 55.21, |
|
"learning_rate": 8.052631578947369e-07, |
|
"loss": 1.4245, |
|
"step": 19490 |
|
}, |
|
{ |
|
"epoch": 55.24, |
|
"learning_rate": 7.894736842105263e-07, |
|
"loss": 1.4234, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 55.27, |
|
"learning_rate": 7.736842105263158e-07, |
|
"loss": 1.4234, |
|
"step": 19510 |
|
}, |
|
{ |
|
"epoch": 55.3, |
|
"learning_rate": 7.578947368421053e-07, |
|
"loss": 1.4215, |
|
"step": 19520 |
|
}, |
|
{ |
|
"epoch": 55.32, |
|
"learning_rate": 7.421052631578948e-07, |
|
"loss": 1.4312, |
|
"step": 19530 |
|
}, |
|
{ |
|
"epoch": 55.35, |
|
"learning_rate": 7.263157894736843e-07, |
|
"loss": 1.423, |
|
"step": 19540 |
|
}, |
|
{ |
|
"epoch": 55.38, |
|
"learning_rate": 7.105263157894737e-07, |
|
"loss": 1.4222, |
|
"step": 19550 |
|
}, |
|
{ |
|
"epoch": 55.41, |
|
"learning_rate": 6.947368421052632e-07, |
|
"loss": 1.4234, |
|
"step": 19560 |
|
}, |
|
{ |
|
"epoch": 55.44, |
|
"learning_rate": 6.789473684210527e-07, |
|
"loss": 1.4215, |
|
"step": 19570 |
|
}, |
|
{ |
|
"epoch": 55.47, |
|
"learning_rate": 6.631578947368421e-07, |
|
"loss": 1.4229, |
|
"step": 19580 |
|
}, |
|
{ |
|
"epoch": 55.49, |
|
"learning_rate": 6.473684210526316e-07, |
|
"loss": 1.4231, |
|
"step": 19590 |
|
}, |
|
{ |
|
"epoch": 55.52, |
|
"learning_rate": 6.315789473684211e-07, |
|
"loss": 1.4225, |
|
"step": 19600 |
|
}, |
|
{ |
|
"epoch": 55.55, |
|
"learning_rate": 6.157894736842106e-07, |
|
"loss": 1.4227, |
|
"step": 19610 |
|
}, |
|
{ |
|
"epoch": 55.58, |
|
"learning_rate": 6.000000000000001e-07, |
|
"loss": 1.4216, |
|
"step": 19620 |
|
}, |
|
{ |
|
"epoch": 55.61, |
|
"learning_rate": 5.842105263157895e-07, |
|
"loss": 1.4228, |
|
"step": 19630 |
|
}, |
|
{ |
|
"epoch": 55.64, |
|
"learning_rate": 5.68421052631579e-07, |
|
"loss": 1.4228, |
|
"step": 19640 |
|
}, |
|
{ |
|
"epoch": 55.66, |
|
"learning_rate": 5.526315789473683e-07, |
|
"loss": 1.4227, |
|
"step": 19650 |
|
}, |
|
{ |
|
"epoch": 55.69, |
|
"learning_rate": 5.368421052631578e-07, |
|
"loss": 1.4222, |
|
"step": 19660 |
|
}, |
|
{ |
|
"epoch": 55.72, |
|
"learning_rate": 5.210526315789473e-07, |
|
"loss": 1.4227, |
|
"step": 19670 |
|
}, |
|
{ |
|
"epoch": 55.75, |
|
"learning_rate": 5.052631578947368e-07, |
|
"loss": 1.4226, |
|
"step": 19680 |
|
}, |
|
{ |
|
"epoch": 55.78, |
|
"learning_rate": 4.894736842105263e-07, |
|
"loss": 1.4226, |
|
"step": 19690 |
|
}, |
|
{ |
|
"epoch": 55.81, |
|
"learning_rate": 4.736842105263158e-07, |
|
"loss": 1.4224, |
|
"step": 19700 |
|
}, |
|
{ |
|
"epoch": 55.83, |
|
"learning_rate": 4.5789473684210523e-07, |
|
"loss": 1.4264, |
|
"step": 19710 |
|
}, |
|
{ |
|
"epoch": 55.86, |
|
"learning_rate": 4.421052631578947e-07, |
|
"loss": 1.4237, |
|
"step": 19720 |
|
}, |
|
{ |
|
"epoch": 55.89, |
|
"learning_rate": 4.263157894736842e-07, |
|
"loss": 1.4225, |
|
"step": 19730 |
|
}, |
|
{ |
|
"epoch": 55.92, |
|
"learning_rate": 4.105263157894737e-07, |
|
"loss": 1.4229, |
|
"step": 19740 |
|
}, |
|
{ |
|
"epoch": 55.95, |
|
"learning_rate": 3.9473684210526315e-07, |
|
"loss": 1.4229, |
|
"step": 19750 |
|
}, |
|
{ |
|
"epoch": 55.97, |
|
"learning_rate": 3.7894736842105264e-07, |
|
"loss": 1.4237, |
|
"step": 19760 |
|
}, |
|
{ |
|
"epoch": 56.01, |
|
"learning_rate": 3.6315789473684213e-07, |
|
"loss": 1.5472, |
|
"step": 19770 |
|
}, |
|
{ |
|
"epoch": 56.03, |
|
"learning_rate": 3.473684210526316e-07, |
|
"loss": 1.4223, |
|
"step": 19780 |
|
}, |
|
{ |
|
"epoch": 56.06, |
|
"learning_rate": 3.3157894736842106e-07, |
|
"loss": 1.4225, |
|
"step": 19790 |
|
}, |
|
{ |
|
"epoch": 56.09, |
|
"learning_rate": 3.1578947368421055e-07, |
|
"loss": 1.4267, |
|
"step": 19800 |
|
}, |
|
{ |
|
"epoch": 56.12, |
|
"learning_rate": 3.0000000000000004e-07, |
|
"loss": 1.423, |
|
"step": 19810 |
|
}, |
|
{ |
|
"epoch": 56.15, |
|
"learning_rate": 2.842105263157895e-07, |
|
"loss": 1.4227, |
|
"step": 19820 |
|
}, |
|
{ |
|
"epoch": 56.18, |
|
"learning_rate": 2.684210526315789e-07, |
|
"loss": 1.4228, |
|
"step": 19830 |
|
}, |
|
{ |
|
"epoch": 56.2, |
|
"learning_rate": 2.526315789473684e-07, |
|
"loss": 1.4227, |
|
"step": 19840 |
|
}, |
|
{ |
|
"epoch": 56.23, |
|
"learning_rate": 2.368421052631579e-07, |
|
"loss": 1.422, |
|
"step": 19850 |
|
}, |
|
{ |
|
"epoch": 56.26, |
|
"learning_rate": 2.2105263157894736e-07, |
|
"loss": 1.4234, |
|
"step": 19860 |
|
}, |
|
{ |
|
"epoch": 56.29, |
|
"learning_rate": 2.0526315789473685e-07, |
|
"loss": 1.4222, |
|
"step": 19870 |
|
}, |
|
{ |
|
"epoch": 56.32, |
|
"learning_rate": 1.8947368421052632e-07, |
|
"loss": 1.4227, |
|
"step": 19880 |
|
}, |
|
{ |
|
"epoch": 56.34, |
|
"learning_rate": 1.736842105263158e-07, |
|
"loss": 1.4303, |
|
"step": 19890 |
|
}, |
|
{ |
|
"epoch": 56.37, |
|
"learning_rate": 1.5789473684210527e-07, |
|
"loss": 1.4238, |
|
"step": 19900 |
|
}, |
|
{ |
|
"epoch": 56.4, |
|
"learning_rate": 1.4210526315789474e-07, |
|
"loss": 1.4224, |
|
"step": 19910 |
|
}, |
|
{ |
|
"epoch": 56.43, |
|
"learning_rate": 1.263157894736842e-07, |
|
"loss": 1.4222, |
|
"step": 19920 |
|
}, |
|
{ |
|
"epoch": 56.46, |
|
"learning_rate": 1.1052631578947368e-07, |
|
"loss": 1.4226, |
|
"step": 19930 |
|
}, |
|
{ |
|
"epoch": 56.49, |
|
"learning_rate": 9.473684210526316e-08, |
|
"loss": 1.4233, |
|
"step": 19940 |
|
}, |
|
{ |
|
"epoch": 56.51, |
|
"learning_rate": 7.894736842105264e-08, |
|
"loss": 1.4236, |
|
"step": 19950 |
|
}, |
|
{ |
|
"epoch": 56.54, |
|
"learning_rate": 6.31578947368421e-08, |
|
"loss": 1.4224, |
|
"step": 19960 |
|
}, |
|
{ |
|
"epoch": 56.57, |
|
"learning_rate": 4.736842105263158e-08, |
|
"loss": 1.4224, |
|
"step": 19970 |
|
}, |
|
{ |
|
"epoch": 56.6, |
|
"learning_rate": 3.157894736842105e-08, |
|
"loss": 1.4238, |
|
"step": 19980 |
|
}, |
|
{ |
|
"epoch": 56.63, |
|
"learning_rate": 1.5789473684210525e-08, |
|
"loss": 1.4241, |
|
"step": 19990 |
|
}, |
|
{ |
|
"epoch": 56.66, |
|
"learning_rate": 0.0, |
|
"loss": 1.4226, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 56.66, |
|
"eval_denotation_accuracy": 0.5793006004945249, |
|
"eval_loss": 2.197631359100342, |
|
"eval_runtime": 322.9471, |
|
"eval_samples_per_second": 8.766, |
|
"eval_steps_per_second": 2.192, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 56.66, |
|
"step": 20000, |
|
"total_flos": 1.1925130485460255e+18, |
|
"train_loss": 1.547329451227188, |
|
"train_runtime": 59809.286, |
|
"train_samples_per_second": 10.701, |
|
"train_steps_per_second": 0.334 |
|
} |
|
], |
|
"max_steps": 20000, |
|
"num_train_epochs": 57, |
|
"total_flos": 1.1925130485460255e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|