| { | |
| "best_metric": 0.7468354430379747, | |
| "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-crop-classification/checkpoint-2932", | |
| "epoch": 3.9972733469665984, | |
| "eval_steps": 500, | |
| "global_step": 2932, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.6735, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.7499, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.7893, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.7297, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.6714, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.664, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 0.7438, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.7032, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.7241, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.7169, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.6711, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.7315, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5.2e-06, | |
| "loss": 0.6819, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.6716, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6e-06, | |
| "loss": 0.759, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.7928, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 0.6941, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 0.7584, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 0.7385, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6385, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.6845, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.8e-06, | |
| "loss": 0.6738, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 0.6623, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 0.6563, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7166, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.04e-05, | |
| "loss": 0.7435, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.0800000000000002e-05, | |
| "loss": 0.6688, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 0.7324, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.16e-05, | |
| "loss": 0.6385, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.776, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2400000000000002e-05, | |
| "loss": 0.6967, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 0.6705, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.3200000000000002e-05, | |
| "loss": 0.7122, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 0.6792, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.4e-05, | |
| "loss": 0.6689, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.4400000000000001e-05, | |
| "loss": 0.7521, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.48e-05, | |
| "loss": 0.7307, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5200000000000002e-05, | |
| "loss": 0.7231, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.5600000000000003e-05, | |
| "loss": 0.5955, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.6321, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.64e-05, | |
| "loss": 0.6292, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 0.7091, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.72e-05, | |
| "loss": 0.6592, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.76e-05, | |
| "loss": 0.7054, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.7367, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8400000000000003e-05, | |
| "loss": 0.7689, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.88e-05, | |
| "loss": 0.6851, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 0.7103, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 0.7006, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7557, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.9917763157894737e-05, | |
| "loss": 0.6855, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.9835526315789476e-05, | |
| "loss": 0.6416, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.975328947368421e-05, | |
| "loss": 0.7053, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.9671052631578947e-05, | |
| "loss": 0.6751, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.9588815789473686e-05, | |
| "loss": 0.7471, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.9506578947368422e-05, | |
| "loss": 0.7891, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.9424342105263157e-05, | |
| "loss": 0.7501, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.9342105263157896e-05, | |
| "loss": 0.6706, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9259868421052632e-05, | |
| "loss": 0.6812, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.917763157894737e-05, | |
| "loss": 0.7809, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.9095394736842106e-05, | |
| "loss": 0.7404, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.9013157894736845e-05, | |
| "loss": 0.7135, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.893092105263158e-05, | |
| "loss": 0.7018, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.8848684210526316e-05, | |
| "loss": 0.7291, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8766447368421055e-05, | |
| "loss": 0.7254, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.868421052631579e-05, | |
| "loss": 0.698, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8601973684210526e-05, | |
| "loss": 0.7054, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.8519736842105265e-05, | |
| "loss": 0.7297, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.84375e-05, | |
| "loss": 0.7312, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.835526315789474e-05, | |
| "loss": 0.679, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.8273026315789475e-05, | |
| "loss": 0.6963, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.819078947368421e-05, | |
| "loss": 0.7223, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.810855263157895e-05, | |
| "loss": 0.7134, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7265055619485999, | |
| "eval_loss": 0.6852419376373291, | |
| "eval_runtime": 68.0133, | |
| "eval_samples_per_second": 38.331, | |
| "eval_steps_per_second": 2.397, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.8026315789473685e-05, | |
| "loss": 0.6766, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.794407894736842e-05, | |
| "loss": 0.6607, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.786184210526316e-05, | |
| "loss": 0.7108, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.7779605263157896e-05, | |
| "loss": 0.6455, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.769736842105263e-05, | |
| "loss": 0.7234, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.761513157894737e-05, | |
| "loss": 0.7256, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.753289473684211e-05, | |
| "loss": 0.7435, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.7450657894736845e-05, | |
| "loss": 0.6775, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.736842105263158e-05, | |
| "loss": 0.686, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.728618421052632e-05, | |
| "loss": 0.6798, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.7203947368421055e-05, | |
| "loss": 0.7085, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.712171052631579e-05, | |
| "loss": 0.6763, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.703947368421053e-05, | |
| "loss": 0.772, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.6957236842105265e-05, | |
| "loss": 0.6988, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.6875e-05, | |
| "loss": 0.7125, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.679276315789474e-05, | |
| "loss": 0.7341, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6710526315789475e-05, | |
| "loss": 0.7137, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.662828947368421e-05, | |
| "loss": 0.758, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.654605263157895e-05, | |
| "loss": 0.6797, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.6463815789473685e-05, | |
| "loss": 0.7042, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.638157894736842e-05, | |
| "loss": 0.6738, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.629934210526316e-05, | |
| "loss": 0.7389, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.6217105263157895e-05, | |
| "loss": 0.6428, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.613486842105263e-05, | |
| "loss": 0.6957, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.605263157894737e-05, | |
| "loss": 0.7889, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.5970394736842108e-05, | |
| "loss": 0.7333, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5888157894736844e-05, | |
| "loss": 0.7317, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.580592105263158e-05, | |
| "loss": 0.7571, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.572368421052632e-05, | |
| "loss": 0.6069, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5641447368421054e-05, | |
| "loss": 0.6884, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.555921052631579e-05, | |
| "loss": 0.708, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.547697368421053e-05, | |
| "loss": 0.6739, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.5394736842105264e-05, | |
| "loss": 0.7046, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.5312500000000003e-05, | |
| "loss": 0.6073, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.5230263157894738e-05, | |
| "loss": 0.6664, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.5148026315789474e-05, | |
| "loss": 0.6721, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.5065789473684211e-05, | |
| "loss": 0.6939, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4983552631578949e-05, | |
| "loss": 0.6383, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4901315789473684e-05, | |
| "loss": 0.6977, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4819078947368421e-05, | |
| "loss": 0.6759, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4736842105263159e-05, | |
| "loss": 0.7071, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.4654605263157896e-05, | |
| "loss": 0.6564, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.4572368421052631e-05, | |
| "loss": 0.7482, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.449013157894737e-05, | |
| "loss": 0.78, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.4407894736842108e-05, | |
| "loss": 0.7113, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.4325657894736845e-05, | |
| "loss": 0.7461, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.424342105263158e-05, | |
| "loss": 0.6659, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.4161184210526318e-05, | |
| "loss": 0.6422, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4078947368421055e-05, | |
| "loss": 0.7345, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.399671052631579e-05, | |
| "loss": 0.6698, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.3914473684210528e-05, | |
| "loss": 0.646, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.3832236842105265e-05, | |
| "loss": 0.6503, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.375e-05, | |
| "loss": 0.6613, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.3667763157894738e-05, | |
| "loss": 0.6948, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.3585526315789475e-05, | |
| "loss": 0.6722, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.350328947368421e-05, | |
| "loss": 0.7016, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.3421052631578948e-05, | |
| "loss": 0.6822, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.3338815789473685e-05, | |
| "loss": 0.6352, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.325657894736842e-05, | |
| "loss": 0.6853, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.3174342105263158e-05, | |
| "loss": 0.6993, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.3092105263157895e-05, | |
| "loss": 0.6799, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.3009868421052632e-05, | |
| "loss": 0.7551, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.2927631578947371e-05, | |
| "loss": 0.6871, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.2845394736842107e-05, | |
| "loss": 0.7272, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.2763157894736844e-05, | |
| "loss": 0.7118, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2680921052631581e-05, | |
| "loss": 0.6774, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.2598684210526317e-05, | |
| "loss": 0.7233, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.2516447368421054e-05, | |
| "loss": 0.7334, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.2434210526315791e-05, | |
| "loss": 0.6322, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.2351973684210527e-05, | |
| "loss": 0.6835, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.2269736842105264e-05, | |
| "loss": 0.7441, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.2187500000000001e-05, | |
| "loss": 0.7206, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.2105263157894737e-05, | |
| "loss": 0.6432, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7349443805140008, | |
| "eval_loss": 0.6505680680274963, | |
| "eval_runtime": 68.2746, | |
| "eval_samples_per_second": 38.184, | |
| "eval_steps_per_second": 2.387, | |
| "step": 1467 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.2023026315789474e-05, | |
| "loss": 0.7163, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.1940789473684212e-05, | |
| "loss": 0.5858, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.1858552631578947e-05, | |
| "loss": 0.6909, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.1776315789473684e-05, | |
| "loss": 0.6648, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.1694078947368422e-05, | |
| "loss": 0.6573, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.1611842105263159e-05, | |
| "loss": 0.6664, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.1529605263157894e-05, | |
| "loss": 0.6735, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.1447368421052632e-05, | |
| "loss": 0.6945, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.136513157894737e-05, | |
| "loss": 0.6979, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.1282894736842108e-05, | |
| "loss": 0.7085, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.1200657894736843e-05, | |
| "loss": 0.6379, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.111842105263158e-05, | |
| "loss": 0.6303, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.1036184210526318e-05, | |
| "loss": 0.6218, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.0953947368421053e-05, | |
| "loss": 0.6561, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.087171052631579e-05, | |
| "loss": 0.721, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.0789473684210528e-05, | |
| "loss": 0.6081, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.0707236842105264e-05, | |
| "loss": 0.6595, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.0625e-05, | |
| "loss": 0.7137, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.0542763157894738e-05, | |
| "loss": 0.6845, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.0460526315789474e-05, | |
| "loss": 0.6858, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.037828947368421e-05, | |
| "loss": 0.8064, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.0296052631578948e-05, | |
| "loss": 0.7143, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.0213815789473684e-05, | |
| "loss": 0.5921, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.0131578947368421e-05, | |
| "loss": 0.7618, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.0049342105263158e-05, | |
| "loss": 0.6845, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 9.967105263157895e-06, | |
| "loss": 0.7679, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.884868421052633e-06, | |
| "loss": 0.6691, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 9.80263157894737e-06, | |
| "loss": 0.7407, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.720394736842105e-06, | |
| "loss": 0.6285, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.638157894736843e-06, | |
| "loss": 0.6886, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.55592105263158e-06, | |
| "loss": 0.6379, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.6129, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.391447368421054e-06, | |
| "loss": 0.6227, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.30921052631579e-06, | |
| "loss": 0.6627, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.226973684210527e-06, | |
| "loss": 0.6865, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.144736842105264e-06, | |
| "loss": 0.6789, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 9.0625e-06, | |
| "loss": 0.6753, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.980263157894737e-06, | |
| "loss": 0.6804, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.898026315789475e-06, | |
| "loss": 0.691, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 8.81578947368421e-06, | |
| "loss": 0.747, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 8.733552631578947e-06, | |
| "loss": 0.6473, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 8.651315789473685e-06, | |
| "loss": 0.6243, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 8.569078947368422e-06, | |
| "loss": 0.6559, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 8.486842105263159e-06, | |
| "loss": 0.7307, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 8.404605263157896e-06, | |
| "loss": 0.652, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 8.322368421052632e-06, | |
| "loss": 0.5953, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 8.24013157894737e-06, | |
| "loss": 0.6136, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 8.157894736842106e-06, | |
| "loss": 0.7463, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 8.075657894736842e-06, | |
| "loss": 0.6598, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.99342105263158e-06, | |
| "loss": 0.6828, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.911184210526316e-06, | |
| "loss": 0.6652, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 7.828947368421054e-06, | |
| "loss": 0.7617, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 7.746710526315791e-06, | |
| "loss": 0.7277, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 7.664473684210527e-06, | |
| "loss": 0.6809, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 7.582236842105264e-06, | |
| "loss": 0.7458, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.6488, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.4177631578947374e-06, | |
| "loss": 0.7741, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.335526315789474e-06, | |
| "loss": 0.6211, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.253289473684211e-06, | |
| "loss": 0.623, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.1710526315789475e-06, | |
| "loss": 0.6192, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 7.088815789473686e-06, | |
| "loss": 0.7201, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.006578947368422e-06, | |
| "loss": 0.6198, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 6.924342105263158e-06, | |
| "loss": 0.7168, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 6.842105263157896e-06, | |
| "loss": 0.6454, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 6.759868421052632e-06, | |
| "loss": 0.7973, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 6.677631578947369e-06, | |
| "loss": 0.6116, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.595394736842106e-06, | |
| "loss": 0.6482, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 6.513157894736842e-06, | |
| "loss": 0.6595, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 6.430921052631579e-06, | |
| "loss": 0.6509, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.348684210526316e-06, | |
| "loss": 0.6936, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.266447368421054e-06, | |
| "loss": 0.667, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 6.18421052631579e-06, | |
| "loss": 0.7169, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 6.101973684210527e-06, | |
| "loss": 0.6574, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.019736842105264e-06, | |
| "loss": 0.6945, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7414652857690832, | |
| "eval_loss": 0.6434159874916077, | |
| "eval_runtime": 73.59, | |
| "eval_samples_per_second": 35.426, | |
| "eval_steps_per_second": 2.215, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.9375e-06, | |
| "loss": 0.6835, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.855263157894738e-06, | |
| "loss": 0.7668, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.773026315789474e-06, | |
| "loss": 0.692, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.69078947368421e-06, | |
| "loss": 0.5897, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.608552631578948e-06, | |
| "loss": 0.6242, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.526315789473685e-06, | |
| "loss": 0.6071, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.444078947368422e-06, | |
| "loss": 0.6479, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.3618421052631585e-06, | |
| "loss": 0.647, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.279605263157895e-06, | |
| "loss": 0.6709, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.197368421052632e-06, | |
| "loss": 0.6383, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 5.115131578947369e-06, | |
| "loss": 0.6083, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 5.032894736842106e-06, | |
| "loss": 0.6279, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.950657894736843e-06, | |
| "loss": 0.6156, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 4.8684210526315795e-06, | |
| "loss": 0.6378, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 4.786184210526316e-06, | |
| "loss": 0.6799, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 4.703947368421053e-06, | |
| "loss": 0.585, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 4.6217105263157896e-06, | |
| "loss": 0.6398, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.539473684210527e-06, | |
| "loss": 0.6599, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 4.457236842105264e-06, | |
| "loss": 0.6582, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 4.3750000000000005e-06, | |
| "loss": 0.644, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 4.292763157894737e-06, | |
| "loss": 0.6174, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 4.210526315789474e-06, | |
| "loss": 0.7334, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 4.128289473684211e-06, | |
| "loss": 0.706, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 4.046052631578948e-06, | |
| "loss": 0.6804, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.963815789473684e-06, | |
| "loss": 0.6367, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.8815789473684214e-06, | |
| "loss": 0.5907, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.799342105263158e-06, | |
| "loss": 0.661, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.717105263157895e-06, | |
| "loss": 0.6839, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.634868421052632e-06, | |
| "loss": 0.6822, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.5526315789473687e-06, | |
| "loss": 0.5974, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.4703947368421056e-06, | |
| "loss": 0.6421, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.388157894736843e-06, | |
| "loss": 0.6496, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.3059210526315792e-06, | |
| "loss": 0.6103, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.223684210526316e-06, | |
| "loss": 0.5875, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.141447368421053e-06, | |
| "loss": 0.7646, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.0592105263157897e-06, | |
| "loss": 0.6155, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 2.976973684210527e-06, | |
| "loss": 0.6922, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 2.8947368421052634e-06, | |
| "loss": 0.6599, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 2.8125e-06, | |
| "loss": 0.6667, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 2.730263157894737e-06, | |
| "loss": 0.6462, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.648026315789474e-06, | |
| "loss": 0.6677, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 2.565789473684211e-06, | |
| "loss": 0.6397, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.4835526315789475e-06, | |
| "loss": 0.6546, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 2.4013157894736843e-06, | |
| "loss": 0.6722, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.3190789473684216e-06, | |
| "loss": 0.6415, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 2.236842105263158e-06, | |
| "loss": 0.6646, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.154605263157895e-06, | |
| "loss": 0.6331, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.0723684210526316e-06, | |
| "loss": 0.6023, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.9901315789473684e-06, | |
| "loss": 0.6192, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.9078947368421057e-06, | |
| "loss": 0.6601, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.8256578947368423e-06, | |
| "loss": 0.6502, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.743421052631579e-06, | |
| "loss": 0.6261, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.661184210526316e-06, | |
| "loss": 0.67, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 0.6308, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.4967105263157896e-06, | |
| "loss": 0.6472, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.4144736842105264e-06, | |
| "loss": 0.647, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.3322368421052635e-06, | |
| "loss": 0.6358, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.6464, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.167763157894737e-06, | |
| "loss": 0.7254, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.0855263157894738e-06, | |
| "loss": 0.6862, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.0032894736842106e-06, | |
| "loss": 0.6115, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 9.210526315789474e-07, | |
| "loss": 0.5889, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 8.388157894736842e-07, | |
| "loss": 0.6563, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 7.565789473684211e-07, | |
| "loss": 0.6694, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.74342105263158e-07, | |
| "loss": 0.5824, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 5.921052631578947e-07, | |
| "loss": 0.6639, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 5.098684210526317e-07, | |
| "loss": 0.5845, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 4.276315789473685e-07, | |
| "loss": 0.6039, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 3.453947368421053e-07, | |
| "loss": 0.6627, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 2.6315789473684213e-07, | |
| "loss": 0.6768, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.8092105263157896e-07, | |
| "loss": 0.6723, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 9.868421052631579e-08, | |
| "loss": 0.5798, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.6447368421052633e-08, | |
| "loss": 0.6312, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7468354430379747, | |
| "eval_loss": 0.627052366733551, | |
| "eval_runtime": 68.6526, | |
| "eval_samples_per_second": 37.974, | |
| "eval_steps_per_second": 2.374, | |
| "step": 2932 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 2932, | |
| "total_flos": 2.3314634867043717e+18, | |
| "train_loss": 0.6816266769439104, | |
| "train_runtime": 3103.1992, | |
| "train_samples_per_second": 30.241, | |
| "train_steps_per_second": 0.945 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2932, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 2.3314634867043717e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |