diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,23087 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 10.0, + "global_step": 38220, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 1.1434, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 0.8908, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 0.8327, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 0.7704, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 0.7429, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 0.7416, + "step": 60 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 0.7539, + "step": 70 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 0.6959, + "step": 80 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 0.717, + "step": 90 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 0.6922, + "step": 100 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 0.6958, + "step": 110 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 0.6727, + "step": 120 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 0.6812, + "step": 130 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 0.6782, + "step": 140 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 0.6778, + "step": 150 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 0.6645, + "step": 160 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 0.6627, + "step": 170 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 0.6547, + "step": 180 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 0.6431, + "step": 190 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 0.6539, + "step": 200 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 0.6533, + "step": 210 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 0.6399, + "step": 220 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 0.6305, + "step": 230 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 0.641, + "step": 240 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 0.6429, + "step": 250 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 0.6336, + "step": 260 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 0.6472, + "step": 270 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 0.6366, + "step": 280 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 0.6508, + "step": 290 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 0.5953, + "step": 300 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 0.6264, + "step": 310 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 0.6243, + "step": 320 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 0.6314, + "step": 330 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 0.63, + "step": 340 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 0.6247, + "step": 350 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 0.6204, + "step": 360 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 0.6158, + "step": 370 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 0.6281, + "step": 380 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 0.6149, + "step": 390 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 0.6236, + "step": 400 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 0.6109, + "step": 410 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 0.6176, + "step": 420 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 0.6102, + "step": 430 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.5758, + "step": 440 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.6204, + "step": 450 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.5835, + "step": 460 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.6161, + "step": 470 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.5945, + "step": 480 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.6199, + "step": 490 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.5952, + "step": 500 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.6038, + "step": 510 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.6137, + "step": 520 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.6131, + "step": 530 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.6225, + "step": 540 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.5931, + "step": 550 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.5881, + "step": 560 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.6024, + "step": 570 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.5813, + "step": 580 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.5926, + "step": 590 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.5862, + "step": 600 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.5937, + "step": 610 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.5938, + "step": 620 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.5907, + "step": 630 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 0.6023, + "step": 640 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 0.5997, + "step": 650 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 0.5643, + "step": 660 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 0.5926, + "step": 670 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 0.5769, + "step": 680 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 0.58, + "step": 690 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 0.5785, + "step": 700 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 0.5786, + "step": 710 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 0.5746, + "step": 720 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 0.5615, + "step": 730 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 0.5845, + "step": 740 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 0.5733, + "step": 750 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 0.5898, + "step": 760 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 0.569, + "step": 770 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 0.5757, + "step": 780 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 0.5662, + "step": 790 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 0.5863, + "step": 800 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 0.5678, + "step": 810 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 0.5786, + "step": 820 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 0.5736, + "step": 830 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 0.5735, + "step": 840 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 0.5615, + "step": 850 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 0.5566, + "step": 860 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 0.57, + "step": 870 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 0.583, + "step": 880 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 0.5508, + "step": 890 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 0.5479, + "step": 900 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 0.566, + "step": 910 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 0.5485, + "step": 920 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 0.5581, + "step": 930 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 0.5456, + "step": 940 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 0.5611, + "step": 950 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 0.5506, + "step": 960 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 0.5448, + "step": 970 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 0.5459, + "step": 980 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 0.5656, + "step": 990 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 0.5475, + "step": 1000 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 0.5554, + "step": 1010 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 0.558, + "step": 1020 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 0.5499, + "step": 1030 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 0.5801, + "step": 1040 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 0.5584, + "step": 1050 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 0.548, + "step": 1060 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 0.5569, + "step": 1070 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 0.5675, + "step": 1080 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 0.5517, + "step": 1090 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 0.5558, + "step": 1100 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 0.5439, + "step": 1110 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 0.5408, + "step": 1120 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 0.5521, + "step": 1130 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 0.5353, + "step": 1140 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 0.5557, + "step": 1150 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 0.5417, + "step": 1160 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 0.5469, + "step": 1170 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 0.5598, + "step": 1180 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 0.5532, + "step": 1190 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 0.5472, + "step": 1200 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 0.5612, + "step": 1210 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 0.5538, + "step": 1220 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 0.5261, + "step": 1230 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 0.5541, + "step": 1240 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 0.5437, + "step": 1250 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 0.5529, + "step": 1260 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 0.5195, + "step": 1270 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 0.5146, + "step": 1280 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 0.5482, + "step": 1290 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 0.5265, + "step": 1300 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 0.5425, + "step": 1310 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 0.5371, + "step": 1320 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 0.5269, + "step": 1330 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 0.5533, + "step": 1340 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 0.5546, + "step": 1350 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 0.5309, + "step": 1360 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 0.5401, + "step": 1370 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 0.5401, + "step": 1380 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 0.55, + "step": 1390 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 0.5387, + "step": 1400 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 0.5323, + "step": 1410 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 0.5497, + "step": 1420 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 0.5156, + "step": 1430 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 0.5382, + "step": 1440 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 0.5191, + "step": 1450 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 0.5426, + "step": 1460 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 0.5397, + "step": 1470 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 0.5491, + "step": 1480 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 0.5027, + "step": 1490 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 0.5234, + "step": 1500 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.5266, + "step": 1510 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.5152, + "step": 1520 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.5337, + "step": 1530 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.5349, + "step": 1540 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 0.5246, + "step": 1550 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 0.5321, + "step": 1560 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 0.5329, + "step": 1570 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 0.5207, + "step": 1580 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 0.5244, + "step": 1590 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 0.5304, + "step": 1600 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 0.5301, + "step": 1610 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 0.5186, + "step": 1620 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 0.5281, + "step": 1630 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 0.5052, + "step": 1640 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 0.5392, + "step": 1650 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 0.5013, + "step": 1660 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 0.5225, + "step": 1670 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 0.5272, + "step": 1680 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 0.5314, + "step": 1690 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 0.5229, + "step": 1700 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 0.5225, + "step": 1710 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 0.5309, + "step": 1720 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 0.5161, + "step": 1730 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 0.522, + "step": 1740 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 0.5073, + "step": 1750 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 0.5263, + "step": 1760 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 0.5304, + "step": 1770 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 0.5123, + "step": 1780 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 0.5021, + "step": 1790 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 0.5059, + "step": 1800 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 0.5107, + "step": 1810 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 0.5251, + "step": 1820 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 0.5168, + "step": 1830 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 0.5277, + "step": 1840 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 0.5081, + "step": 1850 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 0.5214, + "step": 1860 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 0.5021, + "step": 1870 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 0.5203, + "step": 1880 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 0.5128, + "step": 1890 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.5114, + "step": 1900 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.5153, + "step": 1910 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.5281, + "step": 1920 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.5084, + "step": 1930 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.5231, + "step": 1940 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.5275, + "step": 1950 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.5089, + "step": 1960 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 0.5043, + "step": 1970 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 0.5334, + "step": 1980 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 0.5069, + "step": 1990 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 0.4935, + "step": 2000 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 0.5162, + "step": 2010 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 0.497, + "step": 2020 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 0.5046, + "step": 2030 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 0.5077, + "step": 2040 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 0.5115, + "step": 2050 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 0.5044, + "step": 2060 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 0.5149, + "step": 2070 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 0.5136, + "step": 2080 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 0.5061, + "step": 2090 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 0.4954, + "step": 2100 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 0.4877, + "step": 2110 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 0.5132, + "step": 2120 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 0.4977, + "step": 2130 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 0.5227, + "step": 2140 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 0.5082, + "step": 2150 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 0.4965, + "step": 2160 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 0.5031, + "step": 2170 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 0.4987, + "step": 2180 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 0.5033, + "step": 2190 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 0.5073, + "step": 2200 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 0.4928, + "step": 2210 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 0.5123, + "step": 2220 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 0.5061, + "step": 2230 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 0.5268, + "step": 2240 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 0.5077, + "step": 2250 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 0.5175, + "step": 2260 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 0.5111, + "step": 2270 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 0.4874, + "step": 2280 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 0.4912, + "step": 2290 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 0.4988, + "step": 2300 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 0.5128, + "step": 2310 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 0.4889, + "step": 2320 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 0.5126, + "step": 2330 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 0.5181, + "step": 2340 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 0.5024, + "step": 2350 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 0.4772, + "step": 2360 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 0.4859, + "step": 2370 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 0.4898, + "step": 2380 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.4932, + "step": 2390 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.4989, + "step": 2400 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.4742, + "step": 2410 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.4748, + "step": 2420 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 0.4832, + "step": 2430 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 0.4952, + "step": 2440 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 0.4856, + "step": 2450 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 0.5012, + "step": 2460 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 0.4744, + "step": 2470 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 0.484, + "step": 2480 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 0.4986, + "step": 2490 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 0.5023, + "step": 2500 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 0.4928, + "step": 2510 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 0.4759, + "step": 2520 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 0.481, + "step": 2530 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 0.4839, + "step": 2540 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 0.4933, + "step": 2550 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 0.487, + "step": 2560 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 0.4854, + "step": 2570 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 0.484, + "step": 2580 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 0.4953, + "step": 2590 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 0.4903, + "step": 2600 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 0.4859, + "step": 2610 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 0.4773, + "step": 2620 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 0.4698, + "step": 2630 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 0.4852, + "step": 2640 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 0.4877, + "step": 2650 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 0.4847, + "step": 2660 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 0.4869, + "step": 2670 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 0.4907, + "step": 2680 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 0.4868, + "step": 2690 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 0.5049, + "step": 2700 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 0.4849, + "step": 2710 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 0.5119, + "step": 2720 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 0.472, + "step": 2730 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 0.4899, + "step": 2740 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 0.4698, + "step": 2750 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 0.4762, + "step": 2760 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 0.4887, + "step": 2770 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 0.4669, + "step": 2780 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 0.4763, + "step": 2790 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 0.4957, + "step": 2800 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 0.4749, + "step": 2810 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 0.4834, + "step": 2820 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 0.4917, + "step": 2830 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 0.4748, + "step": 2840 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.4945, + "step": 2850 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.4697, + "step": 2860 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.5015, + "step": 2870 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.4772, + "step": 2880 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 0.4802, + "step": 2890 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 0.4793, + "step": 2900 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 0.4859, + "step": 2910 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 0.4735, + "step": 2920 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 0.4811, + "step": 2930 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 0.4831, + "step": 2940 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 0.4912, + "step": 2950 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 0.4875, + "step": 2960 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 0.4985, + "step": 2970 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 0.4672, + "step": 2980 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 0.5021, + "step": 2990 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 0.4897, + "step": 3000 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 0.4894, + "step": 3010 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 0.4749, + "step": 3020 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 0.4813, + "step": 3030 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 0.4907, + "step": 3040 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 0.4703, + "step": 3050 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 0.4744, + "step": 3060 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 0.4964, + "step": 3070 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 0.4903, + "step": 3080 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 0.4877, + "step": 3090 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 0.4853, + "step": 3100 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 0.4642, + "step": 3110 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.4865, + "step": 3120 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.4813, + "step": 3130 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.4703, + "step": 3140 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.47, + "step": 3150 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 0.4945, + "step": 3160 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 0.4962, + "step": 3170 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 0.4696, + "step": 3180 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 0.4739, + "step": 3190 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.4685, + "step": 3200 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.4813, + "step": 3210 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.4715, + "step": 3220 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 0.4813, + "step": 3230 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 0.4974, + "step": 3240 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 0.4853, + "step": 3250 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 0.4941, + "step": 3260 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 0.4836, + "step": 3270 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 0.4864, + "step": 3280 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 0.4929, + "step": 3290 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 0.4796, + "step": 3300 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 0.4706, + "step": 3310 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 0.4914, + "step": 3320 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 0.4896, + "step": 3330 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 0.4809, + "step": 3340 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 0.4722, + "step": 3350 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 0.4719, + "step": 3360 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 0.4969, + "step": 3370 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 0.4601, + "step": 3380 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 0.4701, + "step": 3390 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 0.4617, + "step": 3400 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 0.474, + "step": 3410 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 0.4869, + "step": 3420 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 0.4786, + "step": 3430 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 0.4797, + "step": 3440 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 0.4648, + "step": 3450 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 0.4491, + "step": 3460 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 0.4552, + "step": 3470 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 0.4719, + "step": 3480 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 0.4575, + "step": 3490 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 0.4709, + "step": 3500 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 0.4659, + "step": 3510 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 0.4591, + "step": 3520 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 0.4831, + "step": 3530 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 0.471, + "step": 3540 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 0.4824, + "step": 3550 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 0.457, + "step": 3560 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 0.4767, + "step": 3570 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 0.4699, + "step": 3580 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 0.4844, + "step": 3590 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 0.4612, + "step": 3600 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 0.4729, + "step": 3610 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 0.4596, + "step": 3620 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 0.4717, + "step": 3630 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 0.4783, + "step": 3640 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 0.4744, + "step": 3650 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 0.4667, + "step": 3660 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 0.4781, + "step": 3670 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 0.4752, + "step": 3680 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 0.4787, + "step": 3690 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 0.479, + "step": 3700 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 0.4841, + "step": 3710 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 0.475, + "step": 3720 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 0.4753, + "step": 3730 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 0.4714, + "step": 3740 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 0.4599, + "step": 3750 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 0.4893, + "step": 3760 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 0.4735, + "step": 3770 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 0.48, + "step": 3780 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 0.4659, + "step": 3790 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 0.4674, + "step": 3800 + }, + { + "epoch": 1.0, + "learning_rate": 0.0001, + "loss": 0.481, + "step": 3810 + }, + { + "epoch": 1.0, + "learning_rate": 0.0001, + "loss": 0.462, + "step": 3820 + }, + { + "epoch": 1.0, + "eval_gen_len": 61.6439, + "eval_loss": 0.9571423530578613, + "eval_rouge1": 56.6561, + "eval_rouge2": 34.7459, + "eval_rougeL": 46.3674, + "eval_rougeLsum": 52.2274, + "eval_runtime": 38.1257, + "eval_samples_per_second": 23.055, + "eval_steps_per_second": 0.367, + "step": 3822 + }, + { + "epoch": 1.0, + "learning_rate": 0.0001, + "loss": 0.4131, + "step": 3830 + }, + { + "epoch": 1.0, + "learning_rate": 0.0001, + "loss": 0.4163, + "step": 3840 + }, + { + "epoch": 1.01, + "learning_rate": 0.0001, + "loss": 0.4135, + "step": 3850 + }, + { + "epoch": 1.01, + "learning_rate": 0.0001, + "loss": 0.4011, + "step": 3860 + }, + { + "epoch": 1.01, + "learning_rate": 0.0001, + "loss": 0.4087, + "step": 3870 + }, + { + "epoch": 1.02, + "learning_rate": 0.0001, + "loss": 0.428, + "step": 3880 + }, + { + "epoch": 1.02, + "learning_rate": 0.0001, + "loss": 0.4129, + "step": 3890 + }, + { + "epoch": 1.02, + "learning_rate": 0.0001, + "loss": 0.418, + "step": 3900 + }, + { + "epoch": 1.02, + "learning_rate": 0.0001, + "loss": 0.405, + "step": 3910 + }, + { + "epoch": 1.03, + "learning_rate": 0.0001, + "loss": 0.4091, + "step": 3920 + }, + { + "epoch": 1.03, + "learning_rate": 0.0001, + "loss": 0.433, + "step": 3930 + }, + { + "epoch": 1.03, + "learning_rate": 0.0001, + "loss": 0.4172, + "step": 3940 + }, + { + "epoch": 1.03, + "learning_rate": 0.0001, + "loss": 0.4191, + "step": 3950 + }, + { + "epoch": 1.04, + "learning_rate": 0.0001, + "loss": 0.4137, + "step": 3960 + }, + { + "epoch": 1.04, + "learning_rate": 0.0001, + "loss": 0.4016, + "step": 3970 + }, + { + "epoch": 1.04, + "learning_rate": 0.0001, + "loss": 0.4187, + "step": 3980 + }, + { + "epoch": 1.04, + "learning_rate": 0.0001, + "loss": 0.4292, + "step": 3990 + }, + { + "epoch": 1.05, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 4000 + }, + { + "epoch": 1.05, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 4010 + }, + { + "epoch": 1.05, + "learning_rate": 0.0001, + "loss": 0.4166, + "step": 4020 + }, + { + "epoch": 1.05, + "learning_rate": 0.0001, + "loss": 0.4099, + "step": 4030 + }, + { + "epoch": 1.06, + "learning_rate": 0.0001, + "loss": 0.4172, + "step": 4040 + }, + { + "epoch": 1.06, + "learning_rate": 0.0001, + "loss": 0.4188, + "step": 4050 + }, + { + "epoch": 1.06, + "learning_rate": 0.0001, + "loss": 0.4166, + "step": 4060 + }, + { + "epoch": 1.06, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 4070 + }, + { + "epoch": 1.07, + "learning_rate": 0.0001, + "loss": 0.4068, + "step": 4080 + }, + { + "epoch": 1.07, + "learning_rate": 0.0001, + "loss": 0.4129, + "step": 4090 + }, + { + "epoch": 1.07, + "learning_rate": 0.0001, + "loss": 0.4072, + "step": 4100 + }, + { + "epoch": 1.08, + "learning_rate": 0.0001, + "loss": 0.4062, + "step": 4110 + }, + { + "epoch": 1.08, + "learning_rate": 0.0001, + "loss": 0.4213, + "step": 4120 + }, + { + "epoch": 1.08, + "learning_rate": 0.0001, + "loss": 0.419, + "step": 4130 + }, + { + "epoch": 1.08, + "learning_rate": 0.0001, + "loss": 0.4119, + "step": 4140 + }, + { + "epoch": 1.09, + "learning_rate": 0.0001, + "loss": 0.4065, + "step": 4150 + }, + { + "epoch": 1.09, + "learning_rate": 0.0001, + "loss": 0.4145, + "step": 4160 + }, + { + "epoch": 1.09, + "learning_rate": 0.0001, + "loss": 0.4157, + "step": 4170 + }, + { + "epoch": 1.09, + "learning_rate": 0.0001, + "loss": 0.4069, + "step": 4180 + }, + { + "epoch": 1.1, + "learning_rate": 0.0001, + "loss": 0.411, + "step": 4190 + }, + { + "epoch": 1.1, + "learning_rate": 0.0001, + "loss": 0.4293, + "step": 4200 + }, + { + "epoch": 1.1, + "learning_rate": 0.0001, + "loss": 0.4142, + "step": 4210 + }, + { + "epoch": 1.1, + "learning_rate": 0.0001, + "loss": 0.4278, + "step": 4220 + }, + { + "epoch": 1.11, + "learning_rate": 0.0001, + "loss": 0.4314, + "step": 4230 + }, + { + "epoch": 1.11, + "learning_rate": 0.0001, + "loss": 0.4011, + "step": 4240 + }, + { + "epoch": 1.11, + "learning_rate": 0.0001, + "loss": 0.409, + "step": 4250 + }, + { + "epoch": 1.11, + "learning_rate": 0.0001, + "loss": 0.4311, + "step": 4260 + }, + { + "epoch": 1.12, + "learning_rate": 0.0001, + "loss": 0.424, + "step": 4270 + }, + { + "epoch": 1.12, + "learning_rate": 0.0001, + "loss": 0.4115, + "step": 4280 + }, + { + "epoch": 1.12, + "learning_rate": 0.0001, + "loss": 0.414, + "step": 4290 + }, + { + "epoch": 1.13, + "learning_rate": 0.0001, + "loss": 0.4187, + "step": 4300 + }, + { + "epoch": 1.13, + "learning_rate": 0.0001, + "loss": 0.4072, + "step": 4310 + }, + { + "epoch": 1.13, + "learning_rate": 0.0001, + "loss": 0.419, + "step": 4320 + }, + { + "epoch": 1.13, + "learning_rate": 0.0001, + "loss": 0.4205, + "step": 4330 + }, + { + "epoch": 1.14, + "learning_rate": 0.0001, + "loss": 0.4183, + "step": 4340 + }, + { + "epoch": 1.14, + "learning_rate": 0.0001, + "loss": 0.4114, + "step": 4350 + }, + { + "epoch": 1.14, + "learning_rate": 0.0001, + "loss": 0.4069, + "step": 4360 + }, + { + "epoch": 1.14, + "learning_rate": 0.0001, + "loss": 0.4039, + "step": 4370 + }, + { + "epoch": 1.15, + "learning_rate": 0.0001, + "loss": 0.4141, + "step": 4380 + }, + { + "epoch": 1.15, + "learning_rate": 0.0001, + "loss": 0.417, + "step": 4390 + }, + { + "epoch": 1.15, + "learning_rate": 0.0001, + "loss": 0.4295, + "step": 4400 + }, + { + "epoch": 1.15, + "learning_rate": 0.0001, + "loss": 0.404, + "step": 4410 + }, + { + "epoch": 1.16, + "learning_rate": 0.0001, + "loss": 0.4171, + "step": 4420 + }, + { + "epoch": 1.16, + "learning_rate": 0.0001, + "loss": 0.4156, + "step": 4430 + }, + { + "epoch": 1.16, + "learning_rate": 0.0001, + "loss": 0.4239, + "step": 4440 + }, + { + "epoch": 1.16, + "learning_rate": 0.0001, + "loss": 0.4209, + "step": 4450 + }, + { + "epoch": 1.17, + "learning_rate": 0.0001, + "loss": 0.4051, + "step": 4460 + }, + { + "epoch": 1.17, + "learning_rate": 0.0001, + "loss": 0.4237, + "step": 4470 + }, + { + "epoch": 1.17, + "learning_rate": 0.0001, + "loss": 0.4256, + "step": 4480 + }, + { + "epoch": 1.17, + "learning_rate": 0.0001, + "loss": 0.4109, + "step": 4490 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001, + "loss": 0.4026, + "step": 4500 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001, + "loss": 0.4155, + "step": 4510 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001, + "loss": 0.3901, + "step": 4520 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001, + "loss": 0.432, + "step": 4530 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001, + "loss": 0.4097, + "step": 4540 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001, + "loss": 0.4061, + "step": 4550 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001, + "loss": 0.3863, + "step": 4560 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001, + "loss": 0.4004, + "step": 4570 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001, + "loss": 0.4097, + "step": 4580 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001, + "loss": 0.4018, + "step": 4590 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001, + "loss": 0.4208, + "step": 4600 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001, + "loss": 0.4115, + "step": 4610 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001, + "loss": 0.4255, + "step": 4620 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001, + "loss": 0.4097, + "step": 4630 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001, + "loss": 0.4039, + "step": 4640 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001, + "loss": 0.4028, + "step": 4650 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001, + "loss": 0.415, + "step": 4660 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001, + "loss": 0.4195, + "step": 4670 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001, + "loss": 0.4001, + "step": 4680 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001, + "loss": 0.4223, + "step": 4690 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001, + "loss": 0.4231, + "step": 4700 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001, + "loss": 0.4207, + "step": 4710 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001, + "loss": 0.423, + "step": 4720 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001, + "loss": 0.4104, + "step": 4730 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001, + "loss": 0.411, + "step": 4740 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001, + "loss": 0.4081, + "step": 4750 + }, + { + "epoch": 1.25, + "learning_rate": 0.0001, + "loss": 0.4169, + "step": 4760 + }, + { + "epoch": 1.25, + "learning_rate": 0.0001, + "loss": 0.4118, + "step": 4770 + }, + { + "epoch": 1.25, + "learning_rate": 0.0001, + "loss": 0.4235, + "step": 4780 + }, + { + "epoch": 1.25, + "learning_rate": 0.0001, + "loss": 0.4034, + "step": 4790 + }, + { + "epoch": 1.26, + "learning_rate": 0.0001, + "loss": 0.4013, + "step": 4800 + }, + { + "epoch": 1.26, + "learning_rate": 0.0001, + "loss": 0.4304, + "step": 4810 + }, + { + "epoch": 1.26, + "learning_rate": 0.0001, + "loss": 0.4179, + "step": 4820 + }, + { + "epoch": 1.26, + "learning_rate": 0.0001, + "loss": 0.4164, + "step": 4830 + }, + { + "epoch": 1.27, + "learning_rate": 0.0001, + "loss": 0.3926, + "step": 4840 + }, + { + "epoch": 1.27, + "learning_rate": 0.0001, + "loss": 0.422, + "step": 4850 + }, + { + "epoch": 1.27, + "learning_rate": 0.0001, + "loss": 0.4105, + "step": 4860 + }, + { + "epoch": 1.27, + "learning_rate": 0.0001, + "loss": 0.4059, + "step": 4870 + }, + { + "epoch": 1.28, + "learning_rate": 0.0001, + "loss": 0.4179, + "step": 4880 + }, + { + "epoch": 1.28, + "learning_rate": 0.0001, + "loss": 0.4058, + "step": 4890 + }, + { + "epoch": 1.28, + "learning_rate": 0.0001, + "loss": 0.4104, + "step": 4900 + }, + { + "epoch": 1.28, + "learning_rate": 0.0001, + "loss": 0.4032, + "step": 4910 + }, + { + "epoch": 1.29, + "learning_rate": 0.0001, + "loss": 0.4113, + "step": 4920 + }, + { + "epoch": 1.29, + "learning_rate": 0.0001, + "loss": 0.4036, + "step": 4930 + }, + { + "epoch": 1.29, + "learning_rate": 0.0001, + "loss": 0.4088, + "step": 4940 + }, + { + "epoch": 1.3, + "learning_rate": 0.0001, + "loss": 0.411, + "step": 4950 + }, + { + "epoch": 1.3, + "learning_rate": 0.0001, + "loss": 0.408, + "step": 4960 + }, + { + "epoch": 1.3, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 4970 + }, + { + "epoch": 1.3, + "learning_rate": 0.0001, + "loss": 0.4184, + "step": 4980 + }, + { + "epoch": 1.31, + "learning_rate": 0.0001, + "loss": 0.4034, + "step": 4990 + }, + { + "epoch": 1.31, + "learning_rate": 0.0001, + "loss": 0.3948, + "step": 5000 + }, + { + "epoch": 1.31, + "learning_rate": 0.0001, + "loss": 0.4047, + "step": 5010 + }, + { + "epoch": 1.31, + "learning_rate": 0.0001, + "loss": 0.4008, + "step": 5020 + }, + { + "epoch": 1.32, + "learning_rate": 0.0001, + "loss": 0.397, + "step": 5030 + }, + { + "epoch": 1.32, + "learning_rate": 0.0001, + "loss": 0.4092, + "step": 5040 + }, + { + "epoch": 1.32, + "learning_rate": 0.0001, + "loss": 0.4008, + "step": 5050 + }, + { + "epoch": 1.32, + "learning_rate": 0.0001, + "loss": 0.4055, + "step": 5060 + }, + { + "epoch": 1.33, + "learning_rate": 0.0001, + "loss": 0.4115, + "step": 5070 + }, + { + "epoch": 1.33, + "learning_rate": 0.0001, + "loss": 0.4098, + "step": 5080 + }, + { + "epoch": 1.33, + "learning_rate": 0.0001, + "loss": 0.4167, + "step": 5090 + }, + { + "epoch": 1.33, + "learning_rate": 0.0001, + "loss": 0.4099, + "step": 5100 + }, + { + "epoch": 1.34, + "learning_rate": 0.0001, + "loss": 0.4091, + "step": 5110 + }, + { + "epoch": 1.34, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 5120 + }, + { + "epoch": 1.34, + "learning_rate": 0.0001, + "loss": 0.3934, + "step": 5130 + }, + { + "epoch": 1.34, + "learning_rate": 0.0001, + "loss": 0.4039, + "step": 5140 + }, + { + "epoch": 1.35, + "learning_rate": 0.0001, + "loss": 0.4398, + "step": 5150 + }, + { + "epoch": 1.35, + "learning_rate": 0.0001, + "loss": 0.4046, + "step": 5160 + }, + { + "epoch": 1.35, + "learning_rate": 0.0001, + "loss": 0.4057, + "step": 5170 + }, + { + "epoch": 1.36, + "learning_rate": 0.0001, + "loss": 0.4086, + "step": 5180 + }, + { + "epoch": 1.36, + "learning_rate": 0.0001, + "loss": 0.4219, + "step": 5190 + }, + { + "epoch": 1.36, + "learning_rate": 0.0001, + "loss": 0.4096, + "step": 5200 + }, + { + "epoch": 1.36, + "learning_rate": 0.0001, + "loss": 0.399, + "step": 5210 + }, + { + "epoch": 1.37, + "learning_rate": 0.0001, + "loss": 0.403, + "step": 5220 + }, + { + "epoch": 1.37, + "learning_rate": 0.0001, + "loss": 0.4167, + "step": 5230 + }, + { + "epoch": 1.37, + "learning_rate": 0.0001, + "loss": 0.4144, + "step": 5240 + }, + { + "epoch": 1.37, + "learning_rate": 0.0001, + "loss": 0.3898, + "step": 5250 + }, + { + "epoch": 1.38, + "learning_rate": 0.0001, + "loss": 0.4055, + "step": 5260 + }, + { + "epoch": 1.38, + "learning_rate": 0.0001, + "loss": 0.4049, + "step": 5270 + }, + { + "epoch": 1.38, + "learning_rate": 0.0001, + "loss": 0.3966, + "step": 5280 + }, + { + "epoch": 1.38, + "learning_rate": 0.0001, + "loss": 0.4145, + "step": 5290 + }, + { + "epoch": 1.39, + "learning_rate": 0.0001, + "loss": 0.3974, + "step": 5300 + }, + { + "epoch": 1.39, + "learning_rate": 0.0001, + "loss": 0.415, + "step": 5310 + }, + { + "epoch": 1.39, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 5320 + }, + { + "epoch": 1.39, + "learning_rate": 0.0001, + "loss": 0.3929, + "step": 5330 + }, + { + "epoch": 1.4, + "learning_rate": 0.0001, + "loss": 0.4114, + "step": 5340 + }, + { + "epoch": 1.4, + "learning_rate": 0.0001, + "loss": 0.4075, + "step": 5350 + }, + { + "epoch": 1.4, + "learning_rate": 0.0001, + "loss": 0.3995, + "step": 5360 + }, + { + "epoch": 1.41, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 5370 + }, + { + "epoch": 1.41, + "learning_rate": 0.0001, + "loss": 0.4102, + "step": 5380 + }, + { + "epoch": 1.41, + "learning_rate": 0.0001, + "loss": 0.4101, + "step": 5390 + }, + { + "epoch": 1.41, + "learning_rate": 0.0001, + "loss": 0.3963, + "step": 5400 + }, + { + "epoch": 1.42, + "learning_rate": 0.0001, + "loss": 0.4095, + "step": 5410 + }, + { + "epoch": 1.42, + "learning_rate": 0.0001, + "loss": 0.4144, + "step": 5420 + }, + { + "epoch": 1.42, + "learning_rate": 0.0001, + "loss": 0.421, + "step": 5430 + }, + { + "epoch": 1.42, + "learning_rate": 0.0001, + "loss": 0.4218, + "step": 5440 + }, + { + "epoch": 1.43, + "learning_rate": 0.0001, + "loss": 0.4077, + "step": 5450 + }, + { + "epoch": 1.43, + "learning_rate": 0.0001, + "loss": 0.4094, + "step": 5460 + }, + { + "epoch": 1.43, + "learning_rate": 0.0001, + "loss": 0.4242, + "step": 5470 + }, + { + "epoch": 1.43, + "learning_rate": 0.0001, + "loss": 0.4223, + "step": 5480 + }, + { + "epoch": 1.44, + "learning_rate": 0.0001, + "loss": 0.3987, + "step": 5490 + }, + { + "epoch": 1.44, + "learning_rate": 0.0001, + "loss": 0.4041, + "step": 5500 + }, + { + "epoch": 1.44, + "learning_rate": 0.0001, + "loss": 0.4023, + "step": 5510 + }, + { + "epoch": 1.44, + "learning_rate": 0.0001, + "loss": 0.4019, + "step": 5520 + }, + { + "epoch": 1.45, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 5530 + }, + { + "epoch": 1.45, + "learning_rate": 0.0001, + "loss": 0.3957, + "step": 5540 + }, + { + "epoch": 1.45, + "learning_rate": 0.0001, + "loss": 0.4047, + "step": 5550 + }, + { + "epoch": 1.45, + "learning_rate": 0.0001, + "loss": 0.4112, + "step": 5560 + }, + { + "epoch": 1.46, + "learning_rate": 0.0001, + "loss": 0.4039, + "step": 5570 + }, + { + "epoch": 1.46, + "learning_rate": 0.0001, + "loss": 0.4098, + "step": 5580 + }, + { + "epoch": 1.46, + "learning_rate": 0.0001, + "loss": 0.4144, + "step": 5590 + }, + { + "epoch": 1.47, + "learning_rate": 0.0001, + "loss": 0.4191, + "step": 5600 + }, + { + "epoch": 1.47, + "learning_rate": 0.0001, + "loss": 0.4043, + "step": 5610 + }, + { + "epoch": 1.47, + "learning_rate": 0.0001, + "loss": 0.4087, + "step": 5620 + }, + { + "epoch": 1.47, + "learning_rate": 0.0001, + "loss": 0.4058, + "step": 5630 + }, + { + "epoch": 1.48, + "learning_rate": 0.0001, + "loss": 0.4102, + "step": 5640 + }, + { + "epoch": 1.48, + "learning_rate": 0.0001, + "loss": 0.4022, + "step": 5650 + }, + { + "epoch": 1.48, + "learning_rate": 0.0001, + "loss": 0.4093, + "step": 5660 + }, + { + "epoch": 1.48, + "learning_rate": 0.0001, + "loss": 0.3904, + "step": 5670 + }, + { + "epoch": 1.49, + "learning_rate": 0.0001, + "loss": 0.4018, + "step": 5680 + }, + { + "epoch": 1.49, + "learning_rate": 0.0001, + "loss": 0.4038, + "step": 5690 + }, + { + "epoch": 1.49, + "learning_rate": 0.0001, + "loss": 0.4053, + "step": 5700 + }, + { + "epoch": 1.49, + "learning_rate": 0.0001, + "loss": 0.3861, + "step": 5710 + }, + { + "epoch": 1.5, + "learning_rate": 0.0001, + "loss": 0.4146, + "step": 5720 + }, + { + "epoch": 1.5, + "learning_rate": 0.0001, + "loss": 0.4161, + "step": 5730 + }, + { + "epoch": 1.5, + "learning_rate": 0.0001, + "loss": 0.4095, + "step": 5740 + }, + { + "epoch": 1.5, + "learning_rate": 0.0001, + "loss": 0.4186, + "step": 5750 + }, + { + "epoch": 1.51, + "learning_rate": 0.0001, + "loss": 0.4104, + "step": 5760 + }, + { + "epoch": 1.51, + "learning_rate": 0.0001, + "loss": 0.4057, + "step": 5770 + }, + { + "epoch": 1.51, + "learning_rate": 0.0001, + "loss": 0.4066, + "step": 5780 + }, + { + "epoch": 1.51, + "learning_rate": 0.0001, + "loss": 0.4128, + "step": 5790 + }, + { + "epoch": 1.52, + "learning_rate": 0.0001, + "loss": 0.3914, + "step": 5800 + }, + { + "epoch": 1.52, + "learning_rate": 0.0001, + "loss": 0.3961, + "step": 5810 + }, + { + "epoch": 1.52, + "learning_rate": 0.0001, + "loss": 0.4355, + "step": 5820 + }, + { + "epoch": 1.53, + "learning_rate": 0.0001, + "loss": 0.4003, + "step": 5830 + }, + { + "epoch": 1.53, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 5840 + }, + { + "epoch": 1.53, + "learning_rate": 0.0001, + "loss": 0.3971, + "step": 5850 + }, + { + "epoch": 1.53, + "learning_rate": 0.0001, + "loss": 0.4147, + "step": 5860 + }, + { + "epoch": 1.54, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 5870 + }, + { + "epoch": 1.54, + "learning_rate": 0.0001, + "loss": 0.411, + "step": 5880 + }, + { + "epoch": 1.54, + "learning_rate": 0.0001, + "loss": 0.3975, + "step": 5890 + }, + { + "epoch": 1.54, + "learning_rate": 0.0001, + "loss": 0.4199, + "step": 5900 + }, + { + "epoch": 1.55, + "learning_rate": 0.0001, + "loss": 0.4121, + "step": 5910 + }, + { + "epoch": 1.55, + "learning_rate": 0.0001, + "loss": 0.391, + "step": 5920 + }, + { + "epoch": 1.55, + "learning_rate": 0.0001, + "loss": 0.4131, + "step": 5930 + }, + { + "epoch": 1.55, + "learning_rate": 0.0001, + "loss": 0.3964, + "step": 5940 + }, + { + "epoch": 1.56, + "learning_rate": 0.0001, + "loss": 0.4106, + "step": 5950 + }, + { + "epoch": 1.56, + "learning_rate": 0.0001, + "loss": 0.4079, + "step": 5960 + }, + { + "epoch": 1.56, + "learning_rate": 0.0001, + "loss": 0.4001, + "step": 5970 + }, + { + "epoch": 1.56, + "learning_rate": 0.0001, + "loss": 0.4262, + "step": 5980 + }, + { + "epoch": 1.57, + "learning_rate": 0.0001, + "loss": 0.3974, + "step": 5990 + }, + { + "epoch": 1.57, + "learning_rate": 0.0001, + "loss": 0.3955, + "step": 6000 + }, + { + "epoch": 1.57, + "learning_rate": 0.0001, + "loss": 0.4025, + "step": 6010 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001, + "loss": 0.3967, + "step": 6020 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001, + "loss": 0.4082, + "step": 6030 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001, + "loss": 0.4053, + "step": 6040 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001, + "loss": 0.4156, + "step": 6050 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001, + "loss": 0.4261, + "step": 6060 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001, + "loss": 0.3944, + "step": 6070 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001, + "loss": 0.4089, + "step": 6080 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001, + "loss": 0.419, + "step": 6090 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001, + "loss": 0.423, + "step": 6100 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001, + "loss": 0.4005, + "step": 6110 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001, + "loss": 0.4053, + "step": 6120 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001, + "loss": 0.3933, + "step": 6130 + }, + { + "epoch": 1.61, + "learning_rate": 0.0001, + "loss": 0.4126, + "step": 6140 + }, + { + "epoch": 1.61, + "learning_rate": 0.0001, + "loss": 0.4026, + "step": 6150 + }, + { + "epoch": 1.61, + "learning_rate": 0.0001, + "loss": 0.4225, + "step": 6160 + }, + { + "epoch": 1.61, + "learning_rate": 0.0001, + "loss": 0.401, + "step": 6170 + }, + { + "epoch": 1.62, + "learning_rate": 0.0001, + "loss": 0.4161, + "step": 6180 + }, + { + "epoch": 1.62, + "learning_rate": 0.0001, + "loss": 0.4045, + "step": 6190 + }, + { + "epoch": 1.62, + "learning_rate": 0.0001, + "loss": 0.4159, + "step": 6200 + }, + { + "epoch": 1.62, + "learning_rate": 0.0001, + "loss": 0.4123, + "step": 6210 + }, + { + "epoch": 1.63, + "learning_rate": 0.0001, + "loss": 0.4277, + "step": 6220 + }, + { + "epoch": 1.63, + "learning_rate": 0.0001, + "loss": 0.4205, + "step": 6230 + }, + { + "epoch": 1.63, + "learning_rate": 0.0001, + "loss": 0.4013, + "step": 6240 + }, + { + "epoch": 1.64, + "learning_rate": 0.0001, + "loss": 0.3921, + "step": 6250 + }, + { + "epoch": 1.64, + "learning_rate": 0.0001, + "loss": 0.3934, + "step": 6260 + }, + { + "epoch": 1.64, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 6270 + }, + { + "epoch": 1.64, + "learning_rate": 0.0001, + "loss": 0.4008, + "step": 6280 + }, + { + "epoch": 1.65, + "learning_rate": 0.0001, + "loss": 0.4115, + "step": 6290 + }, + { + "epoch": 1.65, + "learning_rate": 0.0001, + "loss": 0.3982, + "step": 6300 + }, + { + "epoch": 1.65, + "learning_rate": 0.0001, + "loss": 0.4052, + "step": 6310 + }, + { + "epoch": 1.65, + "learning_rate": 0.0001, + "loss": 0.3831, + "step": 6320 + }, + { + "epoch": 1.66, + "learning_rate": 0.0001, + "loss": 0.4073, + "step": 6330 + }, + { + "epoch": 1.66, + "learning_rate": 0.0001, + "loss": 0.4035, + "step": 6340 + }, + { + "epoch": 1.66, + "learning_rate": 0.0001, + "loss": 0.4001, + "step": 6350 + }, + { + "epoch": 1.66, + "learning_rate": 0.0001, + "loss": 0.4156, + "step": 6360 + }, + { + "epoch": 1.67, + "learning_rate": 0.0001, + "loss": 0.3954, + "step": 6370 + }, + { + "epoch": 1.67, + "learning_rate": 0.0001, + "loss": 0.4085, + "step": 6380 + }, + { + "epoch": 1.67, + "learning_rate": 0.0001, + "loss": 0.4112, + "step": 6390 + }, + { + "epoch": 1.67, + "learning_rate": 0.0001, + "loss": 0.4152, + "step": 6400 + }, + { + "epoch": 1.68, + "learning_rate": 0.0001, + "loss": 0.4025, + "step": 6410 + }, + { + "epoch": 1.68, + "learning_rate": 0.0001, + "loss": 0.4004, + "step": 6420 + }, + { + "epoch": 1.68, + "learning_rate": 0.0001, + "loss": 0.4281, + "step": 6430 + }, + { + "epoch": 1.68, + "learning_rate": 0.0001, + "loss": 0.3916, + "step": 6440 + }, + { + "epoch": 1.69, + "learning_rate": 0.0001, + "loss": 0.4241, + "step": 6450 + }, + { + "epoch": 1.69, + "learning_rate": 0.0001, + "loss": 0.4005, + "step": 6460 + }, + { + "epoch": 1.69, + "learning_rate": 0.0001, + "loss": 0.4038, + "step": 6470 + }, + { + "epoch": 1.7, + "learning_rate": 0.0001, + "loss": 0.3981, + "step": 6480 + }, + { + "epoch": 1.7, + "learning_rate": 0.0001, + "loss": 0.4175, + "step": 6490 + }, + { + "epoch": 1.7, + "learning_rate": 0.0001, + "loss": 0.4143, + "step": 6500 + }, + { + "epoch": 1.7, + "learning_rate": 0.0001, + "loss": 0.3968, + "step": 6510 + }, + { + "epoch": 1.71, + "learning_rate": 0.0001, + "loss": 0.4185, + "step": 6520 + }, + { + "epoch": 1.71, + "learning_rate": 0.0001, + "loss": 0.3982, + "step": 6530 + }, + { + "epoch": 1.71, + "learning_rate": 0.0001, + "loss": 0.3946, + "step": 6540 + }, + { + "epoch": 1.71, + "learning_rate": 0.0001, + "loss": 0.4181, + "step": 6550 + }, + { + "epoch": 1.72, + "learning_rate": 0.0001, + "loss": 0.394, + "step": 6560 + }, + { + "epoch": 1.72, + "learning_rate": 0.0001, + "loss": 0.3991, + "step": 6570 + }, + { + "epoch": 1.72, + "learning_rate": 0.0001, + "loss": 0.4, + "step": 6580 + }, + { + "epoch": 1.72, + "learning_rate": 0.0001, + "loss": 0.4077, + "step": 6590 + }, + { + "epoch": 1.73, + "learning_rate": 0.0001, + "loss": 0.3932, + "step": 6600 + }, + { + "epoch": 1.73, + "learning_rate": 0.0001, + "loss": 0.3937, + "step": 6610 + }, + { + "epoch": 1.73, + "learning_rate": 0.0001, + "loss": 0.3886, + "step": 6620 + }, + { + "epoch": 1.73, + "learning_rate": 0.0001, + "loss": 0.4047, + "step": 6630 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001, + "loss": 0.4031, + "step": 6640 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001, + "loss": 0.4111, + "step": 6650 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001, + "loss": 0.3932, + "step": 6660 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001, + "loss": 0.4148, + "step": 6670 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001, + "loss": 0.4054, + "step": 6680 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001, + "loss": 0.3947, + "step": 6690 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001, + "loss": 0.3873, + "step": 6700 + }, + { + "epoch": 1.76, + "learning_rate": 0.0001, + "loss": 0.4057, + "step": 6710 + }, + { + "epoch": 1.76, + "learning_rate": 0.0001, + "loss": 0.4166, + "step": 6720 + }, + { + "epoch": 1.76, + "learning_rate": 0.0001, + "loss": 0.4188, + "step": 6730 + }, + { + "epoch": 1.76, + "learning_rate": 0.0001, + "loss": 0.4132, + "step": 6740 + }, + { + "epoch": 1.77, + "learning_rate": 0.0001, + "loss": 0.3919, + "step": 6750 + }, + { + "epoch": 1.77, + "learning_rate": 0.0001, + "loss": 0.3928, + "step": 6760 + }, + { + "epoch": 1.77, + "learning_rate": 0.0001, + "loss": 0.4093, + "step": 6770 + }, + { + "epoch": 1.77, + "learning_rate": 0.0001, + "loss": 0.4067, + "step": 6780 + }, + { + "epoch": 1.78, + "learning_rate": 0.0001, + "loss": 0.387, + "step": 6790 + }, + { + "epoch": 1.78, + "learning_rate": 0.0001, + "loss": 0.4028, + "step": 6800 + }, + { + "epoch": 1.78, + "learning_rate": 0.0001, + "loss": 0.4019, + "step": 6810 + }, + { + "epoch": 1.78, + "learning_rate": 0.0001, + "loss": 0.3951, + "step": 6820 + }, + { + "epoch": 1.79, + "learning_rate": 0.0001, + "loss": 0.3884, + "step": 6830 + }, + { + "epoch": 1.79, + "learning_rate": 0.0001, + "loss": 0.4262, + "step": 6840 + }, + { + "epoch": 1.79, + "learning_rate": 0.0001, + "loss": 0.3916, + "step": 6850 + }, + { + "epoch": 1.79, + "learning_rate": 0.0001, + "loss": 0.4016, + "step": 6860 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001, + "loss": 0.3883, + "step": 6870 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001, + "loss": 0.3915, + "step": 6880 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001, + "loss": 0.3895, + "step": 6890 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001, + "loss": 0.4014, + "step": 6900 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001, + "loss": 0.4099, + "step": 6910 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001, + "loss": 0.4139, + "step": 6920 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001, + "loss": 0.3952, + "step": 6930 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001, + "loss": 0.4063, + "step": 6940 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001, + "loss": 0.4089, + "step": 6950 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001, + "loss": 0.3947, + "step": 6960 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001, + "loss": 0.3931, + "step": 6970 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001, + "loss": 0.4041, + "step": 6980 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001, + "loss": 0.4012, + "step": 6990 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001, + "loss": 0.3987, + "step": 7000 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001, + "loss": 0.3946, + "step": 7010 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001, + "loss": 0.4145, + "step": 7020 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001, + "loss": 0.424, + "step": 7030 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001, + "loss": 0.403, + "step": 7040 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001, + "loss": 0.417, + "step": 7050 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001, + "loss": 0.3994, + "step": 7060 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001, + "loss": 0.4041, + "step": 7070 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001, + "loss": 0.3957, + "step": 7080 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001, + "loss": 0.4059, + "step": 7090 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001, + "loss": 0.3926, + "step": 7100 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001, + "loss": 0.3887, + "step": 7110 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001, + "loss": 0.4011, + "step": 7120 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001, + "loss": 0.4023, + "step": 7130 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001, + "loss": 0.4045, + "step": 7140 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001, + "loss": 0.4099, + "step": 7150 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001, + "loss": 0.4113, + "step": 7160 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001, + "loss": 0.4112, + "step": 7170 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001, + "loss": 0.4167, + "step": 7180 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001, + "loss": 0.3977, + "step": 7190 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001, + "loss": 0.4083, + "step": 7200 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001, + "loss": 0.4051, + "step": 7210 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001, + "loss": 0.3858, + "step": 7220 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001, + "loss": 0.4043, + "step": 7230 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001, + "loss": 0.394, + "step": 7240 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001, + "loss": 0.3982, + "step": 7250 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001, + "loss": 0.4051, + "step": 7260 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001, + "loss": 0.4113, + "step": 7270 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001, + "loss": 0.396, + "step": 7280 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001, + "loss": 0.4082, + "step": 7290 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001, + "loss": 0.3922, + "step": 7300 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001, + "loss": 0.4097, + "step": 7310 + }, + { + "epoch": 1.92, + "learning_rate": 0.0001, + "loss": 0.4092, + "step": 7320 + }, + { + "epoch": 1.92, + "learning_rate": 0.0001, + "loss": 0.4026, + "step": 7330 + }, + { + "epoch": 1.92, + "learning_rate": 0.0001, + "loss": 0.4084, + "step": 7340 + }, + { + "epoch": 1.92, + "learning_rate": 0.0001, + "loss": 0.4116, + "step": 7350 + }, + { + "epoch": 1.93, + "learning_rate": 0.0001, + "loss": 0.3917, + "step": 7360 + }, + { + "epoch": 1.93, + "learning_rate": 0.0001, + "loss": 0.4041, + "step": 7370 + }, + { + "epoch": 1.93, + "learning_rate": 0.0001, + "loss": 0.4064, + "step": 7380 + }, + { + "epoch": 1.93, + "learning_rate": 0.0001, + "loss": 0.4009, + "step": 7390 + }, + { + "epoch": 1.94, + "learning_rate": 0.0001, + "loss": 0.397, + "step": 7400 + }, + { + "epoch": 1.94, + "learning_rate": 0.0001, + "loss": 0.4163, + "step": 7410 + }, + { + "epoch": 1.94, + "learning_rate": 0.0001, + "loss": 0.3943, + "step": 7420 + }, + { + "epoch": 1.94, + "learning_rate": 0.0001, + "loss": 0.3959, + "step": 7430 + }, + { + "epoch": 1.95, + "learning_rate": 0.0001, + "loss": 0.4006, + "step": 7440 + }, + { + "epoch": 1.95, + "learning_rate": 0.0001, + "loss": 0.4092, + "step": 7450 + }, + { + "epoch": 1.95, + "learning_rate": 0.0001, + "loss": 0.3965, + "step": 7460 + }, + { + "epoch": 1.95, + "learning_rate": 0.0001, + "loss": 0.4006, + "step": 7470 + }, + { + "epoch": 1.96, + "learning_rate": 0.0001, + "loss": 0.3896, + "step": 7480 + }, + { + "epoch": 1.96, + "learning_rate": 0.0001, + "loss": 0.3849, + "step": 7490 + }, + { + "epoch": 1.96, + "learning_rate": 0.0001, + "loss": 0.3956, + "step": 7500 + }, + { + "epoch": 1.96, + "learning_rate": 0.0001, + "loss": 0.3916, + "step": 7510 + }, + { + "epoch": 1.97, + "learning_rate": 0.0001, + "loss": 0.3993, + "step": 7520 + }, + { + "epoch": 1.97, + "learning_rate": 0.0001, + "loss": 0.4134, + "step": 7530 + }, + { + "epoch": 1.97, + "learning_rate": 0.0001, + "loss": 0.398, + "step": 7540 + }, + { + "epoch": 1.98, + "learning_rate": 0.0001, + "loss": 0.4117, + "step": 7550 + }, + { + "epoch": 1.98, + "learning_rate": 0.0001, + "loss": 0.3999, + "step": 7560 + }, + { + "epoch": 1.98, + "learning_rate": 0.0001, + "loss": 0.3978, + "step": 7570 + }, + { + "epoch": 1.98, + "learning_rate": 0.0001, + "loss": 0.3928, + "step": 7580 + }, + { + "epoch": 1.99, + "learning_rate": 0.0001, + "loss": 0.4115, + "step": 7590 + }, + { + "epoch": 1.99, + "learning_rate": 0.0001, + "loss": 0.4012, + "step": 7600 + }, + { + "epoch": 1.99, + "learning_rate": 0.0001, + "loss": 0.3885, + "step": 7610 + }, + { + "epoch": 1.99, + "learning_rate": 0.0001, + "loss": 0.4015, + "step": 7620 + }, + { + "epoch": 2.0, + "learning_rate": 0.0001, + "loss": 0.4107, + "step": 7630 + }, + { + "epoch": 2.0, + "learning_rate": 0.0001, + "loss": 0.3924, + "step": 7640 + }, + { + "epoch": 2.0, + "eval_gen_len": 61.5757, + "eval_loss": 0.9915475845336914, + "eval_rouge1": 56.7247, + "eval_rouge2": 34.6358, + "eval_rougeL": 46.3281, + "eval_rougeLsum": 52.2798, + "eval_runtime": 38.1623, + "eval_samples_per_second": 23.033, + "eval_steps_per_second": 0.367, + "step": 7644 + }, + { + "epoch": 2.0, + "learning_rate": 0.0001, + "loss": 0.3552, + "step": 7650 + }, + { + "epoch": 2.0, + "learning_rate": 0.0001, + "loss": 0.3488, + "step": 7660 + }, + { + "epoch": 2.01, + "learning_rate": 0.0001, + "loss": 0.331, + "step": 7670 + }, + { + "epoch": 2.01, + "learning_rate": 0.0001, + "loss": 0.3472, + "step": 7680 + }, + { + "epoch": 2.01, + "learning_rate": 0.0001, + "loss": 0.342, + "step": 7690 + }, + { + "epoch": 2.01, + "learning_rate": 0.0001, + "loss": 0.3459, + "step": 7700 + }, + { + "epoch": 2.02, + "learning_rate": 0.0001, + "loss": 0.3366, + "step": 7710 + }, + { + "epoch": 2.02, + "learning_rate": 0.0001, + "loss": 0.3496, + "step": 7720 + }, + { + "epoch": 2.02, + "learning_rate": 0.0001, + "loss": 0.3415, + "step": 7730 + }, + { + "epoch": 2.03, + "learning_rate": 0.0001, + "loss": 0.3441, + "step": 7740 + }, + { + "epoch": 2.03, + "learning_rate": 0.0001, + "loss": 0.3469, + "step": 7750 + }, + { + "epoch": 2.03, + "learning_rate": 0.0001, + "loss": 0.3402, + "step": 7760 + }, + { + "epoch": 2.03, + "learning_rate": 0.0001, + "loss": 0.3408, + "step": 7770 + }, + { + "epoch": 2.04, + "learning_rate": 0.0001, + "loss": 0.3345, + "step": 7780 + }, + { + "epoch": 2.04, + "learning_rate": 0.0001, + "loss": 0.3508, + "step": 7790 + }, + { + "epoch": 2.04, + "learning_rate": 0.0001, + "loss": 0.3605, + "step": 7800 + }, + { + "epoch": 2.04, + "learning_rate": 0.0001, + "loss": 0.352, + "step": 7810 + }, + { + "epoch": 2.05, + "learning_rate": 0.0001, + "loss": 0.3474, + "step": 7820 + }, + { + "epoch": 2.05, + "learning_rate": 0.0001, + "loss": 0.3352, + "step": 7830 + }, + { + "epoch": 2.05, + "learning_rate": 0.0001, + "loss": 0.3393, + "step": 7840 + }, + { + "epoch": 2.05, + "learning_rate": 0.0001, + "loss": 0.3542, + "step": 7850 + }, + { + "epoch": 2.06, + "learning_rate": 0.0001, + "loss": 0.3318, + "step": 7860 + }, + { + "epoch": 2.06, + "learning_rate": 0.0001, + "loss": 0.3487, + "step": 7870 + }, + { + "epoch": 2.06, + "learning_rate": 0.0001, + "loss": 0.348, + "step": 7880 + }, + { + "epoch": 2.06, + "learning_rate": 0.0001, + "loss": 0.3517, + "step": 7890 + }, + { + "epoch": 2.07, + "learning_rate": 0.0001, + "loss": 0.3493, + "step": 7900 + }, + { + "epoch": 2.07, + "learning_rate": 0.0001, + "loss": 0.3369, + "step": 7910 + }, + { + "epoch": 2.07, + "learning_rate": 0.0001, + "loss": 0.3532, + "step": 7920 + }, + { + "epoch": 2.07, + "learning_rate": 0.0001, + "loss": 0.3462, + "step": 7930 + }, + { + "epoch": 2.08, + "learning_rate": 0.0001, + "loss": 0.3409, + "step": 7940 + }, + { + "epoch": 2.08, + "learning_rate": 0.0001, + "loss": 0.335, + "step": 7950 + }, + { + "epoch": 2.08, + "learning_rate": 0.0001, + "loss": 0.35, + "step": 7960 + }, + { + "epoch": 2.09, + "learning_rate": 0.0001, + "loss": 0.3356, + "step": 7970 + }, + { + "epoch": 2.09, + "learning_rate": 0.0001, + "loss": 0.3417, + "step": 7980 + }, + { + "epoch": 2.09, + "learning_rate": 0.0001, + "loss": 0.3523, + "step": 7990 + }, + { + "epoch": 2.09, + "learning_rate": 0.0001, + "loss": 0.3323, + "step": 8000 + }, + { + "epoch": 2.1, + "learning_rate": 0.0001, + "loss": 0.3338, + "step": 8010 + }, + { + "epoch": 2.1, + "learning_rate": 0.0001, + "loss": 0.356, + "step": 8020 + }, + { + "epoch": 2.1, + "learning_rate": 0.0001, + "loss": 0.3408, + "step": 8030 + }, + { + "epoch": 2.1, + "learning_rate": 0.0001, + "loss": 0.3437, + "step": 8040 + }, + { + "epoch": 2.11, + "learning_rate": 0.0001, + "loss": 0.3591, + "step": 8050 + }, + { + "epoch": 2.11, + "learning_rate": 0.0001, + "loss": 0.337, + "step": 8060 + }, + { + "epoch": 2.11, + "learning_rate": 0.0001, + "loss": 0.3461, + "step": 8070 + }, + { + "epoch": 2.11, + "learning_rate": 0.0001, + "loss": 0.3496, + "step": 8080 + }, + { + "epoch": 2.12, + "learning_rate": 0.0001, + "loss": 0.3397, + "step": 8090 + }, + { + "epoch": 2.12, + "learning_rate": 0.0001, + "loss": 0.3635, + "step": 8100 + }, + { + "epoch": 2.12, + "learning_rate": 0.0001, + "loss": 0.3537, + "step": 8110 + }, + { + "epoch": 2.12, + "learning_rate": 0.0001, + "loss": 0.3366, + "step": 8120 + }, + { + "epoch": 2.13, + "learning_rate": 0.0001, + "loss": 0.351, + "step": 8130 + }, + { + "epoch": 2.13, + "learning_rate": 0.0001, + "loss": 0.3388, + "step": 8140 + }, + { + "epoch": 2.13, + "learning_rate": 0.0001, + "loss": 0.346, + "step": 8150 + }, + { + "epoch": 2.14, + "learning_rate": 0.0001, + "loss": 0.3543, + "step": 8160 + }, + { + "epoch": 2.14, + "learning_rate": 0.0001, + "loss": 0.3473, + "step": 8170 + }, + { + "epoch": 2.14, + "learning_rate": 0.0001, + "loss": 0.344, + "step": 8180 + }, + { + "epoch": 2.14, + "learning_rate": 0.0001, + "loss": 0.3582, + "step": 8190 + }, + { + "epoch": 2.15, + "learning_rate": 0.0001, + "loss": 0.3542, + "step": 8200 + }, + { + "epoch": 2.15, + "learning_rate": 0.0001, + "loss": 0.3414, + "step": 8210 + }, + { + "epoch": 2.15, + "learning_rate": 0.0001, + "loss": 0.352, + "step": 8220 + }, + { + "epoch": 2.15, + "learning_rate": 0.0001, + "loss": 0.3635, + "step": 8230 + }, + { + "epoch": 2.16, + "learning_rate": 0.0001, + "loss": 0.3431, + "step": 8240 + }, + { + "epoch": 2.16, + "learning_rate": 0.0001, + "loss": 0.3451, + "step": 8250 + }, + { + "epoch": 2.16, + "learning_rate": 0.0001, + "loss": 0.3343, + "step": 8260 + }, + { + "epoch": 2.16, + "learning_rate": 0.0001, + "loss": 0.3446, + "step": 8270 + }, + { + "epoch": 2.17, + "learning_rate": 0.0001, + "loss": 0.3445, + "step": 8280 + }, + { + "epoch": 2.17, + "learning_rate": 0.0001, + "loss": 0.3366, + "step": 8290 + }, + { + "epoch": 2.17, + "learning_rate": 0.0001, + "loss": 0.341, + "step": 8300 + }, + { + "epoch": 2.17, + "learning_rate": 0.0001, + "loss": 0.359, + "step": 8310 + }, + { + "epoch": 2.18, + "learning_rate": 0.0001, + "loss": 0.3472, + "step": 8320 + }, + { + "epoch": 2.18, + "learning_rate": 0.0001, + "loss": 0.3493, + "step": 8330 + }, + { + "epoch": 2.18, + "learning_rate": 0.0001, + "loss": 0.3462, + "step": 8340 + }, + { + "epoch": 2.18, + "learning_rate": 0.0001, + "loss": 0.3404, + "step": 8350 + }, + { + "epoch": 2.19, + "learning_rate": 0.0001, + "loss": 0.3584, + "step": 8360 + }, + { + "epoch": 2.19, + "learning_rate": 0.0001, + "loss": 0.3513, + "step": 8370 + }, + { + "epoch": 2.19, + "learning_rate": 0.0001, + "loss": 0.3584, + "step": 8380 + }, + { + "epoch": 2.2, + "learning_rate": 0.0001, + "loss": 0.3574, + "step": 8390 + }, + { + "epoch": 2.2, + "learning_rate": 0.0001, + "loss": 0.3478, + "step": 8400 + }, + { + "epoch": 2.2, + "learning_rate": 0.0001, + "loss": 0.3604, + "step": 8410 + }, + { + "epoch": 2.2, + "learning_rate": 0.0001, + "loss": 0.3446, + "step": 8420 + }, + { + "epoch": 2.21, + "learning_rate": 0.0001, + "loss": 0.3526, + "step": 8430 + }, + { + "epoch": 2.21, + "learning_rate": 0.0001, + "loss": 0.3397, + "step": 8440 + }, + { + "epoch": 2.21, + "learning_rate": 0.0001, + "loss": 0.3476, + "step": 8450 + }, + { + "epoch": 2.21, + "learning_rate": 0.0001, + "loss": 0.3356, + "step": 8460 + }, + { + "epoch": 2.22, + "learning_rate": 0.0001, + "loss": 0.3349, + "step": 8470 + }, + { + "epoch": 2.22, + "learning_rate": 0.0001, + "loss": 0.3526, + "step": 8480 + }, + { + "epoch": 2.22, + "learning_rate": 0.0001, + "loss": 0.353, + "step": 8490 + }, + { + "epoch": 2.22, + "learning_rate": 0.0001, + "loss": 0.3593, + "step": 8500 + }, + { + "epoch": 2.23, + "learning_rate": 0.0001, + "loss": 0.348, + "step": 8510 + }, + { + "epoch": 2.23, + "learning_rate": 0.0001, + "loss": 0.3488, + "step": 8520 + }, + { + "epoch": 2.23, + "learning_rate": 0.0001, + "loss": 0.3425, + "step": 8530 + }, + { + "epoch": 2.23, + "learning_rate": 0.0001, + "loss": 0.3436, + "step": 8540 + }, + { + "epoch": 2.24, + "learning_rate": 0.0001, + "loss": 0.3406, + "step": 8550 + }, + { + "epoch": 2.24, + "learning_rate": 0.0001, + "loss": 0.3352, + "step": 8560 + }, + { + "epoch": 2.24, + "learning_rate": 0.0001, + "loss": 0.3544, + "step": 8570 + }, + { + "epoch": 2.24, + "learning_rate": 0.0001, + "loss": 0.3512, + "step": 8580 + }, + { + "epoch": 2.25, + "learning_rate": 0.0001, + "loss": 0.3535, + "step": 8590 + }, + { + "epoch": 2.25, + "learning_rate": 0.0001, + "loss": 0.3573, + "step": 8600 + }, + { + "epoch": 2.25, + "learning_rate": 0.0001, + "loss": 0.344, + "step": 8610 + }, + { + "epoch": 2.26, + "learning_rate": 0.0001, + "loss": 0.3628, + "step": 8620 + }, + { + "epoch": 2.26, + "learning_rate": 0.0001, + "loss": 0.363, + "step": 8630 + }, + { + "epoch": 2.26, + "learning_rate": 0.0001, + "loss": 0.3435, + "step": 8640 + }, + { + "epoch": 2.26, + "learning_rate": 0.0001, + "loss": 0.3415, + "step": 8650 + }, + { + "epoch": 2.27, + "learning_rate": 0.0001, + "loss": 0.3524, + "step": 8660 + }, + { + "epoch": 2.27, + "learning_rate": 0.0001, + "loss": 0.3583, + "step": 8670 + }, + { + "epoch": 2.27, + "learning_rate": 0.0001, + "loss": 0.3573, + "step": 8680 + }, + { + "epoch": 2.27, + "learning_rate": 0.0001, + "loss": 0.3284, + "step": 8690 + }, + { + "epoch": 2.28, + "learning_rate": 0.0001, + "loss": 0.332, + "step": 8700 + }, + { + "epoch": 2.28, + "learning_rate": 0.0001, + "loss": 0.3474, + "step": 8710 + }, + { + "epoch": 2.28, + "learning_rate": 0.0001, + "loss": 0.3521, + "step": 8720 + }, + { + "epoch": 2.28, + "learning_rate": 0.0001, + "loss": 0.3323, + "step": 8730 + }, + { + "epoch": 2.29, + "learning_rate": 0.0001, + "loss": 0.3468, + "step": 8740 + }, + { + "epoch": 2.29, + "learning_rate": 0.0001, + "loss": 0.3543, + "step": 8750 + }, + { + "epoch": 2.29, + "learning_rate": 0.0001, + "loss": 0.3539, + "step": 8760 + }, + { + "epoch": 2.29, + "learning_rate": 0.0001, + "loss": 0.3527, + "step": 8770 + }, + { + "epoch": 2.3, + "learning_rate": 0.0001, + "loss": 0.3484, + "step": 8780 + }, + { + "epoch": 2.3, + "learning_rate": 0.0001, + "loss": 0.3445, + "step": 8790 + }, + { + "epoch": 2.3, + "learning_rate": 0.0001, + "loss": 0.357, + "step": 8800 + }, + { + "epoch": 2.31, + "learning_rate": 0.0001, + "loss": 0.3505, + "step": 8810 + }, + { + "epoch": 2.31, + "learning_rate": 0.0001, + "loss": 0.3482, + "step": 8820 + }, + { + "epoch": 2.31, + "learning_rate": 0.0001, + "loss": 0.3345, + "step": 8830 + }, + { + "epoch": 2.31, + "learning_rate": 0.0001, + "loss": 0.3527, + "step": 8840 + }, + { + "epoch": 2.32, + "learning_rate": 0.0001, + "loss": 0.344, + "step": 8850 + }, + { + "epoch": 2.32, + "learning_rate": 0.0001, + "loss": 0.351, + "step": 8860 + }, + { + "epoch": 2.32, + "learning_rate": 0.0001, + "loss": 0.3506, + "step": 8870 + }, + { + "epoch": 2.32, + "learning_rate": 0.0001, + "loss": 0.3546, + "step": 8880 + }, + { + "epoch": 2.33, + "learning_rate": 0.0001, + "loss": 0.35, + "step": 8890 + }, + { + "epoch": 2.33, + "learning_rate": 0.0001, + "loss": 0.3455, + "step": 8900 + }, + { + "epoch": 2.33, + "learning_rate": 0.0001, + "loss": 0.3596, + "step": 8910 + }, + { + "epoch": 2.33, + "learning_rate": 0.0001, + "loss": 0.3537, + "step": 8920 + }, + { + "epoch": 2.34, + "learning_rate": 0.0001, + "loss": 0.3546, + "step": 8930 + }, + { + "epoch": 2.34, + "learning_rate": 0.0001, + "loss": 0.3379, + "step": 8940 + }, + { + "epoch": 2.34, + "learning_rate": 0.0001, + "loss": 0.3547, + "step": 8950 + }, + { + "epoch": 2.34, + "learning_rate": 0.0001, + "loss": 0.3577, + "step": 8960 + }, + { + "epoch": 2.35, + "learning_rate": 0.0001, + "loss": 0.3473, + "step": 8970 + }, + { + "epoch": 2.35, + "learning_rate": 0.0001, + "loss": 0.3343, + "step": 8980 + }, + { + "epoch": 2.35, + "learning_rate": 0.0001, + "loss": 0.3511, + "step": 8990 + }, + { + "epoch": 2.35, + "learning_rate": 0.0001, + "loss": 0.3573, + "step": 9000 + }, + { + "epoch": 2.36, + "learning_rate": 0.0001, + "loss": 0.3454, + "step": 9010 + }, + { + "epoch": 2.36, + "learning_rate": 0.0001, + "loss": 0.3485, + "step": 9020 + }, + { + "epoch": 2.36, + "learning_rate": 0.0001, + "loss": 0.3522, + "step": 9030 + }, + { + "epoch": 2.37, + "learning_rate": 0.0001, + "loss": 0.3424, + "step": 9040 + }, + { + "epoch": 2.37, + "learning_rate": 0.0001, + "loss": 0.3372, + "step": 9050 + }, + { + "epoch": 2.37, + "learning_rate": 0.0001, + "loss": 0.3558, + "step": 9060 + }, + { + "epoch": 2.37, + "learning_rate": 0.0001, + "loss": 0.357, + "step": 9070 + }, + { + "epoch": 2.38, + "learning_rate": 0.0001, + "loss": 0.3414, + "step": 9080 + }, + { + "epoch": 2.38, + "learning_rate": 0.0001, + "loss": 0.3603, + "step": 9090 + }, + { + "epoch": 2.38, + "learning_rate": 0.0001, + "loss": 0.3387, + "step": 9100 + }, + { + "epoch": 2.38, + "learning_rate": 0.0001, + "loss": 0.3538, + "step": 9110 + }, + { + "epoch": 2.39, + "learning_rate": 0.0001, + "loss": 0.3443, + "step": 9120 + }, + { + "epoch": 2.39, + "learning_rate": 0.0001, + "loss": 0.3288, + "step": 9130 + }, + { + "epoch": 2.39, + "learning_rate": 0.0001, + "loss": 0.3583, + "step": 9140 + }, + { + "epoch": 2.39, + "learning_rate": 0.0001, + "loss": 0.3377, + "step": 9150 + }, + { + "epoch": 2.4, + "learning_rate": 0.0001, + "loss": 0.3432, + "step": 9160 + }, + { + "epoch": 2.4, + "learning_rate": 0.0001, + "loss": 0.3505, + "step": 9170 + }, + { + "epoch": 2.4, + "learning_rate": 0.0001, + "loss": 0.3549, + "step": 9180 + }, + { + "epoch": 2.4, + "learning_rate": 0.0001, + "loss": 0.3543, + "step": 9190 + }, + { + "epoch": 2.41, + "learning_rate": 0.0001, + "loss": 0.3407, + "step": 9200 + }, + { + "epoch": 2.41, + "learning_rate": 0.0001, + "loss": 0.3579, + "step": 9210 + }, + { + "epoch": 2.41, + "learning_rate": 0.0001, + "loss": 0.3461, + "step": 9220 + }, + { + "epoch": 2.41, + "learning_rate": 0.0001, + "loss": 0.3455, + "step": 9230 + }, + { + "epoch": 2.42, + "learning_rate": 0.0001, + "loss": 0.3582, + "step": 9240 + }, + { + "epoch": 2.42, + "learning_rate": 0.0001, + "loss": 0.3355, + "step": 9250 + }, + { + "epoch": 2.42, + "learning_rate": 0.0001, + "loss": 0.3503, + "step": 9260 + }, + { + "epoch": 2.43, + "learning_rate": 0.0001, + "loss": 0.3442, + "step": 9270 + }, + { + "epoch": 2.43, + "learning_rate": 0.0001, + "loss": 0.3296, + "step": 9280 + }, + { + "epoch": 2.43, + "learning_rate": 0.0001, + "loss": 0.3439, + "step": 9290 + }, + { + "epoch": 2.43, + "learning_rate": 0.0001, + "loss": 0.3417, + "step": 9300 + }, + { + "epoch": 2.44, + "learning_rate": 0.0001, + "loss": 0.3391, + "step": 9310 + }, + { + "epoch": 2.44, + "learning_rate": 0.0001, + "loss": 0.3525, + "step": 9320 + }, + { + "epoch": 2.44, + "learning_rate": 0.0001, + "loss": 0.3406, + "step": 9330 + }, + { + "epoch": 2.44, + "learning_rate": 0.0001, + "loss": 0.3569, + "step": 9340 + }, + { + "epoch": 2.45, + "learning_rate": 0.0001, + "loss": 0.3484, + "step": 9350 + }, + { + "epoch": 2.45, + "learning_rate": 0.0001, + "loss": 0.358, + "step": 9360 + }, + { + "epoch": 2.45, + "learning_rate": 0.0001, + "loss": 0.3522, + "step": 9370 + }, + { + "epoch": 2.45, + "learning_rate": 0.0001, + "loss": 0.3419, + "step": 9380 + }, + { + "epoch": 2.46, + "learning_rate": 0.0001, + "loss": 0.3505, + "step": 9390 + }, + { + "epoch": 2.46, + "learning_rate": 0.0001, + "loss": 0.3534, + "step": 9400 + }, + { + "epoch": 2.46, + "learning_rate": 0.0001, + "loss": 0.3524, + "step": 9410 + }, + { + "epoch": 2.46, + "learning_rate": 0.0001, + "loss": 0.3448, + "step": 9420 + }, + { + "epoch": 2.47, + "learning_rate": 0.0001, + "loss": 0.3437, + "step": 9430 + }, + { + "epoch": 2.47, + "learning_rate": 0.0001, + "loss": 0.3405, + "step": 9440 + }, + { + "epoch": 2.47, + "learning_rate": 0.0001, + "loss": 0.3419, + "step": 9450 + }, + { + "epoch": 2.48, + "learning_rate": 0.0001, + "loss": 0.3507, + "step": 9460 + }, + { + "epoch": 2.48, + "learning_rate": 0.0001, + "loss": 0.3512, + "step": 9470 + }, + { + "epoch": 2.48, + "learning_rate": 0.0001, + "loss": 0.3487, + "step": 9480 + }, + { + "epoch": 2.48, + "learning_rate": 0.0001, + "loss": 0.3467, + "step": 9490 + }, + { + "epoch": 2.49, + "learning_rate": 0.0001, + "loss": 0.3389, + "step": 9500 + }, + { + "epoch": 2.49, + "learning_rate": 0.0001, + "loss": 0.3647, + "step": 9510 + }, + { + "epoch": 2.49, + "learning_rate": 0.0001, + "loss": 0.3583, + "step": 9520 + }, + { + "epoch": 2.49, + "learning_rate": 0.0001, + "loss": 0.3429, + "step": 9530 + }, + { + "epoch": 2.5, + "learning_rate": 0.0001, + "loss": 0.3541, + "step": 9540 + }, + { + "epoch": 2.5, + "learning_rate": 0.0001, + "loss": 0.3509, + "step": 9550 + }, + { + "epoch": 2.5, + "learning_rate": 0.0001, + "loss": 0.3472, + "step": 9560 + }, + { + "epoch": 2.5, + "learning_rate": 0.0001, + "loss": 0.3385, + "step": 9570 + }, + { + "epoch": 2.51, + "learning_rate": 0.0001, + "loss": 0.3332, + "step": 9580 + }, + { + "epoch": 2.51, + "learning_rate": 0.0001, + "loss": 0.349, + "step": 9590 + }, + { + "epoch": 2.51, + "learning_rate": 0.0001, + "loss": 0.3607, + "step": 9600 + }, + { + "epoch": 2.51, + "learning_rate": 0.0001, + "loss": 0.3435, + "step": 9610 + }, + { + "epoch": 2.52, + "learning_rate": 0.0001, + "loss": 0.3382, + "step": 9620 + }, + { + "epoch": 2.52, + "learning_rate": 0.0001, + "loss": 0.3475, + "step": 9630 + }, + { + "epoch": 2.52, + "learning_rate": 0.0001, + "loss": 0.3593, + "step": 9640 + }, + { + "epoch": 2.52, + "learning_rate": 0.0001, + "loss": 0.3295, + "step": 9650 + }, + { + "epoch": 2.53, + "learning_rate": 0.0001, + "loss": 0.3554, + "step": 9660 + }, + { + "epoch": 2.53, + "learning_rate": 0.0001, + "loss": 0.3348, + "step": 9670 + }, + { + "epoch": 2.53, + "learning_rate": 0.0001, + "loss": 0.3463, + "step": 9680 + }, + { + "epoch": 2.54, + "learning_rate": 0.0001, + "loss": 0.355, + "step": 9690 + }, + { + "epoch": 2.54, + "learning_rate": 0.0001, + "loss": 0.3563, + "step": 9700 + }, + { + "epoch": 2.54, + "learning_rate": 0.0001, + "loss": 0.3497, + "step": 9710 + }, + { + "epoch": 2.54, + "learning_rate": 0.0001, + "loss": 0.3592, + "step": 9720 + }, + { + "epoch": 2.55, + "learning_rate": 0.0001, + "loss": 0.3499, + "step": 9730 + }, + { + "epoch": 2.55, + "learning_rate": 0.0001, + "loss": 0.3376, + "step": 9740 + }, + { + "epoch": 2.55, + "learning_rate": 0.0001, + "loss": 0.3637, + "step": 9750 + }, + { + "epoch": 2.55, + "learning_rate": 0.0001, + "loss": 0.3574, + "step": 9760 + }, + { + "epoch": 2.56, + "learning_rate": 0.0001, + "loss": 0.3498, + "step": 9770 + }, + { + "epoch": 2.56, + "learning_rate": 0.0001, + "loss": 0.349, + "step": 9780 + }, + { + "epoch": 2.56, + "learning_rate": 0.0001, + "loss": 0.3599, + "step": 9790 + }, + { + "epoch": 2.56, + "learning_rate": 0.0001, + "loss": 0.3563, + "step": 9800 + }, + { + "epoch": 2.57, + "learning_rate": 0.0001, + "loss": 0.3446, + "step": 9810 + }, + { + "epoch": 2.57, + "learning_rate": 0.0001, + "loss": 0.3622, + "step": 9820 + }, + { + "epoch": 2.57, + "learning_rate": 0.0001, + "loss": 0.3492, + "step": 9830 + }, + { + "epoch": 2.57, + "learning_rate": 0.0001, + "loss": 0.3606, + "step": 9840 + }, + { + "epoch": 2.58, + "learning_rate": 0.0001, + "loss": 0.346, + "step": 9850 + }, + { + "epoch": 2.58, + "learning_rate": 0.0001, + "loss": 0.3511, + "step": 9860 + }, + { + "epoch": 2.58, + "learning_rate": 0.0001, + "loss": 0.35, + "step": 9870 + }, + { + "epoch": 2.59, + "learning_rate": 0.0001, + "loss": 0.34, + "step": 9880 + }, + { + "epoch": 2.59, + "learning_rate": 0.0001, + "loss": 0.3473, + "step": 9890 + }, + { + "epoch": 2.59, + "learning_rate": 0.0001, + "loss": 0.3482, + "step": 9900 + }, + { + "epoch": 2.59, + "learning_rate": 0.0001, + "loss": 0.3517, + "step": 9910 + }, + { + "epoch": 2.6, + "learning_rate": 0.0001, + "loss": 0.3416, + "step": 9920 + }, + { + "epoch": 2.6, + "learning_rate": 0.0001, + "loss": 0.3538, + "step": 9930 + }, + { + "epoch": 2.6, + "learning_rate": 0.0001, + "loss": 0.3605, + "step": 9940 + }, + { + "epoch": 2.6, + "learning_rate": 0.0001, + "loss": 0.3447, + "step": 9950 + }, + { + "epoch": 2.61, + "learning_rate": 0.0001, + "loss": 0.341, + "step": 9960 + }, + { + "epoch": 2.61, + "learning_rate": 0.0001, + "loss": 0.3482, + "step": 9970 + }, + { + "epoch": 2.61, + "learning_rate": 0.0001, + "loss": 0.3402, + "step": 9980 + }, + { + "epoch": 2.61, + "learning_rate": 0.0001, + "loss": 0.3568, + "step": 9990 + }, + { + "epoch": 2.62, + "learning_rate": 0.0001, + "loss": 0.357, + "step": 10000 + }, + { + "epoch": 2.62, + "learning_rate": 0.0001, + "loss": 0.356, + "step": 10010 + }, + { + "epoch": 2.62, + "learning_rate": 0.0001, + "loss": 0.3614, + "step": 10020 + }, + { + "epoch": 2.62, + "learning_rate": 0.0001, + "loss": 0.3567, + "step": 10030 + }, + { + "epoch": 2.63, + "learning_rate": 0.0001, + "loss": 0.3483, + "step": 10040 + }, + { + "epoch": 2.63, + "learning_rate": 0.0001, + "loss": 0.3434, + "step": 10050 + }, + { + "epoch": 2.63, + "learning_rate": 0.0001, + "loss": 0.3448, + "step": 10060 + }, + { + "epoch": 2.63, + "learning_rate": 0.0001, + "loss": 0.3619, + "step": 10070 + }, + { + "epoch": 2.64, + "learning_rate": 0.0001, + "loss": 0.3356, + "step": 10080 + }, + { + "epoch": 2.64, + "learning_rate": 0.0001, + "loss": 0.3428, + "step": 10090 + }, + { + "epoch": 2.64, + "learning_rate": 0.0001, + "loss": 0.3568, + "step": 10100 + }, + { + "epoch": 2.65, + "learning_rate": 0.0001, + "loss": 0.3452, + "step": 10110 + }, + { + "epoch": 2.65, + "learning_rate": 0.0001, + "loss": 0.352, + "step": 10120 + }, + { + "epoch": 2.65, + "learning_rate": 0.0001, + "loss": 0.3572, + "step": 10130 + }, + { + "epoch": 2.65, + "learning_rate": 0.0001, + "loss": 0.3595, + "step": 10140 + }, + { + "epoch": 2.66, + "learning_rate": 0.0001, + "loss": 0.348, + "step": 10150 + }, + { + "epoch": 2.66, + "learning_rate": 0.0001, + "loss": 0.3585, + "step": 10160 + }, + { + "epoch": 2.66, + "learning_rate": 0.0001, + "loss": 0.3623, + "step": 10170 + }, + { + "epoch": 2.66, + "learning_rate": 0.0001, + "loss": 0.3442, + "step": 10180 + }, + { + "epoch": 2.67, + "learning_rate": 0.0001, + "loss": 0.3546, + "step": 10190 + }, + { + "epoch": 2.67, + "learning_rate": 0.0001, + "loss": 0.3719, + "step": 10200 + }, + { + "epoch": 2.67, + "learning_rate": 0.0001, + "loss": 0.3602, + "step": 10210 + }, + { + "epoch": 2.67, + "learning_rate": 0.0001, + "loss": 0.3604, + "step": 10220 + }, + { + "epoch": 2.68, + "learning_rate": 0.0001, + "loss": 0.3565, + "step": 10230 + }, + { + "epoch": 2.68, + "learning_rate": 0.0001, + "loss": 0.3573, + "step": 10240 + }, + { + "epoch": 2.68, + "learning_rate": 0.0001, + "loss": 0.3416, + "step": 10250 + }, + { + "epoch": 2.68, + "learning_rate": 0.0001, + "loss": 0.3528, + "step": 10260 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001, + "loss": 0.3586, + "step": 10270 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001, + "loss": 0.3559, + "step": 10280 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001, + "loss": 0.3676, + "step": 10290 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001, + "loss": 0.3411, + "step": 10300 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001, + "loss": 0.3419, + "step": 10310 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001, + "loss": 0.3501, + "step": 10320 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001, + "loss": 0.3543, + "step": 10330 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001, + "loss": 0.3442, + "step": 10340 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001, + "loss": 0.3563, + "step": 10350 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001, + "loss": 0.3481, + "step": 10360 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001, + "loss": 0.3541, + "step": 10370 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001, + "loss": 0.3482, + "step": 10380 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001, + "loss": 0.3551, + "step": 10390 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001, + "loss": 0.3551, + "step": 10400 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001, + "loss": 0.3513, + "step": 10410 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001, + "loss": 0.3523, + "step": 10420 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001, + "loss": 0.3494, + "step": 10430 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001, + "loss": 0.3642, + "step": 10440 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001, + "loss": 0.3477, + "step": 10450 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001, + "loss": 0.347, + "step": 10460 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001, + "loss": 0.3568, + "step": 10470 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001, + "loss": 0.3436, + "step": 10480 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001, + "loss": 0.3502, + "step": 10490 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001, + "loss": 0.3476, + "step": 10500 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001, + "loss": 0.3595, + "step": 10510 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001, + "loss": 0.3669, + "step": 10520 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001, + "loss": 0.3379, + "step": 10530 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001, + "loss": 0.3518, + "step": 10540 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001, + "loss": 0.3597, + "step": 10550 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001, + "loss": 0.3456, + "step": 10560 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001, + "loss": 0.3665, + "step": 10570 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001, + "loss": 0.3462, + "step": 10580 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001, + "loss": 0.3621, + "step": 10590 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001, + "loss": 0.3382, + "step": 10600 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001, + "loss": 0.3498, + "step": 10610 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001, + "loss": 0.3513, + "step": 10620 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001, + "loss": 0.3548, + "step": 10630 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001, + "loss": 0.3403, + "step": 10640 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001, + "loss": 0.3534, + "step": 10650 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001, + "loss": 0.3428, + "step": 10660 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001, + "loss": 0.3471, + "step": 10670 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001, + "loss": 0.3529, + "step": 10680 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001, + "loss": 0.3557, + "step": 10690 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001, + "loss": 0.3411, + "step": 10700 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001, + "loss": 0.3479, + "step": 10710 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001, + "loss": 0.3444, + "step": 10720 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001, + "loss": 0.3529, + "step": 10730 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001, + "loss": 0.3473, + "step": 10740 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001, + "loss": 0.3367, + "step": 10750 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001, + "loss": 0.3388, + "step": 10760 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001, + "loss": 0.3514, + "step": 10770 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001, + "loss": 0.343, + "step": 10780 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001, + "loss": 0.3557, + "step": 10790 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001, + "loss": 0.3361, + "step": 10800 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001, + "loss": 0.3543, + "step": 10810 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001, + "loss": 0.357, + "step": 10820 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001, + "loss": 0.3604, + "step": 10830 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001, + "loss": 0.3517, + "step": 10840 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001, + "loss": 0.356, + "step": 10850 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001, + "loss": 0.3491, + "step": 10860 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001, + "loss": 0.3388, + "step": 10870 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001, + "loss": 0.3515, + "step": 10880 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001, + "loss": 0.3347, + "step": 10890 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001, + "loss": 0.3532, + "step": 10900 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001, + "loss": 0.3555, + "step": 10910 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001, + "loss": 0.3577, + "step": 10920 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001, + "loss": 0.3469, + "step": 10930 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001, + "loss": 0.3504, + "step": 10940 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001, + "loss": 0.3487, + "step": 10950 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001, + "loss": 0.3329, + "step": 10960 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001, + "loss": 0.3489, + "step": 10970 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001, + "loss": 0.3465, + "step": 10980 + }, + { + "epoch": 2.88, + "learning_rate": 0.0001, + "loss": 0.3599, + "step": 10990 + }, + { + "epoch": 2.88, + "learning_rate": 0.0001, + "loss": 0.3527, + "step": 11000 + }, + { + "epoch": 2.88, + "learning_rate": 0.0001, + "loss": 0.3532, + "step": 11010 + }, + { + "epoch": 2.88, + "learning_rate": 0.0001, + "loss": 0.3528, + "step": 11020 + }, + { + "epoch": 2.89, + "learning_rate": 0.0001, + "loss": 0.3505, + "step": 11030 + }, + { + "epoch": 2.89, + "learning_rate": 0.0001, + "loss": 0.3418, + "step": 11040 + }, + { + "epoch": 2.89, + "learning_rate": 0.0001, + "loss": 0.342, + "step": 11050 + }, + { + "epoch": 2.89, + "learning_rate": 0.0001, + "loss": 0.3384, + "step": 11060 + }, + { + "epoch": 2.9, + "learning_rate": 0.0001, + "loss": 0.3483, + "step": 11070 + }, + { + "epoch": 2.9, + "learning_rate": 0.0001, + "loss": 0.3547, + "step": 11080 + }, + { + "epoch": 2.9, + "learning_rate": 0.0001, + "loss": 0.3621, + "step": 11090 + }, + { + "epoch": 2.9, + "learning_rate": 0.0001, + "loss": 0.3546, + "step": 11100 + }, + { + "epoch": 2.91, + "learning_rate": 0.0001, + "loss": 0.3538, + "step": 11110 + }, + { + "epoch": 2.91, + "learning_rate": 0.0001, + "loss": 0.3513, + "step": 11120 + }, + { + "epoch": 2.91, + "learning_rate": 0.0001, + "loss": 0.3414, + "step": 11130 + }, + { + "epoch": 2.91, + "learning_rate": 0.0001, + "loss": 0.3489, + "step": 11140 + }, + { + "epoch": 2.92, + "learning_rate": 0.0001, + "loss": 0.3568, + "step": 11150 + }, + { + "epoch": 2.92, + "learning_rate": 0.0001, + "loss": 0.3481, + "step": 11160 + }, + { + "epoch": 2.92, + "learning_rate": 0.0001, + "loss": 0.3461, + "step": 11170 + }, + { + "epoch": 2.93, + "learning_rate": 0.0001, + "loss": 0.3386, + "step": 11180 + }, + { + "epoch": 2.93, + "learning_rate": 0.0001, + "loss": 0.3575, + "step": 11190 + }, + { + "epoch": 2.93, + "learning_rate": 0.0001, + "loss": 0.3518, + "step": 11200 + }, + { + "epoch": 2.93, + "learning_rate": 0.0001, + "loss": 0.3601, + "step": 11210 + }, + { + "epoch": 2.94, + "learning_rate": 0.0001, + "loss": 0.3507, + "step": 11220 + }, + { + "epoch": 2.94, + "learning_rate": 0.0001, + "loss": 0.3554, + "step": 11230 + }, + { + "epoch": 2.94, + "learning_rate": 0.0001, + "loss": 0.351, + "step": 11240 + }, + { + "epoch": 2.94, + "learning_rate": 0.0001, + "loss": 0.3364, + "step": 11250 + }, + { + "epoch": 2.95, + "learning_rate": 0.0001, + "loss": 0.3594, + "step": 11260 + }, + { + "epoch": 2.95, + "learning_rate": 0.0001, + "loss": 0.338, + "step": 11270 + }, + { + "epoch": 2.95, + "learning_rate": 0.0001, + "loss": 0.3487, + "step": 11280 + }, + { + "epoch": 2.95, + "learning_rate": 0.0001, + "loss": 0.3538, + "step": 11290 + }, + { + "epoch": 2.96, + "learning_rate": 0.0001, + "loss": 0.3594, + "step": 11300 + }, + { + "epoch": 2.96, + "learning_rate": 0.0001, + "loss": 0.3673, + "step": 11310 + }, + { + "epoch": 2.96, + "learning_rate": 0.0001, + "loss": 0.3432, + "step": 11320 + }, + { + "epoch": 2.96, + "learning_rate": 0.0001, + "loss": 0.3422, + "step": 11330 + }, + { + "epoch": 2.97, + "learning_rate": 0.0001, + "loss": 0.358, + "step": 11340 + }, + { + "epoch": 2.97, + "learning_rate": 0.0001, + "loss": 0.3542, + "step": 11350 + }, + { + "epoch": 2.97, + "learning_rate": 0.0001, + "loss": 0.3564, + "step": 11360 + }, + { + "epoch": 2.97, + "learning_rate": 0.0001, + "loss": 0.3482, + "step": 11370 + }, + { + "epoch": 2.98, + "learning_rate": 0.0001, + "loss": 0.343, + "step": 11380 + }, + { + "epoch": 2.98, + "learning_rate": 0.0001, + "loss": 0.3456, + "step": 11390 + }, + { + "epoch": 2.98, + "learning_rate": 0.0001, + "loss": 0.3427, + "step": 11400 + }, + { + "epoch": 2.99, + "learning_rate": 0.0001, + "loss": 0.3497, + "step": 11410 + }, + { + "epoch": 2.99, + "learning_rate": 0.0001, + "loss": 0.3498, + "step": 11420 + }, + { + "epoch": 2.99, + "learning_rate": 0.0001, + "loss": 0.3515, + "step": 11430 + }, + { + "epoch": 2.99, + "learning_rate": 0.0001, + "loss": 0.3572, + "step": 11440 + }, + { + "epoch": 3.0, + "learning_rate": 0.0001, + "loss": 0.3461, + "step": 11450 + }, + { + "epoch": 3.0, + "learning_rate": 0.0001, + "loss": 0.3452, + "step": 11460 + }, + { + "epoch": 3.0, + "eval_gen_len": 61.6678, + "eval_loss": 1.0176692008972168, + "eval_rouge1": 57.287, + "eval_rouge2": 35.2249, + "eval_rougeL": 46.6138, + "eval_rougeLsum": 52.748, + "eval_runtime": 38.1217, + "eval_samples_per_second": 23.058, + "eval_steps_per_second": 0.367, + "step": 11466 + }, + { + "epoch": 3.0, + "learning_rate": 0.0001, + "loss": 0.3263, + "step": 11470 + }, + { + "epoch": 3.0, + "learning_rate": 0.0001, + "loss": 0.2949, + "step": 11480 + }, + { + "epoch": 3.01, + "learning_rate": 0.0001, + "loss": 0.3, + "step": 11490 + }, + { + "epoch": 3.01, + "learning_rate": 0.0001, + "loss": 0.2914, + "step": 11500 + }, + { + "epoch": 3.01, + "learning_rate": 0.0001, + "loss": 0.305, + "step": 11510 + }, + { + "epoch": 3.01, + "learning_rate": 0.0001, + "loss": 0.2945, + "step": 11520 + }, + { + "epoch": 3.02, + "learning_rate": 0.0001, + "loss": 0.2934, + "step": 11530 + }, + { + "epoch": 3.02, + "learning_rate": 0.0001, + "loss": 0.3003, + "step": 11540 + }, + { + "epoch": 3.02, + "learning_rate": 0.0001, + "loss": 0.2883, + "step": 11550 + }, + { + "epoch": 3.02, + "learning_rate": 0.0001, + "loss": 0.2888, + "step": 11560 + }, + { + "epoch": 3.03, + "learning_rate": 0.0001, + "loss": 0.2919, + "step": 11570 + }, + { + "epoch": 3.03, + "learning_rate": 0.0001, + "loss": 0.2979, + "step": 11580 + }, + { + "epoch": 3.03, + "learning_rate": 0.0001, + "loss": 0.2906, + "step": 11590 + }, + { + "epoch": 3.04, + "learning_rate": 0.0001, + "loss": 0.3002, + "step": 11600 + }, + { + "epoch": 3.04, + "learning_rate": 0.0001, + "loss": 0.2961, + "step": 11610 + }, + { + "epoch": 3.04, + "learning_rate": 0.0001, + "loss": 0.3054, + "step": 11620 + }, + { + "epoch": 3.04, + "learning_rate": 0.0001, + "loss": 0.294, + "step": 11630 + }, + { + "epoch": 3.05, + "learning_rate": 0.0001, + "loss": 0.2879, + "step": 11640 + }, + { + "epoch": 3.05, + "learning_rate": 0.0001, + "loss": 0.2982, + "step": 11650 + }, + { + "epoch": 3.05, + "learning_rate": 0.0001, + "loss": 0.2931, + "step": 11660 + }, + { + "epoch": 3.05, + "learning_rate": 0.0001, + "loss": 0.3034, + "step": 11670 + }, + { + "epoch": 3.06, + "learning_rate": 0.0001, + "loss": 0.2978, + "step": 11680 + }, + { + "epoch": 3.06, + "learning_rate": 0.0001, + "loss": 0.3069, + "step": 11690 + }, + { + "epoch": 3.06, + "learning_rate": 0.0001, + "loss": 0.2852, + "step": 11700 + }, + { + "epoch": 3.06, + "learning_rate": 0.0001, + "loss": 0.2922, + "step": 11710 + }, + { + "epoch": 3.07, + "learning_rate": 0.0001, + "loss": 0.2937, + "step": 11720 + }, + { + "epoch": 3.07, + "learning_rate": 0.0001, + "loss": 0.278, + "step": 11730 + }, + { + "epoch": 3.07, + "learning_rate": 0.0001, + "loss": 0.3051, + "step": 11740 + }, + { + "epoch": 3.07, + "learning_rate": 0.0001, + "loss": 0.2927, + "step": 11750 + }, + { + "epoch": 3.08, + "learning_rate": 0.0001, + "loss": 0.3012, + "step": 11760 + }, + { + "epoch": 3.08, + "learning_rate": 0.0001, + "loss": 0.2901, + "step": 11770 + }, + { + "epoch": 3.08, + "learning_rate": 0.0001, + "loss": 0.3025, + "step": 11780 + }, + { + "epoch": 3.08, + "learning_rate": 0.0001, + "loss": 0.3085, + "step": 11790 + }, + { + "epoch": 3.09, + "learning_rate": 0.0001, + "loss": 0.2949, + "step": 11800 + }, + { + "epoch": 3.09, + "learning_rate": 0.0001, + "loss": 0.2991, + "step": 11810 + }, + { + "epoch": 3.09, + "learning_rate": 0.0001, + "loss": 0.2973, + "step": 11820 + }, + { + "epoch": 3.1, + "learning_rate": 0.0001, + "loss": 0.302, + "step": 11830 + }, + { + "epoch": 3.1, + "learning_rate": 0.0001, + "loss": 0.2995, + "step": 11840 + }, + { + "epoch": 3.1, + "learning_rate": 0.0001, + "loss": 0.2831, + "step": 11850 + }, + { + "epoch": 3.1, + "learning_rate": 0.0001, + "loss": 0.2959, + "step": 11860 + }, + { + "epoch": 3.11, + "learning_rate": 0.0001, + "loss": 0.3016, + "step": 11870 + }, + { + "epoch": 3.11, + "learning_rate": 0.0001, + "loss": 0.3002, + "step": 11880 + }, + { + "epoch": 3.11, + "learning_rate": 0.0001, + "loss": 0.2963, + "step": 11890 + }, + { + "epoch": 3.11, + "learning_rate": 0.0001, + "loss": 0.2926, + "step": 11900 + }, + { + "epoch": 3.12, + "learning_rate": 0.0001, + "loss": 0.2893, + "step": 11910 + }, + { + "epoch": 3.12, + "learning_rate": 0.0001, + "loss": 0.2987, + "step": 11920 + }, + { + "epoch": 3.12, + "learning_rate": 0.0001, + "loss": 0.3027, + "step": 11930 + }, + { + "epoch": 3.12, + "learning_rate": 0.0001, + "loss": 0.2953, + "step": 11940 + }, + { + "epoch": 3.13, + "learning_rate": 0.0001, + "loss": 0.2969, + "step": 11950 + }, + { + "epoch": 3.13, + "learning_rate": 0.0001, + "loss": 0.2984, + "step": 11960 + }, + { + "epoch": 3.13, + "learning_rate": 0.0001, + "loss": 0.3044, + "step": 11970 + }, + { + "epoch": 3.13, + "learning_rate": 0.0001, + "loss": 0.2943, + "step": 11980 + }, + { + "epoch": 3.14, + "learning_rate": 0.0001, + "loss": 0.2938, + "step": 11990 + }, + { + "epoch": 3.14, + "learning_rate": 0.0001, + "loss": 0.2973, + "step": 12000 + }, + { + "epoch": 3.14, + "learning_rate": 0.0001, + "loss": 0.3037, + "step": 12010 + }, + { + "epoch": 3.14, + "learning_rate": 0.0001, + "loss": 0.307, + "step": 12020 + }, + { + "epoch": 3.15, + "learning_rate": 0.0001, + "loss": 0.2917, + "step": 12030 + }, + { + "epoch": 3.15, + "learning_rate": 0.0001, + "loss": 0.2928, + "step": 12040 + }, + { + "epoch": 3.15, + "learning_rate": 0.0001, + "loss": 0.3021, + "step": 12050 + }, + { + "epoch": 3.16, + "learning_rate": 0.0001, + "loss": 0.2954, + "step": 12060 + }, + { + "epoch": 3.16, + "learning_rate": 0.0001, + "loss": 0.3083, + "step": 12070 + }, + { + "epoch": 3.16, + "learning_rate": 0.0001, + "loss": 0.2946, + "step": 12080 + }, + { + "epoch": 3.16, + "learning_rate": 0.0001, + "loss": 0.3013, + "step": 12090 + }, + { + "epoch": 3.17, + "learning_rate": 0.0001, + "loss": 0.2979, + "step": 12100 + }, + { + "epoch": 3.17, + "learning_rate": 0.0001, + "loss": 0.3007, + "step": 12110 + }, + { + "epoch": 3.17, + "learning_rate": 0.0001, + "loss": 0.3037, + "step": 12120 + }, + { + "epoch": 3.17, + "learning_rate": 0.0001, + "loss": 0.2926, + "step": 12130 + }, + { + "epoch": 3.18, + "learning_rate": 0.0001, + "loss": 0.3001, + "step": 12140 + }, + { + "epoch": 3.18, + "learning_rate": 0.0001, + "loss": 0.3067, + "step": 12150 + }, + { + "epoch": 3.18, + "learning_rate": 0.0001, + "loss": 0.3017, + "step": 12160 + }, + { + "epoch": 3.18, + "learning_rate": 0.0001, + "loss": 0.2978, + "step": 12170 + }, + { + "epoch": 3.19, + "learning_rate": 0.0001, + "loss": 0.3025, + "step": 12180 + }, + { + "epoch": 3.19, + "learning_rate": 0.0001, + "loss": 0.3048, + "step": 12190 + }, + { + "epoch": 3.19, + "learning_rate": 0.0001, + "loss": 0.2983, + "step": 12200 + }, + { + "epoch": 3.19, + "learning_rate": 0.0001, + "loss": 0.3074, + "step": 12210 + }, + { + "epoch": 3.2, + "learning_rate": 0.0001, + "loss": 0.2807, + "step": 12220 + }, + { + "epoch": 3.2, + "learning_rate": 0.0001, + "loss": 0.2969, + "step": 12230 + }, + { + "epoch": 3.2, + "learning_rate": 0.0001, + "loss": 0.3081, + "step": 12240 + }, + { + "epoch": 3.21, + "learning_rate": 0.0001, + "loss": 0.304, + "step": 12250 + }, + { + "epoch": 3.21, + "learning_rate": 0.0001, + "loss": 0.3029, + "step": 12260 + }, + { + "epoch": 3.21, + "learning_rate": 0.0001, + "loss": 0.3093, + "step": 12270 + }, + { + "epoch": 3.21, + "learning_rate": 0.0001, + "loss": 0.2954, + "step": 12280 + }, + { + "epoch": 3.22, + "learning_rate": 0.0001, + "loss": 0.3045, + "step": 12290 + }, + { + "epoch": 3.22, + "learning_rate": 0.0001, + "loss": 0.3079, + "step": 12300 + }, + { + "epoch": 3.22, + "learning_rate": 0.0001, + "loss": 0.3125, + "step": 12310 + }, + { + "epoch": 3.22, + "learning_rate": 0.0001, + "loss": 0.2965, + "step": 12320 + }, + { + "epoch": 3.23, + "learning_rate": 0.0001, + "loss": 0.3043, + "step": 12330 + }, + { + "epoch": 3.23, + "learning_rate": 0.0001, + "loss": 0.3028, + "step": 12340 + }, + { + "epoch": 3.23, + "learning_rate": 0.0001, + "loss": 0.2911, + "step": 12350 + }, + { + "epoch": 3.23, + "learning_rate": 0.0001, + "loss": 0.3046, + "step": 12360 + }, + { + "epoch": 3.24, + "learning_rate": 0.0001, + "loss": 0.2974, + "step": 12370 + }, + { + "epoch": 3.24, + "learning_rate": 0.0001, + "loss": 0.3041, + "step": 12380 + }, + { + "epoch": 3.24, + "learning_rate": 0.0001, + "loss": 0.3028, + "step": 12390 + }, + { + "epoch": 3.24, + "learning_rate": 0.0001, + "loss": 0.3023, + "step": 12400 + }, + { + "epoch": 3.25, + "learning_rate": 0.0001, + "loss": 0.3016, + "step": 12410 + }, + { + "epoch": 3.25, + "learning_rate": 0.0001, + "loss": 0.3032, + "step": 12420 + }, + { + "epoch": 3.25, + "learning_rate": 0.0001, + "loss": 0.2931, + "step": 12430 + }, + { + "epoch": 3.25, + "learning_rate": 0.0001, + "loss": 0.2949, + "step": 12440 + }, + { + "epoch": 3.26, + "learning_rate": 0.0001, + "loss": 0.2984, + "step": 12450 + }, + { + "epoch": 3.26, + "learning_rate": 0.0001, + "loss": 0.2948, + "step": 12460 + }, + { + "epoch": 3.26, + "learning_rate": 0.0001, + "loss": 0.3062, + "step": 12470 + }, + { + "epoch": 3.27, + "learning_rate": 0.0001, + "loss": 0.2986, + "step": 12480 + }, + { + "epoch": 3.27, + "learning_rate": 0.0001, + "loss": 0.2945, + "step": 12490 + }, + { + "epoch": 3.27, + "learning_rate": 0.0001, + "loss": 0.3149, + "step": 12500 + }, + { + "epoch": 3.27, + "learning_rate": 0.0001, + "loss": 0.2979, + "step": 12510 + }, + { + "epoch": 3.28, + "learning_rate": 0.0001, + "loss": 0.3022, + "step": 12520 + }, + { + "epoch": 3.28, + "learning_rate": 0.0001, + "loss": 0.3098, + "step": 12530 + }, + { + "epoch": 3.28, + "learning_rate": 0.0001, + "loss": 0.3078, + "step": 12540 + }, + { + "epoch": 3.28, + "learning_rate": 0.0001, + "loss": 0.296, + "step": 12550 + }, + { + "epoch": 3.29, + "learning_rate": 0.0001, + "loss": 0.2975, + "step": 12560 + }, + { + "epoch": 3.29, + "learning_rate": 0.0001, + "loss": 0.3084, + "step": 12570 + }, + { + "epoch": 3.29, + "learning_rate": 0.0001, + "loss": 0.3008, + "step": 12580 + }, + { + "epoch": 3.29, + "learning_rate": 0.0001, + "loss": 0.3065, + "step": 12590 + }, + { + "epoch": 3.3, + "learning_rate": 0.0001, + "loss": 0.3068, + "step": 12600 + }, + { + "epoch": 3.3, + "learning_rate": 0.0001, + "loss": 0.3056, + "step": 12610 + }, + { + "epoch": 3.3, + "learning_rate": 0.0001, + "loss": 0.3097, + "step": 12620 + }, + { + "epoch": 3.3, + "learning_rate": 0.0001, + "loss": 0.2993, + "step": 12630 + }, + { + "epoch": 3.31, + "learning_rate": 0.0001, + "loss": 0.3103, + "step": 12640 + }, + { + "epoch": 3.31, + "learning_rate": 0.0001, + "loss": 0.2971, + "step": 12650 + }, + { + "epoch": 3.31, + "learning_rate": 0.0001, + "loss": 0.2919, + "step": 12660 + }, + { + "epoch": 3.32, + "learning_rate": 0.0001, + "loss": 0.3061, + "step": 12670 + }, + { + "epoch": 3.32, + "learning_rate": 0.0001, + "loss": 0.2939, + "step": 12680 + }, + { + "epoch": 3.32, + "learning_rate": 0.0001, + "loss": 0.3013, + "step": 12690 + }, + { + "epoch": 3.32, + "learning_rate": 0.0001, + "loss": 0.307, + "step": 12700 + }, + { + "epoch": 3.33, + "learning_rate": 0.0001, + "loss": 0.312, + "step": 12710 + }, + { + "epoch": 3.33, + "learning_rate": 0.0001, + "loss": 0.3101, + "step": 12720 + }, + { + "epoch": 3.33, + "learning_rate": 0.0001, + "loss": 0.3154, + "step": 12730 + }, + { + "epoch": 3.33, + "learning_rate": 0.0001, + "loss": 0.3089, + "step": 12740 + }, + { + "epoch": 3.34, + "learning_rate": 0.0001, + "loss": 0.2994, + "step": 12750 + }, + { + "epoch": 3.34, + "learning_rate": 0.0001, + "loss": 0.3079, + "step": 12760 + }, + { + "epoch": 3.34, + "learning_rate": 0.0001, + "loss": 0.3048, + "step": 12770 + }, + { + "epoch": 3.34, + "learning_rate": 0.0001, + "loss": 0.2955, + "step": 12780 + }, + { + "epoch": 3.35, + "learning_rate": 0.0001, + "loss": 0.3004, + "step": 12790 + }, + { + "epoch": 3.35, + "learning_rate": 0.0001, + "loss": 0.3106, + "step": 12800 + }, + { + "epoch": 3.35, + "learning_rate": 0.0001, + "loss": 0.2989, + "step": 12810 + }, + { + "epoch": 3.35, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 12820 + }, + { + "epoch": 3.36, + "learning_rate": 0.0001, + "loss": 0.2989, + "step": 12830 + }, + { + "epoch": 3.36, + "learning_rate": 0.0001, + "loss": 0.2979, + "step": 12840 + }, + { + "epoch": 3.36, + "learning_rate": 0.0001, + "loss": 0.307, + "step": 12850 + }, + { + "epoch": 3.36, + "learning_rate": 0.0001, + "loss": 0.3133, + "step": 12860 + }, + { + "epoch": 3.37, + "learning_rate": 0.0001, + "loss": 0.3068, + "step": 12870 + }, + { + "epoch": 3.37, + "learning_rate": 0.0001, + "loss": 0.3051, + "step": 12880 + }, + { + "epoch": 3.37, + "learning_rate": 0.0001, + "loss": 0.2972, + "step": 12890 + }, + { + "epoch": 3.38, + "learning_rate": 0.0001, + "loss": 0.3043, + "step": 12900 + }, + { + "epoch": 3.38, + "learning_rate": 0.0001, + "loss": 0.3046, + "step": 12910 + }, + { + "epoch": 3.38, + "learning_rate": 0.0001, + "loss": 0.2992, + "step": 12920 + }, + { + "epoch": 3.38, + "learning_rate": 0.0001, + "loss": 0.2953, + "step": 12930 + }, + { + "epoch": 3.39, + "learning_rate": 0.0001, + "loss": 0.2996, + "step": 12940 + }, + { + "epoch": 3.39, + "learning_rate": 0.0001, + "loss": 0.312, + "step": 12950 + }, + { + "epoch": 3.39, + "learning_rate": 0.0001, + "loss": 0.3071, + "step": 12960 + }, + { + "epoch": 3.39, + "learning_rate": 0.0001, + "loss": 0.3057, + "step": 12970 + }, + { + "epoch": 3.4, + "learning_rate": 0.0001, + "loss": 0.3091, + "step": 12980 + }, + { + "epoch": 3.4, + "learning_rate": 0.0001, + "loss": 0.3094, + "step": 12990 + }, + { + "epoch": 3.4, + "learning_rate": 0.0001, + "loss": 0.3059, + "step": 13000 + }, + { + "epoch": 3.4, + "learning_rate": 0.0001, + "loss": 0.3019, + "step": 13010 + }, + { + "epoch": 3.41, + "learning_rate": 0.0001, + "loss": 0.3103, + "step": 13020 + }, + { + "epoch": 3.41, + "learning_rate": 0.0001, + "loss": 0.3057, + "step": 13030 + }, + { + "epoch": 3.41, + "learning_rate": 0.0001, + "loss": 0.3096, + "step": 13040 + }, + { + "epoch": 3.41, + "learning_rate": 0.0001, + "loss": 0.3061, + "step": 13050 + }, + { + "epoch": 3.42, + "learning_rate": 0.0001, + "loss": 0.2989, + "step": 13060 + }, + { + "epoch": 3.42, + "learning_rate": 0.0001, + "loss": 0.2963, + "step": 13070 + }, + { + "epoch": 3.42, + "learning_rate": 0.0001, + "loss": 0.3109, + "step": 13080 + }, + { + "epoch": 3.42, + "learning_rate": 0.0001, + "loss": 0.2972, + "step": 13090 + }, + { + "epoch": 3.43, + "learning_rate": 0.0001, + "loss": 0.3085, + "step": 13100 + }, + { + "epoch": 3.43, + "learning_rate": 0.0001, + "loss": 0.3041, + "step": 13110 + }, + { + "epoch": 3.43, + "learning_rate": 0.0001, + "loss": 0.2978, + "step": 13120 + }, + { + "epoch": 3.44, + "learning_rate": 0.0001, + "loss": 0.3063, + "step": 13130 + }, + { + "epoch": 3.44, + "learning_rate": 0.0001, + "loss": 0.2994, + "step": 13140 + }, + { + "epoch": 3.44, + "learning_rate": 0.0001, + "loss": 0.2958, + "step": 13150 + }, + { + "epoch": 3.44, + "learning_rate": 0.0001, + "loss": 0.2986, + "step": 13160 + }, + { + "epoch": 3.45, + "learning_rate": 0.0001, + "loss": 0.3155, + "step": 13170 + }, + { + "epoch": 3.45, + "learning_rate": 0.0001, + "loss": 0.3026, + "step": 13180 + }, + { + "epoch": 3.45, + "learning_rate": 0.0001, + "loss": 0.3002, + "step": 13190 + }, + { + "epoch": 3.45, + "learning_rate": 0.0001, + "loss": 0.3041, + "step": 13200 + }, + { + "epoch": 3.46, + "learning_rate": 0.0001, + "loss": 0.3054, + "step": 13210 + }, + { + "epoch": 3.46, + "learning_rate": 0.0001, + "loss": 0.3126, + "step": 13220 + }, + { + "epoch": 3.46, + "learning_rate": 0.0001, + "loss": 0.2893, + "step": 13230 + }, + { + "epoch": 3.46, + "learning_rate": 0.0001, + "loss": 0.3056, + "step": 13240 + }, + { + "epoch": 3.47, + "learning_rate": 0.0001, + "loss": 0.2911, + "step": 13250 + }, + { + "epoch": 3.47, + "learning_rate": 0.0001, + "loss": 0.3059, + "step": 13260 + }, + { + "epoch": 3.47, + "learning_rate": 0.0001, + "loss": 0.3026, + "step": 13270 + }, + { + "epoch": 3.47, + "learning_rate": 0.0001, + "loss": 0.301, + "step": 13280 + }, + { + "epoch": 3.48, + "learning_rate": 0.0001, + "loss": 0.3059, + "step": 13290 + }, + { + "epoch": 3.48, + "learning_rate": 0.0001, + "loss": 0.2945, + "step": 13300 + }, + { + "epoch": 3.48, + "learning_rate": 0.0001, + "loss": 0.2987, + "step": 13310 + }, + { + "epoch": 3.49, + "learning_rate": 0.0001, + "loss": 0.302, + "step": 13320 + }, + { + "epoch": 3.49, + "learning_rate": 0.0001, + "loss": 0.3067, + "step": 13330 + }, + { + "epoch": 3.49, + "learning_rate": 0.0001, + "loss": 0.3045, + "step": 13340 + }, + { + "epoch": 3.49, + "learning_rate": 0.0001, + "loss": 0.3001, + "step": 13350 + }, + { + "epoch": 3.5, + "learning_rate": 0.0001, + "loss": 0.3036, + "step": 13360 + }, + { + "epoch": 3.5, + "learning_rate": 0.0001, + "loss": 0.3096, + "step": 13370 + }, + { + "epoch": 3.5, + "learning_rate": 0.0001, + "loss": 0.3169, + "step": 13380 + }, + { + "epoch": 3.5, + "learning_rate": 0.0001, + "loss": 0.3169, + "step": 13390 + }, + { + "epoch": 3.51, + "learning_rate": 0.0001, + "loss": 0.2933, + "step": 13400 + }, + { + "epoch": 3.51, + "learning_rate": 0.0001, + "loss": 0.3054, + "step": 13410 + }, + { + "epoch": 3.51, + "learning_rate": 0.0001, + "loss": 0.2971, + "step": 13420 + }, + { + "epoch": 3.51, + "learning_rate": 0.0001, + "loss": 0.318, + "step": 13430 + }, + { + "epoch": 3.52, + "learning_rate": 0.0001, + "loss": 0.3043, + "step": 13440 + }, + { + "epoch": 3.52, + "learning_rate": 0.0001, + "loss": 0.3074, + "step": 13450 + }, + { + "epoch": 3.52, + "learning_rate": 0.0001, + "loss": 0.3116, + "step": 13460 + }, + { + "epoch": 3.52, + "learning_rate": 0.0001, + "loss": 0.3158, + "step": 13470 + }, + { + "epoch": 3.53, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 13480 + }, + { + "epoch": 3.53, + "learning_rate": 0.0001, + "loss": 0.3118, + "step": 13490 + }, + { + "epoch": 3.53, + "learning_rate": 0.0001, + "loss": 0.3012, + "step": 13500 + }, + { + "epoch": 3.53, + "learning_rate": 0.0001, + "loss": 0.3001, + "step": 13510 + }, + { + "epoch": 3.54, + "learning_rate": 0.0001, + "loss": 0.3226, + "step": 13520 + }, + { + "epoch": 3.54, + "learning_rate": 0.0001, + "loss": 0.3103, + "step": 13530 + }, + { + "epoch": 3.54, + "learning_rate": 0.0001, + "loss": 0.305, + "step": 13540 + }, + { + "epoch": 3.55, + "learning_rate": 0.0001, + "loss": 0.2912, + "step": 13550 + }, + { + "epoch": 3.55, + "learning_rate": 0.0001, + "loss": 0.3024, + "step": 13560 + }, + { + "epoch": 3.55, + "learning_rate": 0.0001, + "loss": 0.3006, + "step": 13570 + }, + { + "epoch": 3.55, + "learning_rate": 0.0001, + "loss": 0.3114, + "step": 13580 + }, + { + "epoch": 3.56, + "learning_rate": 0.0001, + "loss": 0.3172, + "step": 13590 + }, + { + "epoch": 3.56, + "learning_rate": 0.0001, + "loss": 0.3096, + "step": 13600 + }, + { + "epoch": 3.56, + "learning_rate": 0.0001, + "loss": 0.3049, + "step": 13610 + }, + { + "epoch": 3.56, + "learning_rate": 0.0001, + "loss": 0.3112, + "step": 13620 + }, + { + "epoch": 3.57, + "learning_rate": 0.0001, + "loss": 0.3171, + "step": 13630 + }, + { + "epoch": 3.57, + "learning_rate": 0.0001, + "loss": 0.3071, + "step": 13640 + }, + { + "epoch": 3.57, + "learning_rate": 0.0001, + "loss": 0.3132, + "step": 13650 + }, + { + "epoch": 3.57, + "learning_rate": 0.0001, + "loss": 0.31, + "step": 13660 + }, + { + "epoch": 3.58, + "learning_rate": 0.0001, + "loss": 0.3077, + "step": 13670 + }, + { + "epoch": 3.58, + "learning_rate": 0.0001, + "loss": 0.3024, + "step": 13680 + }, + { + "epoch": 3.58, + "learning_rate": 0.0001, + "loss": 0.3149, + "step": 13690 + }, + { + "epoch": 3.58, + "learning_rate": 0.0001, + "loss": 0.3066, + "step": 13700 + }, + { + "epoch": 3.59, + "learning_rate": 0.0001, + "loss": 0.3083, + "step": 13710 + }, + { + "epoch": 3.59, + "learning_rate": 0.0001, + "loss": 0.3064, + "step": 13720 + }, + { + "epoch": 3.59, + "learning_rate": 0.0001, + "loss": 0.3049, + "step": 13730 + }, + { + "epoch": 3.59, + "learning_rate": 0.0001, + "loss": 0.31, + "step": 13740 + }, + { + "epoch": 3.6, + "learning_rate": 0.0001, + "loss": 0.3102, + "step": 13750 + }, + { + "epoch": 3.6, + "learning_rate": 0.0001, + "loss": 0.3027, + "step": 13760 + }, + { + "epoch": 3.6, + "learning_rate": 0.0001, + "loss": 0.3081, + "step": 13770 + }, + { + "epoch": 3.61, + "learning_rate": 0.0001, + "loss": 0.3064, + "step": 13780 + }, + { + "epoch": 3.61, + "learning_rate": 0.0001, + "loss": 0.3113, + "step": 13790 + }, + { + "epoch": 3.61, + "learning_rate": 0.0001, + "loss": 0.3299, + "step": 13800 + }, + { + "epoch": 3.61, + "learning_rate": 0.0001, + "loss": 0.303, + "step": 13810 + }, + { + "epoch": 3.62, + "learning_rate": 0.0001, + "loss": 0.3113, + "step": 13820 + }, + { + "epoch": 3.62, + "learning_rate": 0.0001, + "loss": 0.3026, + "step": 13830 + }, + { + "epoch": 3.62, + "learning_rate": 0.0001, + "loss": 0.3007, + "step": 13840 + }, + { + "epoch": 3.62, + "learning_rate": 0.0001, + "loss": 0.3101, + "step": 13850 + }, + { + "epoch": 3.63, + "learning_rate": 0.0001, + "loss": 0.3013, + "step": 13860 + }, + { + "epoch": 3.63, + "learning_rate": 0.0001, + "loss": 0.2986, + "step": 13870 + }, + { + "epoch": 3.63, + "learning_rate": 0.0001, + "loss": 0.3023, + "step": 13880 + }, + { + "epoch": 3.63, + "learning_rate": 0.0001, + "loss": 0.3027, + "step": 13890 + }, + { + "epoch": 3.64, + "learning_rate": 0.0001, + "loss": 0.3012, + "step": 13900 + }, + { + "epoch": 3.64, + "learning_rate": 0.0001, + "loss": 0.3163, + "step": 13910 + }, + { + "epoch": 3.64, + "learning_rate": 0.0001, + "loss": 0.3006, + "step": 13920 + }, + { + "epoch": 3.64, + "learning_rate": 0.0001, + "loss": 0.3277, + "step": 13930 + }, + { + "epoch": 3.65, + "learning_rate": 0.0001, + "loss": 0.3, + "step": 13940 + }, + { + "epoch": 3.65, + "learning_rate": 0.0001, + "loss": 0.3109, + "step": 13950 + }, + { + "epoch": 3.65, + "learning_rate": 0.0001, + "loss": 0.3138, + "step": 13960 + }, + { + "epoch": 3.66, + "learning_rate": 0.0001, + "loss": 0.3258, + "step": 13970 + }, + { + "epoch": 3.66, + "learning_rate": 0.0001, + "loss": 0.3125, + "step": 13980 + }, + { + "epoch": 3.66, + "learning_rate": 0.0001, + "loss": 0.3094, + "step": 13990 + }, + { + "epoch": 3.66, + "learning_rate": 0.0001, + "loss": 0.3005, + "step": 14000 + }, + { + "epoch": 3.67, + "learning_rate": 0.0001, + "loss": 0.3015, + "step": 14010 + }, + { + "epoch": 3.67, + "learning_rate": 0.0001, + "loss": 0.3077, + "step": 14020 + }, + { + "epoch": 3.67, + "learning_rate": 0.0001, + "loss": 0.316, + "step": 14030 + }, + { + "epoch": 3.67, + "learning_rate": 0.0001, + "loss": 0.3163, + "step": 14040 + }, + { + "epoch": 3.68, + "learning_rate": 0.0001, + "loss": 0.3077, + "step": 14050 + }, + { + "epoch": 3.68, + "learning_rate": 0.0001, + "loss": 0.3078, + "step": 14060 + }, + { + "epoch": 3.68, + "learning_rate": 0.0001, + "loss": 0.3166, + "step": 14070 + }, + { + "epoch": 3.68, + "learning_rate": 0.0001, + "loss": 0.304, + "step": 14080 + }, + { + "epoch": 3.69, + "learning_rate": 0.0001, + "loss": 0.2997, + "step": 14090 + }, + { + "epoch": 3.69, + "learning_rate": 0.0001, + "loss": 0.3249, + "step": 14100 + }, + { + "epoch": 3.69, + "learning_rate": 0.0001, + "loss": 0.2948, + "step": 14110 + }, + { + "epoch": 3.69, + "learning_rate": 0.0001, + "loss": 0.303, + "step": 14120 + }, + { + "epoch": 3.7, + "learning_rate": 0.0001, + "loss": 0.306, + "step": 14130 + }, + { + "epoch": 3.7, + "learning_rate": 0.0001, + "loss": 0.3088, + "step": 14140 + }, + { + "epoch": 3.7, + "learning_rate": 0.0001, + "loss": 0.317, + "step": 14150 + }, + { + "epoch": 3.7, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 14160 + }, + { + "epoch": 3.71, + "learning_rate": 0.0001, + "loss": 0.3202, + "step": 14170 + }, + { + "epoch": 3.71, + "learning_rate": 0.0001, + "loss": 0.3049, + "step": 14180 + }, + { + "epoch": 3.71, + "learning_rate": 0.0001, + "loss": 0.3095, + "step": 14190 + }, + { + "epoch": 3.72, + "learning_rate": 0.0001, + "loss": 0.3134, + "step": 14200 + }, + { + "epoch": 3.72, + "learning_rate": 0.0001, + "loss": 0.3171, + "step": 14210 + }, + { + "epoch": 3.72, + "learning_rate": 0.0001, + "loss": 0.3079, + "step": 14220 + }, + { + "epoch": 3.72, + "learning_rate": 0.0001, + "loss": 0.3058, + "step": 14230 + }, + { + "epoch": 3.73, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 14240 + }, + { + "epoch": 3.73, + "learning_rate": 0.0001, + "loss": 0.3128, + "step": 14250 + }, + { + "epoch": 3.73, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 14260 + }, + { + "epoch": 3.73, + "learning_rate": 0.0001, + "loss": 0.3028, + "step": 14270 + }, + { + "epoch": 3.74, + "learning_rate": 0.0001, + "loss": 0.3025, + "step": 14280 + }, + { + "epoch": 3.74, + "learning_rate": 0.0001, + "loss": 0.3082, + "step": 14290 + }, + { + "epoch": 3.74, + "learning_rate": 0.0001, + "loss": 0.3164, + "step": 14300 + }, + { + "epoch": 3.74, + "learning_rate": 0.0001, + "loss": 0.2971, + "step": 14310 + }, + { + "epoch": 3.75, + "learning_rate": 0.0001, + "loss": 0.3147, + "step": 14320 + }, + { + "epoch": 3.75, + "learning_rate": 0.0001, + "loss": 0.3099, + "step": 14330 + }, + { + "epoch": 3.75, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 14340 + }, + { + "epoch": 3.75, + "learning_rate": 0.0001, + "loss": 0.3097, + "step": 14350 + }, + { + "epoch": 3.76, + "learning_rate": 0.0001, + "loss": 0.31, + "step": 14360 + }, + { + "epoch": 3.76, + "learning_rate": 0.0001, + "loss": 0.3036, + "step": 14370 + }, + { + "epoch": 3.76, + "learning_rate": 0.0001, + "loss": 0.3068, + "step": 14380 + }, + { + "epoch": 3.77, + "learning_rate": 0.0001, + "loss": 0.3035, + "step": 14390 + }, + { + "epoch": 3.77, + "learning_rate": 0.0001, + "loss": 0.2993, + "step": 14400 + }, + { + "epoch": 3.77, + "learning_rate": 0.0001, + "loss": 0.31, + "step": 14410 + }, + { + "epoch": 3.77, + "learning_rate": 0.0001, + "loss": 0.3112, + "step": 14420 + }, + { + "epoch": 3.78, + "learning_rate": 0.0001, + "loss": 0.3083, + "step": 14430 + }, + { + "epoch": 3.78, + "learning_rate": 0.0001, + "loss": 0.2994, + "step": 14440 + }, + { + "epoch": 3.78, + "learning_rate": 0.0001, + "loss": 0.3151, + "step": 14450 + }, + { + "epoch": 3.78, + "learning_rate": 0.0001, + "loss": 0.3115, + "step": 14460 + }, + { + "epoch": 3.79, + "learning_rate": 0.0001, + "loss": 0.307, + "step": 14470 + }, + { + "epoch": 3.79, + "learning_rate": 0.0001, + "loss": 0.3088, + "step": 14480 + }, + { + "epoch": 3.79, + "learning_rate": 0.0001, + "loss": 0.3069, + "step": 14490 + }, + { + "epoch": 3.79, + "learning_rate": 0.0001, + "loss": 0.3006, + "step": 14500 + }, + { + "epoch": 3.8, + "learning_rate": 0.0001, + "loss": 0.3033, + "step": 14510 + }, + { + "epoch": 3.8, + "learning_rate": 0.0001, + "loss": 0.3132, + "step": 14520 + }, + { + "epoch": 3.8, + "learning_rate": 0.0001, + "loss": 0.311, + "step": 14530 + }, + { + "epoch": 3.8, + "learning_rate": 0.0001, + "loss": 0.3286, + "step": 14540 + }, + { + "epoch": 3.81, + "learning_rate": 0.0001, + "loss": 0.3006, + "step": 14550 + }, + { + "epoch": 3.81, + "learning_rate": 0.0001, + "loss": 0.2988, + "step": 14560 + }, + { + "epoch": 3.81, + "learning_rate": 0.0001, + "loss": 0.3033, + "step": 14570 + }, + { + "epoch": 3.81, + "learning_rate": 0.0001, + "loss": 0.3073, + "step": 14580 + }, + { + "epoch": 3.82, + "learning_rate": 0.0001, + "loss": 0.3132, + "step": 14590 + }, + { + "epoch": 3.82, + "learning_rate": 0.0001, + "loss": 0.3047, + "step": 14600 + }, + { + "epoch": 3.82, + "learning_rate": 0.0001, + "loss": 0.3058, + "step": 14610 + }, + { + "epoch": 3.83, + "learning_rate": 0.0001, + "loss": 0.3021, + "step": 14620 + }, + { + "epoch": 3.83, + "learning_rate": 0.0001, + "loss": 0.3072, + "step": 14630 + }, + { + "epoch": 3.83, + "learning_rate": 0.0001, + "loss": 0.3, + "step": 14640 + }, + { + "epoch": 3.83, + "learning_rate": 0.0001, + "loss": 0.3246, + "step": 14650 + }, + { + "epoch": 3.84, + "learning_rate": 0.0001, + "loss": 0.2984, + "step": 14660 + }, + { + "epoch": 3.84, + "learning_rate": 0.0001, + "loss": 0.3139, + "step": 14670 + }, + { + "epoch": 3.84, + "learning_rate": 0.0001, + "loss": 0.3086, + "step": 14680 + }, + { + "epoch": 3.84, + "learning_rate": 0.0001, + "loss": 0.3093, + "step": 14690 + }, + { + "epoch": 3.85, + "learning_rate": 0.0001, + "loss": 0.3133, + "step": 14700 + }, + { + "epoch": 3.85, + "learning_rate": 0.0001, + "loss": 0.314, + "step": 14710 + }, + { + "epoch": 3.85, + "learning_rate": 0.0001, + "loss": 0.3135, + "step": 14720 + }, + { + "epoch": 3.85, + "learning_rate": 0.0001, + "loss": 0.3152, + "step": 14730 + }, + { + "epoch": 3.86, + "learning_rate": 0.0001, + "loss": 0.2994, + "step": 14740 + }, + { + "epoch": 3.86, + "learning_rate": 0.0001, + "loss": 0.3296, + "step": 14750 + }, + { + "epoch": 3.86, + "learning_rate": 0.0001, + "loss": 0.3003, + "step": 14760 + }, + { + "epoch": 3.86, + "learning_rate": 0.0001, + "loss": 0.3212, + "step": 14770 + }, + { + "epoch": 3.87, + "learning_rate": 0.0001, + "loss": 0.3149, + "step": 14780 + }, + { + "epoch": 3.87, + "learning_rate": 0.0001, + "loss": 0.3083, + "step": 14790 + }, + { + "epoch": 3.87, + "learning_rate": 0.0001, + "loss": 0.3108, + "step": 14800 + }, + { + "epoch": 3.87, + "learning_rate": 0.0001, + "loss": 0.299, + "step": 14810 + }, + { + "epoch": 3.88, + "learning_rate": 0.0001, + "loss": 0.291, + "step": 14820 + }, + { + "epoch": 3.88, + "learning_rate": 0.0001, + "loss": 0.3059, + "step": 14830 + }, + { + "epoch": 3.88, + "learning_rate": 0.0001, + "loss": 0.3124, + "step": 14840 + }, + { + "epoch": 3.89, + "learning_rate": 0.0001, + "loss": 0.3181, + "step": 14850 + }, + { + "epoch": 3.89, + "learning_rate": 0.0001, + "loss": 0.3016, + "step": 14860 + }, + { + "epoch": 3.89, + "learning_rate": 0.0001, + "loss": 0.3123, + "step": 14870 + }, + { + "epoch": 3.89, + "learning_rate": 0.0001, + "loss": 0.3096, + "step": 14880 + }, + { + "epoch": 3.9, + "learning_rate": 0.0001, + "loss": 0.3077, + "step": 14890 + }, + { + "epoch": 3.9, + "learning_rate": 0.0001, + "loss": 0.3012, + "step": 14900 + }, + { + "epoch": 3.9, + "learning_rate": 0.0001, + "loss": 0.3123, + "step": 14910 + }, + { + "epoch": 3.9, + "learning_rate": 0.0001, + "loss": 0.312, + "step": 14920 + }, + { + "epoch": 3.91, + "learning_rate": 0.0001, + "loss": 0.3157, + "step": 14930 + }, + { + "epoch": 3.91, + "learning_rate": 0.0001, + "loss": 0.318, + "step": 14940 + }, + { + "epoch": 3.91, + "learning_rate": 0.0001, + "loss": 0.3099, + "step": 14950 + }, + { + "epoch": 3.91, + "learning_rate": 0.0001, + "loss": 0.3214, + "step": 14960 + }, + { + "epoch": 3.92, + "learning_rate": 0.0001, + "loss": 0.3166, + "step": 14970 + }, + { + "epoch": 3.92, + "learning_rate": 0.0001, + "loss": 0.2989, + "step": 14980 + }, + { + "epoch": 3.92, + "learning_rate": 0.0001, + "loss": 0.3081, + "step": 14990 + }, + { + "epoch": 3.92, + "learning_rate": 0.0001, + "loss": 0.3096, + "step": 15000 + }, + { + "epoch": 3.93, + "learning_rate": 0.0001, + "loss": 0.3187, + "step": 15010 + }, + { + "epoch": 3.93, + "learning_rate": 0.0001, + "loss": 0.3129, + "step": 15020 + }, + { + "epoch": 3.93, + "learning_rate": 0.0001, + "loss": 0.2977, + "step": 15030 + }, + { + "epoch": 3.94, + "learning_rate": 0.0001, + "loss": 0.3203, + "step": 15040 + }, + { + "epoch": 3.94, + "learning_rate": 0.0001, + "loss": 0.304, + "step": 15050 + }, + { + "epoch": 3.94, + "learning_rate": 0.0001, + "loss": 0.318, + "step": 15060 + }, + { + "epoch": 3.94, + "learning_rate": 0.0001, + "loss": 0.3215, + "step": 15070 + }, + { + "epoch": 3.95, + "learning_rate": 0.0001, + "loss": 0.3208, + "step": 15080 + }, + { + "epoch": 3.95, + "learning_rate": 0.0001, + "loss": 0.3049, + "step": 15090 + }, + { + "epoch": 3.95, + "learning_rate": 0.0001, + "loss": 0.3066, + "step": 15100 + }, + { + "epoch": 3.95, + "learning_rate": 0.0001, + "loss": 0.3101, + "step": 15110 + }, + { + "epoch": 3.96, + "learning_rate": 0.0001, + "loss": 0.3189, + "step": 15120 + }, + { + "epoch": 3.96, + "learning_rate": 0.0001, + "loss": 0.3068, + "step": 15130 + }, + { + "epoch": 3.96, + "learning_rate": 0.0001, + "loss": 0.3019, + "step": 15140 + }, + { + "epoch": 3.96, + "learning_rate": 0.0001, + "loss": 0.3079, + "step": 15150 + }, + { + "epoch": 3.97, + "learning_rate": 0.0001, + "loss": 0.3013, + "step": 15160 + }, + { + "epoch": 3.97, + "learning_rate": 0.0001, + "loss": 0.3128, + "step": 15170 + }, + { + "epoch": 3.97, + "learning_rate": 0.0001, + "loss": 0.3108, + "step": 15180 + }, + { + "epoch": 3.97, + "learning_rate": 0.0001, + "loss": 0.3046, + "step": 15190 + }, + { + "epoch": 3.98, + "learning_rate": 0.0001, + "loss": 0.316, + "step": 15200 + }, + { + "epoch": 3.98, + "learning_rate": 0.0001, + "loss": 0.296, + "step": 15210 + }, + { + "epoch": 3.98, + "learning_rate": 0.0001, + "loss": 0.3118, + "step": 15220 + }, + { + "epoch": 3.98, + "learning_rate": 0.0001, + "loss": 0.3137, + "step": 15230 + }, + { + "epoch": 3.99, + "learning_rate": 0.0001, + "loss": 0.3086, + "step": 15240 + }, + { + "epoch": 3.99, + "learning_rate": 0.0001, + "loss": 0.3102, + "step": 15250 + }, + { + "epoch": 3.99, + "learning_rate": 0.0001, + "loss": 0.299, + "step": 15260 + }, + { + "epoch": 4.0, + "learning_rate": 0.0001, + "loss": 0.3167, + "step": 15270 + }, + { + "epoch": 4.0, + "learning_rate": 0.0001, + "loss": 0.3123, + "step": 15280 + }, + { + "epoch": 4.0, + "eval_gen_len": 61.711, + "eval_loss": 1.057576060295105, + "eval_rouge1": 56.7373, + "eval_rouge2": 34.8953, + "eval_rougeL": 46.3477, + "eval_rougeLsum": 52.3934, + "eval_runtime": 39.8843, + "eval_samples_per_second": 22.039, + "eval_steps_per_second": 0.351, + "step": 15288 + }, + { + "epoch": 4.0, + "learning_rate": 0.0001, + "loss": 0.288, + "step": 15290 + }, + { + "epoch": 4.0, + "learning_rate": 0.0001, + "loss": 0.2515, + "step": 15300 + }, + { + "epoch": 4.01, + "learning_rate": 0.0001, + "loss": 0.2603, + "step": 15310 + }, + { + "epoch": 4.01, + "learning_rate": 0.0001, + "loss": 0.2555, + "step": 15320 + }, + { + "epoch": 4.01, + "learning_rate": 0.0001, + "loss": 0.2598, + "step": 15330 + }, + { + "epoch": 4.01, + "learning_rate": 0.0001, + "loss": 0.2556, + "step": 15340 + }, + { + "epoch": 4.02, + "learning_rate": 0.0001, + "loss": 0.2638, + "step": 15350 + }, + { + "epoch": 4.02, + "learning_rate": 0.0001, + "loss": 0.2499, + "step": 15360 + }, + { + "epoch": 4.02, + "learning_rate": 0.0001, + "loss": 0.2495, + "step": 15370 + }, + { + "epoch": 4.02, + "learning_rate": 0.0001, + "loss": 0.2546, + "step": 15380 + }, + { + "epoch": 4.03, + "learning_rate": 0.0001, + "loss": 0.2656, + "step": 15390 + }, + { + "epoch": 4.03, + "learning_rate": 0.0001, + "loss": 0.2513, + "step": 15400 + }, + { + "epoch": 4.03, + "learning_rate": 0.0001, + "loss": 0.2508, + "step": 15410 + }, + { + "epoch": 4.03, + "learning_rate": 0.0001, + "loss": 0.2515, + "step": 15420 + }, + { + "epoch": 4.04, + "learning_rate": 0.0001, + "loss": 0.2655, + "step": 15430 + }, + { + "epoch": 4.04, + "learning_rate": 0.0001, + "loss": 0.2599, + "step": 15440 + }, + { + "epoch": 4.04, + "learning_rate": 0.0001, + "loss": 0.259, + "step": 15450 + }, + { + "epoch": 4.05, + "learning_rate": 0.0001, + "loss": 0.26, + "step": 15460 + }, + { + "epoch": 4.05, + "learning_rate": 0.0001, + "loss": 0.2414, + "step": 15470 + }, + { + "epoch": 4.05, + "learning_rate": 0.0001, + "loss": 0.2577, + "step": 15480 + }, + { + "epoch": 4.05, + "learning_rate": 0.0001, + "loss": 0.262, + "step": 15490 + }, + { + "epoch": 4.06, + "learning_rate": 0.0001, + "loss": 0.2533, + "step": 15500 + }, + { + "epoch": 4.06, + "learning_rate": 0.0001, + "loss": 0.2654, + "step": 15510 + }, + { + "epoch": 4.06, + "learning_rate": 0.0001, + "loss": 0.2594, + "step": 15520 + }, + { + "epoch": 4.06, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 15530 + }, + { + "epoch": 4.07, + "learning_rate": 0.0001, + "loss": 0.2607, + "step": 15540 + }, + { + "epoch": 4.07, + "learning_rate": 0.0001, + "loss": 0.2574, + "step": 15550 + }, + { + "epoch": 4.07, + "learning_rate": 0.0001, + "loss": 0.2611, + "step": 15560 + }, + { + "epoch": 4.07, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 15570 + }, + { + "epoch": 4.08, + "learning_rate": 0.0001, + "loss": 0.2711, + "step": 15580 + }, + { + "epoch": 4.08, + "learning_rate": 0.0001, + "loss": 0.2599, + "step": 15590 + }, + { + "epoch": 4.08, + "learning_rate": 0.0001, + "loss": 0.2586, + "step": 15600 + }, + { + "epoch": 4.08, + "learning_rate": 0.0001, + "loss": 0.2676, + "step": 15610 + }, + { + "epoch": 4.09, + "learning_rate": 0.0001, + "loss": 0.2615, + "step": 15620 + }, + { + "epoch": 4.09, + "learning_rate": 0.0001, + "loss": 0.2563, + "step": 15630 + }, + { + "epoch": 4.09, + "learning_rate": 0.0001, + "loss": 0.2712, + "step": 15640 + }, + { + "epoch": 4.09, + "learning_rate": 0.0001, + "loss": 0.2653, + "step": 15650 + }, + { + "epoch": 4.1, + "learning_rate": 0.0001, + "loss": 0.2546, + "step": 15660 + }, + { + "epoch": 4.1, + "learning_rate": 0.0001, + "loss": 0.256, + "step": 15670 + }, + { + "epoch": 4.1, + "learning_rate": 0.0001, + "loss": 0.2587, + "step": 15680 + }, + { + "epoch": 4.11, + "learning_rate": 0.0001, + "loss": 0.2597, + "step": 15690 + }, + { + "epoch": 4.11, + "learning_rate": 0.0001, + "loss": 0.263, + "step": 15700 + }, + { + "epoch": 4.11, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 15710 + }, + { + "epoch": 4.11, + "learning_rate": 0.0001, + "loss": 0.2622, + "step": 15720 + }, + { + "epoch": 4.12, + "learning_rate": 0.0001, + "loss": 0.267, + "step": 15730 + }, + { + "epoch": 4.12, + "learning_rate": 0.0001, + "loss": 0.264, + "step": 15740 + }, + { + "epoch": 4.12, + "learning_rate": 0.0001, + "loss": 0.2557, + "step": 15750 + }, + { + "epoch": 4.12, + "learning_rate": 0.0001, + "loss": 0.2603, + "step": 15760 + }, + { + "epoch": 4.13, + "learning_rate": 0.0001, + "loss": 0.26, + "step": 15770 + }, + { + "epoch": 4.13, + "learning_rate": 0.0001, + "loss": 0.2631, + "step": 15780 + }, + { + "epoch": 4.13, + "learning_rate": 0.0001, + "loss": 0.2699, + "step": 15790 + }, + { + "epoch": 4.13, + "learning_rate": 0.0001, + "loss": 0.2617, + "step": 15800 + }, + { + "epoch": 4.14, + "learning_rate": 0.0001, + "loss": 0.2684, + "step": 15810 + }, + { + "epoch": 4.14, + "learning_rate": 0.0001, + "loss": 0.2664, + "step": 15820 + }, + { + "epoch": 4.14, + "learning_rate": 0.0001, + "loss": 0.2665, + "step": 15830 + }, + { + "epoch": 4.14, + "learning_rate": 0.0001, + "loss": 0.268, + "step": 15840 + }, + { + "epoch": 4.15, + "learning_rate": 0.0001, + "loss": 0.2713, + "step": 15850 + }, + { + "epoch": 4.15, + "learning_rate": 0.0001, + "loss": 0.2583, + "step": 15860 + }, + { + "epoch": 4.15, + "learning_rate": 0.0001, + "loss": 0.2669, + "step": 15870 + }, + { + "epoch": 4.15, + "learning_rate": 0.0001, + "loss": 0.2615, + "step": 15880 + }, + { + "epoch": 4.16, + "learning_rate": 0.0001, + "loss": 0.2597, + "step": 15890 + }, + { + "epoch": 4.16, + "learning_rate": 0.0001, + "loss": 0.2527, + "step": 15900 + }, + { + "epoch": 4.16, + "learning_rate": 0.0001, + "loss": 0.2681, + "step": 15910 + }, + { + "epoch": 4.17, + "learning_rate": 0.0001, + "loss": 0.2615, + "step": 15920 + }, + { + "epoch": 4.17, + "learning_rate": 0.0001, + "loss": 0.2584, + "step": 15930 + }, + { + "epoch": 4.17, + "learning_rate": 0.0001, + "loss": 0.2629, + "step": 15940 + }, + { + "epoch": 4.17, + "learning_rate": 0.0001, + "loss": 0.2597, + "step": 15950 + }, + { + "epoch": 4.18, + "learning_rate": 0.0001, + "loss": 0.2683, + "step": 15960 + }, + { + "epoch": 4.18, + "learning_rate": 0.0001, + "loss": 0.2617, + "step": 15970 + }, + { + "epoch": 4.18, + "learning_rate": 0.0001, + "loss": 0.2607, + "step": 15980 + }, + { + "epoch": 4.18, + "learning_rate": 0.0001, + "loss": 0.2666, + "step": 15990 + }, + { + "epoch": 4.19, + "learning_rate": 0.0001, + "loss": 0.2639, + "step": 16000 + }, + { + "epoch": 4.19, + "learning_rate": 0.0001, + "loss": 0.2645, + "step": 16010 + }, + { + "epoch": 4.19, + "learning_rate": 0.0001, + "loss": 0.2648, + "step": 16020 + }, + { + "epoch": 4.19, + "learning_rate": 0.0001, + "loss": 0.2625, + "step": 16030 + }, + { + "epoch": 4.2, + "learning_rate": 0.0001, + "loss": 0.2559, + "step": 16040 + }, + { + "epoch": 4.2, + "learning_rate": 0.0001, + "loss": 0.2587, + "step": 16050 + }, + { + "epoch": 4.2, + "learning_rate": 0.0001, + "loss": 0.2605, + "step": 16060 + }, + { + "epoch": 4.2, + "learning_rate": 0.0001, + "loss": 0.2595, + "step": 16070 + }, + { + "epoch": 4.21, + "learning_rate": 0.0001, + "loss": 0.263, + "step": 16080 + }, + { + "epoch": 4.21, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 16090 + }, + { + "epoch": 4.21, + "learning_rate": 0.0001, + "loss": 0.2661, + "step": 16100 + }, + { + "epoch": 4.22, + "learning_rate": 0.0001, + "loss": 0.2552, + "step": 16110 + }, + { + "epoch": 4.22, + "learning_rate": 0.0001, + "loss": 0.2678, + "step": 16120 + }, + { + "epoch": 4.22, + "learning_rate": 0.0001, + "loss": 0.2606, + "step": 16130 + }, + { + "epoch": 4.22, + "learning_rate": 0.0001, + "loss": 0.2656, + "step": 16140 + }, + { + "epoch": 4.23, + "learning_rate": 0.0001, + "loss": 0.2732, + "step": 16150 + }, + { + "epoch": 4.23, + "learning_rate": 0.0001, + "loss": 0.2654, + "step": 16160 + }, + { + "epoch": 4.23, + "learning_rate": 0.0001, + "loss": 0.2622, + "step": 16170 + }, + { + "epoch": 4.23, + "learning_rate": 0.0001, + "loss": 0.27, + "step": 16180 + }, + { + "epoch": 4.24, + "learning_rate": 0.0001, + "loss": 0.2713, + "step": 16190 + }, + { + "epoch": 4.24, + "learning_rate": 0.0001, + "loss": 0.2641, + "step": 16200 + }, + { + "epoch": 4.24, + "learning_rate": 0.0001, + "loss": 0.2687, + "step": 16210 + }, + { + "epoch": 4.24, + "learning_rate": 0.0001, + "loss": 0.2587, + "step": 16220 + }, + { + "epoch": 4.25, + "learning_rate": 0.0001, + "loss": 0.2589, + "step": 16230 + }, + { + "epoch": 4.25, + "learning_rate": 0.0001, + "loss": 0.2668, + "step": 16240 + }, + { + "epoch": 4.25, + "learning_rate": 0.0001, + "loss": 0.2634, + "step": 16250 + }, + { + "epoch": 4.25, + "learning_rate": 0.0001, + "loss": 0.2769, + "step": 16260 + }, + { + "epoch": 4.26, + "learning_rate": 0.0001, + "loss": 0.2688, + "step": 16270 + }, + { + "epoch": 4.26, + "learning_rate": 0.0001, + "loss": 0.2542, + "step": 16280 + }, + { + "epoch": 4.26, + "learning_rate": 0.0001, + "loss": 0.2648, + "step": 16290 + }, + { + "epoch": 4.26, + "learning_rate": 0.0001, + "loss": 0.2675, + "step": 16300 + }, + { + "epoch": 4.27, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 16310 + }, + { + "epoch": 4.27, + "learning_rate": 0.0001, + "loss": 0.2716, + "step": 16320 + }, + { + "epoch": 4.27, + "learning_rate": 0.0001, + "loss": 0.2603, + "step": 16330 + }, + { + "epoch": 4.28, + "learning_rate": 0.0001, + "loss": 0.2569, + "step": 16340 + }, + { + "epoch": 4.28, + "learning_rate": 0.0001, + "loss": 0.2651, + "step": 16350 + }, + { + "epoch": 4.28, + "learning_rate": 0.0001, + "loss": 0.2706, + "step": 16360 + }, + { + "epoch": 4.28, + "learning_rate": 0.0001, + "loss": 0.2675, + "step": 16370 + }, + { + "epoch": 4.29, + "learning_rate": 0.0001, + "loss": 0.2732, + "step": 16380 + }, + { + "epoch": 4.29, + "learning_rate": 0.0001, + "loss": 0.2698, + "step": 16390 + }, + { + "epoch": 4.29, + "learning_rate": 0.0001, + "loss": 0.2787, + "step": 16400 + }, + { + "epoch": 4.29, + "learning_rate": 0.0001, + "loss": 0.2653, + "step": 16410 + }, + { + "epoch": 4.3, + "learning_rate": 0.0001, + "loss": 0.2702, + "step": 16420 + }, + { + "epoch": 4.3, + "learning_rate": 0.0001, + "loss": 0.2599, + "step": 16430 + }, + { + "epoch": 4.3, + "learning_rate": 0.0001, + "loss": 0.2644, + "step": 16440 + }, + { + "epoch": 4.3, + "learning_rate": 0.0001, + "loss": 0.2656, + "step": 16450 + }, + { + "epoch": 4.31, + "learning_rate": 0.0001, + "loss": 0.2674, + "step": 16460 + }, + { + "epoch": 4.31, + "learning_rate": 0.0001, + "loss": 0.2667, + "step": 16470 + }, + { + "epoch": 4.31, + "learning_rate": 0.0001, + "loss": 0.2676, + "step": 16480 + }, + { + "epoch": 4.31, + "learning_rate": 0.0001, + "loss": 0.2565, + "step": 16490 + }, + { + "epoch": 4.32, + "learning_rate": 0.0001, + "loss": 0.2687, + "step": 16500 + }, + { + "epoch": 4.32, + "learning_rate": 0.0001, + "loss": 0.2615, + "step": 16510 + }, + { + "epoch": 4.32, + "learning_rate": 0.0001, + "loss": 0.2709, + "step": 16520 + }, + { + "epoch": 4.32, + "learning_rate": 0.0001, + "loss": 0.2739, + "step": 16530 + }, + { + "epoch": 4.33, + "learning_rate": 0.0001, + "loss": 0.2749, + "step": 16540 + }, + { + "epoch": 4.33, + "learning_rate": 0.0001, + "loss": 0.2703, + "step": 16550 + }, + { + "epoch": 4.33, + "learning_rate": 0.0001, + "loss": 0.2575, + "step": 16560 + }, + { + "epoch": 4.34, + "learning_rate": 0.0001, + "loss": 0.2652, + "step": 16570 + }, + { + "epoch": 4.34, + "learning_rate": 0.0001, + "loss": 0.271, + "step": 16580 + }, + { + "epoch": 4.34, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 16590 + }, + { + "epoch": 4.34, + "learning_rate": 0.0001, + "loss": 0.2678, + "step": 16600 + }, + { + "epoch": 4.35, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 16610 + }, + { + "epoch": 4.35, + "learning_rate": 0.0001, + "loss": 0.2746, + "step": 16620 + }, + { + "epoch": 4.35, + "learning_rate": 0.0001, + "loss": 0.2686, + "step": 16630 + }, + { + "epoch": 4.35, + "learning_rate": 0.0001, + "loss": 0.2699, + "step": 16640 + }, + { + "epoch": 4.36, + "learning_rate": 0.0001, + "loss": 0.2782, + "step": 16650 + }, + { + "epoch": 4.36, + "learning_rate": 0.0001, + "loss": 0.2726, + "step": 16660 + }, + { + "epoch": 4.36, + "learning_rate": 0.0001, + "loss": 0.2646, + "step": 16670 + }, + { + "epoch": 4.36, + "learning_rate": 0.0001, + "loss": 0.2687, + "step": 16680 + }, + { + "epoch": 4.37, + "learning_rate": 0.0001, + "loss": 0.2637, + "step": 16690 + }, + { + "epoch": 4.37, + "learning_rate": 0.0001, + "loss": 0.2614, + "step": 16700 + }, + { + "epoch": 4.37, + "learning_rate": 0.0001, + "loss": 0.257, + "step": 16710 + }, + { + "epoch": 4.37, + "learning_rate": 0.0001, + "loss": 0.266, + "step": 16720 + }, + { + "epoch": 4.38, + "learning_rate": 0.0001, + "loss": 0.273, + "step": 16730 + }, + { + "epoch": 4.38, + "learning_rate": 0.0001, + "loss": 0.2716, + "step": 16740 + }, + { + "epoch": 4.38, + "learning_rate": 0.0001, + "loss": 0.2646, + "step": 16750 + }, + { + "epoch": 4.39, + "learning_rate": 0.0001, + "loss": 0.2717, + "step": 16760 + }, + { + "epoch": 4.39, + "learning_rate": 0.0001, + "loss": 0.2672, + "step": 16770 + }, + { + "epoch": 4.39, + "learning_rate": 0.0001, + "loss": 0.267, + "step": 16780 + }, + { + "epoch": 4.39, + "learning_rate": 0.0001, + "loss": 0.264, + "step": 16790 + }, + { + "epoch": 4.4, + "learning_rate": 0.0001, + "loss": 0.2669, + "step": 16800 + }, + { + "epoch": 4.4, + "learning_rate": 0.0001, + "loss": 0.2762, + "step": 16810 + }, + { + "epoch": 4.4, + "learning_rate": 0.0001, + "loss": 0.2606, + "step": 16820 + }, + { + "epoch": 4.4, + "learning_rate": 0.0001, + "loss": 0.2526, + "step": 16830 + }, + { + "epoch": 4.41, + "learning_rate": 0.0001, + "loss": 0.2629, + "step": 16840 + }, + { + "epoch": 4.41, + "learning_rate": 0.0001, + "loss": 0.2758, + "step": 16850 + }, + { + "epoch": 4.41, + "learning_rate": 0.0001, + "loss": 0.2735, + "step": 16860 + }, + { + "epoch": 4.41, + "learning_rate": 0.0001, + "loss": 0.2686, + "step": 16870 + }, + { + "epoch": 4.42, + "learning_rate": 0.0001, + "loss": 0.2676, + "step": 16880 + }, + { + "epoch": 4.42, + "learning_rate": 0.0001, + "loss": 0.2653, + "step": 16890 + }, + { + "epoch": 4.42, + "learning_rate": 0.0001, + "loss": 0.273, + "step": 16900 + }, + { + "epoch": 4.42, + "learning_rate": 0.0001, + "loss": 0.2669, + "step": 16910 + }, + { + "epoch": 4.43, + "learning_rate": 0.0001, + "loss": 0.2665, + "step": 16920 + }, + { + "epoch": 4.43, + "learning_rate": 0.0001, + "loss": 0.2704, + "step": 16930 + }, + { + "epoch": 4.43, + "learning_rate": 0.0001, + "loss": 0.2655, + "step": 16940 + }, + { + "epoch": 4.43, + "learning_rate": 0.0001, + "loss": 0.2661, + "step": 16950 + }, + { + "epoch": 4.44, + "learning_rate": 0.0001, + "loss": 0.2653, + "step": 16960 + }, + { + "epoch": 4.44, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 16970 + }, + { + "epoch": 4.44, + "learning_rate": 0.0001, + "loss": 0.2672, + "step": 16980 + }, + { + "epoch": 4.45, + "learning_rate": 0.0001, + "loss": 0.2669, + "step": 16990 + }, + { + "epoch": 4.45, + "learning_rate": 0.0001, + "loss": 0.2663, + "step": 17000 + }, + { + "epoch": 4.45, + "learning_rate": 0.0001, + "loss": 0.2652, + "step": 17010 + }, + { + "epoch": 4.45, + "learning_rate": 0.0001, + "loss": 0.2749, + "step": 17020 + }, + { + "epoch": 4.46, + "learning_rate": 0.0001, + "loss": 0.2749, + "step": 17030 + }, + { + "epoch": 4.46, + "learning_rate": 0.0001, + "loss": 0.2755, + "step": 17040 + }, + { + "epoch": 4.46, + "learning_rate": 0.0001, + "loss": 0.2676, + "step": 17050 + }, + { + "epoch": 4.46, + "learning_rate": 0.0001, + "loss": 0.2677, + "step": 17060 + }, + { + "epoch": 4.47, + "learning_rate": 0.0001, + "loss": 0.2678, + "step": 17070 + }, + { + "epoch": 4.47, + "learning_rate": 0.0001, + "loss": 0.2719, + "step": 17080 + }, + { + "epoch": 4.47, + "learning_rate": 0.0001, + "loss": 0.2733, + "step": 17090 + }, + { + "epoch": 4.47, + "learning_rate": 0.0001, + "loss": 0.2654, + "step": 17100 + }, + { + "epoch": 4.48, + "learning_rate": 0.0001, + "loss": 0.2657, + "step": 17110 + }, + { + "epoch": 4.48, + "learning_rate": 0.0001, + "loss": 0.2642, + "step": 17120 + }, + { + "epoch": 4.48, + "learning_rate": 0.0001, + "loss": 0.2669, + "step": 17130 + }, + { + "epoch": 4.48, + "learning_rate": 0.0001, + "loss": 0.2776, + "step": 17140 + }, + { + "epoch": 4.49, + "learning_rate": 0.0001, + "loss": 0.2719, + "step": 17150 + }, + { + "epoch": 4.49, + "learning_rate": 0.0001, + "loss": 0.2705, + "step": 17160 + }, + { + "epoch": 4.49, + "learning_rate": 0.0001, + "loss": 0.2698, + "step": 17170 + }, + { + "epoch": 4.5, + "learning_rate": 0.0001, + "loss": 0.2661, + "step": 17180 + }, + { + "epoch": 4.5, + "learning_rate": 0.0001, + "loss": 0.2782, + "step": 17190 + }, + { + "epoch": 4.5, + "learning_rate": 0.0001, + "loss": 0.261, + "step": 17200 + }, + { + "epoch": 4.5, + "learning_rate": 0.0001, + "loss": 0.2848, + "step": 17210 + }, + { + "epoch": 4.51, + "learning_rate": 0.0001, + "loss": 0.2729, + "step": 17220 + }, + { + "epoch": 4.51, + "learning_rate": 0.0001, + "loss": 0.2776, + "step": 17230 + }, + { + "epoch": 4.51, + "learning_rate": 0.0001, + "loss": 0.2741, + "step": 17240 + }, + { + "epoch": 4.51, + "learning_rate": 0.0001, + "loss": 0.2697, + "step": 17250 + }, + { + "epoch": 4.52, + "learning_rate": 0.0001, + "loss": 0.2704, + "step": 17260 + }, + { + "epoch": 4.52, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 17270 + }, + { + "epoch": 4.52, + "learning_rate": 0.0001, + "loss": 0.2719, + "step": 17280 + }, + { + "epoch": 4.52, + "learning_rate": 0.0001, + "loss": 0.2802, + "step": 17290 + }, + { + "epoch": 4.53, + "learning_rate": 0.0001, + "loss": 0.2661, + "step": 17300 + }, + { + "epoch": 4.53, + "learning_rate": 0.0001, + "loss": 0.2725, + "step": 17310 + }, + { + "epoch": 4.53, + "learning_rate": 0.0001, + "loss": 0.2642, + "step": 17320 + }, + { + "epoch": 4.53, + "learning_rate": 0.0001, + "loss": 0.2708, + "step": 17330 + }, + { + "epoch": 4.54, + "learning_rate": 0.0001, + "loss": 0.2668, + "step": 17340 + }, + { + "epoch": 4.54, + "learning_rate": 0.0001, + "loss": 0.2655, + "step": 17350 + }, + { + "epoch": 4.54, + "learning_rate": 0.0001, + "loss": 0.2704, + "step": 17360 + }, + { + "epoch": 4.54, + "learning_rate": 0.0001, + "loss": 0.2671, + "step": 17370 + }, + { + "epoch": 4.55, + "learning_rate": 0.0001, + "loss": 0.2743, + "step": 17380 + }, + { + "epoch": 4.55, + "learning_rate": 0.0001, + "loss": 0.2663, + "step": 17390 + }, + { + "epoch": 4.55, + "learning_rate": 0.0001, + "loss": 0.2632, + "step": 17400 + }, + { + "epoch": 4.56, + "learning_rate": 0.0001, + "loss": 0.2774, + "step": 17410 + }, + { + "epoch": 4.56, + "learning_rate": 0.0001, + "loss": 0.2654, + "step": 17420 + }, + { + "epoch": 4.56, + "learning_rate": 0.0001, + "loss": 0.2677, + "step": 17430 + }, + { + "epoch": 4.56, + "learning_rate": 0.0001, + "loss": 0.2684, + "step": 17440 + }, + { + "epoch": 4.57, + "learning_rate": 0.0001, + "loss": 0.2689, + "step": 17450 + }, + { + "epoch": 4.57, + "learning_rate": 0.0001, + "loss": 0.2719, + "step": 17460 + }, + { + "epoch": 4.57, + "learning_rate": 0.0001, + "loss": 0.2719, + "step": 17470 + }, + { + "epoch": 4.57, + "learning_rate": 0.0001, + "loss": 0.271, + "step": 17480 + }, + { + "epoch": 4.58, + "learning_rate": 0.0001, + "loss": 0.273, + "step": 17490 + }, + { + "epoch": 4.58, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 17500 + }, + { + "epoch": 4.58, + "learning_rate": 0.0001, + "loss": 0.261, + "step": 17510 + }, + { + "epoch": 4.58, + "learning_rate": 0.0001, + "loss": 0.2611, + "step": 17520 + }, + { + "epoch": 4.59, + "learning_rate": 0.0001, + "loss": 0.2718, + "step": 17530 + }, + { + "epoch": 4.59, + "learning_rate": 0.0001, + "loss": 0.2792, + "step": 17540 + }, + { + "epoch": 4.59, + "learning_rate": 0.0001, + "loss": 0.2795, + "step": 17550 + }, + { + "epoch": 4.59, + "learning_rate": 0.0001, + "loss": 0.2749, + "step": 17560 + }, + { + "epoch": 4.6, + "learning_rate": 0.0001, + "loss": 0.2727, + "step": 17570 + }, + { + "epoch": 4.6, + "learning_rate": 0.0001, + "loss": 0.2667, + "step": 17580 + }, + { + "epoch": 4.6, + "learning_rate": 0.0001, + "loss": 0.2694, + "step": 17590 + }, + { + "epoch": 4.6, + "learning_rate": 0.0001, + "loss": 0.2746, + "step": 17600 + }, + { + "epoch": 4.61, + "learning_rate": 0.0001, + "loss": 0.2724, + "step": 17610 + }, + { + "epoch": 4.61, + "learning_rate": 0.0001, + "loss": 0.2688, + "step": 17620 + }, + { + "epoch": 4.61, + "learning_rate": 0.0001, + "loss": 0.2683, + "step": 17630 + }, + { + "epoch": 4.62, + "learning_rate": 0.0001, + "loss": 0.2628, + "step": 17640 + }, + { + "epoch": 4.62, + "learning_rate": 0.0001, + "loss": 0.2737, + "step": 17650 + }, + { + "epoch": 4.62, + "learning_rate": 0.0001, + "loss": 0.2671, + "step": 17660 + }, + { + "epoch": 4.62, + "learning_rate": 0.0001, + "loss": 0.2589, + "step": 17670 + }, + { + "epoch": 4.63, + "learning_rate": 0.0001, + "loss": 0.2633, + "step": 17680 + }, + { + "epoch": 4.63, + "learning_rate": 0.0001, + "loss": 0.2642, + "step": 17690 + }, + { + "epoch": 4.63, + "learning_rate": 0.0001, + "loss": 0.2737, + "step": 17700 + }, + { + "epoch": 4.63, + "learning_rate": 0.0001, + "loss": 0.2643, + "step": 17710 + }, + { + "epoch": 4.64, + "learning_rate": 0.0001, + "loss": 0.2646, + "step": 17720 + }, + { + "epoch": 4.64, + "learning_rate": 0.0001, + "loss": 0.2598, + "step": 17730 + }, + { + "epoch": 4.64, + "learning_rate": 0.0001, + "loss": 0.2727, + "step": 17740 + }, + { + "epoch": 4.64, + "learning_rate": 0.0001, + "loss": 0.2687, + "step": 17750 + }, + { + "epoch": 4.65, + "learning_rate": 0.0001, + "loss": 0.2741, + "step": 17760 + }, + { + "epoch": 4.65, + "learning_rate": 0.0001, + "loss": 0.2806, + "step": 17770 + }, + { + "epoch": 4.65, + "learning_rate": 0.0001, + "loss": 0.2668, + "step": 17780 + }, + { + "epoch": 4.65, + "learning_rate": 0.0001, + "loss": 0.2731, + "step": 17790 + }, + { + "epoch": 4.66, + "learning_rate": 0.0001, + "loss": 0.2644, + "step": 17800 + }, + { + "epoch": 4.66, + "learning_rate": 0.0001, + "loss": 0.2787, + "step": 17810 + }, + { + "epoch": 4.66, + "learning_rate": 0.0001, + "loss": 0.2648, + "step": 17820 + }, + { + "epoch": 4.67, + "learning_rate": 0.0001, + "loss": 0.2731, + "step": 17830 + }, + { + "epoch": 4.67, + "learning_rate": 0.0001, + "loss": 0.2664, + "step": 17840 + }, + { + "epoch": 4.67, + "learning_rate": 0.0001, + "loss": 0.2734, + "step": 17850 + }, + { + "epoch": 4.67, + "learning_rate": 0.0001, + "loss": 0.271, + "step": 17860 + }, + { + "epoch": 4.68, + "learning_rate": 0.0001, + "loss": 0.2796, + "step": 17870 + }, + { + "epoch": 4.68, + "learning_rate": 0.0001, + "loss": 0.2688, + "step": 17880 + }, + { + "epoch": 4.68, + "learning_rate": 0.0001, + "loss": 0.2753, + "step": 17890 + }, + { + "epoch": 4.68, + "learning_rate": 0.0001, + "loss": 0.2755, + "step": 17900 + }, + { + "epoch": 4.69, + "learning_rate": 0.0001, + "loss": 0.2733, + "step": 17910 + }, + { + "epoch": 4.69, + "learning_rate": 0.0001, + "loss": 0.2793, + "step": 17920 + }, + { + "epoch": 4.69, + "learning_rate": 0.0001, + "loss": 0.2727, + "step": 17930 + }, + { + "epoch": 4.69, + "learning_rate": 0.0001, + "loss": 0.2742, + "step": 17940 + }, + { + "epoch": 4.7, + "learning_rate": 0.0001, + "loss": 0.2775, + "step": 17950 + }, + { + "epoch": 4.7, + "learning_rate": 0.0001, + "loss": 0.2771, + "step": 17960 + }, + { + "epoch": 4.7, + "learning_rate": 0.0001, + "loss": 0.275, + "step": 17970 + }, + { + "epoch": 4.7, + "learning_rate": 0.0001, + "loss": 0.2763, + "step": 17980 + }, + { + "epoch": 4.71, + "learning_rate": 0.0001, + "loss": 0.2637, + "step": 17990 + }, + { + "epoch": 4.71, + "learning_rate": 0.0001, + "loss": 0.2774, + "step": 18000 + }, + { + "epoch": 4.71, + "learning_rate": 0.0001, + "loss": 0.2723, + "step": 18010 + }, + { + "epoch": 4.71, + "learning_rate": 0.0001, + "loss": 0.2679, + "step": 18020 + }, + { + "epoch": 4.72, + "learning_rate": 0.0001, + "loss": 0.2632, + "step": 18030 + }, + { + "epoch": 4.72, + "learning_rate": 0.0001, + "loss": 0.2646, + "step": 18040 + }, + { + "epoch": 4.72, + "learning_rate": 0.0001, + "loss": 0.2706, + "step": 18050 + }, + { + "epoch": 4.73, + "learning_rate": 0.0001, + "loss": 0.2726, + "step": 18060 + }, + { + "epoch": 4.73, + "learning_rate": 0.0001, + "loss": 0.2676, + "step": 18070 + }, + { + "epoch": 4.73, + "learning_rate": 0.0001, + "loss": 0.2706, + "step": 18080 + }, + { + "epoch": 4.73, + "learning_rate": 0.0001, + "loss": 0.2616, + "step": 18090 + }, + { + "epoch": 4.74, + "learning_rate": 0.0001, + "loss": 0.2775, + "step": 18100 + }, + { + "epoch": 4.74, + "learning_rate": 0.0001, + "loss": 0.2583, + "step": 18110 + }, + { + "epoch": 4.74, + "learning_rate": 0.0001, + "loss": 0.2828, + "step": 18120 + }, + { + "epoch": 4.74, + "learning_rate": 0.0001, + "loss": 0.2877, + "step": 18130 + }, + { + "epoch": 4.75, + "learning_rate": 0.0001, + "loss": 0.2713, + "step": 18140 + }, + { + "epoch": 4.75, + "learning_rate": 0.0001, + "loss": 0.2762, + "step": 18150 + }, + { + "epoch": 4.75, + "learning_rate": 0.0001, + "loss": 0.2723, + "step": 18160 + }, + { + "epoch": 4.75, + "learning_rate": 0.0001, + "loss": 0.2578, + "step": 18170 + }, + { + "epoch": 4.76, + "learning_rate": 0.0001, + "loss": 0.271, + "step": 18180 + }, + { + "epoch": 4.76, + "learning_rate": 0.0001, + "loss": 0.2849, + "step": 18190 + }, + { + "epoch": 4.76, + "learning_rate": 0.0001, + "loss": 0.2853, + "step": 18200 + }, + { + "epoch": 4.76, + "learning_rate": 0.0001, + "loss": 0.2768, + "step": 18210 + }, + { + "epoch": 4.77, + "learning_rate": 0.0001, + "loss": 0.2646, + "step": 18220 + }, + { + "epoch": 4.77, + "learning_rate": 0.0001, + "loss": 0.2549, + "step": 18230 + }, + { + "epoch": 4.77, + "learning_rate": 0.0001, + "loss": 0.2771, + "step": 18240 + }, + { + "epoch": 4.77, + "learning_rate": 0.0001, + "loss": 0.2713, + "step": 18250 + }, + { + "epoch": 4.78, + "learning_rate": 0.0001, + "loss": 0.283, + "step": 18260 + }, + { + "epoch": 4.78, + "learning_rate": 0.0001, + "loss": 0.2608, + "step": 18270 + }, + { + "epoch": 4.78, + "learning_rate": 0.0001, + "loss": 0.265, + "step": 18280 + }, + { + "epoch": 4.79, + "learning_rate": 0.0001, + "loss": 0.2727, + "step": 18290 + }, + { + "epoch": 4.79, + "learning_rate": 0.0001, + "loss": 0.263, + "step": 18300 + }, + { + "epoch": 4.79, + "learning_rate": 0.0001, + "loss": 0.2713, + "step": 18310 + }, + { + "epoch": 4.79, + "learning_rate": 0.0001, + "loss": 0.2752, + "step": 18320 + }, + { + "epoch": 4.8, + "learning_rate": 0.0001, + "loss": 0.269, + "step": 18330 + }, + { + "epoch": 4.8, + "learning_rate": 0.0001, + "loss": 0.2748, + "step": 18340 + }, + { + "epoch": 4.8, + "learning_rate": 0.0001, + "loss": 0.2718, + "step": 18350 + }, + { + "epoch": 4.8, + "learning_rate": 0.0001, + "loss": 0.2688, + "step": 18360 + }, + { + "epoch": 4.81, + "learning_rate": 0.0001, + "loss": 0.2679, + "step": 18370 + }, + { + "epoch": 4.81, + "learning_rate": 0.0001, + "loss": 0.2797, + "step": 18380 + }, + { + "epoch": 4.81, + "learning_rate": 0.0001, + "loss": 0.2763, + "step": 18390 + }, + { + "epoch": 4.81, + "learning_rate": 0.0001, + "loss": 0.2747, + "step": 18400 + }, + { + "epoch": 4.82, + "learning_rate": 0.0001, + "loss": 0.284, + "step": 18410 + }, + { + "epoch": 4.82, + "learning_rate": 0.0001, + "loss": 0.2725, + "step": 18420 + }, + { + "epoch": 4.82, + "learning_rate": 0.0001, + "loss": 0.2685, + "step": 18430 + }, + { + "epoch": 4.82, + "learning_rate": 0.0001, + "loss": 0.2704, + "step": 18440 + }, + { + "epoch": 4.83, + "learning_rate": 0.0001, + "loss": 0.2745, + "step": 18450 + }, + { + "epoch": 4.83, + "learning_rate": 0.0001, + "loss": 0.2721, + "step": 18460 + }, + { + "epoch": 4.83, + "learning_rate": 0.0001, + "loss": 0.2884, + "step": 18470 + }, + { + "epoch": 4.84, + "learning_rate": 0.0001, + "loss": 0.2697, + "step": 18480 + }, + { + "epoch": 4.84, + "learning_rate": 0.0001, + "loss": 0.2768, + "step": 18490 + }, + { + "epoch": 4.84, + "learning_rate": 0.0001, + "loss": 0.2725, + "step": 18500 + }, + { + "epoch": 4.84, + "learning_rate": 0.0001, + "loss": 0.277, + "step": 18510 + }, + { + "epoch": 4.85, + "learning_rate": 0.0001, + "loss": 0.2895, + "step": 18520 + }, + { + "epoch": 4.85, + "learning_rate": 0.0001, + "loss": 0.2808, + "step": 18530 + }, + { + "epoch": 4.85, + "learning_rate": 0.0001, + "loss": 0.2752, + "step": 18540 + }, + { + "epoch": 4.85, + "learning_rate": 0.0001, + "loss": 0.2748, + "step": 18550 + }, + { + "epoch": 4.86, + "learning_rate": 0.0001, + "loss": 0.2796, + "step": 18560 + }, + { + "epoch": 4.86, + "learning_rate": 0.0001, + "loss": 0.2757, + "step": 18570 + }, + { + "epoch": 4.86, + "learning_rate": 0.0001, + "loss": 0.2753, + "step": 18580 + }, + { + "epoch": 4.86, + "learning_rate": 0.0001, + "loss": 0.283, + "step": 18590 + }, + { + "epoch": 4.87, + "learning_rate": 0.0001, + "loss": 0.2881, + "step": 18600 + }, + { + "epoch": 4.87, + "learning_rate": 0.0001, + "loss": 0.2724, + "step": 18610 + }, + { + "epoch": 4.87, + "learning_rate": 0.0001, + "loss": 0.2783, + "step": 18620 + }, + { + "epoch": 4.87, + "learning_rate": 0.0001, + "loss": 0.2726, + "step": 18630 + }, + { + "epoch": 4.88, + "learning_rate": 0.0001, + "loss": 0.2866, + "step": 18640 + }, + { + "epoch": 4.88, + "learning_rate": 0.0001, + "loss": 0.2774, + "step": 18650 + }, + { + "epoch": 4.88, + "learning_rate": 0.0001, + "loss": 0.2764, + "step": 18660 + }, + { + "epoch": 4.88, + "learning_rate": 0.0001, + "loss": 0.2733, + "step": 18670 + }, + { + "epoch": 4.89, + "learning_rate": 0.0001, + "loss": 0.2759, + "step": 18680 + }, + { + "epoch": 4.89, + "learning_rate": 0.0001, + "loss": 0.2629, + "step": 18690 + }, + { + "epoch": 4.89, + "learning_rate": 0.0001, + "loss": 0.2666, + "step": 18700 + }, + { + "epoch": 4.9, + "learning_rate": 0.0001, + "loss": 0.2829, + "step": 18710 + }, + { + "epoch": 4.9, + "learning_rate": 0.0001, + "loss": 0.2836, + "step": 18720 + }, + { + "epoch": 4.9, + "learning_rate": 0.0001, + "loss": 0.27, + "step": 18730 + }, + { + "epoch": 4.9, + "learning_rate": 0.0001, + "loss": 0.2773, + "step": 18740 + }, + { + "epoch": 4.91, + "learning_rate": 0.0001, + "loss": 0.2854, + "step": 18750 + }, + { + "epoch": 4.91, + "learning_rate": 0.0001, + "loss": 0.266, + "step": 18760 + }, + { + "epoch": 4.91, + "learning_rate": 0.0001, + "loss": 0.2736, + "step": 18770 + }, + { + "epoch": 4.91, + "learning_rate": 0.0001, + "loss": 0.2793, + "step": 18780 + }, + { + "epoch": 4.92, + "learning_rate": 0.0001, + "loss": 0.2615, + "step": 18790 + }, + { + "epoch": 4.92, + "learning_rate": 0.0001, + "loss": 0.2685, + "step": 18800 + }, + { + "epoch": 4.92, + "learning_rate": 0.0001, + "loss": 0.2722, + "step": 18810 + }, + { + "epoch": 4.92, + "learning_rate": 0.0001, + "loss": 0.2732, + "step": 18820 + }, + { + "epoch": 4.93, + "learning_rate": 0.0001, + "loss": 0.2744, + "step": 18830 + }, + { + "epoch": 4.93, + "learning_rate": 0.0001, + "loss": 0.2673, + "step": 18840 + }, + { + "epoch": 4.93, + "learning_rate": 0.0001, + "loss": 0.2775, + "step": 18850 + }, + { + "epoch": 4.93, + "learning_rate": 0.0001, + "loss": 0.27, + "step": 18860 + }, + { + "epoch": 4.94, + "learning_rate": 0.0001, + "loss": 0.2603, + "step": 18870 + }, + { + "epoch": 4.94, + "learning_rate": 0.0001, + "loss": 0.2717, + "step": 18880 + }, + { + "epoch": 4.94, + "learning_rate": 0.0001, + "loss": 0.2825, + "step": 18890 + }, + { + "epoch": 4.95, + "learning_rate": 0.0001, + "loss": 0.2779, + "step": 18900 + }, + { + "epoch": 4.95, + "learning_rate": 0.0001, + "loss": 0.2754, + "step": 18910 + }, + { + "epoch": 4.95, + "learning_rate": 0.0001, + "loss": 0.2805, + "step": 18920 + }, + { + "epoch": 4.95, + "learning_rate": 0.0001, + "loss": 0.2703, + "step": 18930 + }, + { + "epoch": 4.96, + "learning_rate": 0.0001, + "loss": 0.2751, + "step": 18940 + }, + { + "epoch": 4.96, + "learning_rate": 0.0001, + "loss": 0.2776, + "step": 18950 + }, + { + "epoch": 4.96, + "learning_rate": 0.0001, + "loss": 0.2636, + "step": 18960 + }, + { + "epoch": 4.96, + "learning_rate": 0.0001, + "loss": 0.2694, + "step": 18970 + }, + { + "epoch": 4.97, + "learning_rate": 0.0001, + "loss": 0.272, + "step": 18980 + }, + { + "epoch": 4.97, + "learning_rate": 0.0001, + "loss": 0.282, + "step": 18990 + }, + { + "epoch": 4.97, + "learning_rate": 0.0001, + "loss": 0.2793, + "step": 19000 + }, + { + "epoch": 4.97, + "learning_rate": 0.0001, + "loss": 0.2827, + "step": 19010 + }, + { + "epoch": 4.98, + "learning_rate": 0.0001, + "loss": 0.2671, + "step": 19020 + }, + { + "epoch": 4.98, + "learning_rate": 0.0001, + "loss": 0.2709, + "step": 19030 + }, + { + "epoch": 4.98, + "learning_rate": 0.0001, + "loss": 0.2836, + "step": 19040 + }, + { + "epoch": 4.98, + "learning_rate": 0.0001, + "loss": 0.2785, + "step": 19050 + }, + { + "epoch": 4.99, + "learning_rate": 0.0001, + "loss": 0.274, + "step": 19060 + }, + { + "epoch": 4.99, + "learning_rate": 0.0001, + "loss": 0.2753, + "step": 19070 + }, + { + "epoch": 4.99, + "learning_rate": 0.0001, + "loss": 0.2716, + "step": 19080 + }, + { + "epoch": 4.99, + "learning_rate": 0.0001, + "loss": 0.2682, + "step": 19090 + }, + { + "epoch": 5.0, + "learning_rate": 0.0001, + "loss": 0.268, + "step": 19100 + }, + { + "epoch": 5.0, + "learning_rate": 0.0001, + "loss": 0.2652, + "step": 19110 + }, + { + "epoch": 5.0, + "eval_gen_len": 61.7281, + "eval_loss": 1.108832836151123, + "eval_rouge1": 57.2217, + "eval_rouge2": 35.1923, + "eval_rougeL": 46.6928, + "eval_rougeLsum": 52.7666, + "eval_runtime": 38.7944, + "eval_samples_per_second": 22.658, + "eval_steps_per_second": 0.361, + "step": 19110 + }, + { + "epoch": 5.0, + "learning_rate": 0.0001, + "loss": 0.2285, + "step": 19120 + }, + { + "epoch": 5.01, + "learning_rate": 0.0001, + "loss": 0.2231, + "step": 19130 + }, + { + "epoch": 5.01, + "learning_rate": 0.0001, + "loss": 0.2254, + "step": 19140 + }, + { + "epoch": 5.01, + "learning_rate": 0.0001, + "loss": 0.2278, + "step": 19150 + }, + { + "epoch": 5.01, + "learning_rate": 0.0001, + "loss": 0.222, + "step": 19160 + }, + { + "epoch": 5.02, + "learning_rate": 0.0001, + "loss": 0.2349, + "step": 19170 + }, + { + "epoch": 5.02, + "learning_rate": 0.0001, + "loss": 0.2332, + "step": 19180 + }, + { + "epoch": 5.02, + "learning_rate": 0.0001, + "loss": 0.2234, + "step": 19190 + }, + { + "epoch": 5.02, + "learning_rate": 0.0001, + "loss": 0.2223, + "step": 19200 + }, + { + "epoch": 5.03, + "learning_rate": 0.0001, + "loss": 0.2291, + "step": 19210 + }, + { + "epoch": 5.03, + "learning_rate": 0.0001, + "loss": 0.223, + "step": 19220 + }, + { + "epoch": 5.03, + "learning_rate": 0.0001, + "loss": 0.2224, + "step": 19230 + }, + { + "epoch": 5.03, + "learning_rate": 0.0001, + "loss": 0.239, + "step": 19240 + }, + { + "epoch": 5.04, + "learning_rate": 0.0001, + "loss": 0.229, + "step": 19250 + }, + { + "epoch": 5.04, + "learning_rate": 0.0001, + "loss": 0.2224, + "step": 19260 + }, + { + "epoch": 5.04, + "learning_rate": 0.0001, + "loss": 0.2273, + "step": 19270 + }, + { + "epoch": 5.04, + "learning_rate": 0.0001, + "loss": 0.2274, + "step": 19280 + }, + { + "epoch": 5.05, + "learning_rate": 0.0001, + "loss": 0.2262, + "step": 19290 + }, + { + "epoch": 5.05, + "learning_rate": 0.0001, + "loss": 0.2214, + "step": 19300 + }, + { + "epoch": 5.05, + "learning_rate": 0.0001, + "loss": 0.2316, + "step": 19310 + }, + { + "epoch": 5.05, + "learning_rate": 0.0001, + "loss": 0.2304, + "step": 19320 + }, + { + "epoch": 5.06, + "learning_rate": 0.0001, + "loss": 0.2331, + "step": 19330 + }, + { + "epoch": 5.06, + "learning_rate": 0.0001, + "loss": 0.233, + "step": 19340 + }, + { + "epoch": 5.06, + "learning_rate": 0.0001, + "loss": 0.2205, + "step": 19350 + }, + { + "epoch": 5.07, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 19360 + }, + { + "epoch": 5.07, + "learning_rate": 0.0001, + "loss": 0.2272, + "step": 19370 + }, + { + "epoch": 5.07, + "learning_rate": 0.0001, + "loss": 0.2348, + "step": 19380 + }, + { + "epoch": 5.07, + "learning_rate": 0.0001, + "loss": 0.2294, + "step": 19390 + }, + { + "epoch": 5.08, + "learning_rate": 0.0001, + "loss": 0.2202, + "step": 19400 + }, + { + "epoch": 5.08, + "learning_rate": 0.0001, + "loss": 0.2325, + "step": 19410 + }, + { + "epoch": 5.08, + "learning_rate": 0.0001, + "loss": 0.2315, + "step": 19420 + }, + { + "epoch": 5.08, + "learning_rate": 0.0001, + "loss": 0.2313, + "step": 19430 + }, + { + "epoch": 5.09, + "learning_rate": 0.0001, + "loss": 0.2255, + "step": 19440 + }, + { + "epoch": 5.09, + "learning_rate": 0.0001, + "loss": 0.2234, + "step": 19450 + }, + { + "epoch": 5.09, + "learning_rate": 0.0001, + "loss": 0.2234, + "step": 19460 + }, + { + "epoch": 5.09, + "learning_rate": 0.0001, + "loss": 0.2249, + "step": 19470 + }, + { + "epoch": 5.1, + "learning_rate": 0.0001, + "loss": 0.2269, + "step": 19480 + }, + { + "epoch": 5.1, + "learning_rate": 0.0001, + "loss": 0.2233, + "step": 19490 + }, + { + "epoch": 5.1, + "learning_rate": 0.0001, + "loss": 0.2282, + "step": 19500 + }, + { + "epoch": 5.1, + "learning_rate": 0.0001, + "loss": 0.2204, + "step": 19510 + }, + { + "epoch": 5.11, + "learning_rate": 0.0001, + "loss": 0.2312, + "step": 19520 + }, + { + "epoch": 5.11, + "learning_rate": 0.0001, + "loss": 0.2385, + "step": 19530 + }, + { + "epoch": 5.11, + "learning_rate": 0.0001, + "loss": 0.2287, + "step": 19540 + }, + { + "epoch": 5.12, + "learning_rate": 0.0001, + "loss": 0.2395, + "step": 19550 + }, + { + "epoch": 5.12, + "learning_rate": 0.0001, + "loss": 0.2344, + "step": 19560 + }, + { + "epoch": 5.12, + "learning_rate": 0.0001, + "loss": 0.2287, + "step": 19570 + }, + { + "epoch": 5.12, + "learning_rate": 0.0001, + "loss": 0.2248, + "step": 19580 + }, + { + "epoch": 5.13, + "learning_rate": 0.0001, + "loss": 0.2174, + "step": 19590 + }, + { + "epoch": 5.13, + "learning_rate": 0.0001, + "loss": 0.2321, + "step": 19600 + }, + { + "epoch": 5.13, + "learning_rate": 0.0001, + "loss": 0.2318, + "step": 19610 + }, + { + "epoch": 5.13, + "learning_rate": 0.0001, + "loss": 0.2304, + "step": 19620 + }, + { + "epoch": 5.14, + "learning_rate": 0.0001, + "loss": 0.226, + "step": 19630 + }, + { + "epoch": 5.14, + "learning_rate": 0.0001, + "loss": 0.2316, + "step": 19640 + }, + { + "epoch": 5.14, + "learning_rate": 0.0001, + "loss": 0.2321, + "step": 19650 + }, + { + "epoch": 5.14, + "learning_rate": 0.0001, + "loss": 0.2323, + "step": 19660 + }, + { + "epoch": 5.15, + "learning_rate": 0.0001, + "loss": 0.2275, + "step": 19670 + }, + { + "epoch": 5.15, + "learning_rate": 0.0001, + "loss": 0.2255, + "step": 19680 + }, + { + "epoch": 5.15, + "learning_rate": 0.0001, + "loss": 0.2177, + "step": 19690 + }, + { + "epoch": 5.15, + "learning_rate": 0.0001, + "loss": 0.2357, + "step": 19700 + }, + { + "epoch": 5.16, + "learning_rate": 0.0001, + "loss": 0.2329, + "step": 19710 + }, + { + "epoch": 5.16, + "learning_rate": 0.0001, + "loss": 0.2233, + "step": 19720 + }, + { + "epoch": 5.16, + "learning_rate": 0.0001, + "loss": 0.2282, + "step": 19730 + }, + { + "epoch": 5.16, + "learning_rate": 0.0001, + "loss": 0.2289, + "step": 19740 + }, + { + "epoch": 5.17, + "learning_rate": 0.0001, + "loss": 0.2339, + "step": 19750 + }, + { + "epoch": 5.17, + "learning_rate": 0.0001, + "loss": 0.2281, + "step": 19760 + }, + { + "epoch": 5.17, + "learning_rate": 0.0001, + "loss": 0.2271, + "step": 19770 + }, + { + "epoch": 5.18, + "learning_rate": 0.0001, + "loss": 0.2404, + "step": 19780 + }, + { + "epoch": 5.18, + "learning_rate": 0.0001, + "loss": 0.2333, + "step": 19790 + }, + { + "epoch": 5.18, + "learning_rate": 0.0001, + "loss": 0.2422, + "step": 19800 + }, + { + "epoch": 5.18, + "learning_rate": 0.0001, + "loss": 0.2346, + "step": 19810 + }, + { + "epoch": 5.19, + "learning_rate": 0.0001, + "loss": 0.2284, + "step": 19820 + }, + { + "epoch": 5.19, + "learning_rate": 0.0001, + "loss": 0.2398, + "step": 19830 + }, + { + "epoch": 5.19, + "learning_rate": 0.0001, + "loss": 0.2351, + "step": 19840 + }, + { + "epoch": 5.19, + "learning_rate": 0.0001, + "loss": 0.2222, + "step": 19850 + }, + { + "epoch": 5.2, + "learning_rate": 0.0001, + "loss": 0.2318, + "step": 19860 + }, + { + "epoch": 5.2, + "learning_rate": 0.0001, + "loss": 0.2274, + "step": 19870 + }, + { + "epoch": 5.2, + "learning_rate": 0.0001, + "loss": 0.2413, + "step": 19880 + }, + { + "epoch": 5.2, + "learning_rate": 0.0001, + "loss": 0.2448, + "step": 19890 + }, + { + "epoch": 5.21, + "learning_rate": 0.0001, + "loss": 0.2444, + "step": 19900 + }, + { + "epoch": 5.21, + "learning_rate": 0.0001, + "loss": 0.2274, + "step": 19910 + }, + { + "epoch": 5.21, + "learning_rate": 0.0001, + "loss": 0.2354, + "step": 19920 + }, + { + "epoch": 5.21, + "learning_rate": 0.0001, + "loss": 0.2299, + "step": 19930 + }, + { + "epoch": 5.22, + "learning_rate": 0.0001, + "loss": 0.2417, + "step": 19940 + }, + { + "epoch": 5.22, + "learning_rate": 0.0001, + "loss": 0.226, + "step": 19950 + }, + { + "epoch": 5.22, + "learning_rate": 0.0001, + "loss": 0.2378, + "step": 19960 + }, + { + "epoch": 5.23, + "learning_rate": 0.0001, + "loss": 0.2268, + "step": 19970 + }, + { + "epoch": 5.23, + "learning_rate": 0.0001, + "loss": 0.225, + "step": 19980 + }, + { + "epoch": 5.23, + "learning_rate": 0.0001, + "loss": 0.2288, + "step": 19990 + }, + { + "epoch": 5.23, + "learning_rate": 0.0001, + "loss": 0.2406, + "step": 20000 + }, + { + "epoch": 5.24, + "learning_rate": 0.0001, + "loss": 0.2306, + "step": 20010 + }, + { + "epoch": 5.24, + "learning_rate": 0.0001, + "loss": 0.2302, + "step": 20020 + }, + { + "epoch": 5.24, + "learning_rate": 0.0001, + "loss": 0.2234, + "step": 20030 + }, + { + "epoch": 5.24, + "learning_rate": 0.0001, + "loss": 0.2372, + "step": 20040 + }, + { + "epoch": 5.25, + "learning_rate": 0.0001, + "loss": 0.2299, + "step": 20050 + }, + { + "epoch": 5.25, + "learning_rate": 0.0001, + "loss": 0.241, + "step": 20060 + }, + { + "epoch": 5.25, + "learning_rate": 0.0001, + "loss": 0.2407, + "step": 20070 + }, + { + "epoch": 5.25, + "learning_rate": 0.0001, + "loss": 0.2428, + "step": 20080 + }, + { + "epoch": 5.26, + "learning_rate": 0.0001, + "loss": 0.2321, + "step": 20090 + }, + { + "epoch": 5.26, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 20100 + }, + { + "epoch": 5.26, + "learning_rate": 0.0001, + "loss": 0.2471, + "step": 20110 + }, + { + "epoch": 5.26, + "learning_rate": 0.0001, + "loss": 0.2378, + "step": 20120 + }, + { + "epoch": 5.27, + "learning_rate": 0.0001, + "loss": 0.2318, + "step": 20130 + }, + { + "epoch": 5.27, + "learning_rate": 0.0001, + "loss": 0.2297, + "step": 20140 + }, + { + "epoch": 5.27, + "learning_rate": 0.0001, + "loss": 0.2326, + "step": 20150 + }, + { + "epoch": 5.27, + "learning_rate": 0.0001, + "loss": 0.2294, + "step": 20160 + }, + { + "epoch": 5.28, + "learning_rate": 0.0001, + "loss": 0.235, + "step": 20170 + }, + { + "epoch": 5.28, + "learning_rate": 0.0001, + "loss": 0.2306, + "step": 20180 + }, + { + "epoch": 5.28, + "learning_rate": 0.0001, + "loss": 0.2283, + "step": 20190 + }, + { + "epoch": 5.29, + "learning_rate": 0.0001, + "loss": 0.2334, + "step": 20200 + }, + { + "epoch": 5.29, + "learning_rate": 0.0001, + "loss": 0.2324, + "step": 20210 + }, + { + "epoch": 5.29, + "learning_rate": 0.0001, + "loss": 0.2349, + "step": 20220 + }, + { + "epoch": 5.29, + "learning_rate": 0.0001, + "loss": 0.2336, + "step": 20230 + }, + { + "epoch": 5.3, + "learning_rate": 0.0001, + "loss": 0.2373, + "step": 20240 + }, + { + "epoch": 5.3, + "learning_rate": 0.0001, + "loss": 0.231, + "step": 20250 + }, + { + "epoch": 5.3, + "learning_rate": 0.0001, + "loss": 0.2281, + "step": 20260 + }, + { + "epoch": 5.3, + "learning_rate": 0.0001, + "loss": 0.2269, + "step": 20270 + }, + { + "epoch": 5.31, + "learning_rate": 0.0001, + "loss": 0.2306, + "step": 20280 + }, + { + "epoch": 5.31, + "learning_rate": 0.0001, + "loss": 0.235, + "step": 20290 + }, + { + "epoch": 5.31, + "learning_rate": 0.0001, + "loss": 0.2283, + "step": 20300 + }, + { + "epoch": 5.31, + "learning_rate": 0.0001, + "loss": 0.2276, + "step": 20310 + }, + { + "epoch": 5.32, + "learning_rate": 0.0001, + "loss": 0.2399, + "step": 20320 + }, + { + "epoch": 5.32, + "learning_rate": 0.0001, + "loss": 0.2398, + "step": 20330 + }, + { + "epoch": 5.32, + "learning_rate": 0.0001, + "loss": 0.2303, + "step": 20340 + }, + { + "epoch": 5.32, + "learning_rate": 0.0001, + "loss": 0.2374, + "step": 20350 + }, + { + "epoch": 5.33, + "learning_rate": 0.0001, + "loss": 0.227, + "step": 20360 + }, + { + "epoch": 5.33, + "learning_rate": 0.0001, + "loss": 0.2338, + "step": 20370 + }, + { + "epoch": 5.33, + "learning_rate": 0.0001, + "loss": 0.24, + "step": 20380 + }, + { + "epoch": 5.33, + "learning_rate": 0.0001, + "loss": 0.237, + "step": 20390 + }, + { + "epoch": 5.34, + "learning_rate": 0.0001, + "loss": 0.2317, + "step": 20400 + }, + { + "epoch": 5.34, + "learning_rate": 0.0001, + "loss": 0.2338, + "step": 20410 + }, + { + "epoch": 5.34, + "learning_rate": 0.0001, + "loss": 0.2317, + "step": 20420 + }, + { + "epoch": 5.35, + "learning_rate": 0.0001, + "loss": 0.2348, + "step": 20430 + }, + { + "epoch": 5.35, + "learning_rate": 0.0001, + "loss": 0.2442, + "step": 20440 + }, + { + "epoch": 5.35, + "learning_rate": 0.0001, + "loss": 0.2268, + "step": 20450 + }, + { + "epoch": 5.35, + "learning_rate": 0.0001, + "loss": 0.2384, + "step": 20460 + }, + { + "epoch": 5.36, + "learning_rate": 0.0001, + "loss": 0.2425, + "step": 20470 + }, + { + "epoch": 5.36, + "learning_rate": 0.0001, + "loss": 0.2472, + "step": 20480 + }, + { + "epoch": 5.36, + "learning_rate": 0.0001, + "loss": 0.2322, + "step": 20490 + }, + { + "epoch": 5.36, + "learning_rate": 0.0001, + "loss": 0.2405, + "step": 20500 + }, + { + "epoch": 5.37, + "learning_rate": 0.0001, + "loss": 0.2368, + "step": 20510 + }, + { + "epoch": 5.37, + "learning_rate": 0.0001, + "loss": 0.2414, + "step": 20520 + }, + { + "epoch": 5.37, + "learning_rate": 0.0001, + "loss": 0.2455, + "step": 20530 + }, + { + "epoch": 5.37, + "learning_rate": 0.0001, + "loss": 0.2362, + "step": 20540 + }, + { + "epoch": 5.38, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 20550 + }, + { + "epoch": 5.38, + "learning_rate": 0.0001, + "loss": 0.2302, + "step": 20560 + }, + { + "epoch": 5.38, + "learning_rate": 0.0001, + "loss": 0.2344, + "step": 20570 + }, + { + "epoch": 5.38, + "learning_rate": 0.0001, + "loss": 0.237, + "step": 20580 + }, + { + "epoch": 5.39, + "learning_rate": 0.0001, + "loss": 0.2412, + "step": 20590 + }, + { + "epoch": 5.39, + "learning_rate": 0.0001, + "loss": 0.2443, + "step": 20600 + }, + { + "epoch": 5.39, + "learning_rate": 0.0001, + "loss": 0.2447, + "step": 20610 + }, + { + "epoch": 5.4, + "learning_rate": 0.0001, + "loss": 0.24, + "step": 20620 + }, + { + "epoch": 5.4, + "learning_rate": 0.0001, + "loss": 0.2411, + "step": 20630 + }, + { + "epoch": 5.4, + "learning_rate": 0.0001, + "loss": 0.2336, + "step": 20640 + }, + { + "epoch": 5.4, + "learning_rate": 0.0001, + "loss": 0.2447, + "step": 20650 + }, + { + "epoch": 5.41, + "learning_rate": 0.0001, + "loss": 0.236, + "step": 20660 + }, + { + "epoch": 5.41, + "learning_rate": 0.0001, + "loss": 0.2405, + "step": 20670 + }, + { + "epoch": 5.41, + "learning_rate": 0.0001, + "loss": 0.2325, + "step": 20680 + }, + { + "epoch": 5.41, + "learning_rate": 0.0001, + "loss": 0.2368, + "step": 20690 + }, + { + "epoch": 5.42, + "learning_rate": 0.0001, + "loss": 0.2375, + "step": 20700 + }, + { + "epoch": 5.42, + "learning_rate": 0.0001, + "loss": 0.2301, + "step": 20710 + }, + { + "epoch": 5.42, + "learning_rate": 0.0001, + "loss": 0.2296, + "step": 20720 + }, + { + "epoch": 5.42, + "learning_rate": 0.0001, + "loss": 0.228, + "step": 20730 + }, + { + "epoch": 5.43, + "learning_rate": 0.0001, + "loss": 0.2403, + "step": 20740 + }, + { + "epoch": 5.43, + "learning_rate": 0.0001, + "loss": 0.2351, + "step": 20750 + }, + { + "epoch": 5.43, + "learning_rate": 0.0001, + "loss": 0.2494, + "step": 20760 + }, + { + "epoch": 5.43, + "learning_rate": 0.0001, + "loss": 0.2419, + "step": 20770 + }, + { + "epoch": 5.44, + "learning_rate": 0.0001, + "loss": 0.24, + "step": 20780 + }, + { + "epoch": 5.44, + "learning_rate": 0.0001, + "loss": 0.2363, + "step": 20790 + }, + { + "epoch": 5.44, + "learning_rate": 0.0001, + "loss": 0.2339, + "step": 20800 + }, + { + "epoch": 5.44, + "learning_rate": 0.0001, + "loss": 0.2324, + "step": 20810 + }, + { + "epoch": 5.45, + "learning_rate": 0.0001, + "loss": 0.2345, + "step": 20820 + }, + { + "epoch": 5.45, + "learning_rate": 0.0001, + "loss": 0.2484, + "step": 20830 + }, + { + "epoch": 5.45, + "learning_rate": 0.0001, + "loss": 0.2281, + "step": 20840 + }, + { + "epoch": 5.46, + "learning_rate": 0.0001, + "loss": 0.2298, + "step": 20850 + }, + { + "epoch": 5.46, + "learning_rate": 0.0001, + "loss": 0.2378, + "step": 20860 + }, + { + "epoch": 5.46, + "learning_rate": 0.0001, + "loss": 0.2379, + "step": 20870 + }, + { + "epoch": 5.46, + "learning_rate": 0.0001, + "loss": 0.2357, + "step": 20880 + }, + { + "epoch": 5.47, + "learning_rate": 0.0001, + "loss": 0.2445, + "step": 20890 + }, + { + "epoch": 5.47, + "learning_rate": 0.0001, + "loss": 0.2398, + "step": 20900 + }, + { + "epoch": 5.47, + "learning_rate": 0.0001, + "loss": 0.2338, + "step": 20910 + }, + { + "epoch": 5.47, + "learning_rate": 0.0001, + "loss": 0.2444, + "step": 20920 + }, + { + "epoch": 5.48, + "learning_rate": 0.0001, + "loss": 0.2417, + "step": 20930 + }, + { + "epoch": 5.48, + "learning_rate": 0.0001, + "loss": 0.2384, + "step": 20940 + }, + { + "epoch": 5.48, + "learning_rate": 0.0001, + "loss": 0.2395, + "step": 20950 + }, + { + "epoch": 5.48, + "learning_rate": 0.0001, + "loss": 0.2387, + "step": 20960 + }, + { + "epoch": 5.49, + "learning_rate": 0.0001, + "loss": 0.2453, + "step": 20970 + }, + { + "epoch": 5.49, + "learning_rate": 0.0001, + "loss": 0.2351, + "step": 20980 + }, + { + "epoch": 5.49, + "learning_rate": 0.0001, + "loss": 0.2303, + "step": 20990 + }, + { + "epoch": 5.49, + "learning_rate": 0.0001, + "loss": 0.2384, + "step": 21000 + }, + { + "epoch": 5.5, + "learning_rate": 0.0001, + "loss": 0.2366, + "step": 21010 + }, + { + "epoch": 5.5, + "learning_rate": 0.0001, + "loss": 0.2397, + "step": 21020 + }, + { + "epoch": 5.5, + "learning_rate": 0.0001, + "loss": 0.2375, + "step": 21030 + }, + { + "epoch": 5.5, + "learning_rate": 0.0001, + "loss": 0.2342, + "step": 21040 + }, + { + "epoch": 5.51, + "learning_rate": 0.0001, + "loss": 0.2482, + "step": 21050 + }, + { + "epoch": 5.51, + "learning_rate": 0.0001, + "loss": 0.2341, + "step": 21060 + }, + { + "epoch": 5.51, + "learning_rate": 0.0001, + "loss": 0.2376, + "step": 21070 + }, + { + "epoch": 5.52, + "learning_rate": 0.0001, + "loss": 0.2461, + "step": 21080 + }, + { + "epoch": 5.52, + "learning_rate": 0.0001, + "loss": 0.2329, + "step": 21090 + }, + { + "epoch": 5.52, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 21100 + }, + { + "epoch": 5.52, + "learning_rate": 0.0001, + "loss": 0.234, + "step": 21110 + }, + { + "epoch": 5.53, + "learning_rate": 0.0001, + "loss": 0.2464, + "step": 21120 + }, + { + "epoch": 5.53, + "learning_rate": 0.0001, + "loss": 0.238, + "step": 21130 + }, + { + "epoch": 5.53, + "learning_rate": 0.0001, + "loss": 0.242, + "step": 21140 + }, + { + "epoch": 5.53, + "learning_rate": 0.0001, + "loss": 0.2472, + "step": 21150 + }, + { + "epoch": 5.54, + "learning_rate": 0.0001, + "loss": 0.2396, + "step": 21160 + }, + { + "epoch": 5.54, + "learning_rate": 0.0001, + "loss": 0.2404, + "step": 21170 + }, + { + "epoch": 5.54, + "learning_rate": 0.0001, + "loss": 0.2416, + "step": 21180 + }, + { + "epoch": 5.54, + "learning_rate": 0.0001, + "loss": 0.2369, + "step": 21190 + }, + { + "epoch": 5.55, + "learning_rate": 0.0001, + "loss": 0.2404, + "step": 21200 + }, + { + "epoch": 5.55, + "learning_rate": 0.0001, + "loss": 0.2409, + "step": 21210 + }, + { + "epoch": 5.55, + "learning_rate": 0.0001, + "loss": 0.2345, + "step": 21220 + }, + { + "epoch": 5.55, + "learning_rate": 0.0001, + "loss": 0.2408, + "step": 21230 + }, + { + "epoch": 5.56, + "learning_rate": 0.0001, + "loss": 0.2449, + "step": 21240 + }, + { + "epoch": 5.56, + "learning_rate": 0.0001, + "loss": 0.2445, + "step": 21250 + }, + { + "epoch": 5.56, + "learning_rate": 0.0001, + "loss": 0.2368, + "step": 21260 + }, + { + "epoch": 5.57, + "learning_rate": 0.0001, + "loss": 0.2444, + "step": 21270 + }, + { + "epoch": 5.57, + "learning_rate": 0.0001, + "loss": 0.2496, + "step": 21280 + }, + { + "epoch": 5.57, + "learning_rate": 0.0001, + "loss": 0.2468, + "step": 21290 + }, + { + "epoch": 5.57, + "learning_rate": 0.0001, + "loss": 0.2484, + "step": 21300 + }, + { + "epoch": 5.58, + "learning_rate": 0.0001, + "loss": 0.239, + "step": 21310 + }, + { + "epoch": 5.58, + "learning_rate": 0.0001, + "loss": 0.2311, + "step": 21320 + }, + { + "epoch": 5.58, + "learning_rate": 0.0001, + "loss": 0.2358, + "step": 21330 + }, + { + "epoch": 5.58, + "learning_rate": 0.0001, + "loss": 0.2352, + "step": 21340 + }, + { + "epoch": 5.59, + "learning_rate": 0.0001, + "loss": 0.2577, + "step": 21350 + }, + { + "epoch": 5.59, + "learning_rate": 0.0001, + "loss": 0.2438, + "step": 21360 + }, + { + "epoch": 5.59, + "learning_rate": 0.0001, + "loss": 0.2429, + "step": 21370 + }, + { + "epoch": 5.59, + "learning_rate": 0.0001, + "loss": 0.2298, + "step": 21380 + }, + { + "epoch": 5.6, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 21390 + }, + { + "epoch": 5.6, + "learning_rate": 0.0001, + "loss": 0.2405, + "step": 21400 + }, + { + "epoch": 5.6, + "learning_rate": 0.0001, + "loss": 0.2424, + "step": 21410 + }, + { + "epoch": 5.6, + "learning_rate": 0.0001, + "loss": 0.2371, + "step": 21420 + }, + { + "epoch": 5.61, + "learning_rate": 0.0001, + "loss": 0.2427, + "step": 21430 + }, + { + "epoch": 5.61, + "learning_rate": 0.0001, + "loss": 0.2426, + "step": 21440 + }, + { + "epoch": 5.61, + "learning_rate": 0.0001, + "loss": 0.239, + "step": 21450 + }, + { + "epoch": 5.61, + "learning_rate": 0.0001, + "loss": 0.2457, + "step": 21460 + }, + { + "epoch": 5.62, + "learning_rate": 0.0001, + "loss": 0.2438, + "step": 21470 + }, + { + "epoch": 5.62, + "learning_rate": 0.0001, + "loss": 0.242, + "step": 21480 + }, + { + "epoch": 5.62, + "learning_rate": 0.0001, + "loss": 0.2309, + "step": 21490 + }, + { + "epoch": 5.63, + "learning_rate": 0.0001, + "loss": 0.2486, + "step": 21500 + }, + { + "epoch": 5.63, + "learning_rate": 0.0001, + "loss": 0.2474, + "step": 21510 + }, + { + "epoch": 5.63, + "learning_rate": 0.0001, + "loss": 0.2419, + "step": 21520 + }, + { + "epoch": 5.63, + "learning_rate": 0.0001, + "loss": 0.2457, + "step": 21530 + }, + { + "epoch": 5.64, + "learning_rate": 0.0001, + "loss": 0.2332, + "step": 21540 + }, + { + "epoch": 5.64, + "learning_rate": 0.0001, + "loss": 0.2325, + "step": 21550 + }, + { + "epoch": 5.64, + "learning_rate": 0.0001, + "loss": 0.2385, + "step": 21560 + }, + { + "epoch": 5.64, + "learning_rate": 0.0001, + "loss": 0.2461, + "step": 21570 + }, + { + "epoch": 5.65, + "learning_rate": 0.0001, + "loss": 0.2389, + "step": 21580 + }, + { + "epoch": 5.65, + "learning_rate": 0.0001, + "loss": 0.2468, + "step": 21590 + }, + { + "epoch": 5.65, + "learning_rate": 0.0001, + "loss": 0.2503, + "step": 21600 + }, + { + "epoch": 5.65, + "learning_rate": 0.0001, + "loss": 0.2431, + "step": 21610 + }, + { + "epoch": 5.66, + "learning_rate": 0.0001, + "loss": 0.2427, + "step": 21620 + }, + { + "epoch": 5.66, + "learning_rate": 0.0001, + "loss": 0.2274, + "step": 21630 + }, + { + "epoch": 5.66, + "learning_rate": 0.0001, + "loss": 0.2445, + "step": 21640 + }, + { + "epoch": 5.66, + "learning_rate": 0.0001, + "loss": 0.2436, + "step": 21650 + }, + { + "epoch": 5.67, + "learning_rate": 0.0001, + "loss": 0.2398, + "step": 21660 + }, + { + "epoch": 5.67, + "learning_rate": 0.0001, + "loss": 0.2473, + "step": 21670 + }, + { + "epoch": 5.67, + "learning_rate": 0.0001, + "loss": 0.2375, + "step": 21680 + }, + { + "epoch": 5.68, + "learning_rate": 0.0001, + "loss": 0.2429, + "step": 21690 + }, + { + "epoch": 5.68, + "learning_rate": 0.0001, + "loss": 0.2466, + "step": 21700 + }, + { + "epoch": 5.68, + "learning_rate": 0.0001, + "loss": 0.2402, + "step": 21710 + }, + { + "epoch": 5.68, + "learning_rate": 0.0001, + "loss": 0.2379, + "step": 21720 + }, + { + "epoch": 5.69, + "learning_rate": 0.0001, + "loss": 0.2428, + "step": 21730 + }, + { + "epoch": 5.69, + "learning_rate": 0.0001, + "loss": 0.2338, + "step": 21740 + }, + { + "epoch": 5.69, + "learning_rate": 0.0001, + "loss": 0.2413, + "step": 21750 + }, + { + "epoch": 5.69, + "learning_rate": 0.0001, + "loss": 0.25, + "step": 21760 + }, + { + "epoch": 5.7, + "learning_rate": 0.0001, + "loss": 0.2411, + "step": 21770 + }, + { + "epoch": 5.7, + "learning_rate": 0.0001, + "loss": 0.2427, + "step": 21780 + }, + { + "epoch": 5.7, + "learning_rate": 0.0001, + "loss": 0.2452, + "step": 21790 + }, + { + "epoch": 5.7, + "learning_rate": 0.0001, + "loss": 0.2403, + "step": 21800 + }, + { + "epoch": 5.71, + "learning_rate": 0.0001, + "loss": 0.238, + "step": 21810 + }, + { + "epoch": 5.71, + "learning_rate": 0.0001, + "loss": 0.2416, + "step": 21820 + }, + { + "epoch": 5.71, + "learning_rate": 0.0001, + "loss": 0.2403, + "step": 21830 + }, + { + "epoch": 5.71, + "learning_rate": 0.0001, + "loss": 0.2405, + "step": 21840 + }, + { + "epoch": 5.72, + "learning_rate": 0.0001, + "loss": 0.2343, + "step": 21850 + }, + { + "epoch": 5.72, + "learning_rate": 0.0001, + "loss": 0.2318, + "step": 21860 + }, + { + "epoch": 5.72, + "learning_rate": 0.0001, + "loss": 0.2429, + "step": 21870 + }, + { + "epoch": 5.72, + "learning_rate": 0.0001, + "loss": 0.2437, + "step": 21880 + }, + { + "epoch": 5.73, + "learning_rate": 0.0001, + "loss": 0.2482, + "step": 21890 + }, + { + "epoch": 5.73, + "learning_rate": 0.0001, + "loss": 0.2366, + "step": 21900 + }, + { + "epoch": 5.73, + "learning_rate": 0.0001, + "loss": 0.241, + "step": 21910 + }, + { + "epoch": 5.74, + "learning_rate": 0.0001, + "loss": 0.2343, + "step": 21920 + }, + { + "epoch": 5.74, + "learning_rate": 0.0001, + "loss": 0.2334, + "step": 21930 + }, + { + "epoch": 5.74, + "learning_rate": 0.0001, + "loss": 0.2367, + "step": 21940 + }, + { + "epoch": 5.74, + "learning_rate": 0.0001, + "loss": 0.2346, + "step": 21950 + }, + { + "epoch": 5.75, + "learning_rate": 0.0001, + "loss": 0.2443, + "step": 21960 + }, + { + "epoch": 5.75, + "learning_rate": 0.0001, + "loss": 0.2478, + "step": 21970 + }, + { + "epoch": 5.75, + "learning_rate": 0.0001, + "loss": 0.2522, + "step": 21980 + }, + { + "epoch": 5.75, + "learning_rate": 0.0001, + "loss": 0.2309, + "step": 21990 + }, + { + "epoch": 5.76, + "learning_rate": 0.0001, + "loss": 0.2434, + "step": 22000 + }, + { + "epoch": 5.76, + "learning_rate": 0.0001, + "loss": 0.2484, + "step": 22010 + }, + { + "epoch": 5.76, + "learning_rate": 0.0001, + "loss": 0.2415, + "step": 22020 + }, + { + "epoch": 5.76, + "learning_rate": 0.0001, + "loss": 0.2371, + "step": 22030 + }, + { + "epoch": 5.77, + "learning_rate": 0.0001, + "loss": 0.2401, + "step": 22040 + }, + { + "epoch": 5.77, + "learning_rate": 0.0001, + "loss": 0.2383, + "step": 22050 + }, + { + "epoch": 5.77, + "learning_rate": 0.0001, + "loss": 0.2314, + "step": 22060 + }, + { + "epoch": 5.77, + "learning_rate": 0.0001, + "loss": 0.2448, + "step": 22070 + }, + { + "epoch": 5.78, + "learning_rate": 0.0001, + "loss": 0.2419, + "step": 22080 + }, + { + "epoch": 5.78, + "learning_rate": 0.0001, + "loss": 0.2418, + "step": 22090 + }, + { + "epoch": 5.78, + "learning_rate": 0.0001, + "loss": 0.2378, + "step": 22100 + }, + { + "epoch": 5.78, + "learning_rate": 0.0001, + "loss": 0.2454, + "step": 22110 + }, + { + "epoch": 5.79, + "learning_rate": 0.0001, + "loss": 0.249, + "step": 22120 + }, + { + "epoch": 5.79, + "learning_rate": 0.0001, + "loss": 0.2467, + "step": 22130 + }, + { + "epoch": 5.79, + "learning_rate": 0.0001, + "loss": 0.2452, + "step": 22140 + }, + { + "epoch": 5.8, + "learning_rate": 0.0001, + "loss": 0.2404, + "step": 22150 + }, + { + "epoch": 5.8, + "learning_rate": 0.0001, + "loss": 0.2443, + "step": 22160 + }, + { + "epoch": 5.8, + "learning_rate": 0.0001, + "loss": 0.2546, + "step": 22170 + }, + { + "epoch": 5.8, + "learning_rate": 0.0001, + "loss": 0.2344, + "step": 22180 + }, + { + "epoch": 5.81, + "learning_rate": 0.0001, + "loss": 0.2427, + "step": 22190 + }, + { + "epoch": 5.81, + "learning_rate": 0.0001, + "loss": 0.2418, + "step": 22200 + }, + { + "epoch": 5.81, + "learning_rate": 0.0001, + "loss": 0.2428, + "step": 22210 + }, + { + "epoch": 5.81, + "learning_rate": 0.0001, + "loss": 0.2491, + "step": 22220 + }, + { + "epoch": 5.82, + "learning_rate": 0.0001, + "loss": 0.2437, + "step": 22230 + }, + { + "epoch": 5.82, + "learning_rate": 0.0001, + "loss": 0.2401, + "step": 22240 + }, + { + "epoch": 5.82, + "learning_rate": 0.0001, + "loss": 0.2419, + "step": 22250 + }, + { + "epoch": 5.82, + "learning_rate": 0.0001, + "loss": 0.2391, + "step": 22260 + }, + { + "epoch": 5.83, + "learning_rate": 0.0001, + "loss": 0.2478, + "step": 22270 + }, + { + "epoch": 5.83, + "learning_rate": 0.0001, + "loss": 0.2395, + "step": 22280 + }, + { + "epoch": 5.83, + "learning_rate": 0.0001, + "loss": 0.2526, + "step": 22290 + }, + { + "epoch": 5.83, + "learning_rate": 0.0001, + "loss": 0.2452, + "step": 22300 + }, + { + "epoch": 5.84, + "learning_rate": 0.0001, + "loss": 0.2405, + "step": 22310 + }, + { + "epoch": 5.84, + "learning_rate": 0.0001, + "loss": 0.2406, + "step": 22320 + }, + { + "epoch": 5.84, + "learning_rate": 0.0001, + "loss": 0.2303, + "step": 22330 + }, + { + "epoch": 5.85, + "learning_rate": 0.0001, + "loss": 0.2454, + "step": 22340 + }, + { + "epoch": 5.85, + "learning_rate": 0.0001, + "loss": 0.2448, + "step": 22350 + }, + { + "epoch": 5.85, + "learning_rate": 0.0001, + "loss": 0.238, + "step": 22360 + }, + { + "epoch": 5.85, + "learning_rate": 0.0001, + "loss": 0.2418, + "step": 22370 + }, + { + "epoch": 5.86, + "learning_rate": 0.0001, + "loss": 0.2411, + "step": 22380 + }, + { + "epoch": 5.86, + "learning_rate": 0.0001, + "loss": 0.2433, + "step": 22390 + }, + { + "epoch": 5.86, + "learning_rate": 0.0001, + "loss": 0.2474, + "step": 22400 + }, + { + "epoch": 5.86, + "learning_rate": 0.0001, + "loss": 0.2459, + "step": 22410 + }, + { + "epoch": 5.87, + "learning_rate": 0.0001, + "loss": 0.2419, + "step": 22420 + }, + { + "epoch": 5.87, + "learning_rate": 0.0001, + "loss": 0.2434, + "step": 22430 + }, + { + "epoch": 5.87, + "learning_rate": 0.0001, + "loss": 0.238, + "step": 22440 + }, + { + "epoch": 5.87, + "learning_rate": 0.0001, + "loss": 0.2494, + "step": 22450 + }, + { + "epoch": 5.88, + "learning_rate": 0.0001, + "loss": 0.2538, + "step": 22460 + }, + { + "epoch": 5.88, + "learning_rate": 0.0001, + "loss": 0.2502, + "step": 22470 + }, + { + "epoch": 5.88, + "learning_rate": 0.0001, + "loss": 0.2528, + "step": 22480 + }, + { + "epoch": 5.88, + "learning_rate": 0.0001, + "loss": 0.2519, + "step": 22490 + }, + { + "epoch": 5.89, + "learning_rate": 0.0001, + "loss": 0.255, + "step": 22500 + }, + { + "epoch": 5.89, + "learning_rate": 0.0001, + "loss": 0.2441, + "step": 22510 + }, + { + "epoch": 5.89, + "learning_rate": 0.0001, + "loss": 0.2335, + "step": 22520 + }, + { + "epoch": 5.89, + "learning_rate": 0.0001, + "loss": 0.2554, + "step": 22530 + }, + { + "epoch": 5.9, + "learning_rate": 0.0001, + "loss": 0.244, + "step": 22540 + }, + { + "epoch": 5.9, + "learning_rate": 0.0001, + "loss": 0.2476, + "step": 22550 + }, + { + "epoch": 5.9, + "learning_rate": 0.0001, + "loss": 0.2464, + "step": 22560 + }, + { + "epoch": 5.91, + "learning_rate": 0.0001, + "loss": 0.2457, + "step": 22570 + }, + { + "epoch": 5.91, + "learning_rate": 0.0001, + "loss": 0.2467, + "step": 22580 + }, + { + "epoch": 5.91, + "learning_rate": 0.0001, + "loss": 0.2518, + "step": 22590 + }, + { + "epoch": 5.91, + "learning_rate": 0.0001, + "loss": 0.2504, + "step": 22600 + }, + { + "epoch": 5.92, + "learning_rate": 0.0001, + "loss": 0.2488, + "step": 22610 + }, + { + "epoch": 5.92, + "learning_rate": 0.0001, + "loss": 0.2421, + "step": 22620 + }, + { + "epoch": 5.92, + "learning_rate": 0.0001, + "loss": 0.2305, + "step": 22630 + }, + { + "epoch": 5.92, + "learning_rate": 0.0001, + "loss": 0.2463, + "step": 22640 + }, + { + "epoch": 5.93, + "learning_rate": 0.0001, + "loss": 0.2421, + "step": 22650 + }, + { + "epoch": 5.93, + "learning_rate": 0.0001, + "loss": 0.2386, + "step": 22660 + }, + { + "epoch": 5.93, + "learning_rate": 0.0001, + "loss": 0.245, + "step": 22670 + }, + { + "epoch": 5.93, + "learning_rate": 0.0001, + "loss": 0.2523, + "step": 22680 + }, + { + "epoch": 5.94, + "learning_rate": 0.0001, + "loss": 0.2512, + "step": 22690 + }, + { + "epoch": 5.94, + "learning_rate": 0.0001, + "loss": 0.2411, + "step": 22700 + }, + { + "epoch": 5.94, + "learning_rate": 0.0001, + "loss": 0.2424, + "step": 22710 + }, + { + "epoch": 5.94, + "learning_rate": 0.0001, + "loss": 0.2422, + "step": 22720 + }, + { + "epoch": 5.95, + "learning_rate": 0.0001, + "loss": 0.246, + "step": 22730 + }, + { + "epoch": 5.95, + "learning_rate": 0.0001, + "loss": 0.2486, + "step": 22740 + }, + { + "epoch": 5.95, + "learning_rate": 0.0001, + "loss": 0.2463, + "step": 22750 + }, + { + "epoch": 5.95, + "learning_rate": 0.0001, + "loss": 0.2395, + "step": 22760 + }, + { + "epoch": 5.96, + "learning_rate": 0.0001, + "loss": 0.2549, + "step": 22770 + }, + { + "epoch": 5.96, + "learning_rate": 0.0001, + "loss": 0.2472, + "step": 22780 + }, + { + "epoch": 5.96, + "learning_rate": 0.0001, + "loss": 0.254, + "step": 22790 + }, + { + "epoch": 5.97, + "learning_rate": 0.0001, + "loss": 0.2451, + "step": 22800 + }, + { + "epoch": 5.97, + "learning_rate": 0.0001, + "loss": 0.2492, + "step": 22810 + }, + { + "epoch": 5.97, + "learning_rate": 0.0001, + "loss": 0.2435, + "step": 22820 + }, + { + "epoch": 5.97, + "learning_rate": 0.0001, + "loss": 0.241, + "step": 22830 + }, + { + "epoch": 5.98, + "learning_rate": 0.0001, + "loss": 0.2446, + "step": 22840 + }, + { + "epoch": 5.98, + "learning_rate": 0.0001, + "loss": 0.2562, + "step": 22850 + }, + { + "epoch": 5.98, + "learning_rate": 0.0001, + "loss": 0.2514, + "step": 22860 + }, + { + "epoch": 5.98, + "learning_rate": 0.0001, + "loss": 0.2349, + "step": 22870 + }, + { + "epoch": 5.99, + "learning_rate": 0.0001, + "loss": 0.2501, + "step": 22880 + }, + { + "epoch": 5.99, + "learning_rate": 0.0001, + "loss": 0.2458, + "step": 22890 + }, + { + "epoch": 5.99, + "learning_rate": 0.0001, + "loss": 0.2428, + "step": 22900 + }, + { + "epoch": 5.99, + "learning_rate": 0.0001, + "loss": 0.2454, + "step": 22910 + }, + { + "epoch": 6.0, + "learning_rate": 0.0001, + "loss": 0.2591, + "step": 22920 + }, + { + "epoch": 6.0, + "learning_rate": 0.0001, + "loss": 0.2498, + "step": 22930 + }, + { + "epoch": 6.0, + "eval_gen_len": 61.8942, + "eval_loss": 1.1511127948760986, + "eval_rouge1": 56.7504, + "eval_rouge2": 34.6807, + "eval_rougeL": 46.0248, + "eval_rougeLsum": 52.2035, + "eval_runtime": 36.4344, + "eval_samples_per_second": 24.126, + "eval_steps_per_second": 0.384, + "step": 22932 + }, + { + "epoch": 6.0, + "learning_rate": 0.0001, + "loss": 0.205, + "step": 22940 + }, + { + "epoch": 6.0, + "learning_rate": 0.0001, + "loss": 0.1979, + "step": 22950 + }, + { + "epoch": 6.01, + "learning_rate": 0.0001, + "loss": 0.1994, + "step": 22960 + }, + { + "epoch": 6.01, + "learning_rate": 0.0001, + "loss": 0.2014, + "step": 22970 + }, + { + "epoch": 6.01, + "learning_rate": 0.0001, + "loss": 0.2054, + "step": 22980 + }, + { + "epoch": 6.02, + "learning_rate": 0.0001, + "loss": 0.2016, + "step": 22990 + }, + { + "epoch": 6.02, + "learning_rate": 0.0001, + "loss": 0.2027, + "step": 23000 + }, + { + "epoch": 6.02, + "learning_rate": 0.0001, + "loss": 0.2011, + "step": 23010 + }, + { + "epoch": 6.02, + "learning_rate": 0.0001, + "loss": 0.1983, + "step": 23020 + }, + { + "epoch": 6.03, + "learning_rate": 0.0001, + "loss": 0.2001, + "step": 23030 + }, + { + "epoch": 6.03, + "learning_rate": 0.0001, + "loss": 0.2065, + "step": 23040 + }, + { + "epoch": 6.03, + "learning_rate": 0.0001, + "loss": 0.1932, + "step": 23050 + }, + { + "epoch": 6.03, + "learning_rate": 0.0001, + "loss": 0.1911, + "step": 23060 + }, + { + "epoch": 6.04, + "learning_rate": 0.0001, + "loss": 0.1944, + "step": 23070 + }, + { + "epoch": 6.04, + "learning_rate": 0.0001, + "loss": 0.2023, + "step": 23080 + }, + { + "epoch": 6.04, + "learning_rate": 0.0001, + "loss": 0.203, + "step": 23090 + }, + { + "epoch": 6.04, + "learning_rate": 0.0001, + "loss": 0.1984, + "step": 23100 + }, + { + "epoch": 6.05, + "learning_rate": 0.0001, + "loss": 0.2052, + "step": 23110 + }, + { + "epoch": 6.05, + "learning_rate": 0.0001, + "loss": 0.1995, + "step": 23120 + }, + { + "epoch": 6.05, + "learning_rate": 0.0001, + "loss": 0.1974, + "step": 23130 + }, + { + "epoch": 6.05, + "learning_rate": 0.0001, + "loss": 0.2049, + "step": 23140 + }, + { + "epoch": 6.06, + "learning_rate": 0.0001, + "loss": 0.1976, + "step": 23150 + }, + { + "epoch": 6.06, + "learning_rate": 0.0001, + "loss": 0.2028, + "step": 23160 + }, + { + "epoch": 6.06, + "learning_rate": 0.0001, + "loss": 0.2032, + "step": 23170 + }, + { + "epoch": 6.06, + "learning_rate": 0.0001, + "loss": 0.2032, + "step": 23180 + }, + { + "epoch": 6.07, + "learning_rate": 0.0001, + "loss": 0.2002, + "step": 23190 + }, + { + "epoch": 6.07, + "learning_rate": 0.0001, + "loss": 0.2067, + "step": 23200 + }, + { + "epoch": 6.07, + "learning_rate": 0.0001, + "loss": 0.202, + "step": 23210 + }, + { + "epoch": 6.08, + "learning_rate": 0.0001, + "loss": 0.2051, + "step": 23220 + }, + { + "epoch": 6.08, + "learning_rate": 0.0001, + "loss": 0.2028, + "step": 23230 + }, + { + "epoch": 6.08, + "learning_rate": 0.0001, + "loss": 0.2015, + "step": 23240 + }, + { + "epoch": 6.08, + "learning_rate": 0.0001, + "loss": 0.2059, + "step": 23250 + }, + { + "epoch": 6.09, + "learning_rate": 0.0001, + "loss": 0.2022, + "step": 23260 + }, + { + "epoch": 6.09, + "learning_rate": 0.0001, + "loss": 0.1951, + "step": 23270 + }, + { + "epoch": 6.09, + "learning_rate": 0.0001, + "loss": 0.1977, + "step": 23280 + }, + { + "epoch": 6.09, + "learning_rate": 0.0001, + "loss": 0.2052, + "step": 23290 + }, + { + "epoch": 6.1, + "learning_rate": 0.0001, + "loss": 0.199, + "step": 23300 + }, + { + "epoch": 6.1, + "learning_rate": 0.0001, + "loss": 0.2051, + "step": 23310 + }, + { + "epoch": 6.1, + "learning_rate": 0.0001, + "loss": 0.2055, + "step": 23320 + }, + { + "epoch": 6.1, + "learning_rate": 0.0001, + "loss": 0.1952, + "step": 23330 + }, + { + "epoch": 6.11, + "learning_rate": 0.0001, + "loss": 0.2055, + "step": 23340 + }, + { + "epoch": 6.11, + "learning_rate": 0.0001, + "loss": 0.2084, + "step": 23350 + }, + { + "epoch": 6.11, + "learning_rate": 0.0001, + "loss": 0.2031, + "step": 23360 + }, + { + "epoch": 6.11, + "learning_rate": 0.0001, + "loss": 0.199, + "step": 23370 + }, + { + "epoch": 6.12, + "learning_rate": 0.0001, + "loss": 0.2048, + "step": 23380 + }, + { + "epoch": 6.12, + "learning_rate": 0.0001, + "loss": 0.2028, + "step": 23390 + }, + { + "epoch": 6.12, + "learning_rate": 0.0001, + "loss": 0.2002, + "step": 23400 + }, + { + "epoch": 6.13, + "learning_rate": 0.0001, + "loss": 0.1952, + "step": 23410 + }, + { + "epoch": 6.13, + "learning_rate": 0.0001, + "loss": 0.2044, + "step": 23420 + }, + { + "epoch": 6.13, + "learning_rate": 0.0001, + "loss": 0.2065, + "step": 23430 + }, + { + "epoch": 6.13, + "learning_rate": 0.0001, + "loss": 0.2026, + "step": 23440 + }, + { + "epoch": 6.14, + "learning_rate": 0.0001, + "loss": 0.2032, + "step": 23450 + }, + { + "epoch": 6.14, + "learning_rate": 0.0001, + "loss": 0.2029, + "step": 23460 + }, + { + "epoch": 6.14, + "learning_rate": 0.0001, + "loss": 0.2075, + "step": 23470 + }, + { + "epoch": 6.14, + "learning_rate": 0.0001, + "loss": 0.2156, + "step": 23480 + }, + { + "epoch": 6.15, + "learning_rate": 0.0001, + "loss": 0.2016, + "step": 23490 + }, + { + "epoch": 6.15, + "learning_rate": 0.0001, + "loss": 0.2013, + "step": 23500 + }, + { + "epoch": 6.15, + "learning_rate": 0.0001, + "loss": 0.2108, + "step": 23510 + }, + { + "epoch": 6.15, + "learning_rate": 0.0001, + "loss": 0.1994, + "step": 23520 + }, + { + "epoch": 6.16, + "learning_rate": 0.0001, + "loss": 0.2064, + "step": 23530 + }, + { + "epoch": 6.16, + "learning_rate": 0.0001, + "loss": 0.2026, + "step": 23540 + }, + { + "epoch": 6.16, + "learning_rate": 0.0001, + "loss": 0.1994, + "step": 23550 + }, + { + "epoch": 6.16, + "learning_rate": 0.0001, + "loss": 0.204, + "step": 23560 + }, + { + "epoch": 6.17, + "learning_rate": 0.0001, + "loss": 0.2067, + "step": 23570 + }, + { + "epoch": 6.17, + "learning_rate": 0.0001, + "loss": 0.1987, + "step": 23580 + }, + { + "epoch": 6.17, + "learning_rate": 0.0001, + "loss": 0.2047, + "step": 23590 + }, + { + "epoch": 6.17, + "learning_rate": 0.0001, + "loss": 0.2047, + "step": 23600 + }, + { + "epoch": 6.18, + "learning_rate": 0.0001, + "loss": 0.2046, + "step": 23610 + }, + { + "epoch": 6.18, + "learning_rate": 0.0001, + "loss": 0.2084, + "step": 23620 + }, + { + "epoch": 6.18, + "learning_rate": 0.0001, + "loss": 0.2074, + "step": 23630 + }, + { + "epoch": 6.19, + "learning_rate": 0.0001, + "loss": 0.2108, + "step": 23640 + }, + { + "epoch": 6.19, + "learning_rate": 0.0001, + "loss": 0.2072, + "step": 23650 + }, + { + "epoch": 6.19, + "learning_rate": 0.0001, + "loss": 0.2073, + "step": 23660 + }, + { + "epoch": 6.19, + "learning_rate": 0.0001, + "loss": 0.2057, + "step": 23670 + }, + { + "epoch": 6.2, + "learning_rate": 0.0001, + "loss": 0.2028, + "step": 23680 + }, + { + "epoch": 6.2, + "learning_rate": 0.0001, + "loss": 0.2103, + "step": 23690 + }, + { + "epoch": 6.2, + "learning_rate": 0.0001, + "loss": 0.2065, + "step": 23700 + }, + { + "epoch": 6.2, + "learning_rate": 0.0001, + "loss": 0.2103, + "step": 23710 + }, + { + "epoch": 6.21, + "learning_rate": 0.0001, + "loss": 0.2053, + "step": 23720 + }, + { + "epoch": 6.21, + "learning_rate": 0.0001, + "loss": 0.2076, + "step": 23730 + }, + { + "epoch": 6.21, + "learning_rate": 0.0001, + "loss": 0.2043, + "step": 23740 + }, + { + "epoch": 6.21, + "learning_rate": 0.0001, + "loss": 0.1973, + "step": 23750 + }, + { + "epoch": 6.22, + "learning_rate": 0.0001, + "loss": 0.1999, + "step": 23760 + }, + { + "epoch": 6.22, + "learning_rate": 0.0001, + "loss": 0.2008, + "step": 23770 + }, + { + "epoch": 6.22, + "learning_rate": 0.0001, + "loss": 0.2082, + "step": 23780 + }, + { + "epoch": 6.22, + "learning_rate": 0.0001, + "loss": 0.21, + "step": 23790 + }, + { + "epoch": 6.23, + "learning_rate": 0.0001, + "loss": 0.1965, + "step": 23800 + }, + { + "epoch": 6.23, + "learning_rate": 0.0001, + "loss": 0.2035, + "step": 23810 + }, + { + "epoch": 6.23, + "learning_rate": 0.0001, + "loss": 0.215, + "step": 23820 + }, + { + "epoch": 6.23, + "learning_rate": 0.0001, + "loss": 0.2083, + "step": 23830 + }, + { + "epoch": 6.24, + "learning_rate": 0.0001, + "loss": 0.2019, + "step": 23840 + }, + { + "epoch": 6.24, + "learning_rate": 0.0001, + "loss": 0.2041, + "step": 23850 + }, + { + "epoch": 6.24, + "learning_rate": 0.0001, + "loss": 0.2027, + "step": 23860 + }, + { + "epoch": 6.25, + "learning_rate": 0.0001, + "loss": 0.2116, + "step": 23870 + }, + { + "epoch": 6.25, + "learning_rate": 0.0001, + "loss": 0.2027, + "step": 23880 + }, + { + "epoch": 6.25, + "learning_rate": 0.0001, + "loss": 0.2063, + "step": 23890 + }, + { + "epoch": 6.25, + "learning_rate": 0.0001, + "loss": 0.2072, + "step": 23900 + }, + { + "epoch": 6.26, + "learning_rate": 0.0001, + "loss": 0.2094, + "step": 23910 + }, + { + "epoch": 6.26, + "learning_rate": 0.0001, + "loss": 0.2073, + "step": 23920 + }, + { + "epoch": 6.26, + "learning_rate": 0.0001, + "loss": 0.2045, + "step": 23930 + }, + { + "epoch": 6.26, + "learning_rate": 0.0001, + "loss": 0.2098, + "step": 23940 + }, + { + "epoch": 6.27, + "learning_rate": 0.0001, + "loss": 0.2057, + "step": 23950 + }, + { + "epoch": 6.27, + "learning_rate": 0.0001, + "loss": 0.2105, + "step": 23960 + }, + { + "epoch": 6.27, + "learning_rate": 0.0001, + "loss": 0.2063, + "step": 23970 + }, + { + "epoch": 6.27, + "learning_rate": 0.0001, + "loss": 0.2084, + "step": 23980 + }, + { + "epoch": 6.28, + "learning_rate": 0.0001, + "loss": 0.212, + "step": 23990 + }, + { + "epoch": 6.28, + "learning_rate": 0.0001, + "loss": 0.1992, + "step": 24000 + }, + { + "epoch": 6.28, + "learning_rate": 0.0001, + "loss": 0.1961, + "step": 24010 + }, + { + "epoch": 6.28, + "learning_rate": 0.0001, + "loss": 0.2157, + "step": 24020 + }, + { + "epoch": 6.29, + "learning_rate": 0.0001, + "loss": 0.2072, + "step": 24030 + }, + { + "epoch": 6.29, + "learning_rate": 0.0001, + "loss": 0.2025, + "step": 24040 + }, + { + "epoch": 6.29, + "learning_rate": 0.0001, + "loss": 0.2044, + "step": 24050 + }, + { + "epoch": 6.3, + "learning_rate": 0.0001, + "loss": 0.2094, + "step": 24060 + }, + { + "epoch": 6.3, + "learning_rate": 0.0001, + "loss": 0.2041, + "step": 24070 + }, + { + "epoch": 6.3, + "learning_rate": 0.0001, + "loss": 0.2086, + "step": 24080 + }, + { + "epoch": 6.3, + "learning_rate": 0.0001, + "loss": 0.2013, + "step": 24090 + }, + { + "epoch": 6.31, + "learning_rate": 0.0001, + "loss": 0.2106, + "step": 24100 + }, + { + "epoch": 6.31, + "learning_rate": 0.0001, + "loss": 0.1999, + "step": 24110 + }, + { + "epoch": 6.31, + "learning_rate": 0.0001, + "loss": 0.2023, + "step": 24120 + }, + { + "epoch": 6.31, + "learning_rate": 0.0001, + "loss": 0.206, + "step": 24130 + }, + { + "epoch": 6.32, + "learning_rate": 0.0001, + "loss": 0.2058, + "step": 24140 + }, + { + "epoch": 6.32, + "learning_rate": 0.0001, + "loss": 0.2089, + "step": 24150 + }, + { + "epoch": 6.32, + "learning_rate": 0.0001, + "loss": 0.1977, + "step": 24160 + }, + { + "epoch": 6.32, + "learning_rate": 0.0001, + "loss": 0.2118, + "step": 24170 + }, + { + "epoch": 6.33, + "learning_rate": 0.0001, + "loss": 0.2203, + "step": 24180 + }, + { + "epoch": 6.33, + "learning_rate": 0.0001, + "loss": 0.2102, + "step": 24190 + }, + { + "epoch": 6.33, + "learning_rate": 0.0001, + "loss": 0.1975, + "step": 24200 + }, + { + "epoch": 6.33, + "learning_rate": 0.0001, + "loss": 0.2113, + "step": 24210 + }, + { + "epoch": 6.34, + "learning_rate": 0.0001, + "loss": 0.2097, + "step": 24220 + }, + { + "epoch": 6.34, + "learning_rate": 0.0001, + "loss": 0.2098, + "step": 24230 + }, + { + "epoch": 6.34, + "learning_rate": 0.0001, + "loss": 0.2133, + "step": 24240 + }, + { + "epoch": 6.34, + "learning_rate": 0.0001, + "loss": 0.2231, + "step": 24250 + }, + { + "epoch": 6.35, + "learning_rate": 0.0001, + "loss": 0.2071, + "step": 24260 + }, + { + "epoch": 6.35, + "learning_rate": 0.0001, + "loss": 0.2019, + "step": 24270 + }, + { + "epoch": 6.35, + "learning_rate": 0.0001, + "loss": 0.2082, + "step": 24280 + }, + { + "epoch": 6.36, + "learning_rate": 0.0001, + "loss": 0.1957, + "step": 24290 + }, + { + "epoch": 6.36, + "learning_rate": 0.0001, + "loss": 0.2044, + "step": 24300 + }, + { + "epoch": 6.36, + "learning_rate": 0.0001, + "loss": 0.2157, + "step": 24310 + }, + { + "epoch": 6.36, + "learning_rate": 0.0001, + "loss": 0.2057, + "step": 24320 + }, + { + "epoch": 6.37, + "learning_rate": 0.0001, + "loss": 0.2087, + "step": 24330 + }, + { + "epoch": 6.37, + "learning_rate": 0.0001, + "loss": 0.209, + "step": 24340 + }, + { + "epoch": 6.37, + "learning_rate": 0.0001, + "loss": 0.2108, + "step": 24350 + }, + { + "epoch": 6.37, + "learning_rate": 0.0001, + "loss": 0.2142, + "step": 24360 + }, + { + "epoch": 6.38, + "learning_rate": 0.0001, + "loss": 0.2081, + "step": 24370 + }, + { + "epoch": 6.38, + "learning_rate": 0.0001, + "loss": 0.2048, + "step": 24380 + }, + { + "epoch": 6.38, + "learning_rate": 0.0001, + "loss": 0.2115, + "step": 24390 + }, + { + "epoch": 6.38, + "learning_rate": 0.0001, + "loss": 0.2022, + "step": 24400 + }, + { + "epoch": 6.39, + "learning_rate": 0.0001, + "loss": 0.2082, + "step": 24410 + }, + { + "epoch": 6.39, + "learning_rate": 0.0001, + "loss": 0.2128, + "step": 24420 + }, + { + "epoch": 6.39, + "learning_rate": 0.0001, + "loss": 0.2003, + "step": 24430 + }, + { + "epoch": 6.39, + "learning_rate": 0.0001, + "loss": 0.2145, + "step": 24440 + }, + { + "epoch": 6.4, + "learning_rate": 0.0001, + "loss": 0.2118, + "step": 24450 + }, + { + "epoch": 6.4, + "learning_rate": 0.0001, + "loss": 0.219, + "step": 24460 + }, + { + "epoch": 6.4, + "learning_rate": 0.0001, + "loss": 0.2089, + "step": 24470 + }, + { + "epoch": 6.41, + "learning_rate": 0.0001, + "loss": 0.2171, + "step": 24480 + }, + { + "epoch": 6.41, + "learning_rate": 0.0001, + "loss": 0.2092, + "step": 24490 + }, + { + "epoch": 6.41, + "learning_rate": 0.0001, + "loss": 0.2025, + "step": 24500 + }, + { + "epoch": 6.41, + "learning_rate": 0.0001, + "loss": 0.1957, + "step": 24510 + }, + { + "epoch": 6.42, + "learning_rate": 0.0001, + "loss": 0.2111, + "step": 24520 + }, + { + "epoch": 6.42, + "learning_rate": 0.0001, + "loss": 0.2231, + "step": 24530 + }, + { + "epoch": 6.42, + "learning_rate": 0.0001, + "loss": 0.2146, + "step": 24540 + }, + { + "epoch": 6.42, + "learning_rate": 0.0001, + "loss": 0.2072, + "step": 24550 + }, + { + "epoch": 6.43, + "learning_rate": 0.0001, + "loss": 0.2109, + "step": 24560 + }, + { + "epoch": 6.43, + "learning_rate": 0.0001, + "loss": 0.2042, + "step": 24570 + }, + { + "epoch": 6.43, + "learning_rate": 0.0001, + "loss": 0.2053, + "step": 24580 + }, + { + "epoch": 6.43, + "learning_rate": 0.0001, + "loss": 0.211, + "step": 24590 + }, + { + "epoch": 6.44, + "learning_rate": 0.0001, + "loss": 0.214, + "step": 24600 + }, + { + "epoch": 6.44, + "learning_rate": 0.0001, + "loss": 0.2125, + "step": 24610 + }, + { + "epoch": 6.44, + "learning_rate": 0.0001, + "loss": 0.2186, + "step": 24620 + }, + { + "epoch": 6.44, + "learning_rate": 0.0001, + "loss": 0.2128, + "step": 24630 + }, + { + "epoch": 6.45, + "learning_rate": 0.0001, + "loss": 0.2167, + "step": 24640 + }, + { + "epoch": 6.45, + "learning_rate": 0.0001, + "loss": 0.2089, + "step": 24650 + }, + { + "epoch": 6.45, + "learning_rate": 0.0001, + "loss": 0.2014, + "step": 24660 + }, + { + "epoch": 6.45, + "learning_rate": 0.0001, + "loss": 0.204, + "step": 24670 + }, + { + "epoch": 6.46, + "learning_rate": 0.0001, + "loss": 0.2069, + "step": 24680 + }, + { + "epoch": 6.46, + "learning_rate": 0.0001, + "loss": 0.2139, + "step": 24690 + }, + { + "epoch": 6.46, + "learning_rate": 0.0001, + "loss": 0.2122, + "step": 24700 + }, + { + "epoch": 6.47, + "learning_rate": 0.0001, + "loss": 0.2121, + "step": 24710 + }, + { + "epoch": 6.47, + "learning_rate": 0.0001, + "loss": 0.216, + "step": 24720 + }, + { + "epoch": 6.47, + "learning_rate": 0.0001, + "loss": 0.2109, + "step": 24730 + }, + { + "epoch": 6.47, + "learning_rate": 0.0001, + "loss": 0.2136, + "step": 24740 + }, + { + "epoch": 6.48, + "learning_rate": 0.0001, + "loss": 0.2053, + "step": 24750 + }, + { + "epoch": 6.48, + "learning_rate": 0.0001, + "loss": 0.2068, + "step": 24760 + }, + { + "epoch": 6.48, + "learning_rate": 0.0001, + "loss": 0.2073, + "step": 24770 + }, + { + "epoch": 6.48, + "learning_rate": 0.0001, + "loss": 0.2118, + "step": 24780 + }, + { + "epoch": 6.49, + "learning_rate": 0.0001, + "loss": 0.2131, + "step": 24790 + }, + { + "epoch": 6.49, + "learning_rate": 0.0001, + "loss": 0.2096, + "step": 24800 + }, + { + "epoch": 6.49, + "learning_rate": 0.0001, + "loss": 0.214, + "step": 24810 + }, + { + "epoch": 6.49, + "learning_rate": 0.0001, + "loss": 0.2188, + "step": 24820 + }, + { + "epoch": 6.5, + "learning_rate": 0.0001, + "loss": 0.2161, + "step": 24830 + }, + { + "epoch": 6.5, + "learning_rate": 0.0001, + "loss": 0.2163, + "step": 24840 + }, + { + "epoch": 6.5, + "learning_rate": 0.0001, + "loss": 0.2119, + "step": 24850 + }, + { + "epoch": 6.5, + "learning_rate": 0.0001, + "loss": 0.2149, + "step": 24860 + }, + { + "epoch": 6.51, + "learning_rate": 0.0001, + "loss": 0.2117, + "step": 24870 + }, + { + "epoch": 6.51, + "learning_rate": 0.0001, + "loss": 0.2109, + "step": 24880 + }, + { + "epoch": 6.51, + "learning_rate": 0.0001, + "loss": 0.2127, + "step": 24890 + }, + { + "epoch": 6.51, + "learning_rate": 0.0001, + "loss": 0.2133, + "step": 24900 + }, + { + "epoch": 6.52, + "learning_rate": 0.0001, + "loss": 0.2073, + "step": 24910 + }, + { + "epoch": 6.52, + "learning_rate": 0.0001, + "loss": 0.2147, + "step": 24920 + }, + { + "epoch": 6.52, + "learning_rate": 0.0001, + "loss": 0.2122, + "step": 24930 + }, + { + "epoch": 6.53, + "learning_rate": 0.0001, + "loss": 0.2068, + "step": 24940 + }, + { + "epoch": 6.53, + "learning_rate": 0.0001, + "loss": 0.2197, + "step": 24950 + }, + { + "epoch": 6.53, + "learning_rate": 0.0001, + "loss": 0.2085, + "step": 24960 + }, + { + "epoch": 6.53, + "learning_rate": 0.0001, + "loss": 0.2113, + "step": 24970 + }, + { + "epoch": 6.54, + "learning_rate": 0.0001, + "loss": 0.2173, + "step": 24980 + }, + { + "epoch": 6.54, + "learning_rate": 0.0001, + "loss": 0.214, + "step": 24990 + }, + { + "epoch": 6.54, + "learning_rate": 0.0001, + "loss": 0.214, + "step": 25000 + }, + { + "epoch": 6.54, + "learning_rate": 0.0001, + "loss": 0.216, + "step": 25010 + }, + { + "epoch": 6.55, + "learning_rate": 0.0001, + "loss": 0.2166, + "step": 25020 + }, + { + "epoch": 6.55, + "learning_rate": 0.0001, + "loss": 0.2257, + "step": 25030 + }, + { + "epoch": 6.55, + "learning_rate": 0.0001, + "loss": 0.2148, + "step": 25040 + }, + { + "epoch": 6.55, + "learning_rate": 0.0001, + "loss": 0.2201, + "step": 25050 + }, + { + "epoch": 6.56, + "learning_rate": 0.0001, + "loss": 0.2141, + "step": 25060 + }, + { + "epoch": 6.56, + "learning_rate": 0.0001, + "loss": 0.2142, + "step": 25070 + }, + { + "epoch": 6.56, + "learning_rate": 0.0001, + "loss": 0.209, + "step": 25080 + }, + { + "epoch": 6.56, + "learning_rate": 0.0001, + "loss": 0.2132, + "step": 25090 + }, + { + "epoch": 6.57, + "learning_rate": 0.0001, + "loss": 0.2199, + "step": 25100 + }, + { + "epoch": 6.57, + "learning_rate": 0.0001, + "loss": 0.2117, + "step": 25110 + }, + { + "epoch": 6.57, + "learning_rate": 0.0001, + "loss": 0.2103, + "step": 25120 + }, + { + "epoch": 6.58, + "learning_rate": 0.0001, + "loss": 0.2088, + "step": 25130 + }, + { + "epoch": 6.58, + "learning_rate": 0.0001, + "loss": 0.2111, + "step": 25140 + }, + { + "epoch": 6.58, + "learning_rate": 0.0001, + "loss": 0.2154, + "step": 25150 + }, + { + "epoch": 6.58, + "learning_rate": 0.0001, + "loss": 0.2048, + "step": 25160 + }, + { + "epoch": 6.59, + "learning_rate": 0.0001, + "loss": 0.2161, + "step": 25170 + }, + { + "epoch": 6.59, + "learning_rate": 0.0001, + "loss": 0.2206, + "step": 25180 + }, + { + "epoch": 6.59, + "learning_rate": 0.0001, + "loss": 0.2142, + "step": 25190 + }, + { + "epoch": 6.59, + "learning_rate": 0.0001, + "loss": 0.2095, + "step": 25200 + }, + { + "epoch": 6.6, + "learning_rate": 0.0001, + "loss": 0.2161, + "step": 25210 + }, + { + "epoch": 6.6, + "learning_rate": 0.0001, + "loss": 0.2049, + "step": 25220 + }, + { + "epoch": 6.6, + "learning_rate": 0.0001, + "loss": 0.2167, + "step": 25230 + }, + { + "epoch": 6.6, + "learning_rate": 0.0001, + "loss": 0.208, + "step": 25240 + }, + { + "epoch": 6.61, + "learning_rate": 0.0001, + "loss": 0.2112, + "step": 25250 + }, + { + "epoch": 6.61, + "learning_rate": 0.0001, + "loss": 0.2133, + "step": 25260 + }, + { + "epoch": 6.61, + "learning_rate": 0.0001, + "loss": 0.213, + "step": 25270 + }, + { + "epoch": 6.61, + "learning_rate": 0.0001, + "loss": 0.2143, + "step": 25280 + }, + { + "epoch": 6.62, + "learning_rate": 0.0001, + "loss": 0.2151, + "step": 25290 + }, + { + "epoch": 6.62, + "learning_rate": 0.0001, + "loss": 0.215, + "step": 25300 + }, + { + "epoch": 6.62, + "learning_rate": 0.0001, + "loss": 0.2143, + "step": 25310 + }, + { + "epoch": 6.62, + "learning_rate": 0.0001, + "loss": 0.2035, + "step": 25320 + }, + { + "epoch": 6.63, + "learning_rate": 0.0001, + "loss": 0.2151, + "step": 25330 + }, + { + "epoch": 6.63, + "learning_rate": 0.0001, + "loss": 0.2153, + "step": 25340 + }, + { + "epoch": 6.63, + "learning_rate": 0.0001, + "loss": 0.2125, + "step": 25350 + }, + { + "epoch": 6.64, + "learning_rate": 0.0001, + "loss": 0.2197, + "step": 25360 + }, + { + "epoch": 6.64, + "learning_rate": 0.0001, + "loss": 0.2105, + "step": 25370 + }, + { + "epoch": 6.64, + "learning_rate": 0.0001, + "loss": 0.2115, + "step": 25380 + }, + { + "epoch": 6.64, + "learning_rate": 0.0001, + "loss": 0.2154, + "step": 25390 + }, + { + "epoch": 6.65, + "learning_rate": 0.0001, + "loss": 0.2194, + "step": 25400 + }, + { + "epoch": 6.65, + "learning_rate": 0.0001, + "loss": 0.2278, + "step": 25410 + }, + { + "epoch": 6.65, + "learning_rate": 0.0001, + "loss": 0.2142, + "step": 25420 + }, + { + "epoch": 6.65, + "learning_rate": 0.0001, + "loss": 0.2178, + "step": 25430 + }, + { + "epoch": 6.66, + "learning_rate": 0.0001, + "loss": 0.2106, + "step": 25440 + }, + { + "epoch": 6.66, + "learning_rate": 0.0001, + "loss": 0.2081, + "step": 25450 + }, + { + "epoch": 6.66, + "learning_rate": 0.0001, + "loss": 0.2114, + "step": 25460 + }, + { + "epoch": 6.66, + "learning_rate": 0.0001, + "loss": 0.2141, + "step": 25470 + }, + { + "epoch": 6.67, + "learning_rate": 0.0001, + "loss": 0.2126, + "step": 25480 + }, + { + "epoch": 6.67, + "learning_rate": 0.0001, + "loss": 0.22, + "step": 25490 + }, + { + "epoch": 6.67, + "learning_rate": 0.0001, + "loss": 0.2134, + "step": 25500 + }, + { + "epoch": 6.67, + "learning_rate": 0.0001, + "loss": 0.2121, + "step": 25510 + }, + { + "epoch": 6.68, + "learning_rate": 0.0001, + "loss": 0.2164, + "step": 25520 + }, + { + "epoch": 6.68, + "learning_rate": 0.0001, + "loss": 0.2108, + "step": 25530 + }, + { + "epoch": 6.68, + "learning_rate": 0.0001, + "loss": 0.217, + "step": 25540 + }, + { + "epoch": 6.68, + "learning_rate": 0.0001, + "loss": 0.2161, + "step": 25550 + }, + { + "epoch": 6.69, + "learning_rate": 0.0001, + "loss": 0.2198, + "step": 25560 + }, + { + "epoch": 6.69, + "learning_rate": 0.0001, + "loss": 0.2149, + "step": 25570 + }, + { + "epoch": 6.69, + "learning_rate": 0.0001, + "loss": 0.2052, + "step": 25580 + }, + { + "epoch": 6.7, + "learning_rate": 0.0001, + "loss": 0.2123, + "step": 25590 + }, + { + "epoch": 6.7, + "learning_rate": 0.0001, + "loss": 0.2076, + "step": 25600 + }, + { + "epoch": 6.7, + "learning_rate": 0.0001, + "loss": 0.2196, + "step": 25610 + }, + { + "epoch": 6.7, + "learning_rate": 0.0001, + "loss": 0.213, + "step": 25620 + }, + { + "epoch": 6.71, + "learning_rate": 0.0001, + "loss": 0.2167, + "step": 25630 + }, + { + "epoch": 6.71, + "learning_rate": 0.0001, + "loss": 0.2181, + "step": 25640 + }, + { + "epoch": 6.71, + "learning_rate": 0.0001, + "loss": 0.2162, + "step": 25650 + }, + { + "epoch": 6.71, + "learning_rate": 0.0001, + "loss": 0.2268, + "step": 25660 + }, + { + "epoch": 6.72, + "learning_rate": 0.0001, + "loss": 0.2158, + "step": 25670 + }, + { + "epoch": 6.72, + "learning_rate": 0.0001, + "loss": 0.2133, + "step": 25680 + }, + { + "epoch": 6.72, + "learning_rate": 0.0001, + "loss": 0.2148, + "step": 25690 + }, + { + "epoch": 6.72, + "learning_rate": 0.0001, + "loss": 0.2156, + "step": 25700 + }, + { + "epoch": 6.73, + "learning_rate": 0.0001, + "loss": 0.2101, + "step": 25710 + }, + { + "epoch": 6.73, + "learning_rate": 0.0001, + "loss": 0.2223, + "step": 25720 + }, + { + "epoch": 6.73, + "learning_rate": 0.0001, + "loss": 0.2104, + "step": 25730 + }, + { + "epoch": 6.73, + "learning_rate": 0.0001, + "loss": 0.2176, + "step": 25740 + }, + { + "epoch": 6.74, + "learning_rate": 0.0001, + "loss": 0.2225, + "step": 25750 + }, + { + "epoch": 6.74, + "learning_rate": 0.0001, + "loss": 0.2151, + "step": 25760 + }, + { + "epoch": 6.74, + "learning_rate": 0.0001, + "loss": 0.2204, + "step": 25770 + }, + { + "epoch": 6.75, + "learning_rate": 0.0001, + "loss": 0.2144, + "step": 25780 + }, + { + "epoch": 6.75, + "learning_rate": 0.0001, + "loss": 0.2151, + "step": 25790 + }, + { + "epoch": 6.75, + "learning_rate": 0.0001, + "loss": 0.2229, + "step": 25800 + }, + { + "epoch": 6.75, + "learning_rate": 0.0001, + "loss": 0.2196, + "step": 25810 + }, + { + "epoch": 6.76, + "learning_rate": 0.0001, + "loss": 0.2115, + "step": 25820 + }, + { + "epoch": 6.76, + "learning_rate": 0.0001, + "loss": 0.2208, + "step": 25830 + }, + { + "epoch": 6.76, + "learning_rate": 0.0001, + "loss": 0.2107, + "step": 25840 + }, + { + "epoch": 6.76, + "learning_rate": 0.0001, + "loss": 0.222, + "step": 25850 + }, + { + "epoch": 6.77, + "learning_rate": 0.0001, + "loss": 0.2248, + "step": 25860 + }, + { + "epoch": 6.77, + "learning_rate": 0.0001, + "loss": 0.2208, + "step": 25870 + }, + { + "epoch": 6.77, + "learning_rate": 0.0001, + "loss": 0.2134, + "step": 25880 + }, + { + "epoch": 6.77, + "learning_rate": 0.0001, + "loss": 0.2196, + "step": 25890 + }, + { + "epoch": 6.78, + "learning_rate": 0.0001, + "loss": 0.2143, + "step": 25900 + }, + { + "epoch": 6.78, + "learning_rate": 0.0001, + "loss": 0.2154, + "step": 25910 + }, + { + "epoch": 6.78, + "learning_rate": 0.0001, + "loss": 0.2208, + "step": 25920 + }, + { + "epoch": 6.78, + "learning_rate": 0.0001, + "loss": 0.2132, + "step": 25930 + }, + { + "epoch": 6.79, + "learning_rate": 0.0001, + "loss": 0.2105, + "step": 25940 + }, + { + "epoch": 6.79, + "learning_rate": 0.0001, + "loss": 0.2194, + "step": 25950 + }, + { + "epoch": 6.79, + "learning_rate": 0.0001, + "loss": 0.2121, + "step": 25960 + }, + { + "epoch": 6.79, + "learning_rate": 0.0001, + "loss": 0.2163, + "step": 25970 + }, + { + "epoch": 6.8, + "learning_rate": 0.0001, + "loss": 0.2157, + "step": 25980 + }, + { + "epoch": 6.8, + "learning_rate": 0.0001, + "loss": 0.2149, + "step": 25990 + }, + { + "epoch": 6.8, + "learning_rate": 0.0001, + "loss": 0.2088, + "step": 26000 + }, + { + "epoch": 6.81, + "learning_rate": 0.0001, + "loss": 0.2193, + "step": 26010 + }, + { + "epoch": 6.81, + "learning_rate": 0.0001, + "loss": 0.2165, + "step": 26020 + }, + { + "epoch": 6.81, + "learning_rate": 0.0001, + "loss": 0.2164, + "step": 26030 + }, + { + "epoch": 6.81, + "learning_rate": 0.0001, + "loss": 0.2121, + "step": 26040 + }, + { + "epoch": 6.82, + "learning_rate": 0.0001, + "loss": 0.2116, + "step": 26050 + }, + { + "epoch": 6.82, + "learning_rate": 0.0001, + "loss": 0.2218, + "step": 26060 + }, + { + "epoch": 6.82, + "learning_rate": 0.0001, + "loss": 0.218, + "step": 26070 + }, + { + "epoch": 6.82, + "learning_rate": 0.0001, + "loss": 0.2195, + "step": 26080 + }, + { + "epoch": 6.83, + "learning_rate": 0.0001, + "loss": 0.2095, + "step": 26090 + }, + { + "epoch": 6.83, + "learning_rate": 0.0001, + "loss": 0.2183, + "step": 26100 + }, + { + "epoch": 6.83, + "learning_rate": 0.0001, + "loss": 0.2137, + "step": 26110 + }, + { + "epoch": 6.83, + "learning_rate": 0.0001, + "loss": 0.215, + "step": 26120 + }, + { + "epoch": 6.84, + "learning_rate": 0.0001, + "loss": 0.219, + "step": 26130 + }, + { + "epoch": 6.84, + "learning_rate": 0.0001, + "loss": 0.2183, + "step": 26140 + }, + { + "epoch": 6.84, + "learning_rate": 0.0001, + "loss": 0.2283, + "step": 26150 + }, + { + "epoch": 6.84, + "learning_rate": 0.0001, + "loss": 0.2209, + "step": 26160 + }, + { + "epoch": 6.85, + "learning_rate": 0.0001, + "loss": 0.2192, + "step": 26170 + }, + { + "epoch": 6.85, + "learning_rate": 0.0001, + "loss": 0.2295, + "step": 26180 + }, + { + "epoch": 6.85, + "learning_rate": 0.0001, + "loss": 0.2198, + "step": 26190 + }, + { + "epoch": 6.86, + "learning_rate": 0.0001, + "loss": 0.2233, + "step": 26200 + }, + { + "epoch": 6.86, + "learning_rate": 0.0001, + "loss": 0.2125, + "step": 26210 + }, + { + "epoch": 6.86, + "learning_rate": 0.0001, + "loss": 0.2118, + "step": 26220 + }, + { + "epoch": 6.86, + "learning_rate": 0.0001, + "loss": 0.2135, + "step": 26230 + }, + { + "epoch": 6.87, + "learning_rate": 0.0001, + "loss": 0.2202, + "step": 26240 + }, + { + "epoch": 6.87, + "learning_rate": 0.0001, + "loss": 0.2163, + "step": 26250 + }, + { + "epoch": 6.87, + "learning_rate": 0.0001, + "loss": 0.2193, + "step": 26260 + }, + { + "epoch": 6.87, + "learning_rate": 0.0001, + "loss": 0.218, + "step": 26270 + }, + { + "epoch": 6.88, + "learning_rate": 0.0001, + "loss": 0.2184, + "step": 26280 + }, + { + "epoch": 6.88, + "learning_rate": 0.0001, + "loss": 0.2176, + "step": 26290 + }, + { + "epoch": 6.88, + "learning_rate": 0.0001, + "loss": 0.2102, + "step": 26300 + }, + { + "epoch": 6.88, + "learning_rate": 0.0001, + "loss": 0.2131, + "step": 26310 + }, + { + "epoch": 6.89, + "learning_rate": 0.0001, + "loss": 0.2149, + "step": 26320 + }, + { + "epoch": 6.89, + "learning_rate": 0.0001, + "loss": 0.222, + "step": 26330 + }, + { + "epoch": 6.89, + "learning_rate": 0.0001, + "loss": 0.2226, + "step": 26340 + }, + { + "epoch": 6.89, + "learning_rate": 0.0001, + "loss": 0.2228, + "step": 26350 + }, + { + "epoch": 6.9, + "learning_rate": 0.0001, + "loss": 0.2169, + "step": 26360 + }, + { + "epoch": 6.9, + "learning_rate": 0.0001, + "loss": 0.2114, + "step": 26370 + }, + { + "epoch": 6.9, + "learning_rate": 0.0001, + "loss": 0.2197, + "step": 26380 + }, + { + "epoch": 6.9, + "learning_rate": 0.0001, + "loss": 0.215, + "step": 26390 + }, + { + "epoch": 6.91, + "learning_rate": 0.0001, + "loss": 0.223, + "step": 26400 + }, + { + "epoch": 6.91, + "learning_rate": 0.0001, + "loss": 0.2217, + "step": 26410 + }, + { + "epoch": 6.91, + "learning_rate": 0.0001, + "loss": 0.2195, + "step": 26420 + }, + { + "epoch": 6.92, + "learning_rate": 0.0001, + "loss": 0.2207, + "step": 26430 + }, + { + "epoch": 6.92, + "learning_rate": 0.0001, + "loss": 0.22, + "step": 26440 + }, + { + "epoch": 6.92, + "learning_rate": 0.0001, + "loss": 0.2211, + "step": 26450 + }, + { + "epoch": 6.92, + "learning_rate": 0.0001, + "loss": 0.2171, + "step": 26460 + }, + { + "epoch": 6.93, + "learning_rate": 0.0001, + "loss": 0.2241, + "step": 26470 + }, + { + "epoch": 6.93, + "learning_rate": 0.0001, + "loss": 0.2069, + "step": 26480 + }, + { + "epoch": 6.93, + "learning_rate": 0.0001, + "loss": 0.2276, + "step": 26490 + }, + { + "epoch": 6.93, + "learning_rate": 0.0001, + "loss": 0.222, + "step": 26500 + }, + { + "epoch": 6.94, + "learning_rate": 0.0001, + "loss": 0.231, + "step": 26510 + }, + { + "epoch": 6.94, + "learning_rate": 0.0001, + "loss": 0.2179, + "step": 26520 + }, + { + "epoch": 6.94, + "learning_rate": 0.0001, + "loss": 0.22, + "step": 26530 + }, + { + "epoch": 6.94, + "learning_rate": 0.0001, + "loss": 0.2173, + "step": 26540 + }, + { + "epoch": 6.95, + "learning_rate": 0.0001, + "loss": 0.2312, + "step": 26550 + }, + { + "epoch": 6.95, + "learning_rate": 0.0001, + "loss": 0.2144, + "step": 26560 + }, + { + "epoch": 6.95, + "learning_rate": 0.0001, + "loss": 0.2317, + "step": 26570 + }, + { + "epoch": 6.95, + "learning_rate": 0.0001, + "loss": 0.2173, + "step": 26580 + }, + { + "epoch": 6.96, + "learning_rate": 0.0001, + "loss": 0.2181, + "step": 26590 + }, + { + "epoch": 6.96, + "learning_rate": 0.0001, + "loss": 0.2167, + "step": 26600 + }, + { + "epoch": 6.96, + "learning_rate": 0.0001, + "loss": 0.2187, + "step": 26610 + }, + { + "epoch": 6.96, + "learning_rate": 0.0001, + "loss": 0.2261, + "step": 26620 + }, + { + "epoch": 6.97, + "learning_rate": 0.0001, + "loss": 0.2182, + "step": 26630 + }, + { + "epoch": 6.97, + "learning_rate": 0.0001, + "loss": 0.2134, + "step": 26640 + }, + { + "epoch": 6.97, + "learning_rate": 0.0001, + "loss": 0.2288, + "step": 26650 + }, + { + "epoch": 6.98, + "learning_rate": 0.0001, + "loss": 0.2226, + "step": 26660 + }, + { + "epoch": 6.98, + "learning_rate": 0.0001, + "loss": 0.2197, + "step": 26670 + }, + { + "epoch": 6.98, + "learning_rate": 0.0001, + "loss": 0.2147, + "step": 26680 + }, + { + "epoch": 6.98, + "learning_rate": 0.0001, + "loss": 0.2167, + "step": 26690 + }, + { + "epoch": 6.99, + "learning_rate": 0.0001, + "loss": 0.2198, + "step": 26700 + }, + { + "epoch": 6.99, + "learning_rate": 0.0001, + "loss": 0.2178, + "step": 26710 + }, + { + "epoch": 6.99, + "learning_rate": 0.0001, + "loss": 0.2213, + "step": 26720 + }, + { + "epoch": 6.99, + "learning_rate": 0.0001, + "loss": 0.2148, + "step": 26730 + }, + { + "epoch": 7.0, + "learning_rate": 0.0001, + "loss": 0.2195, + "step": 26740 + }, + { + "epoch": 7.0, + "learning_rate": 0.0001, + "loss": 0.2101, + "step": 26750 + }, + { + "epoch": 7.0, + "eval_gen_len": 61.6894, + "eval_loss": 1.2173724174499512, + "eval_rouge1": 56.9312, + "eval_rouge2": 34.8013, + "eval_rougeL": 46.3631, + "eval_rougeLsum": 52.5898, + "eval_runtime": 39.1959, + "eval_samples_per_second": 22.426, + "eval_steps_per_second": 0.357, + "step": 26754 + }, + { + "epoch": 7.0, + "learning_rate": 0.0001, + "loss": 0.195, + "step": 26760 + }, + { + "epoch": 7.0, + "learning_rate": 0.0001, + "loss": 0.1764, + "step": 26770 + }, + { + "epoch": 7.01, + "learning_rate": 0.0001, + "loss": 0.185, + "step": 26780 + }, + { + "epoch": 7.01, + "learning_rate": 0.0001, + "loss": 0.1729, + "step": 26790 + }, + { + "epoch": 7.01, + "learning_rate": 0.0001, + "loss": 0.1741, + "step": 26800 + }, + { + "epoch": 7.01, + "learning_rate": 0.0001, + "loss": 0.1732, + "step": 26810 + }, + { + "epoch": 7.02, + "learning_rate": 0.0001, + "loss": 0.1786, + "step": 26820 + }, + { + "epoch": 7.02, + "learning_rate": 0.0001, + "loss": 0.1796, + "step": 26830 + }, + { + "epoch": 7.02, + "learning_rate": 0.0001, + "loss": 0.1729, + "step": 26840 + }, + { + "epoch": 7.03, + "learning_rate": 0.0001, + "loss": 0.1755, + "step": 26850 + }, + { + "epoch": 7.03, + "learning_rate": 0.0001, + "loss": 0.1787, + "step": 26860 + }, + { + "epoch": 7.03, + "learning_rate": 0.0001, + "loss": 0.1799, + "step": 26870 + }, + { + "epoch": 7.03, + "learning_rate": 0.0001, + "loss": 0.1851, + "step": 26880 + }, + { + "epoch": 7.04, + "learning_rate": 0.0001, + "loss": 0.182, + "step": 26890 + }, + { + "epoch": 7.04, + "learning_rate": 0.0001, + "loss": 0.1786, + "step": 26900 + }, + { + "epoch": 7.04, + "learning_rate": 0.0001, + "loss": 0.1859, + "step": 26910 + }, + { + "epoch": 7.04, + "learning_rate": 0.0001, + "loss": 0.1813, + "step": 26920 + }, + { + "epoch": 7.05, + "learning_rate": 0.0001, + "loss": 0.182, + "step": 26930 + }, + { + "epoch": 7.05, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 26940 + }, + { + "epoch": 7.05, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 26950 + }, + { + "epoch": 7.05, + "learning_rate": 0.0001, + "loss": 0.1742, + "step": 26960 + }, + { + "epoch": 7.06, + "learning_rate": 0.0001, + "loss": 0.1803, + "step": 26970 + }, + { + "epoch": 7.06, + "learning_rate": 0.0001, + "loss": 0.1787, + "step": 26980 + }, + { + "epoch": 7.06, + "learning_rate": 0.0001, + "loss": 0.1792, + "step": 26990 + }, + { + "epoch": 7.06, + "learning_rate": 0.0001, + "loss": 0.1775, + "step": 27000 + }, + { + "epoch": 7.07, + "learning_rate": 0.0001, + "loss": 0.1772, + "step": 27010 + }, + { + "epoch": 7.07, + "learning_rate": 0.0001, + "loss": 0.1823, + "step": 27020 + }, + { + "epoch": 7.07, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 27030 + }, + { + "epoch": 7.07, + "learning_rate": 0.0001, + "loss": 0.1746, + "step": 27040 + }, + { + "epoch": 7.08, + "learning_rate": 0.0001, + "loss": 0.1812, + "step": 27050 + }, + { + "epoch": 7.08, + "learning_rate": 0.0001, + "loss": 0.1843, + "step": 27060 + }, + { + "epoch": 7.08, + "learning_rate": 0.0001, + "loss": 0.1834, + "step": 27070 + }, + { + "epoch": 7.09, + "learning_rate": 0.0001, + "loss": 0.1794, + "step": 27080 + }, + { + "epoch": 7.09, + "learning_rate": 0.0001, + "loss": 0.1732, + "step": 27090 + }, + { + "epoch": 7.09, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 27100 + }, + { + "epoch": 7.09, + "learning_rate": 0.0001, + "loss": 0.1805, + "step": 27110 + }, + { + "epoch": 7.1, + "learning_rate": 0.0001, + "loss": 0.1828, + "step": 27120 + }, + { + "epoch": 7.1, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 27130 + }, + { + "epoch": 7.1, + "learning_rate": 0.0001, + "loss": 0.1752, + "step": 27140 + }, + { + "epoch": 7.1, + "learning_rate": 0.0001, + "loss": 0.1787, + "step": 27150 + }, + { + "epoch": 7.11, + "learning_rate": 0.0001, + "loss": 0.178, + "step": 27160 + }, + { + "epoch": 7.11, + "learning_rate": 0.0001, + "loss": 0.1801, + "step": 27170 + }, + { + "epoch": 7.11, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 27180 + }, + { + "epoch": 7.11, + "learning_rate": 0.0001, + "loss": 0.1751, + "step": 27190 + }, + { + "epoch": 7.12, + "learning_rate": 0.0001, + "loss": 0.1769, + "step": 27200 + }, + { + "epoch": 7.12, + "learning_rate": 0.0001, + "loss": 0.1864, + "step": 27210 + }, + { + "epoch": 7.12, + "learning_rate": 0.0001, + "loss": 0.1785, + "step": 27220 + }, + { + "epoch": 7.12, + "learning_rate": 0.0001, + "loss": 0.1708, + "step": 27230 + }, + { + "epoch": 7.13, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 27240 + }, + { + "epoch": 7.13, + "learning_rate": 0.0001, + "loss": 0.1836, + "step": 27250 + }, + { + "epoch": 7.13, + "learning_rate": 0.0001, + "loss": 0.1821, + "step": 27260 + }, + { + "epoch": 7.14, + "learning_rate": 0.0001, + "loss": 0.1858, + "step": 27270 + }, + { + "epoch": 7.14, + "learning_rate": 0.0001, + "loss": 0.1836, + "step": 27280 + }, + { + "epoch": 7.14, + "learning_rate": 0.0001, + "loss": 0.181, + "step": 27290 + }, + { + "epoch": 7.14, + "learning_rate": 0.0001, + "loss": 0.1835, + "step": 27300 + }, + { + "epoch": 7.15, + "learning_rate": 0.0001, + "loss": 0.188, + "step": 27310 + }, + { + "epoch": 7.15, + "learning_rate": 0.0001, + "loss": 0.1879, + "step": 27320 + }, + { + "epoch": 7.15, + "learning_rate": 0.0001, + "loss": 0.1787, + "step": 27330 + }, + { + "epoch": 7.15, + "learning_rate": 0.0001, + "loss": 0.1754, + "step": 27340 + }, + { + "epoch": 7.16, + "learning_rate": 0.0001, + "loss": 0.1684, + "step": 27350 + }, + { + "epoch": 7.16, + "learning_rate": 0.0001, + "loss": 0.1826, + "step": 27360 + }, + { + "epoch": 7.16, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 27370 + }, + { + "epoch": 7.16, + "learning_rate": 0.0001, + "loss": 0.1838, + "step": 27380 + }, + { + "epoch": 7.17, + "learning_rate": 0.0001, + "loss": 0.1894, + "step": 27390 + }, + { + "epoch": 7.17, + "learning_rate": 0.0001, + "loss": 0.1864, + "step": 27400 + }, + { + "epoch": 7.17, + "learning_rate": 0.0001, + "loss": 0.1836, + "step": 27410 + }, + { + "epoch": 7.17, + "learning_rate": 0.0001, + "loss": 0.1807, + "step": 27420 + }, + { + "epoch": 7.18, + "learning_rate": 0.0001, + "loss": 0.1821, + "step": 27430 + }, + { + "epoch": 7.18, + "learning_rate": 0.0001, + "loss": 0.1921, + "step": 27440 + }, + { + "epoch": 7.18, + "learning_rate": 0.0001, + "loss": 0.1837, + "step": 27450 + }, + { + "epoch": 7.18, + "learning_rate": 0.0001, + "loss": 0.1911, + "step": 27460 + }, + { + "epoch": 7.19, + "learning_rate": 0.0001, + "loss": 0.1877, + "step": 27470 + }, + { + "epoch": 7.19, + "learning_rate": 0.0001, + "loss": 0.1802, + "step": 27480 + }, + { + "epoch": 7.19, + "learning_rate": 0.0001, + "loss": 0.1847, + "step": 27490 + }, + { + "epoch": 7.2, + "learning_rate": 0.0001, + "loss": 0.1733, + "step": 27500 + }, + { + "epoch": 7.2, + "learning_rate": 0.0001, + "loss": 0.1858, + "step": 27510 + }, + { + "epoch": 7.2, + "learning_rate": 0.0001, + "loss": 0.1893, + "step": 27520 + }, + { + "epoch": 7.2, + "learning_rate": 0.0001, + "loss": 0.1838, + "step": 27530 + }, + { + "epoch": 7.21, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 27540 + }, + { + "epoch": 7.21, + "learning_rate": 0.0001, + "loss": 0.1795, + "step": 27550 + }, + { + "epoch": 7.21, + "learning_rate": 0.0001, + "loss": 0.1906, + "step": 27560 + }, + { + "epoch": 7.21, + "learning_rate": 0.0001, + "loss": 0.1862, + "step": 27570 + }, + { + "epoch": 7.22, + "learning_rate": 0.0001, + "loss": 0.1818, + "step": 27580 + }, + { + "epoch": 7.22, + "learning_rate": 0.0001, + "loss": 0.1919, + "step": 27590 + }, + { + "epoch": 7.22, + "learning_rate": 0.0001, + "loss": 0.1821, + "step": 27600 + }, + { + "epoch": 7.22, + "learning_rate": 0.0001, + "loss": 0.1837, + "step": 27610 + }, + { + "epoch": 7.23, + "learning_rate": 0.0001, + "loss": 0.178, + "step": 27620 + }, + { + "epoch": 7.23, + "learning_rate": 0.0001, + "loss": 0.1784, + "step": 27630 + }, + { + "epoch": 7.23, + "learning_rate": 0.0001, + "loss": 0.18, + "step": 27640 + }, + { + "epoch": 7.23, + "learning_rate": 0.0001, + "loss": 0.1826, + "step": 27650 + }, + { + "epoch": 7.24, + "learning_rate": 0.0001, + "loss": 0.1797, + "step": 27660 + }, + { + "epoch": 7.24, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 27670 + }, + { + "epoch": 7.24, + "learning_rate": 0.0001, + "loss": 0.1841, + "step": 27680 + }, + { + "epoch": 7.24, + "learning_rate": 0.0001, + "loss": 0.1758, + "step": 27690 + }, + { + "epoch": 7.25, + "learning_rate": 0.0001, + "loss": 0.1872, + "step": 27700 + }, + { + "epoch": 7.25, + "learning_rate": 0.0001, + "loss": 0.186, + "step": 27710 + }, + { + "epoch": 7.25, + "learning_rate": 0.0001, + "loss": 0.1833, + "step": 27720 + }, + { + "epoch": 7.26, + "learning_rate": 0.0001, + "loss": 0.1875, + "step": 27730 + }, + { + "epoch": 7.26, + "learning_rate": 0.0001, + "loss": 0.1891, + "step": 27740 + }, + { + "epoch": 7.26, + "learning_rate": 0.0001, + "loss": 0.1866, + "step": 27750 + }, + { + "epoch": 7.26, + "learning_rate": 0.0001, + "loss": 0.192, + "step": 27760 + }, + { + "epoch": 7.27, + "learning_rate": 0.0001, + "loss": 0.182, + "step": 27770 + }, + { + "epoch": 7.27, + "learning_rate": 0.0001, + "loss": 0.1881, + "step": 27780 + }, + { + "epoch": 7.27, + "learning_rate": 0.0001, + "loss": 0.1825, + "step": 27790 + }, + { + "epoch": 7.27, + "learning_rate": 0.0001, + "loss": 0.1898, + "step": 27800 + }, + { + "epoch": 7.28, + "learning_rate": 0.0001, + "loss": 0.1876, + "step": 27810 + }, + { + "epoch": 7.28, + "learning_rate": 0.0001, + "loss": 0.1751, + "step": 27820 + }, + { + "epoch": 7.28, + "learning_rate": 0.0001, + "loss": 0.1919, + "step": 27830 + }, + { + "epoch": 7.28, + "learning_rate": 0.0001, + "loss": 0.1919, + "step": 27840 + }, + { + "epoch": 7.29, + "learning_rate": 0.0001, + "loss": 0.1832, + "step": 27850 + }, + { + "epoch": 7.29, + "learning_rate": 0.0001, + "loss": 0.1818, + "step": 27860 + }, + { + "epoch": 7.29, + "learning_rate": 0.0001, + "loss": 0.1868, + "step": 27870 + }, + { + "epoch": 7.29, + "learning_rate": 0.0001, + "loss": 0.1751, + "step": 27880 + }, + { + "epoch": 7.3, + "learning_rate": 0.0001, + "loss": 0.1839, + "step": 27890 + }, + { + "epoch": 7.3, + "learning_rate": 0.0001, + "loss": 0.1886, + "step": 27900 + }, + { + "epoch": 7.3, + "learning_rate": 0.0001, + "loss": 0.1787, + "step": 27910 + }, + { + "epoch": 7.31, + "learning_rate": 0.0001, + "loss": 0.1807, + "step": 27920 + }, + { + "epoch": 7.31, + "learning_rate": 0.0001, + "loss": 0.1843, + "step": 27930 + }, + { + "epoch": 7.31, + "learning_rate": 0.0001, + "loss": 0.1861, + "step": 27940 + }, + { + "epoch": 7.31, + "learning_rate": 0.0001, + "loss": 0.1815, + "step": 27950 + }, + { + "epoch": 7.32, + "learning_rate": 0.0001, + "loss": 0.1859, + "step": 27960 + }, + { + "epoch": 7.32, + "learning_rate": 0.0001, + "loss": 0.1826, + "step": 27970 + }, + { + "epoch": 7.32, + "learning_rate": 0.0001, + "loss": 0.1751, + "step": 27980 + }, + { + "epoch": 7.32, + "learning_rate": 0.0001, + "loss": 0.1889, + "step": 27990 + }, + { + "epoch": 7.33, + "learning_rate": 0.0001, + "loss": 0.1867, + "step": 28000 + }, + { + "epoch": 7.33, + "learning_rate": 0.0001, + "loss": 0.1925, + "step": 28010 + }, + { + "epoch": 7.33, + "learning_rate": 0.0001, + "loss": 0.189, + "step": 28020 + }, + { + "epoch": 7.33, + "learning_rate": 0.0001, + "loss": 0.1911, + "step": 28030 + }, + { + "epoch": 7.34, + "learning_rate": 0.0001, + "loss": 0.1901, + "step": 28040 + }, + { + "epoch": 7.34, + "learning_rate": 0.0001, + "loss": 0.1946, + "step": 28050 + }, + { + "epoch": 7.34, + "learning_rate": 0.0001, + "loss": 0.1923, + "step": 28060 + }, + { + "epoch": 7.34, + "learning_rate": 0.0001, + "loss": 0.1846, + "step": 28070 + }, + { + "epoch": 7.35, + "learning_rate": 0.0001, + "loss": 0.1809, + "step": 28080 + }, + { + "epoch": 7.35, + "learning_rate": 0.0001, + "loss": 0.186, + "step": 28090 + }, + { + "epoch": 7.35, + "learning_rate": 0.0001, + "loss": 0.1858, + "step": 28100 + }, + { + "epoch": 7.35, + "learning_rate": 0.0001, + "loss": 0.1908, + "step": 28110 + }, + { + "epoch": 7.36, + "learning_rate": 0.0001, + "loss": 0.185, + "step": 28120 + }, + { + "epoch": 7.36, + "learning_rate": 0.0001, + "loss": 0.1859, + "step": 28130 + }, + { + "epoch": 7.36, + "learning_rate": 0.0001, + "loss": 0.185, + "step": 28140 + }, + { + "epoch": 7.37, + "learning_rate": 0.0001, + "loss": 0.1853, + "step": 28150 + }, + { + "epoch": 7.37, + "learning_rate": 0.0001, + "loss": 0.1868, + "step": 28160 + }, + { + "epoch": 7.37, + "learning_rate": 0.0001, + "loss": 0.1892, + "step": 28170 + }, + { + "epoch": 7.37, + "learning_rate": 0.0001, + "loss": 0.1877, + "step": 28180 + }, + { + "epoch": 7.38, + "learning_rate": 0.0001, + "loss": 0.1824, + "step": 28190 + }, + { + "epoch": 7.38, + "learning_rate": 0.0001, + "loss": 0.1782, + "step": 28200 + }, + { + "epoch": 7.38, + "learning_rate": 0.0001, + "loss": 0.1813, + "step": 28210 + }, + { + "epoch": 7.38, + "learning_rate": 0.0001, + "loss": 0.191, + "step": 28220 + }, + { + "epoch": 7.39, + "learning_rate": 0.0001, + "loss": 0.199, + "step": 28230 + }, + { + "epoch": 7.39, + "learning_rate": 0.0001, + "loss": 0.1878, + "step": 28240 + }, + { + "epoch": 7.39, + "learning_rate": 0.0001, + "loss": 0.1934, + "step": 28250 + }, + { + "epoch": 7.39, + "learning_rate": 0.0001, + "loss": 0.1832, + "step": 28260 + }, + { + "epoch": 7.4, + "learning_rate": 0.0001, + "loss": 0.19, + "step": 28270 + }, + { + "epoch": 7.4, + "learning_rate": 0.0001, + "loss": 0.1859, + "step": 28280 + }, + { + "epoch": 7.4, + "learning_rate": 0.0001, + "loss": 0.1884, + "step": 28290 + }, + { + "epoch": 7.4, + "learning_rate": 0.0001, + "loss": 0.192, + "step": 28300 + }, + { + "epoch": 7.41, + "learning_rate": 0.0001, + "loss": 0.1817, + "step": 28310 + }, + { + "epoch": 7.41, + "learning_rate": 0.0001, + "loss": 0.1892, + "step": 28320 + }, + { + "epoch": 7.41, + "learning_rate": 0.0001, + "loss": 0.1872, + "step": 28330 + }, + { + "epoch": 7.41, + "learning_rate": 0.0001, + "loss": 0.1983, + "step": 28340 + }, + { + "epoch": 7.42, + "learning_rate": 0.0001, + "loss": 0.1876, + "step": 28350 + }, + { + "epoch": 7.42, + "learning_rate": 0.0001, + "loss": 0.1889, + "step": 28360 + }, + { + "epoch": 7.42, + "learning_rate": 0.0001, + "loss": 0.1943, + "step": 28370 + }, + { + "epoch": 7.43, + "learning_rate": 0.0001, + "loss": 0.1851, + "step": 28380 + }, + { + "epoch": 7.43, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 28390 + }, + { + "epoch": 7.43, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 28400 + }, + { + "epoch": 7.43, + "learning_rate": 0.0001, + "loss": 0.1885, + "step": 28410 + }, + { + "epoch": 7.44, + "learning_rate": 0.0001, + "loss": 0.1857, + "step": 28420 + }, + { + "epoch": 7.44, + "learning_rate": 0.0001, + "loss": 0.1916, + "step": 28430 + }, + { + "epoch": 7.44, + "learning_rate": 0.0001, + "loss": 0.1871, + "step": 28440 + }, + { + "epoch": 7.44, + "learning_rate": 0.0001, + "loss": 0.1961, + "step": 28450 + }, + { + "epoch": 7.45, + "learning_rate": 0.0001, + "loss": 0.1847, + "step": 28460 + }, + { + "epoch": 7.45, + "learning_rate": 0.0001, + "loss": 0.1941, + "step": 28470 + }, + { + "epoch": 7.45, + "learning_rate": 0.0001, + "loss": 0.1874, + "step": 28480 + }, + { + "epoch": 7.45, + "learning_rate": 0.0001, + "loss": 0.19, + "step": 28490 + }, + { + "epoch": 7.46, + "learning_rate": 0.0001, + "loss": 0.1969, + "step": 28500 + }, + { + "epoch": 7.46, + "learning_rate": 0.0001, + "loss": 0.1916, + "step": 28510 + }, + { + "epoch": 7.46, + "learning_rate": 0.0001, + "loss": 0.193, + "step": 28520 + }, + { + "epoch": 7.46, + "learning_rate": 0.0001, + "loss": 0.1924, + "step": 28530 + }, + { + "epoch": 7.47, + "learning_rate": 0.0001, + "loss": 0.1929, + "step": 28540 + }, + { + "epoch": 7.47, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 28550 + }, + { + "epoch": 7.47, + "learning_rate": 0.0001, + "loss": 0.1868, + "step": 28560 + }, + { + "epoch": 7.48, + "learning_rate": 0.0001, + "loss": 0.1899, + "step": 28570 + }, + { + "epoch": 7.48, + "learning_rate": 0.0001, + "loss": 0.181, + "step": 28580 + }, + { + "epoch": 7.48, + "learning_rate": 0.0001, + "loss": 0.1926, + "step": 28590 + }, + { + "epoch": 7.48, + "learning_rate": 0.0001, + "loss": 0.1856, + "step": 28600 + }, + { + "epoch": 7.49, + "learning_rate": 0.0001, + "loss": 0.1868, + "step": 28610 + }, + { + "epoch": 7.49, + "learning_rate": 0.0001, + "loss": 0.1775, + "step": 28620 + }, + { + "epoch": 7.49, + "learning_rate": 0.0001, + "loss": 0.1891, + "step": 28630 + }, + { + "epoch": 7.49, + "learning_rate": 0.0001, + "loss": 0.2002, + "step": 28640 + }, + { + "epoch": 7.5, + "learning_rate": 0.0001, + "loss": 0.1901, + "step": 28650 + }, + { + "epoch": 7.5, + "learning_rate": 0.0001, + "loss": 0.1889, + "step": 28660 + }, + { + "epoch": 7.5, + "learning_rate": 0.0001, + "loss": 0.1896, + "step": 28670 + }, + { + "epoch": 7.5, + "learning_rate": 0.0001, + "loss": 0.1866, + "step": 28680 + }, + { + "epoch": 7.51, + "learning_rate": 0.0001, + "loss": 0.1876, + "step": 28690 + }, + { + "epoch": 7.51, + "learning_rate": 0.0001, + "loss": 0.1916, + "step": 28700 + }, + { + "epoch": 7.51, + "learning_rate": 0.0001, + "loss": 0.1883, + "step": 28710 + }, + { + "epoch": 7.51, + "learning_rate": 0.0001, + "loss": 0.1979, + "step": 28720 + }, + { + "epoch": 7.52, + "learning_rate": 0.0001, + "loss": 0.1961, + "step": 28730 + }, + { + "epoch": 7.52, + "learning_rate": 0.0001, + "loss": 0.1922, + "step": 28740 + }, + { + "epoch": 7.52, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 28750 + }, + { + "epoch": 7.52, + "learning_rate": 0.0001, + "loss": 0.1965, + "step": 28760 + }, + { + "epoch": 7.53, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 28770 + }, + { + "epoch": 7.53, + "learning_rate": 0.0001, + "loss": 0.19, + "step": 28780 + }, + { + "epoch": 7.53, + "learning_rate": 0.0001, + "loss": 0.1878, + "step": 28790 + }, + { + "epoch": 7.54, + "learning_rate": 0.0001, + "loss": 0.1854, + "step": 28800 + }, + { + "epoch": 7.54, + "learning_rate": 0.0001, + "loss": 0.1921, + "step": 28810 + }, + { + "epoch": 7.54, + "learning_rate": 0.0001, + "loss": 0.1919, + "step": 28820 + }, + { + "epoch": 7.54, + "learning_rate": 0.0001, + "loss": 0.1892, + "step": 28830 + }, + { + "epoch": 7.55, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 28840 + }, + { + "epoch": 7.55, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 28850 + }, + { + "epoch": 7.55, + "learning_rate": 0.0001, + "loss": 0.1863, + "step": 28860 + }, + { + "epoch": 7.55, + "learning_rate": 0.0001, + "loss": 0.1842, + "step": 28870 + }, + { + "epoch": 7.56, + "learning_rate": 0.0001, + "loss": 0.1897, + "step": 28880 + }, + { + "epoch": 7.56, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 28890 + }, + { + "epoch": 7.56, + "learning_rate": 0.0001, + "loss": 0.1926, + "step": 28900 + }, + { + "epoch": 7.56, + "learning_rate": 0.0001, + "loss": 0.1896, + "step": 28910 + }, + { + "epoch": 7.57, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 28920 + }, + { + "epoch": 7.57, + "learning_rate": 0.0001, + "loss": 0.1871, + "step": 28930 + }, + { + "epoch": 7.57, + "learning_rate": 0.0001, + "loss": 0.1873, + "step": 28940 + }, + { + "epoch": 7.57, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 28950 + }, + { + "epoch": 7.58, + "learning_rate": 0.0001, + "loss": 0.1868, + "step": 28960 + }, + { + "epoch": 7.58, + "learning_rate": 0.0001, + "loss": 0.1865, + "step": 28970 + }, + { + "epoch": 7.58, + "learning_rate": 0.0001, + "loss": 0.1858, + "step": 28980 + }, + { + "epoch": 7.59, + "learning_rate": 0.0001, + "loss": 0.1966, + "step": 28990 + }, + { + "epoch": 7.59, + "learning_rate": 0.0001, + "loss": 0.1919, + "step": 29000 + }, + { + "epoch": 7.59, + "learning_rate": 0.0001, + "loss": 0.1837, + "step": 29010 + }, + { + "epoch": 7.59, + "learning_rate": 0.0001, + "loss": 0.1898, + "step": 29020 + }, + { + "epoch": 7.6, + "learning_rate": 0.0001, + "loss": 0.188, + "step": 29030 + }, + { + "epoch": 7.6, + "learning_rate": 0.0001, + "loss": 0.1801, + "step": 29040 + }, + { + "epoch": 7.6, + "learning_rate": 0.0001, + "loss": 0.1876, + "step": 29050 + }, + { + "epoch": 7.6, + "learning_rate": 0.0001, + "loss": 0.1922, + "step": 29060 + }, + { + "epoch": 7.61, + "learning_rate": 0.0001, + "loss": 0.191, + "step": 29070 + }, + { + "epoch": 7.61, + "learning_rate": 0.0001, + "loss": 0.1927, + "step": 29080 + }, + { + "epoch": 7.61, + "learning_rate": 0.0001, + "loss": 0.1896, + "step": 29090 + }, + { + "epoch": 7.61, + "learning_rate": 0.0001, + "loss": 0.1915, + "step": 29100 + }, + { + "epoch": 7.62, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 29110 + }, + { + "epoch": 7.62, + "learning_rate": 0.0001, + "loss": 0.1966, + "step": 29120 + }, + { + "epoch": 7.62, + "learning_rate": 0.0001, + "loss": 0.1856, + "step": 29130 + }, + { + "epoch": 7.62, + "learning_rate": 0.0001, + "loss": 0.1803, + "step": 29140 + }, + { + "epoch": 7.63, + "learning_rate": 0.0001, + "loss": 0.1965, + "step": 29150 + }, + { + "epoch": 7.63, + "learning_rate": 0.0001, + "loss": 0.1936, + "step": 29160 + }, + { + "epoch": 7.63, + "learning_rate": 0.0001, + "loss": 0.1967, + "step": 29170 + }, + { + "epoch": 7.63, + "learning_rate": 0.0001, + "loss": 0.1904, + "step": 29180 + }, + { + "epoch": 7.64, + "learning_rate": 0.0001, + "loss": 0.1901, + "step": 29190 + }, + { + "epoch": 7.64, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 29200 + }, + { + "epoch": 7.64, + "learning_rate": 0.0001, + "loss": 0.1867, + "step": 29210 + }, + { + "epoch": 7.65, + "learning_rate": 0.0001, + "loss": 0.193, + "step": 29220 + }, + { + "epoch": 7.65, + "learning_rate": 0.0001, + "loss": 0.1957, + "step": 29230 + }, + { + "epoch": 7.65, + "learning_rate": 0.0001, + "loss": 0.1947, + "step": 29240 + }, + { + "epoch": 7.65, + "learning_rate": 0.0001, + "loss": 0.1928, + "step": 29250 + }, + { + "epoch": 7.66, + "learning_rate": 0.0001, + "loss": 0.1913, + "step": 29260 + }, + { + "epoch": 7.66, + "learning_rate": 0.0001, + "loss": 0.1954, + "step": 29270 + }, + { + "epoch": 7.66, + "learning_rate": 0.0001, + "loss": 0.1937, + "step": 29280 + }, + { + "epoch": 7.66, + "learning_rate": 0.0001, + "loss": 0.197, + "step": 29290 + }, + { + "epoch": 7.67, + "learning_rate": 0.0001, + "loss": 0.1913, + "step": 29300 + }, + { + "epoch": 7.67, + "learning_rate": 0.0001, + "loss": 0.1922, + "step": 29310 + }, + { + "epoch": 7.67, + "learning_rate": 0.0001, + "loss": 0.1938, + "step": 29320 + }, + { + "epoch": 7.67, + "learning_rate": 0.0001, + "loss": 0.1897, + "step": 29330 + }, + { + "epoch": 7.68, + "learning_rate": 0.0001, + "loss": 0.1892, + "step": 29340 + }, + { + "epoch": 7.68, + "learning_rate": 0.0001, + "loss": 0.1851, + "step": 29350 + }, + { + "epoch": 7.68, + "learning_rate": 0.0001, + "loss": 0.1862, + "step": 29360 + }, + { + "epoch": 7.68, + "learning_rate": 0.0001, + "loss": 0.1998, + "step": 29370 + }, + { + "epoch": 7.69, + "learning_rate": 0.0001, + "loss": 0.1954, + "step": 29380 + }, + { + "epoch": 7.69, + "learning_rate": 0.0001, + "loss": 0.1852, + "step": 29390 + }, + { + "epoch": 7.69, + "learning_rate": 0.0001, + "loss": 0.2049, + "step": 29400 + }, + { + "epoch": 7.69, + "learning_rate": 0.0001, + "loss": 0.201, + "step": 29410 + }, + { + "epoch": 7.7, + "learning_rate": 0.0001, + "loss": 0.1894, + "step": 29420 + }, + { + "epoch": 7.7, + "learning_rate": 0.0001, + "loss": 0.193, + "step": 29430 + }, + { + "epoch": 7.7, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 29440 + }, + { + "epoch": 7.71, + "learning_rate": 0.0001, + "loss": 0.1955, + "step": 29450 + }, + { + "epoch": 7.71, + "learning_rate": 0.0001, + "loss": 0.1924, + "step": 29460 + }, + { + "epoch": 7.71, + "learning_rate": 0.0001, + "loss": 0.1907, + "step": 29470 + }, + { + "epoch": 7.71, + "learning_rate": 0.0001, + "loss": 0.1893, + "step": 29480 + }, + { + "epoch": 7.72, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 29490 + }, + { + "epoch": 7.72, + "learning_rate": 0.0001, + "loss": 0.19, + "step": 29500 + }, + { + "epoch": 7.72, + "learning_rate": 0.0001, + "loss": 0.1888, + "step": 29510 + }, + { + "epoch": 7.72, + "learning_rate": 0.0001, + "loss": 0.1916, + "step": 29520 + }, + { + "epoch": 7.73, + "learning_rate": 0.0001, + "loss": 0.1888, + "step": 29530 + }, + { + "epoch": 7.73, + "learning_rate": 0.0001, + "loss": 0.1986, + "step": 29540 + }, + { + "epoch": 7.73, + "learning_rate": 0.0001, + "loss": 0.2062, + "step": 29550 + }, + { + "epoch": 7.73, + "learning_rate": 0.0001, + "loss": 0.2038, + "step": 29560 + }, + { + "epoch": 7.74, + "learning_rate": 0.0001, + "loss": 0.1958, + "step": 29570 + }, + { + "epoch": 7.74, + "learning_rate": 0.0001, + "loss": 0.1924, + "step": 29580 + }, + { + "epoch": 7.74, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 29590 + }, + { + "epoch": 7.74, + "learning_rate": 0.0001, + "loss": 0.1954, + "step": 29600 + }, + { + "epoch": 7.75, + "learning_rate": 0.0001, + "loss": 0.1914, + "step": 29610 + }, + { + "epoch": 7.75, + "learning_rate": 0.0001, + "loss": 0.1986, + "step": 29620 + }, + { + "epoch": 7.75, + "learning_rate": 0.0001, + "loss": 0.1957, + "step": 29630 + }, + { + "epoch": 7.76, + "learning_rate": 0.0001, + "loss": 0.1841, + "step": 29640 + }, + { + "epoch": 7.76, + "learning_rate": 0.0001, + "loss": 0.1878, + "step": 29650 + }, + { + "epoch": 7.76, + "learning_rate": 0.0001, + "loss": 0.1953, + "step": 29660 + }, + { + "epoch": 7.76, + "learning_rate": 0.0001, + "loss": 0.1964, + "step": 29670 + }, + { + "epoch": 7.77, + "learning_rate": 0.0001, + "loss": 0.1891, + "step": 29680 + }, + { + "epoch": 7.77, + "learning_rate": 0.0001, + "loss": 0.1976, + "step": 29690 + }, + { + "epoch": 7.77, + "learning_rate": 0.0001, + "loss": 0.1931, + "step": 29700 + }, + { + "epoch": 7.77, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 29710 + }, + { + "epoch": 7.78, + "learning_rate": 0.0001, + "loss": 0.1959, + "step": 29720 + }, + { + "epoch": 7.78, + "learning_rate": 0.0001, + "loss": 0.202, + "step": 29730 + }, + { + "epoch": 7.78, + "learning_rate": 0.0001, + "loss": 0.1946, + "step": 29740 + }, + { + "epoch": 7.78, + "learning_rate": 0.0001, + "loss": 0.1985, + "step": 29750 + }, + { + "epoch": 7.79, + "learning_rate": 0.0001, + "loss": 0.1992, + "step": 29760 + }, + { + "epoch": 7.79, + "learning_rate": 0.0001, + "loss": 0.1861, + "step": 29770 + }, + { + "epoch": 7.79, + "learning_rate": 0.0001, + "loss": 0.1824, + "step": 29780 + }, + { + "epoch": 7.79, + "learning_rate": 0.0001, + "loss": 0.1887, + "step": 29790 + }, + { + "epoch": 7.8, + "learning_rate": 0.0001, + "loss": 0.1979, + "step": 29800 + }, + { + "epoch": 7.8, + "learning_rate": 0.0001, + "loss": 0.1948, + "step": 29810 + }, + { + "epoch": 7.8, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 29820 + }, + { + "epoch": 7.8, + "learning_rate": 0.0001, + "loss": 0.1912, + "step": 29830 + }, + { + "epoch": 7.81, + "learning_rate": 0.0001, + "loss": 0.2025, + "step": 29840 + }, + { + "epoch": 7.81, + "learning_rate": 0.0001, + "loss": 0.188, + "step": 29850 + }, + { + "epoch": 7.81, + "learning_rate": 0.0001, + "loss": 0.2032, + "step": 29860 + }, + { + "epoch": 7.82, + "learning_rate": 0.0001, + "loss": 0.1881, + "step": 29870 + }, + { + "epoch": 7.82, + "learning_rate": 0.0001, + "loss": 0.1987, + "step": 29880 + }, + { + "epoch": 7.82, + "learning_rate": 0.0001, + "loss": 0.1995, + "step": 29890 + }, + { + "epoch": 7.82, + "learning_rate": 0.0001, + "loss": 0.1941, + "step": 29900 + }, + { + "epoch": 7.83, + "learning_rate": 0.0001, + "loss": 0.197, + "step": 29910 + }, + { + "epoch": 7.83, + "learning_rate": 0.0001, + "loss": 0.1977, + "step": 29920 + }, + { + "epoch": 7.83, + "learning_rate": 0.0001, + "loss": 0.2029, + "step": 29930 + }, + { + "epoch": 7.83, + "learning_rate": 0.0001, + "loss": 0.1988, + "step": 29940 + }, + { + "epoch": 7.84, + "learning_rate": 0.0001, + "loss": 0.2014, + "step": 29950 + }, + { + "epoch": 7.84, + "learning_rate": 0.0001, + "loss": 0.1978, + "step": 29960 + }, + { + "epoch": 7.84, + "learning_rate": 0.0001, + "loss": 0.1974, + "step": 29970 + }, + { + "epoch": 7.84, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 29980 + }, + { + "epoch": 7.85, + "learning_rate": 0.0001, + "loss": 0.1996, + "step": 29990 + }, + { + "epoch": 7.85, + "learning_rate": 0.0001, + "loss": 0.2008, + "step": 30000 + }, + { + "epoch": 7.85, + "learning_rate": 0.0001, + "loss": 0.1903, + "step": 30010 + }, + { + "epoch": 7.85, + "learning_rate": 0.0001, + "loss": 0.1922, + "step": 30020 + }, + { + "epoch": 7.86, + "learning_rate": 0.0001, + "loss": 0.2, + "step": 30030 + }, + { + "epoch": 7.86, + "learning_rate": 0.0001, + "loss": 0.1935, + "step": 30040 + }, + { + "epoch": 7.86, + "learning_rate": 0.0001, + "loss": 0.1978, + "step": 30050 + }, + { + "epoch": 7.86, + "learning_rate": 0.0001, + "loss": 0.2002, + "step": 30060 + }, + { + "epoch": 7.87, + "learning_rate": 0.0001, + "loss": 0.1906, + "step": 30070 + }, + { + "epoch": 7.87, + "learning_rate": 0.0001, + "loss": 0.191, + "step": 30080 + }, + { + "epoch": 7.87, + "learning_rate": 0.0001, + "loss": 0.2056, + "step": 30090 + }, + { + "epoch": 7.88, + "learning_rate": 0.0001, + "loss": 0.2005, + "step": 30100 + }, + { + "epoch": 7.88, + "learning_rate": 0.0001, + "loss": 0.1893, + "step": 30110 + }, + { + "epoch": 7.88, + "learning_rate": 0.0001, + "loss": 0.1877, + "step": 30120 + }, + { + "epoch": 7.88, + "learning_rate": 0.0001, + "loss": 0.1977, + "step": 30130 + }, + { + "epoch": 7.89, + "learning_rate": 0.0001, + "loss": 0.2019, + "step": 30140 + }, + { + "epoch": 7.89, + "learning_rate": 0.0001, + "loss": 0.1891, + "step": 30150 + }, + { + "epoch": 7.89, + "learning_rate": 0.0001, + "loss": 0.198, + "step": 30160 + }, + { + "epoch": 7.89, + "learning_rate": 0.0001, + "loss": 0.1896, + "step": 30170 + }, + { + "epoch": 7.9, + "learning_rate": 0.0001, + "loss": 0.1966, + "step": 30180 + }, + { + "epoch": 7.9, + "learning_rate": 0.0001, + "loss": 0.1883, + "step": 30190 + }, + { + "epoch": 7.9, + "learning_rate": 0.0001, + "loss": 0.1947, + "step": 30200 + }, + { + "epoch": 7.9, + "learning_rate": 0.0001, + "loss": 0.2013, + "step": 30210 + }, + { + "epoch": 7.91, + "learning_rate": 0.0001, + "loss": 0.1931, + "step": 30220 + }, + { + "epoch": 7.91, + "learning_rate": 0.0001, + "loss": 0.1862, + "step": 30230 + }, + { + "epoch": 7.91, + "learning_rate": 0.0001, + "loss": 0.1927, + "step": 30240 + }, + { + "epoch": 7.91, + "learning_rate": 0.0001, + "loss": 0.1893, + "step": 30250 + }, + { + "epoch": 7.92, + "learning_rate": 0.0001, + "loss": 0.2012, + "step": 30260 + }, + { + "epoch": 7.92, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 30270 + }, + { + "epoch": 7.92, + "learning_rate": 0.0001, + "loss": 0.193, + "step": 30280 + }, + { + "epoch": 7.93, + "learning_rate": 0.0001, + "loss": 0.1943, + "step": 30290 + }, + { + "epoch": 7.93, + "learning_rate": 0.0001, + "loss": 0.1936, + "step": 30300 + }, + { + "epoch": 7.93, + "learning_rate": 0.0001, + "loss": 0.1896, + "step": 30310 + }, + { + "epoch": 7.93, + "learning_rate": 0.0001, + "loss": 0.193, + "step": 30320 + }, + { + "epoch": 7.94, + "learning_rate": 0.0001, + "loss": 0.1894, + "step": 30330 + }, + { + "epoch": 7.94, + "learning_rate": 0.0001, + "loss": 0.1974, + "step": 30340 + }, + { + "epoch": 7.94, + "learning_rate": 0.0001, + "loss": 0.1874, + "step": 30350 + }, + { + "epoch": 7.94, + "learning_rate": 0.0001, + "loss": 0.1875, + "step": 30360 + }, + { + "epoch": 7.95, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 30370 + }, + { + "epoch": 7.95, + "learning_rate": 0.0001, + "loss": 0.2011, + "step": 30380 + }, + { + "epoch": 7.95, + "learning_rate": 0.0001, + "loss": 0.1889, + "step": 30390 + }, + { + "epoch": 7.95, + "learning_rate": 0.0001, + "loss": 0.1966, + "step": 30400 + }, + { + "epoch": 7.96, + "learning_rate": 0.0001, + "loss": 0.1979, + "step": 30410 + }, + { + "epoch": 7.96, + "learning_rate": 0.0001, + "loss": 0.195, + "step": 30420 + }, + { + "epoch": 7.96, + "learning_rate": 0.0001, + "loss": 0.1961, + "step": 30430 + }, + { + "epoch": 7.96, + "learning_rate": 0.0001, + "loss": 0.1974, + "step": 30440 + }, + { + "epoch": 7.97, + "learning_rate": 0.0001, + "loss": 0.1951, + "step": 30450 + }, + { + "epoch": 7.97, + "learning_rate": 0.0001, + "loss": 0.1895, + "step": 30460 + }, + { + "epoch": 7.97, + "learning_rate": 0.0001, + "loss": 0.1892, + "step": 30470 + }, + { + "epoch": 7.97, + "learning_rate": 0.0001, + "loss": 0.1867, + "step": 30480 + }, + { + "epoch": 7.98, + "learning_rate": 0.0001, + "loss": 0.1964, + "step": 30490 + }, + { + "epoch": 7.98, + "learning_rate": 0.0001, + "loss": 0.1982, + "step": 30500 + }, + { + "epoch": 7.98, + "learning_rate": 0.0001, + "loss": 0.2012, + "step": 30510 + }, + { + "epoch": 7.99, + "learning_rate": 0.0001, + "loss": 0.1939, + "step": 30520 + }, + { + "epoch": 7.99, + "learning_rate": 0.0001, + "loss": 0.1972, + "step": 30530 + }, + { + "epoch": 7.99, + "learning_rate": 0.0001, + "loss": 0.1933, + "step": 30540 + }, + { + "epoch": 7.99, + "learning_rate": 0.0001, + "loss": 0.1906, + "step": 30550 + }, + { + "epoch": 8.0, + "learning_rate": 0.0001, + "loss": 0.1957, + "step": 30560 + }, + { + "epoch": 8.0, + "learning_rate": 0.0001, + "loss": 0.1964, + "step": 30570 + }, + { + "epoch": 8.0, + "eval_gen_len": 61.6985, + "eval_loss": 1.2625230550765991, + "eval_rouge1": 57.0616, + "eval_rouge2": 34.9092, + "eval_rougeL": 46.5325, + "eval_rougeLsum": 52.6338, + "eval_runtime": 36.4577, + "eval_samples_per_second": 24.11, + "eval_steps_per_second": 0.384, + "step": 30576 + }, + { + "epoch": 8.0, + "learning_rate": 0.0001, + "loss": 0.1764, + "step": 30580 + }, + { + "epoch": 8.0, + "learning_rate": 0.0001, + "loss": 0.1616, + "step": 30590 + }, + { + "epoch": 8.01, + "learning_rate": 0.0001, + "loss": 0.1544, + "step": 30600 + }, + { + "epoch": 8.01, + "learning_rate": 0.0001, + "loss": 0.1621, + "step": 30610 + }, + { + "epoch": 8.01, + "learning_rate": 0.0001, + "loss": 0.1618, + "step": 30620 + }, + { + "epoch": 8.01, + "learning_rate": 0.0001, + "loss": 0.1603, + "step": 30630 + }, + { + "epoch": 8.02, + "learning_rate": 0.0001, + "loss": 0.1565, + "step": 30640 + }, + { + "epoch": 8.02, + "learning_rate": 0.0001, + "loss": 0.1591, + "step": 30650 + }, + { + "epoch": 8.02, + "learning_rate": 0.0001, + "loss": 0.1627, + "step": 30660 + }, + { + "epoch": 8.02, + "learning_rate": 0.0001, + "loss": 0.1611, + "step": 30670 + }, + { + "epoch": 8.03, + "learning_rate": 0.0001, + "loss": 0.1559, + "step": 30680 + }, + { + "epoch": 8.03, + "learning_rate": 0.0001, + "loss": 0.1565, + "step": 30690 + }, + { + "epoch": 8.03, + "learning_rate": 0.0001, + "loss": 0.1503, + "step": 30700 + }, + { + "epoch": 8.04, + "learning_rate": 0.0001, + "loss": 0.1566, + "step": 30710 + }, + { + "epoch": 8.04, + "learning_rate": 0.0001, + "loss": 0.1559, + "step": 30720 + }, + { + "epoch": 8.04, + "learning_rate": 0.0001, + "loss": 0.1632, + "step": 30730 + }, + { + "epoch": 8.04, + "learning_rate": 0.0001, + "loss": 0.1628, + "step": 30740 + }, + { + "epoch": 8.05, + "learning_rate": 0.0001, + "loss": 0.1581, + "step": 30750 + }, + { + "epoch": 8.05, + "learning_rate": 0.0001, + "loss": 0.1607, + "step": 30760 + }, + { + "epoch": 8.05, + "learning_rate": 0.0001, + "loss": 0.1615, + "step": 30770 + }, + { + "epoch": 8.05, + "learning_rate": 0.0001, + "loss": 0.1606, + "step": 30780 + }, + { + "epoch": 8.06, + "learning_rate": 0.0001, + "loss": 0.1519, + "step": 30790 + }, + { + "epoch": 8.06, + "learning_rate": 0.0001, + "loss": 0.1546, + "step": 30800 + }, + { + "epoch": 8.06, + "learning_rate": 0.0001, + "loss": 0.1536, + "step": 30810 + }, + { + "epoch": 8.06, + "learning_rate": 0.0001, + "loss": 0.1504, + "step": 30820 + }, + { + "epoch": 8.07, + "learning_rate": 0.0001, + "loss": 0.1631, + "step": 30830 + }, + { + "epoch": 8.07, + "learning_rate": 0.0001, + "loss": 0.1638, + "step": 30840 + }, + { + "epoch": 8.07, + "learning_rate": 0.0001, + "loss": 0.1621, + "step": 30850 + }, + { + "epoch": 8.07, + "learning_rate": 0.0001, + "loss": 0.1584, + "step": 30860 + }, + { + "epoch": 8.08, + "learning_rate": 0.0001, + "loss": 0.1587, + "step": 30870 + }, + { + "epoch": 8.08, + "learning_rate": 0.0001, + "loss": 0.154, + "step": 30880 + }, + { + "epoch": 8.08, + "learning_rate": 0.0001, + "loss": 0.1597, + "step": 30890 + }, + { + "epoch": 8.08, + "learning_rate": 0.0001, + "loss": 0.1621, + "step": 30900 + }, + { + "epoch": 8.09, + "learning_rate": 0.0001, + "loss": 0.1526, + "step": 30910 + }, + { + "epoch": 8.09, + "learning_rate": 0.0001, + "loss": 0.1605, + "step": 30920 + }, + { + "epoch": 8.09, + "learning_rate": 0.0001, + "loss": 0.167, + "step": 30930 + }, + { + "epoch": 8.1, + "learning_rate": 0.0001, + "loss": 0.1667, + "step": 30940 + }, + { + "epoch": 8.1, + "learning_rate": 0.0001, + "loss": 0.1531, + "step": 30950 + }, + { + "epoch": 8.1, + "learning_rate": 0.0001, + "loss": 0.1524, + "step": 30960 + }, + { + "epoch": 8.1, + "learning_rate": 0.0001, + "loss": 0.1607, + "step": 30970 + }, + { + "epoch": 8.11, + "learning_rate": 0.0001, + "loss": 0.1652, + "step": 30980 + }, + { + "epoch": 8.11, + "learning_rate": 0.0001, + "loss": 0.1588, + "step": 30990 + }, + { + "epoch": 8.11, + "learning_rate": 0.0001, + "loss": 0.1608, + "step": 31000 + }, + { + "epoch": 8.11, + "learning_rate": 0.0001, + "loss": 0.1667, + "step": 31010 + }, + { + "epoch": 8.12, + "learning_rate": 0.0001, + "loss": 0.1614, + "step": 31020 + }, + { + "epoch": 8.12, + "learning_rate": 0.0001, + "loss": 0.1583, + "step": 31030 + }, + { + "epoch": 8.12, + "learning_rate": 0.0001, + "loss": 0.1571, + "step": 31040 + }, + { + "epoch": 8.12, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 31050 + }, + { + "epoch": 8.13, + "learning_rate": 0.0001, + "loss": 0.1612, + "step": 31060 + }, + { + "epoch": 8.13, + "learning_rate": 0.0001, + "loss": 0.162, + "step": 31070 + }, + { + "epoch": 8.13, + "learning_rate": 0.0001, + "loss": 0.163, + "step": 31080 + }, + { + "epoch": 8.13, + "learning_rate": 0.0001, + "loss": 0.1647, + "step": 31090 + }, + { + "epoch": 8.14, + "learning_rate": 0.0001, + "loss": 0.1596, + "step": 31100 + }, + { + "epoch": 8.14, + "learning_rate": 0.0001, + "loss": 0.1694, + "step": 31110 + }, + { + "epoch": 8.14, + "learning_rate": 0.0001, + "loss": 0.163, + "step": 31120 + }, + { + "epoch": 8.14, + "learning_rate": 0.0001, + "loss": 0.1716, + "step": 31130 + }, + { + "epoch": 8.15, + "learning_rate": 0.0001, + "loss": 0.1677, + "step": 31140 + }, + { + "epoch": 8.15, + "learning_rate": 0.0001, + "loss": 0.1617, + "step": 31150 + }, + { + "epoch": 8.15, + "learning_rate": 0.0001, + "loss": 0.1607, + "step": 31160 + }, + { + "epoch": 8.16, + "learning_rate": 0.0001, + "loss": 0.1574, + "step": 31170 + }, + { + "epoch": 8.16, + "learning_rate": 0.0001, + "loss": 0.1595, + "step": 31180 + }, + { + "epoch": 8.16, + "learning_rate": 0.0001, + "loss": 0.167, + "step": 31190 + }, + { + "epoch": 8.16, + "learning_rate": 0.0001, + "loss": 0.1648, + "step": 31200 + }, + { + "epoch": 8.17, + "learning_rate": 0.0001, + "loss": 0.1651, + "step": 31210 + }, + { + "epoch": 8.17, + "learning_rate": 0.0001, + "loss": 0.1599, + "step": 31220 + }, + { + "epoch": 8.17, + "learning_rate": 0.0001, + "loss": 0.1585, + "step": 31230 + }, + { + "epoch": 8.17, + "learning_rate": 0.0001, + "loss": 0.164, + "step": 31240 + }, + { + "epoch": 8.18, + "learning_rate": 0.0001, + "loss": 0.1564, + "step": 31250 + }, + { + "epoch": 8.18, + "learning_rate": 0.0001, + "loss": 0.1573, + "step": 31260 + }, + { + "epoch": 8.18, + "learning_rate": 0.0001, + "loss": 0.1605, + "step": 31270 + }, + { + "epoch": 8.18, + "learning_rate": 0.0001, + "loss": 0.1596, + "step": 31280 + }, + { + "epoch": 8.19, + "learning_rate": 0.0001, + "loss": 0.1613, + "step": 31290 + }, + { + "epoch": 8.19, + "learning_rate": 0.0001, + "loss": 0.166, + "step": 31300 + }, + { + "epoch": 8.19, + "learning_rate": 0.0001, + "loss": 0.1634, + "step": 31310 + }, + { + "epoch": 8.19, + "learning_rate": 0.0001, + "loss": 0.1579, + "step": 31320 + }, + { + "epoch": 8.2, + "learning_rate": 0.0001, + "loss": 0.1694, + "step": 31330 + }, + { + "epoch": 8.2, + "learning_rate": 0.0001, + "loss": 0.1645, + "step": 31340 + }, + { + "epoch": 8.2, + "learning_rate": 0.0001, + "loss": 0.1625, + "step": 31350 + }, + { + "epoch": 8.21, + "learning_rate": 0.0001, + "loss": 0.163, + "step": 31360 + }, + { + "epoch": 8.21, + "learning_rate": 0.0001, + "loss": 0.1626, + "step": 31370 + }, + { + "epoch": 8.21, + "learning_rate": 0.0001, + "loss": 0.1626, + "step": 31380 + }, + { + "epoch": 8.21, + "learning_rate": 0.0001, + "loss": 0.1627, + "step": 31390 + }, + { + "epoch": 8.22, + "learning_rate": 0.0001, + "loss": 0.1581, + "step": 31400 + }, + { + "epoch": 8.22, + "learning_rate": 0.0001, + "loss": 0.1667, + "step": 31410 + }, + { + "epoch": 8.22, + "learning_rate": 0.0001, + "loss": 0.1666, + "step": 31420 + }, + { + "epoch": 8.22, + "learning_rate": 0.0001, + "loss": 0.1652, + "step": 31430 + }, + { + "epoch": 8.23, + "learning_rate": 0.0001, + "loss": 0.1613, + "step": 31440 + }, + { + "epoch": 8.23, + "learning_rate": 0.0001, + "loss": 0.1637, + "step": 31450 + }, + { + "epoch": 8.23, + "learning_rate": 0.0001, + "loss": 0.1624, + "step": 31460 + }, + { + "epoch": 8.23, + "learning_rate": 0.0001, + "loss": 0.1576, + "step": 31470 + }, + { + "epoch": 8.24, + "learning_rate": 0.0001, + "loss": 0.1628, + "step": 31480 + }, + { + "epoch": 8.24, + "learning_rate": 0.0001, + "loss": 0.1674, + "step": 31490 + }, + { + "epoch": 8.24, + "learning_rate": 0.0001, + "loss": 0.1605, + "step": 31500 + }, + { + "epoch": 8.24, + "learning_rate": 0.0001, + "loss": 0.1558, + "step": 31510 + }, + { + "epoch": 8.25, + "learning_rate": 0.0001, + "loss": 0.1569, + "step": 31520 + }, + { + "epoch": 8.25, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 31530 + }, + { + "epoch": 8.25, + "learning_rate": 0.0001, + "loss": 0.1741, + "step": 31540 + }, + { + "epoch": 8.25, + "learning_rate": 0.0001, + "loss": 0.1701, + "step": 31550 + }, + { + "epoch": 8.26, + "learning_rate": 0.0001, + "loss": 0.1611, + "step": 31560 + }, + { + "epoch": 8.26, + "learning_rate": 0.0001, + "loss": 0.1621, + "step": 31570 + }, + { + "epoch": 8.26, + "learning_rate": 0.0001, + "loss": 0.1637, + "step": 31580 + }, + { + "epoch": 8.27, + "learning_rate": 0.0001, + "loss": 0.1734, + "step": 31590 + }, + { + "epoch": 8.27, + "learning_rate": 0.0001, + "loss": 0.1621, + "step": 31600 + }, + { + "epoch": 8.27, + "learning_rate": 0.0001, + "loss": 0.1634, + "step": 31610 + }, + { + "epoch": 8.27, + "learning_rate": 0.0001, + "loss": 0.1682, + "step": 31620 + }, + { + "epoch": 8.28, + "learning_rate": 0.0001, + "loss": 0.162, + "step": 31630 + }, + { + "epoch": 8.28, + "learning_rate": 0.0001, + "loss": 0.1714, + "step": 31640 + }, + { + "epoch": 8.28, + "learning_rate": 0.0001, + "loss": 0.1663, + "step": 31650 + }, + { + "epoch": 8.28, + "learning_rate": 0.0001, + "loss": 0.1615, + "step": 31660 + }, + { + "epoch": 8.29, + "learning_rate": 0.0001, + "loss": 0.1602, + "step": 31670 + }, + { + "epoch": 8.29, + "learning_rate": 0.0001, + "loss": 0.1661, + "step": 31680 + }, + { + "epoch": 8.29, + "learning_rate": 0.0001, + "loss": 0.1647, + "step": 31690 + }, + { + "epoch": 8.29, + "learning_rate": 0.0001, + "loss": 0.1641, + "step": 31700 + }, + { + "epoch": 8.3, + "learning_rate": 0.0001, + "loss": 0.1653, + "step": 31710 + }, + { + "epoch": 8.3, + "learning_rate": 0.0001, + "loss": 0.1622, + "step": 31720 + }, + { + "epoch": 8.3, + "learning_rate": 0.0001, + "loss": 0.1703, + "step": 31730 + }, + { + "epoch": 8.3, + "learning_rate": 0.0001, + "loss": 0.1689, + "step": 31740 + }, + { + "epoch": 8.31, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 31750 + }, + { + "epoch": 8.31, + "learning_rate": 0.0001, + "loss": 0.1677, + "step": 31760 + }, + { + "epoch": 8.31, + "learning_rate": 0.0001, + "loss": 0.1614, + "step": 31770 + }, + { + "epoch": 8.32, + "learning_rate": 0.0001, + "loss": 0.1711, + "step": 31780 + }, + { + "epoch": 8.32, + "learning_rate": 0.0001, + "loss": 0.1674, + "step": 31790 + }, + { + "epoch": 8.32, + "learning_rate": 0.0001, + "loss": 0.1658, + "step": 31800 + }, + { + "epoch": 8.32, + "learning_rate": 0.0001, + "loss": 0.1673, + "step": 31810 + }, + { + "epoch": 8.33, + "learning_rate": 0.0001, + "loss": 0.1698, + "step": 31820 + }, + { + "epoch": 8.33, + "learning_rate": 0.0001, + "loss": 0.1687, + "step": 31830 + }, + { + "epoch": 8.33, + "learning_rate": 0.0001, + "loss": 0.1692, + "step": 31840 + }, + { + "epoch": 8.33, + "learning_rate": 0.0001, + "loss": 0.164, + "step": 31850 + }, + { + "epoch": 8.34, + "learning_rate": 0.0001, + "loss": 0.1688, + "step": 31860 + }, + { + "epoch": 8.34, + "learning_rate": 0.0001, + "loss": 0.1654, + "step": 31870 + }, + { + "epoch": 8.34, + "learning_rate": 0.0001, + "loss": 0.1609, + "step": 31880 + }, + { + "epoch": 8.34, + "learning_rate": 0.0001, + "loss": 0.1682, + "step": 31890 + }, + { + "epoch": 8.35, + "learning_rate": 0.0001, + "loss": 0.1696, + "step": 31900 + }, + { + "epoch": 8.35, + "learning_rate": 0.0001, + "loss": 0.1666, + "step": 31910 + }, + { + "epoch": 8.35, + "learning_rate": 0.0001, + "loss": 0.167, + "step": 31920 + }, + { + "epoch": 8.35, + "learning_rate": 0.0001, + "loss": 0.1682, + "step": 31930 + }, + { + "epoch": 8.36, + "learning_rate": 0.0001, + "loss": 0.1739, + "step": 31940 + }, + { + "epoch": 8.36, + "learning_rate": 0.0001, + "loss": 0.1644, + "step": 31950 + }, + { + "epoch": 8.36, + "learning_rate": 0.0001, + "loss": 0.1712, + "step": 31960 + }, + { + "epoch": 8.36, + "learning_rate": 0.0001, + "loss": 0.166, + "step": 31970 + }, + { + "epoch": 8.37, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 31980 + }, + { + "epoch": 8.37, + "learning_rate": 0.0001, + "loss": 0.1687, + "step": 31990 + }, + { + "epoch": 8.37, + "learning_rate": 0.0001, + "loss": 0.1655, + "step": 32000 + }, + { + "epoch": 8.38, + "learning_rate": 0.0001, + "loss": 0.1631, + "step": 32010 + }, + { + "epoch": 8.38, + "learning_rate": 0.0001, + "loss": 0.1664, + "step": 32020 + }, + { + "epoch": 8.38, + "learning_rate": 0.0001, + "loss": 0.1672, + "step": 32030 + }, + { + "epoch": 8.38, + "learning_rate": 0.0001, + "loss": 0.1685, + "step": 32040 + }, + { + "epoch": 8.39, + "learning_rate": 0.0001, + "loss": 0.1678, + "step": 32050 + }, + { + "epoch": 8.39, + "learning_rate": 0.0001, + "loss": 0.1657, + "step": 32060 + }, + { + "epoch": 8.39, + "learning_rate": 0.0001, + "loss": 0.1656, + "step": 32070 + }, + { + "epoch": 8.39, + "learning_rate": 0.0001, + "loss": 0.1681, + "step": 32080 + }, + { + "epoch": 8.4, + "learning_rate": 0.0001, + "loss": 0.1645, + "step": 32090 + }, + { + "epoch": 8.4, + "learning_rate": 0.0001, + "loss": 0.1708, + "step": 32100 + }, + { + "epoch": 8.4, + "learning_rate": 0.0001, + "loss": 0.1601, + "step": 32110 + }, + { + "epoch": 8.4, + "learning_rate": 0.0001, + "loss": 0.1643, + "step": 32120 + }, + { + "epoch": 8.41, + "learning_rate": 0.0001, + "loss": 0.1628, + "step": 32130 + }, + { + "epoch": 8.41, + "learning_rate": 0.0001, + "loss": 0.1768, + "step": 32140 + }, + { + "epoch": 8.41, + "learning_rate": 0.0001, + "loss": 0.1742, + "step": 32150 + }, + { + "epoch": 8.41, + "learning_rate": 0.0001, + "loss": 0.1645, + "step": 32160 + }, + { + "epoch": 8.42, + "learning_rate": 0.0001, + "loss": 0.1635, + "step": 32170 + }, + { + "epoch": 8.42, + "learning_rate": 0.0001, + "loss": 0.1749, + "step": 32180 + }, + { + "epoch": 8.42, + "learning_rate": 0.0001, + "loss": 0.1652, + "step": 32190 + }, + { + "epoch": 8.42, + "learning_rate": 0.0001, + "loss": 0.1682, + "step": 32200 + }, + { + "epoch": 8.43, + "learning_rate": 0.0001, + "loss": 0.1749, + "step": 32210 + }, + { + "epoch": 8.43, + "learning_rate": 0.0001, + "loss": 0.1701, + "step": 32220 + }, + { + "epoch": 8.43, + "learning_rate": 0.0001, + "loss": 0.1669, + "step": 32230 + }, + { + "epoch": 8.44, + "learning_rate": 0.0001, + "loss": 0.1696, + "step": 32240 + }, + { + "epoch": 8.44, + "learning_rate": 0.0001, + "loss": 0.1599, + "step": 32250 + }, + { + "epoch": 8.44, + "learning_rate": 0.0001, + "loss": 0.1743, + "step": 32260 + }, + { + "epoch": 8.44, + "learning_rate": 0.0001, + "loss": 0.1659, + "step": 32270 + }, + { + "epoch": 8.45, + "learning_rate": 0.0001, + "loss": 0.1624, + "step": 32280 + }, + { + "epoch": 8.45, + "learning_rate": 0.0001, + "loss": 0.1665, + "step": 32290 + }, + { + "epoch": 8.45, + "learning_rate": 0.0001, + "loss": 0.1567, + "step": 32300 + }, + { + "epoch": 8.45, + "learning_rate": 0.0001, + "loss": 0.1728, + "step": 32310 + }, + { + "epoch": 8.46, + "learning_rate": 0.0001, + "loss": 0.1669, + "step": 32320 + }, + { + "epoch": 8.46, + "learning_rate": 0.0001, + "loss": 0.1677, + "step": 32330 + }, + { + "epoch": 8.46, + "learning_rate": 0.0001, + "loss": 0.1673, + "step": 32340 + }, + { + "epoch": 8.46, + "learning_rate": 0.0001, + "loss": 0.164, + "step": 32350 + }, + { + "epoch": 8.47, + "learning_rate": 0.0001, + "loss": 0.1689, + "step": 32360 + }, + { + "epoch": 8.47, + "learning_rate": 0.0001, + "loss": 0.1664, + "step": 32370 + }, + { + "epoch": 8.47, + "learning_rate": 0.0001, + "loss": 0.172, + "step": 32380 + }, + { + "epoch": 8.47, + "learning_rate": 0.0001, + "loss": 0.1737, + "step": 32390 + }, + { + "epoch": 8.48, + "learning_rate": 0.0001, + "loss": 0.1624, + "step": 32400 + }, + { + "epoch": 8.48, + "learning_rate": 0.0001, + "loss": 0.1606, + "step": 32410 + }, + { + "epoch": 8.48, + "learning_rate": 0.0001, + "loss": 0.1658, + "step": 32420 + }, + { + "epoch": 8.49, + "learning_rate": 0.0001, + "loss": 0.1704, + "step": 32430 + }, + { + "epoch": 8.49, + "learning_rate": 0.0001, + "loss": 0.1749, + "step": 32440 + }, + { + "epoch": 8.49, + "learning_rate": 0.0001, + "loss": 0.1695, + "step": 32450 + }, + { + "epoch": 8.49, + "learning_rate": 0.0001, + "loss": 0.1686, + "step": 32460 + }, + { + "epoch": 8.5, + "learning_rate": 0.0001, + "loss": 0.1649, + "step": 32470 + }, + { + "epoch": 8.5, + "learning_rate": 0.0001, + "loss": 0.1608, + "step": 32480 + }, + { + "epoch": 8.5, + "learning_rate": 0.0001, + "loss": 0.1688, + "step": 32490 + }, + { + "epoch": 8.5, + "learning_rate": 0.0001, + "loss": 0.1801, + "step": 32500 + }, + { + "epoch": 8.51, + "learning_rate": 0.0001, + "loss": 0.171, + "step": 32510 + }, + { + "epoch": 8.51, + "learning_rate": 0.0001, + "loss": 0.1712, + "step": 32520 + }, + { + "epoch": 8.51, + "learning_rate": 0.0001, + "loss": 0.17, + "step": 32530 + }, + { + "epoch": 8.51, + "learning_rate": 0.0001, + "loss": 0.1748, + "step": 32540 + }, + { + "epoch": 8.52, + "learning_rate": 0.0001, + "loss": 0.1714, + "step": 32550 + }, + { + "epoch": 8.52, + "learning_rate": 0.0001, + "loss": 0.1792, + "step": 32560 + }, + { + "epoch": 8.52, + "learning_rate": 0.0001, + "loss": 0.1783, + "step": 32570 + }, + { + "epoch": 8.52, + "learning_rate": 0.0001, + "loss": 0.1706, + "step": 32580 + }, + { + "epoch": 8.53, + "learning_rate": 0.0001, + "loss": 0.1693, + "step": 32590 + }, + { + "epoch": 8.53, + "learning_rate": 0.0001, + "loss": 0.1719, + "step": 32600 + }, + { + "epoch": 8.53, + "learning_rate": 0.0001, + "loss": 0.1741, + "step": 32610 + }, + { + "epoch": 8.53, + "learning_rate": 0.0001, + "loss": 0.1711, + "step": 32620 + }, + { + "epoch": 8.54, + "learning_rate": 0.0001, + "loss": 0.169, + "step": 32630 + }, + { + "epoch": 8.54, + "learning_rate": 0.0001, + "loss": 0.1674, + "step": 32640 + }, + { + "epoch": 8.54, + "learning_rate": 0.0001, + "loss": 0.1642, + "step": 32650 + }, + { + "epoch": 8.55, + "learning_rate": 0.0001, + "loss": 0.1724, + "step": 32660 + }, + { + "epoch": 8.55, + "learning_rate": 0.0001, + "loss": 0.1668, + "step": 32670 + }, + { + "epoch": 8.55, + "learning_rate": 0.0001, + "loss": 0.1635, + "step": 32680 + }, + { + "epoch": 8.55, + "learning_rate": 0.0001, + "loss": 0.1676, + "step": 32690 + }, + { + "epoch": 8.56, + "learning_rate": 0.0001, + "loss": 0.1718, + "step": 32700 + }, + { + "epoch": 8.56, + "learning_rate": 0.0001, + "loss": 0.1642, + "step": 32710 + }, + { + "epoch": 8.56, + "learning_rate": 0.0001, + "loss": 0.1766, + "step": 32720 + }, + { + "epoch": 8.56, + "learning_rate": 0.0001, + "loss": 0.179, + "step": 32730 + }, + { + "epoch": 8.57, + "learning_rate": 0.0001, + "loss": 0.1698, + "step": 32740 + }, + { + "epoch": 8.57, + "learning_rate": 0.0001, + "loss": 0.1671, + "step": 32750 + }, + { + "epoch": 8.57, + "learning_rate": 0.0001, + "loss": 0.1699, + "step": 32760 + }, + { + "epoch": 8.57, + "learning_rate": 0.0001, + "loss": 0.1766, + "step": 32770 + }, + { + "epoch": 8.58, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 32780 + }, + { + "epoch": 8.58, + "learning_rate": 0.0001, + "loss": 0.1691, + "step": 32790 + }, + { + "epoch": 8.58, + "learning_rate": 0.0001, + "loss": 0.1725, + "step": 32800 + }, + { + "epoch": 8.58, + "learning_rate": 0.0001, + "loss": 0.1759, + "step": 32810 + }, + { + "epoch": 8.59, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 32820 + }, + { + "epoch": 8.59, + "learning_rate": 0.0001, + "loss": 0.1629, + "step": 32830 + }, + { + "epoch": 8.59, + "learning_rate": 0.0001, + "loss": 0.1666, + "step": 32840 + }, + { + "epoch": 8.59, + "learning_rate": 0.0001, + "loss": 0.1656, + "step": 32850 + }, + { + "epoch": 8.6, + "learning_rate": 0.0001, + "loss": 0.1696, + "step": 32860 + }, + { + "epoch": 8.6, + "learning_rate": 0.0001, + "loss": 0.169, + "step": 32870 + }, + { + "epoch": 8.6, + "learning_rate": 0.0001, + "loss": 0.169, + "step": 32880 + }, + { + "epoch": 8.61, + "learning_rate": 0.0001, + "loss": 0.1694, + "step": 32890 + }, + { + "epoch": 8.61, + "learning_rate": 0.0001, + "loss": 0.1707, + "step": 32900 + }, + { + "epoch": 8.61, + "learning_rate": 0.0001, + "loss": 0.1781, + "step": 32910 + }, + { + "epoch": 8.61, + "learning_rate": 0.0001, + "loss": 0.1785, + "step": 32920 + }, + { + "epoch": 8.62, + "learning_rate": 0.0001, + "loss": 0.17, + "step": 32930 + }, + { + "epoch": 8.62, + "learning_rate": 0.0001, + "loss": 0.1654, + "step": 32940 + }, + { + "epoch": 8.62, + "learning_rate": 0.0001, + "loss": 0.1732, + "step": 32950 + }, + { + "epoch": 8.62, + "learning_rate": 0.0001, + "loss": 0.1644, + "step": 32960 + }, + { + "epoch": 8.63, + "learning_rate": 0.0001, + "loss": 0.1743, + "step": 32970 + }, + { + "epoch": 8.63, + "learning_rate": 0.0001, + "loss": 0.1757, + "step": 32980 + }, + { + "epoch": 8.63, + "learning_rate": 0.0001, + "loss": 0.1684, + "step": 32990 + }, + { + "epoch": 8.63, + "learning_rate": 0.0001, + "loss": 0.1705, + "step": 33000 + }, + { + "epoch": 8.64, + "learning_rate": 0.0001, + "loss": 0.1718, + "step": 33010 + }, + { + "epoch": 8.64, + "learning_rate": 0.0001, + "loss": 0.1681, + "step": 33020 + }, + { + "epoch": 8.64, + "learning_rate": 0.0001, + "loss": 0.1724, + "step": 33030 + }, + { + "epoch": 8.64, + "learning_rate": 0.0001, + "loss": 0.1748, + "step": 33040 + }, + { + "epoch": 8.65, + "learning_rate": 0.0001, + "loss": 0.1671, + "step": 33050 + }, + { + "epoch": 8.65, + "learning_rate": 0.0001, + "loss": 0.1753, + "step": 33060 + }, + { + "epoch": 8.65, + "learning_rate": 0.0001, + "loss": 0.176, + "step": 33070 + }, + { + "epoch": 8.66, + "learning_rate": 0.0001, + "loss": 0.1766, + "step": 33080 + }, + { + "epoch": 8.66, + "learning_rate": 0.0001, + "loss": 0.1754, + "step": 33090 + }, + { + "epoch": 8.66, + "learning_rate": 0.0001, + "loss": 0.1719, + "step": 33100 + }, + { + "epoch": 8.66, + "learning_rate": 0.0001, + "loss": 0.1783, + "step": 33110 + }, + { + "epoch": 8.67, + "learning_rate": 0.0001, + "loss": 0.1637, + "step": 33120 + }, + { + "epoch": 8.67, + "learning_rate": 0.0001, + "loss": 0.1698, + "step": 33130 + }, + { + "epoch": 8.67, + "learning_rate": 0.0001, + "loss": 0.1804, + "step": 33140 + }, + { + "epoch": 8.67, + "learning_rate": 0.0001, + "loss": 0.1775, + "step": 33150 + }, + { + "epoch": 8.68, + "learning_rate": 0.0001, + "loss": 0.1733, + "step": 33160 + }, + { + "epoch": 8.68, + "learning_rate": 0.0001, + "loss": 0.164, + "step": 33170 + }, + { + "epoch": 8.68, + "learning_rate": 0.0001, + "loss": 0.1717, + "step": 33180 + }, + { + "epoch": 8.68, + "learning_rate": 0.0001, + "loss": 0.1663, + "step": 33190 + }, + { + "epoch": 8.69, + "learning_rate": 0.0001, + "loss": 0.1762, + "step": 33200 + }, + { + "epoch": 8.69, + "learning_rate": 0.0001, + "loss": 0.1753, + "step": 33210 + }, + { + "epoch": 8.69, + "learning_rate": 0.0001, + "loss": 0.1704, + "step": 33220 + }, + { + "epoch": 8.69, + "learning_rate": 0.0001, + "loss": 0.171, + "step": 33230 + }, + { + "epoch": 8.7, + "learning_rate": 0.0001, + "loss": 0.1797, + "step": 33240 + }, + { + "epoch": 8.7, + "learning_rate": 0.0001, + "loss": 0.1752, + "step": 33250 + }, + { + "epoch": 8.7, + "learning_rate": 0.0001, + "loss": 0.1666, + "step": 33260 + }, + { + "epoch": 8.7, + "learning_rate": 0.0001, + "loss": 0.1703, + "step": 33270 + }, + { + "epoch": 8.71, + "learning_rate": 0.0001, + "loss": 0.1683, + "step": 33280 + }, + { + "epoch": 8.71, + "learning_rate": 0.0001, + "loss": 0.1793, + "step": 33290 + }, + { + "epoch": 8.71, + "learning_rate": 0.0001, + "loss": 0.1757, + "step": 33300 + }, + { + "epoch": 8.72, + "learning_rate": 0.0001, + "loss": 0.1752, + "step": 33310 + }, + { + "epoch": 8.72, + "learning_rate": 0.0001, + "loss": 0.1821, + "step": 33320 + }, + { + "epoch": 8.72, + "learning_rate": 0.0001, + "loss": 0.1769, + "step": 33330 + }, + { + "epoch": 8.72, + "learning_rate": 0.0001, + "loss": 0.1669, + "step": 33340 + }, + { + "epoch": 8.73, + "learning_rate": 0.0001, + "loss": 0.1625, + "step": 33350 + }, + { + "epoch": 8.73, + "learning_rate": 0.0001, + "loss": 0.1807, + "step": 33360 + }, + { + "epoch": 8.73, + "learning_rate": 0.0001, + "loss": 0.1747, + "step": 33370 + }, + { + "epoch": 8.73, + "learning_rate": 0.0001, + "loss": 0.1761, + "step": 33380 + }, + { + "epoch": 8.74, + "learning_rate": 0.0001, + "loss": 0.176, + "step": 33390 + }, + { + "epoch": 8.74, + "learning_rate": 0.0001, + "loss": 0.176, + "step": 33400 + }, + { + "epoch": 8.74, + "learning_rate": 0.0001, + "loss": 0.1685, + "step": 33410 + }, + { + "epoch": 8.74, + "learning_rate": 0.0001, + "loss": 0.1756, + "step": 33420 + }, + { + "epoch": 8.75, + "learning_rate": 0.0001, + "loss": 0.1744, + "step": 33430 + }, + { + "epoch": 8.75, + "learning_rate": 0.0001, + "loss": 0.1764, + "step": 33440 + }, + { + "epoch": 8.75, + "learning_rate": 0.0001, + "loss": 0.1843, + "step": 33450 + }, + { + "epoch": 8.75, + "learning_rate": 0.0001, + "loss": 0.1784, + "step": 33460 + }, + { + "epoch": 8.76, + "learning_rate": 0.0001, + "loss": 0.1698, + "step": 33470 + }, + { + "epoch": 8.76, + "learning_rate": 0.0001, + "loss": 0.1727, + "step": 33480 + }, + { + "epoch": 8.76, + "learning_rate": 0.0001, + "loss": 0.1769, + "step": 33490 + }, + { + "epoch": 8.77, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 33500 + }, + { + "epoch": 8.77, + "learning_rate": 0.0001, + "loss": 0.18, + "step": 33510 + }, + { + "epoch": 8.77, + "learning_rate": 0.0001, + "loss": 0.1681, + "step": 33520 + }, + { + "epoch": 8.77, + "learning_rate": 0.0001, + "loss": 0.1775, + "step": 33530 + }, + { + "epoch": 8.78, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 33540 + }, + { + "epoch": 8.78, + "learning_rate": 0.0001, + "loss": 0.1799, + "step": 33550 + }, + { + "epoch": 8.78, + "learning_rate": 0.0001, + "loss": 0.1654, + "step": 33560 + }, + { + "epoch": 8.78, + "learning_rate": 0.0001, + "loss": 0.1695, + "step": 33570 + }, + { + "epoch": 8.79, + "learning_rate": 0.0001, + "loss": 0.173, + "step": 33580 + }, + { + "epoch": 8.79, + "learning_rate": 0.0001, + "loss": 0.1777, + "step": 33590 + }, + { + "epoch": 8.79, + "learning_rate": 0.0001, + "loss": 0.1724, + "step": 33600 + }, + { + "epoch": 8.79, + "learning_rate": 0.0001, + "loss": 0.1764, + "step": 33610 + }, + { + "epoch": 8.8, + "learning_rate": 0.0001, + "loss": 0.1707, + "step": 33620 + }, + { + "epoch": 8.8, + "learning_rate": 0.0001, + "loss": 0.1728, + "step": 33630 + }, + { + "epoch": 8.8, + "learning_rate": 0.0001, + "loss": 0.174, + "step": 33640 + }, + { + "epoch": 8.8, + "learning_rate": 0.0001, + "loss": 0.1732, + "step": 33650 + }, + { + "epoch": 8.81, + "learning_rate": 0.0001, + "loss": 0.1779, + "step": 33660 + }, + { + "epoch": 8.81, + "learning_rate": 0.0001, + "loss": 0.1718, + "step": 33670 + }, + { + "epoch": 8.81, + "learning_rate": 0.0001, + "loss": 0.1739, + "step": 33680 + }, + { + "epoch": 8.81, + "learning_rate": 0.0001, + "loss": 0.1719, + "step": 33690 + }, + { + "epoch": 8.82, + "learning_rate": 0.0001, + "loss": 0.174, + "step": 33700 + }, + { + "epoch": 8.82, + "learning_rate": 0.0001, + "loss": 0.1687, + "step": 33710 + }, + { + "epoch": 8.82, + "learning_rate": 0.0001, + "loss": 0.1757, + "step": 33720 + }, + { + "epoch": 8.83, + "learning_rate": 0.0001, + "loss": 0.1739, + "step": 33730 + }, + { + "epoch": 8.83, + "learning_rate": 0.0001, + "loss": 0.1671, + "step": 33740 + }, + { + "epoch": 8.83, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 33750 + }, + { + "epoch": 8.83, + "learning_rate": 0.0001, + "loss": 0.1761, + "step": 33760 + }, + { + "epoch": 8.84, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 33770 + }, + { + "epoch": 8.84, + "learning_rate": 0.0001, + "loss": 0.1697, + "step": 33780 + }, + { + "epoch": 8.84, + "learning_rate": 0.0001, + "loss": 0.1714, + "step": 33790 + }, + { + "epoch": 8.84, + "learning_rate": 0.0001, + "loss": 0.1738, + "step": 33800 + }, + { + "epoch": 8.85, + "learning_rate": 0.0001, + "loss": 0.1757, + "step": 33810 + }, + { + "epoch": 8.85, + "learning_rate": 0.0001, + "loss": 0.1716, + "step": 33820 + }, + { + "epoch": 8.85, + "learning_rate": 0.0001, + "loss": 0.1738, + "step": 33830 + }, + { + "epoch": 8.85, + "learning_rate": 0.0001, + "loss": 0.1771, + "step": 33840 + }, + { + "epoch": 8.86, + "learning_rate": 0.0001, + "loss": 0.1765, + "step": 33850 + }, + { + "epoch": 8.86, + "learning_rate": 0.0001, + "loss": 0.1698, + "step": 33860 + }, + { + "epoch": 8.86, + "learning_rate": 0.0001, + "loss": 0.1726, + "step": 33870 + }, + { + "epoch": 8.86, + "learning_rate": 0.0001, + "loss": 0.177, + "step": 33880 + }, + { + "epoch": 8.87, + "learning_rate": 0.0001, + "loss": 0.1687, + "step": 33890 + }, + { + "epoch": 8.87, + "learning_rate": 0.0001, + "loss": 0.1789, + "step": 33900 + }, + { + "epoch": 8.87, + "learning_rate": 0.0001, + "loss": 0.1794, + "step": 33910 + }, + { + "epoch": 8.87, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 33920 + }, + { + "epoch": 8.88, + "learning_rate": 0.0001, + "loss": 0.1719, + "step": 33930 + }, + { + "epoch": 8.88, + "learning_rate": 0.0001, + "loss": 0.1731, + "step": 33940 + }, + { + "epoch": 8.88, + "learning_rate": 0.0001, + "loss": 0.1661, + "step": 33950 + }, + { + "epoch": 8.89, + "learning_rate": 0.0001, + "loss": 0.1833, + "step": 33960 + }, + { + "epoch": 8.89, + "learning_rate": 0.0001, + "loss": 0.1688, + "step": 33970 + }, + { + "epoch": 8.89, + "learning_rate": 0.0001, + "loss": 0.1701, + "step": 33980 + }, + { + "epoch": 8.89, + "learning_rate": 0.0001, + "loss": 0.1772, + "step": 33990 + }, + { + "epoch": 8.9, + "learning_rate": 0.0001, + "loss": 0.1815, + "step": 34000 + }, + { + "epoch": 8.9, + "learning_rate": 0.0001, + "loss": 0.1747, + "step": 34010 + }, + { + "epoch": 8.9, + "learning_rate": 0.0001, + "loss": 0.1721, + "step": 34020 + }, + { + "epoch": 8.9, + "learning_rate": 0.0001, + "loss": 0.1713, + "step": 34030 + }, + { + "epoch": 8.91, + "learning_rate": 0.0001, + "loss": 0.1748, + "step": 34040 + }, + { + "epoch": 8.91, + "learning_rate": 0.0001, + "loss": 0.1769, + "step": 34050 + }, + { + "epoch": 8.91, + "learning_rate": 0.0001, + "loss": 0.1721, + "step": 34060 + }, + { + "epoch": 8.91, + "learning_rate": 0.0001, + "loss": 0.1741, + "step": 34070 + }, + { + "epoch": 8.92, + "learning_rate": 0.0001, + "loss": 0.1674, + "step": 34080 + }, + { + "epoch": 8.92, + "learning_rate": 0.0001, + "loss": 0.1782, + "step": 34090 + }, + { + "epoch": 8.92, + "learning_rate": 0.0001, + "loss": 0.1808, + "step": 34100 + }, + { + "epoch": 8.92, + "learning_rate": 0.0001, + "loss": 0.1803, + "step": 34110 + }, + { + "epoch": 8.93, + "learning_rate": 0.0001, + "loss": 0.1848, + "step": 34120 + }, + { + "epoch": 8.93, + "learning_rate": 0.0001, + "loss": 0.176, + "step": 34130 + }, + { + "epoch": 8.93, + "learning_rate": 0.0001, + "loss": 0.1773, + "step": 34140 + }, + { + "epoch": 8.94, + "learning_rate": 0.0001, + "loss": 0.1713, + "step": 34150 + }, + { + "epoch": 8.94, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 34160 + }, + { + "epoch": 8.94, + "learning_rate": 0.0001, + "loss": 0.1801, + "step": 34170 + }, + { + "epoch": 8.94, + "learning_rate": 0.0001, + "loss": 0.1797, + "step": 34180 + }, + { + "epoch": 8.95, + "learning_rate": 0.0001, + "loss": 0.1826, + "step": 34190 + }, + { + "epoch": 8.95, + "learning_rate": 0.0001, + "loss": 0.1768, + "step": 34200 + }, + { + "epoch": 8.95, + "learning_rate": 0.0001, + "loss": 0.1778, + "step": 34210 + }, + { + "epoch": 8.95, + "learning_rate": 0.0001, + "loss": 0.1749, + "step": 34220 + }, + { + "epoch": 8.96, + "learning_rate": 0.0001, + "loss": 0.1764, + "step": 34230 + }, + { + "epoch": 8.96, + "learning_rate": 0.0001, + "loss": 0.169, + "step": 34240 + }, + { + "epoch": 8.96, + "learning_rate": 0.0001, + "loss": 0.1829, + "step": 34250 + }, + { + "epoch": 8.96, + "learning_rate": 0.0001, + "loss": 0.1767, + "step": 34260 + }, + { + "epoch": 8.97, + "learning_rate": 0.0001, + "loss": 0.1758, + "step": 34270 + }, + { + "epoch": 8.97, + "learning_rate": 0.0001, + "loss": 0.173, + "step": 34280 + }, + { + "epoch": 8.97, + "learning_rate": 0.0001, + "loss": 0.1671, + "step": 34290 + }, + { + "epoch": 8.97, + "learning_rate": 0.0001, + "loss": 0.1803, + "step": 34300 + }, + { + "epoch": 8.98, + "learning_rate": 0.0001, + "loss": 0.1811, + "step": 34310 + }, + { + "epoch": 8.98, + "learning_rate": 0.0001, + "loss": 0.1765, + "step": 34320 + }, + { + "epoch": 8.98, + "learning_rate": 0.0001, + "loss": 0.1751, + "step": 34330 + }, + { + "epoch": 8.98, + "learning_rate": 0.0001, + "loss": 0.1691, + "step": 34340 + }, + { + "epoch": 8.99, + "learning_rate": 0.0001, + "loss": 0.1798, + "step": 34350 + }, + { + "epoch": 8.99, + "learning_rate": 0.0001, + "loss": 0.1793, + "step": 34360 + }, + { + "epoch": 8.99, + "learning_rate": 0.0001, + "loss": 0.1732, + "step": 34370 + }, + { + "epoch": 9.0, + "learning_rate": 0.0001, + "loss": 0.1791, + "step": 34380 + }, + { + "epoch": 9.0, + "learning_rate": 0.0001, + "loss": 0.1746, + "step": 34390 + }, + { + "epoch": 9.0, + "eval_gen_len": 61.7361, + "eval_loss": 1.3171662092208862, + "eval_rouge1": 57.2251, + "eval_rouge2": 35.2062, + "eval_rougeL": 46.6761, + "eval_rougeLsum": 52.905, + "eval_runtime": 33.7432, + "eval_samples_per_second": 26.05, + "eval_steps_per_second": 0.415, + "step": 34398 + }, + { + "epoch": 9.0, + "learning_rate": 0.0001, + "loss": 0.1738, + "step": 34400 + }, + { + "epoch": 9.0, + "learning_rate": 0.0001, + "loss": 0.1412, + "step": 34410 + }, + { + "epoch": 9.01, + "learning_rate": 0.0001, + "loss": 0.1413, + "step": 34420 + }, + { + "epoch": 9.01, + "learning_rate": 0.0001, + "loss": 0.1485, + "step": 34430 + }, + { + "epoch": 9.01, + "learning_rate": 0.0001, + "loss": 0.1404, + "step": 34440 + }, + { + "epoch": 9.01, + "learning_rate": 0.0001, + "loss": 0.1421, + "step": 34450 + }, + { + "epoch": 9.02, + "learning_rate": 0.0001, + "loss": 0.1457, + "step": 34460 + }, + { + "epoch": 9.02, + "learning_rate": 0.0001, + "loss": 0.1346, + "step": 34470 + }, + { + "epoch": 9.02, + "learning_rate": 0.0001, + "loss": 0.1477, + "step": 34480 + }, + { + "epoch": 9.02, + "learning_rate": 0.0001, + "loss": 0.1431, + "step": 34490 + }, + { + "epoch": 9.03, + "learning_rate": 0.0001, + "loss": 0.1447, + "step": 34500 + }, + { + "epoch": 9.03, + "learning_rate": 0.0001, + "loss": 0.144, + "step": 34510 + }, + { + "epoch": 9.03, + "learning_rate": 0.0001, + "loss": 0.1439, + "step": 34520 + }, + { + "epoch": 9.03, + "learning_rate": 0.0001, + "loss": 0.1491, + "step": 34530 + }, + { + "epoch": 9.04, + "learning_rate": 0.0001, + "loss": 0.1454, + "step": 34540 + }, + { + "epoch": 9.04, + "learning_rate": 0.0001, + "loss": 0.1366, + "step": 34550 + }, + { + "epoch": 9.04, + "learning_rate": 0.0001, + "loss": 0.1408, + "step": 34560 + }, + { + "epoch": 9.05, + "learning_rate": 0.0001, + "loss": 0.1342, + "step": 34570 + }, + { + "epoch": 9.05, + "learning_rate": 0.0001, + "loss": 0.1453, + "step": 34580 + }, + { + "epoch": 9.05, + "learning_rate": 0.0001, + "loss": 0.1416, + "step": 34590 + }, + { + "epoch": 9.05, + "learning_rate": 0.0001, + "loss": 0.1361, + "step": 34600 + }, + { + "epoch": 9.06, + "learning_rate": 0.0001, + "loss": 0.1448, + "step": 34610 + }, + { + "epoch": 9.06, + "learning_rate": 0.0001, + "loss": 0.148, + "step": 34620 + }, + { + "epoch": 9.06, + "learning_rate": 0.0001, + "loss": 0.146, + "step": 34630 + }, + { + "epoch": 9.06, + "learning_rate": 0.0001, + "loss": 0.1365, + "step": 34640 + }, + { + "epoch": 9.07, + "learning_rate": 0.0001, + "loss": 0.1469, + "step": 34650 + }, + { + "epoch": 9.07, + "learning_rate": 0.0001, + "loss": 0.1452, + "step": 34660 + }, + { + "epoch": 9.07, + "learning_rate": 0.0001, + "loss": 0.1429, + "step": 34670 + }, + { + "epoch": 9.07, + "learning_rate": 0.0001, + "loss": 0.1431, + "step": 34680 + }, + { + "epoch": 9.08, + "learning_rate": 0.0001, + "loss": 0.1432, + "step": 34690 + }, + { + "epoch": 9.08, + "learning_rate": 0.0001, + "loss": 0.1517, + "step": 34700 + }, + { + "epoch": 9.08, + "learning_rate": 0.0001, + "loss": 0.1468, + "step": 34710 + }, + { + "epoch": 9.08, + "learning_rate": 0.0001, + "loss": 0.1433, + "step": 34720 + }, + { + "epoch": 9.09, + "learning_rate": 0.0001, + "loss": 0.1425, + "step": 34730 + }, + { + "epoch": 9.09, + "learning_rate": 0.0001, + "loss": 0.1478, + "step": 34740 + }, + { + "epoch": 9.09, + "learning_rate": 0.0001, + "loss": 0.1403, + "step": 34750 + }, + { + "epoch": 9.09, + "learning_rate": 0.0001, + "loss": 0.1476, + "step": 34760 + }, + { + "epoch": 9.1, + "learning_rate": 0.0001, + "loss": 0.1452, + "step": 34770 + }, + { + "epoch": 9.1, + "learning_rate": 0.0001, + "loss": 0.1396, + "step": 34780 + }, + { + "epoch": 9.1, + "learning_rate": 0.0001, + "loss": 0.1416, + "step": 34790 + }, + { + "epoch": 9.11, + "learning_rate": 0.0001, + "loss": 0.1468, + "step": 34800 + }, + { + "epoch": 9.11, + "learning_rate": 0.0001, + "loss": 0.1429, + "step": 34810 + }, + { + "epoch": 9.11, + "learning_rate": 0.0001, + "loss": 0.1394, + "step": 34820 + }, + { + "epoch": 9.11, + "learning_rate": 0.0001, + "loss": 0.1457, + "step": 34830 + }, + { + "epoch": 9.12, + "learning_rate": 0.0001, + "loss": 0.1408, + "step": 34840 + }, + { + "epoch": 9.12, + "learning_rate": 0.0001, + "loss": 0.1413, + "step": 34850 + }, + { + "epoch": 9.12, + "learning_rate": 0.0001, + "loss": 0.1415, + "step": 34860 + }, + { + "epoch": 9.12, + "learning_rate": 0.0001, + "loss": 0.146, + "step": 34870 + }, + { + "epoch": 9.13, + "learning_rate": 0.0001, + "loss": 0.1449, + "step": 34880 + }, + { + "epoch": 9.13, + "learning_rate": 0.0001, + "loss": 0.1439, + "step": 34890 + }, + { + "epoch": 9.13, + "learning_rate": 0.0001, + "loss": 0.1449, + "step": 34900 + }, + { + "epoch": 9.13, + "learning_rate": 0.0001, + "loss": 0.1425, + "step": 34910 + }, + { + "epoch": 9.14, + "learning_rate": 0.0001, + "loss": 0.1413, + "step": 34920 + }, + { + "epoch": 9.14, + "learning_rate": 0.0001, + "loss": 0.1493, + "step": 34930 + }, + { + "epoch": 9.14, + "learning_rate": 0.0001, + "loss": 0.1442, + "step": 34940 + }, + { + "epoch": 9.14, + "learning_rate": 0.0001, + "loss": 0.1463, + "step": 34950 + }, + { + "epoch": 9.15, + "learning_rate": 0.0001, + "loss": 0.1427, + "step": 34960 + }, + { + "epoch": 9.15, + "learning_rate": 0.0001, + "loss": 0.1533, + "step": 34970 + }, + { + "epoch": 9.15, + "learning_rate": 0.0001, + "loss": 0.1415, + "step": 34980 + }, + { + "epoch": 9.15, + "learning_rate": 0.0001, + "loss": 0.1468, + "step": 34990 + }, + { + "epoch": 9.16, + "learning_rate": 0.0001, + "loss": 0.1464, + "step": 35000 + }, + { + "epoch": 9.16, + "learning_rate": 0.0001, + "loss": 0.1444, + "step": 35010 + }, + { + "epoch": 9.16, + "learning_rate": 0.0001, + "loss": 0.1466, + "step": 35020 + }, + { + "epoch": 9.17, + "learning_rate": 0.0001, + "loss": 0.1385, + "step": 35030 + }, + { + "epoch": 9.17, + "learning_rate": 0.0001, + "loss": 0.1398, + "step": 35040 + }, + { + "epoch": 9.17, + "learning_rate": 0.0001, + "loss": 0.1465, + "step": 35050 + }, + { + "epoch": 9.17, + "learning_rate": 0.0001, + "loss": 0.145, + "step": 35060 + }, + { + "epoch": 9.18, + "learning_rate": 0.0001, + "loss": 0.1437, + "step": 35070 + }, + { + "epoch": 9.18, + "learning_rate": 0.0001, + "loss": 0.1486, + "step": 35080 + }, + { + "epoch": 9.18, + "learning_rate": 0.0001, + "loss": 0.1414, + "step": 35090 + }, + { + "epoch": 9.18, + "learning_rate": 0.0001, + "loss": 0.1512, + "step": 35100 + }, + { + "epoch": 9.19, + "learning_rate": 0.0001, + "loss": 0.1468, + "step": 35110 + }, + { + "epoch": 9.19, + "learning_rate": 0.0001, + "loss": 0.1403, + "step": 35120 + }, + { + "epoch": 9.19, + "learning_rate": 0.0001, + "loss": 0.1437, + "step": 35130 + }, + { + "epoch": 9.19, + "learning_rate": 0.0001, + "loss": 0.1497, + "step": 35140 + }, + { + "epoch": 9.2, + "learning_rate": 0.0001, + "loss": 0.1492, + "step": 35150 + }, + { + "epoch": 9.2, + "learning_rate": 0.0001, + "loss": 0.1399, + "step": 35160 + }, + { + "epoch": 9.2, + "learning_rate": 0.0001, + "loss": 0.1461, + "step": 35170 + }, + { + "epoch": 9.2, + "learning_rate": 0.0001, + "loss": 0.1458, + "step": 35180 + }, + { + "epoch": 9.21, + "learning_rate": 0.0001, + "loss": 0.1436, + "step": 35190 + }, + { + "epoch": 9.21, + "learning_rate": 0.0001, + "loss": 0.146, + "step": 35200 + }, + { + "epoch": 9.21, + "learning_rate": 0.0001, + "loss": 0.1438, + "step": 35210 + }, + { + "epoch": 9.22, + "learning_rate": 0.0001, + "loss": 0.1436, + "step": 35220 + }, + { + "epoch": 9.22, + "learning_rate": 0.0001, + "loss": 0.1482, + "step": 35230 + }, + { + "epoch": 9.22, + "learning_rate": 0.0001, + "loss": 0.141, + "step": 35240 + }, + { + "epoch": 9.22, + "learning_rate": 0.0001, + "loss": 0.1492, + "step": 35250 + }, + { + "epoch": 9.23, + "learning_rate": 0.0001, + "loss": 0.146, + "step": 35260 + }, + { + "epoch": 9.23, + "learning_rate": 0.0001, + "loss": 0.1462, + "step": 35270 + }, + { + "epoch": 9.23, + "learning_rate": 0.0001, + "loss": 0.1436, + "step": 35280 + }, + { + "epoch": 9.23, + "learning_rate": 0.0001, + "loss": 0.1446, + "step": 35290 + }, + { + "epoch": 9.24, + "learning_rate": 0.0001, + "loss": 0.1513, + "step": 35300 + }, + { + "epoch": 9.24, + "learning_rate": 0.0001, + "loss": 0.1494, + "step": 35310 + }, + { + "epoch": 9.24, + "learning_rate": 0.0001, + "loss": 0.1442, + "step": 35320 + }, + { + "epoch": 9.24, + "learning_rate": 0.0001, + "loss": 0.1554, + "step": 35330 + }, + { + "epoch": 9.25, + "learning_rate": 0.0001, + "loss": 0.1476, + "step": 35340 + }, + { + "epoch": 9.25, + "learning_rate": 0.0001, + "loss": 0.1489, + "step": 35350 + }, + { + "epoch": 9.25, + "learning_rate": 0.0001, + "loss": 0.1429, + "step": 35360 + }, + { + "epoch": 9.25, + "learning_rate": 0.0001, + "loss": 0.1483, + "step": 35370 + }, + { + "epoch": 9.26, + "learning_rate": 0.0001, + "loss": 0.149, + "step": 35380 + }, + { + "epoch": 9.26, + "learning_rate": 0.0001, + "loss": 0.1457, + "step": 35390 + }, + { + "epoch": 9.26, + "learning_rate": 0.0001, + "loss": 0.1482, + "step": 35400 + }, + { + "epoch": 9.26, + "learning_rate": 0.0001, + "loss": 0.1537, + "step": 35410 + }, + { + "epoch": 9.27, + "learning_rate": 0.0001, + "loss": 0.1475, + "step": 35420 + }, + { + "epoch": 9.27, + "learning_rate": 0.0001, + "loss": 0.1476, + "step": 35430 + }, + { + "epoch": 9.27, + "learning_rate": 0.0001, + "loss": 0.1515, + "step": 35440 + }, + { + "epoch": 9.28, + "learning_rate": 0.0001, + "loss": 0.1498, + "step": 35450 + }, + { + "epoch": 9.28, + "learning_rate": 0.0001, + "loss": 0.1556, + "step": 35460 + }, + { + "epoch": 9.28, + "learning_rate": 0.0001, + "loss": 0.1479, + "step": 35470 + }, + { + "epoch": 9.28, + "learning_rate": 0.0001, + "loss": 0.1511, + "step": 35480 + }, + { + "epoch": 9.29, + "learning_rate": 0.0001, + "loss": 0.147, + "step": 35490 + }, + { + "epoch": 9.29, + "learning_rate": 0.0001, + "loss": 0.152, + "step": 35500 + }, + { + "epoch": 9.29, + "learning_rate": 0.0001, + "loss": 0.1453, + "step": 35510 + }, + { + "epoch": 9.29, + "learning_rate": 0.0001, + "loss": 0.1476, + "step": 35520 + }, + { + "epoch": 9.3, + "learning_rate": 0.0001, + "loss": 0.1498, + "step": 35530 + }, + { + "epoch": 9.3, + "learning_rate": 0.0001, + "loss": 0.1426, + "step": 35540 + }, + { + "epoch": 9.3, + "learning_rate": 0.0001, + "loss": 0.1519, + "step": 35550 + }, + { + "epoch": 9.3, + "learning_rate": 0.0001, + "loss": 0.1453, + "step": 35560 + }, + { + "epoch": 9.31, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 35570 + }, + { + "epoch": 9.31, + "learning_rate": 0.0001, + "loss": 0.1452, + "step": 35580 + }, + { + "epoch": 9.31, + "learning_rate": 0.0001, + "loss": 0.1431, + "step": 35590 + }, + { + "epoch": 9.31, + "learning_rate": 0.0001, + "loss": 0.1555, + "step": 35600 + }, + { + "epoch": 9.32, + "learning_rate": 0.0001, + "loss": 0.1463, + "step": 35610 + }, + { + "epoch": 9.32, + "learning_rate": 0.0001, + "loss": 0.1488, + "step": 35620 + }, + { + "epoch": 9.32, + "learning_rate": 0.0001, + "loss": 0.1521, + "step": 35630 + }, + { + "epoch": 9.32, + "learning_rate": 0.0001, + "loss": 0.1474, + "step": 35640 + }, + { + "epoch": 9.33, + "learning_rate": 0.0001, + "loss": 0.143, + "step": 35650 + }, + { + "epoch": 9.33, + "learning_rate": 0.0001, + "loss": 0.1438, + "step": 35660 + }, + { + "epoch": 9.33, + "learning_rate": 0.0001, + "loss": 0.148, + "step": 35670 + }, + { + "epoch": 9.34, + "learning_rate": 0.0001, + "loss": 0.1542, + "step": 35680 + }, + { + "epoch": 9.34, + "learning_rate": 0.0001, + "loss": 0.1468, + "step": 35690 + }, + { + "epoch": 9.34, + "learning_rate": 0.0001, + "loss": 0.1485, + "step": 35700 + }, + { + "epoch": 9.34, + "learning_rate": 0.0001, + "loss": 0.1523, + "step": 35710 + }, + { + "epoch": 9.35, + "learning_rate": 0.0001, + "loss": 0.1537, + "step": 35720 + }, + { + "epoch": 9.35, + "learning_rate": 0.0001, + "loss": 0.1548, + "step": 35730 + }, + { + "epoch": 9.35, + "learning_rate": 0.0001, + "loss": 0.1491, + "step": 35740 + }, + { + "epoch": 9.35, + "learning_rate": 0.0001, + "loss": 0.1533, + "step": 35750 + }, + { + "epoch": 9.36, + "learning_rate": 0.0001, + "loss": 0.1466, + "step": 35760 + }, + { + "epoch": 9.36, + "learning_rate": 0.0001, + "loss": 0.1503, + "step": 35770 + }, + { + "epoch": 9.36, + "learning_rate": 0.0001, + "loss": 0.1461, + "step": 35780 + }, + { + "epoch": 9.36, + "learning_rate": 0.0001, + "loss": 0.147, + "step": 35790 + }, + { + "epoch": 9.37, + "learning_rate": 0.0001, + "loss": 0.1474, + "step": 35800 + }, + { + "epoch": 9.37, + "learning_rate": 0.0001, + "loss": 0.1518, + "step": 35810 + }, + { + "epoch": 9.37, + "learning_rate": 0.0001, + "loss": 0.1465, + "step": 35820 + }, + { + "epoch": 9.37, + "learning_rate": 0.0001, + "loss": 0.1525, + "step": 35830 + }, + { + "epoch": 9.38, + "learning_rate": 0.0001, + "loss": 0.1453, + "step": 35840 + }, + { + "epoch": 9.38, + "learning_rate": 0.0001, + "loss": 0.1486, + "step": 35850 + }, + { + "epoch": 9.38, + "learning_rate": 0.0001, + "loss": 0.1536, + "step": 35860 + }, + { + "epoch": 9.39, + "learning_rate": 0.0001, + "loss": 0.1505, + "step": 35870 + }, + { + "epoch": 9.39, + "learning_rate": 0.0001, + "loss": 0.1458, + "step": 35880 + }, + { + "epoch": 9.39, + "learning_rate": 0.0001, + "loss": 0.1523, + "step": 35890 + }, + { + "epoch": 9.39, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 35900 + }, + { + "epoch": 9.4, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 35910 + }, + { + "epoch": 9.4, + "learning_rate": 0.0001, + "loss": 0.1456, + "step": 35920 + }, + { + "epoch": 9.4, + "learning_rate": 0.0001, + "loss": 0.155, + "step": 35930 + }, + { + "epoch": 9.4, + "learning_rate": 0.0001, + "loss": 0.1506, + "step": 35940 + }, + { + "epoch": 9.41, + "learning_rate": 0.0001, + "loss": 0.1515, + "step": 35950 + }, + { + "epoch": 9.41, + "learning_rate": 0.0001, + "loss": 0.1526, + "step": 35960 + }, + { + "epoch": 9.41, + "learning_rate": 0.0001, + "loss": 0.1572, + "step": 35970 + }, + { + "epoch": 9.41, + "learning_rate": 0.0001, + "loss": 0.1545, + "step": 35980 + }, + { + "epoch": 9.42, + "learning_rate": 0.0001, + "loss": 0.1566, + "step": 35990 + }, + { + "epoch": 9.42, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 36000 + }, + { + "epoch": 9.42, + "learning_rate": 0.0001, + "loss": 0.1548, + "step": 36010 + }, + { + "epoch": 9.42, + "learning_rate": 0.0001, + "loss": 0.1516, + "step": 36020 + }, + { + "epoch": 9.43, + "learning_rate": 0.0001, + "loss": 0.1449, + "step": 36030 + }, + { + "epoch": 9.43, + "learning_rate": 0.0001, + "loss": 0.1524, + "step": 36040 + }, + { + "epoch": 9.43, + "learning_rate": 0.0001, + "loss": 0.1467, + "step": 36050 + }, + { + "epoch": 9.43, + "learning_rate": 0.0001, + "loss": 0.1518, + "step": 36060 + }, + { + "epoch": 9.44, + "learning_rate": 0.0001, + "loss": 0.1434, + "step": 36070 + }, + { + "epoch": 9.44, + "learning_rate": 0.0001, + "loss": 0.1485, + "step": 36080 + }, + { + "epoch": 9.44, + "learning_rate": 0.0001, + "loss": 0.1487, + "step": 36090 + }, + { + "epoch": 9.45, + "learning_rate": 0.0001, + "loss": 0.1534, + "step": 36100 + }, + { + "epoch": 9.45, + "learning_rate": 0.0001, + "loss": 0.1568, + "step": 36110 + }, + { + "epoch": 9.45, + "learning_rate": 0.0001, + "loss": 0.1489, + "step": 36120 + }, + { + "epoch": 9.45, + "learning_rate": 0.0001, + "loss": 0.1419, + "step": 36130 + }, + { + "epoch": 9.46, + "learning_rate": 0.0001, + "loss": 0.1534, + "step": 36140 + }, + { + "epoch": 9.46, + "learning_rate": 0.0001, + "loss": 0.1538, + "step": 36150 + }, + { + "epoch": 9.46, + "learning_rate": 0.0001, + "loss": 0.1532, + "step": 36160 + }, + { + "epoch": 9.46, + "learning_rate": 0.0001, + "loss": 0.1555, + "step": 36170 + }, + { + "epoch": 9.47, + "learning_rate": 0.0001, + "loss": 0.1527, + "step": 36180 + }, + { + "epoch": 9.47, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 36190 + }, + { + "epoch": 9.47, + "learning_rate": 0.0001, + "loss": 0.1559, + "step": 36200 + }, + { + "epoch": 9.47, + "learning_rate": 0.0001, + "loss": 0.1482, + "step": 36210 + }, + { + "epoch": 9.48, + "learning_rate": 0.0001, + "loss": 0.149, + "step": 36220 + }, + { + "epoch": 9.48, + "learning_rate": 0.0001, + "loss": 0.1481, + "step": 36230 + }, + { + "epoch": 9.48, + "learning_rate": 0.0001, + "loss": 0.1509, + "step": 36240 + }, + { + "epoch": 9.48, + "learning_rate": 0.0001, + "loss": 0.1498, + "step": 36250 + }, + { + "epoch": 9.49, + "learning_rate": 0.0001, + "loss": 0.1507, + "step": 36260 + }, + { + "epoch": 9.49, + "learning_rate": 0.0001, + "loss": 0.1522, + "step": 36270 + }, + { + "epoch": 9.49, + "learning_rate": 0.0001, + "loss": 0.1494, + "step": 36280 + }, + { + "epoch": 9.5, + "learning_rate": 0.0001, + "loss": 0.1531, + "step": 36290 + }, + { + "epoch": 9.5, + "learning_rate": 0.0001, + "loss": 0.152, + "step": 36300 + }, + { + "epoch": 9.5, + "learning_rate": 0.0001, + "loss": 0.1516, + "step": 36310 + }, + { + "epoch": 9.5, + "learning_rate": 0.0001, + "loss": 0.1498, + "step": 36320 + }, + { + "epoch": 9.51, + "learning_rate": 0.0001, + "loss": 0.1554, + "step": 36330 + }, + { + "epoch": 9.51, + "learning_rate": 0.0001, + "loss": 0.153, + "step": 36340 + }, + { + "epoch": 9.51, + "learning_rate": 0.0001, + "loss": 0.148, + "step": 36350 + }, + { + "epoch": 9.51, + "learning_rate": 0.0001, + "loss": 0.1463, + "step": 36360 + }, + { + "epoch": 9.52, + "learning_rate": 0.0001, + "loss": 0.1494, + "step": 36370 + }, + { + "epoch": 9.52, + "learning_rate": 0.0001, + "loss": 0.1523, + "step": 36380 + }, + { + "epoch": 9.52, + "learning_rate": 0.0001, + "loss": 0.1552, + "step": 36390 + }, + { + "epoch": 9.52, + "learning_rate": 0.0001, + "loss": 0.1538, + "step": 36400 + }, + { + "epoch": 9.53, + "learning_rate": 0.0001, + "loss": 0.1543, + "step": 36410 + }, + { + "epoch": 9.53, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 36420 + }, + { + "epoch": 9.53, + "learning_rate": 0.0001, + "loss": 0.1556, + "step": 36430 + }, + { + "epoch": 9.53, + "learning_rate": 0.0001, + "loss": 0.1506, + "step": 36440 + }, + { + "epoch": 9.54, + "learning_rate": 0.0001, + "loss": 0.1576, + "step": 36450 + }, + { + "epoch": 9.54, + "learning_rate": 0.0001, + "loss": 0.159, + "step": 36460 + }, + { + "epoch": 9.54, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 36470 + }, + { + "epoch": 9.54, + "learning_rate": 0.0001, + "loss": 0.1557, + "step": 36480 + }, + { + "epoch": 9.55, + "learning_rate": 0.0001, + "loss": 0.1573, + "step": 36490 + }, + { + "epoch": 9.55, + "learning_rate": 0.0001, + "loss": 0.1537, + "step": 36500 + }, + { + "epoch": 9.55, + "learning_rate": 0.0001, + "loss": 0.156, + "step": 36510 + }, + { + "epoch": 9.56, + "learning_rate": 0.0001, + "loss": 0.157, + "step": 36520 + }, + { + "epoch": 9.56, + "learning_rate": 0.0001, + "loss": 0.1519, + "step": 36530 + }, + { + "epoch": 9.56, + "learning_rate": 0.0001, + "loss": 0.1485, + "step": 36540 + }, + { + "epoch": 9.56, + "learning_rate": 0.0001, + "loss": 0.1565, + "step": 36550 + }, + { + "epoch": 9.57, + "learning_rate": 0.0001, + "loss": 0.1525, + "step": 36560 + }, + { + "epoch": 9.57, + "learning_rate": 0.0001, + "loss": 0.1538, + "step": 36570 + }, + { + "epoch": 9.57, + "learning_rate": 0.0001, + "loss": 0.1451, + "step": 36580 + }, + { + "epoch": 9.57, + "learning_rate": 0.0001, + "loss": 0.1504, + "step": 36590 + }, + { + "epoch": 9.58, + "learning_rate": 0.0001, + "loss": 0.1575, + "step": 36600 + }, + { + "epoch": 9.58, + "learning_rate": 0.0001, + "loss": 0.1555, + "step": 36610 + }, + { + "epoch": 9.58, + "learning_rate": 0.0001, + "loss": 0.159, + "step": 36620 + }, + { + "epoch": 9.58, + "learning_rate": 0.0001, + "loss": 0.1545, + "step": 36630 + }, + { + "epoch": 9.59, + "learning_rate": 0.0001, + "loss": 0.1562, + "step": 36640 + }, + { + "epoch": 9.59, + "learning_rate": 0.0001, + "loss": 0.1485, + "step": 36650 + }, + { + "epoch": 9.59, + "learning_rate": 0.0001, + "loss": 0.1537, + "step": 36660 + }, + { + "epoch": 9.59, + "learning_rate": 0.0001, + "loss": 0.1569, + "step": 36670 + }, + { + "epoch": 9.6, + "learning_rate": 0.0001, + "loss": 0.1524, + "step": 36680 + }, + { + "epoch": 9.6, + "learning_rate": 0.0001, + "loss": 0.1507, + "step": 36690 + }, + { + "epoch": 9.6, + "learning_rate": 0.0001, + "loss": 0.1444, + "step": 36700 + }, + { + "epoch": 9.6, + "learning_rate": 0.0001, + "loss": 0.155, + "step": 36710 + }, + { + "epoch": 9.61, + "learning_rate": 0.0001, + "loss": 0.1511, + "step": 36720 + }, + { + "epoch": 9.61, + "learning_rate": 0.0001, + "loss": 0.1508, + "step": 36730 + }, + { + "epoch": 9.61, + "learning_rate": 0.0001, + "loss": 0.1554, + "step": 36740 + }, + { + "epoch": 9.62, + "learning_rate": 0.0001, + "loss": 0.1565, + "step": 36750 + }, + { + "epoch": 9.62, + "learning_rate": 0.0001, + "loss": 0.1639, + "step": 36760 + }, + { + "epoch": 9.62, + "learning_rate": 0.0001, + "loss": 0.1538, + "step": 36770 + }, + { + "epoch": 9.62, + "learning_rate": 0.0001, + "loss": 0.1578, + "step": 36780 + }, + { + "epoch": 9.63, + "learning_rate": 0.0001, + "loss": 0.159, + "step": 36790 + }, + { + "epoch": 9.63, + "learning_rate": 0.0001, + "loss": 0.1512, + "step": 36800 + }, + { + "epoch": 9.63, + "learning_rate": 0.0001, + "loss": 0.1569, + "step": 36810 + }, + { + "epoch": 9.63, + "learning_rate": 0.0001, + "loss": 0.1623, + "step": 36820 + }, + { + "epoch": 9.64, + "learning_rate": 0.0001, + "loss": 0.15, + "step": 36830 + }, + { + "epoch": 9.64, + "learning_rate": 0.0001, + "loss": 0.1611, + "step": 36840 + }, + { + "epoch": 9.64, + "learning_rate": 0.0001, + "loss": 0.155, + "step": 36850 + }, + { + "epoch": 9.64, + "learning_rate": 0.0001, + "loss": 0.1557, + "step": 36860 + }, + { + "epoch": 9.65, + "learning_rate": 0.0001, + "loss": 0.1579, + "step": 36870 + }, + { + "epoch": 9.65, + "learning_rate": 0.0001, + "loss": 0.1543, + "step": 36880 + }, + { + "epoch": 9.65, + "learning_rate": 0.0001, + "loss": 0.154, + "step": 36890 + }, + { + "epoch": 9.65, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 36900 + }, + { + "epoch": 9.66, + "learning_rate": 0.0001, + "loss": 0.1489, + "step": 36910 + }, + { + "epoch": 9.66, + "learning_rate": 0.0001, + "loss": 0.1619, + "step": 36920 + }, + { + "epoch": 9.66, + "learning_rate": 0.0001, + "loss": 0.1573, + "step": 36930 + }, + { + "epoch": 9.67, + "learning_rate": 0.0001, + "loss": 0.1566, + "step": 36940 + }, + { + "epoch": 9.67, + "learning_rate": 0.0001, + "loss": 0.1569, + "step": 36950 + }, + { + "epoch": 9.67, + "learning_rate": 0.0001, + "loss": 0.156, + "step": 36960 + }, + { + "epoch": 9.67, + "learning_rate": 0.0001, + "loss": 0.1534, + "step": 36970 + }, + { + "epoch": 9.68, + "learning_rate": 0.0001, + "loss": 0.1561, + "step": 36980 + }, + { + "epoch": 9.68, + "learning_rate": 0.0001, + "loss": 0.1585, + "step": 36990 + }, + { + "epoch": 9.68, + "learning_rate": 0.0001, + "loss": 0.1557, + "step": 37000 + }, + { + "epoch": 9.68, + "learning_rate": 0.0001, + "loss": 0.1541, + "step": 37010 + }, + { + "epoch": 9.69, + "learning_rate": 0.0001, + "loss": 0.1496, + "step": 37020 + }, + { + "epoch": 9.69, + "learning_rate": 0.0001, + "loss": 0.1535, + "step": 37030 + }, + { + "epoch": 9.69, + "learning_rate": 0.0001, + "loss": 0.1516, + "step": 37040 + }, + { + "epoch": 9.69, + "learning_rate": 0.0001, + "loss": 0.1603, + "step": 37050 + }, + { + "epoch": 9.7, + "learning_rate": 0.0001, + "loss": 0.1641, + "step": 37060 + }, + { + "epoch": 9.7, + "learning_rate": 0.0001, + "loss": 0.1532, + "step": 37070 + }, + { + "epoch": 9.7, + "learning_rate": 0.0001, + "loss": 0.1505, + "step": 37080 + }, + { + "epoch": 9.7, + "learning_rate": 0.0001, + "loss": 0.1566, + "step": 37090 + }, + { + "epoch": 9.71, + "learning_rate": 0.0001, + "loss": 0.1536, + "step": 37100 + }, + { + "epoch": 9.71, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 37110 + }, + { + "epoch": 9.71, + "learning_rate": 0.0001, + "loss": 0.1633, + "step": 37120 + }, + { + "epoch": 9.71, + "learning_rate": 0.0001, + "loss": 0.1596, + "step": 37130 + }, + { + "epoch": 9.72, + "learning_rate": 0.0001, + "loss": 0.1577, + "step": 37140 + }, + { + "epoch": 9.72, + "learning_rate": 0.0001, + "loss": 0.1495, + "step": 37150 + }, + { + "epoch": 9.72, + "learning_rate": 0.0001, + "loss": 0.1541, + "step": 37160 + }, + { + "epoch": 9.73, + "learning_rate": 0.0001, + "loss": 0.1558, + "step": 37170 + }, + { + "epoch": 9.73, + "learning_rate": 0.0001, + "loss": 0.1587, + "step": 37180 + }, + { + "epoch": 9.73, + "learning_rate": 0.0001, + "loss": 0.1566, + "step": 37190 + }, + { + "epoch": 9.73, + "learning_rate": 0.0001, + "loss": 0.1546, + "step": 37200 + }, + { + "epoch": 9.74, + "learning_rate": 0.0001, + "loss": 0.1594, + "step": 37210 + }, + { + "epoch": 9.74, + "learning_rate": 0.0001, + "loss": 0.1531, + "step": 37220 + }, + { + "epoch": 9.74, + "learning_rate": 0.0001, + "loss": 0.161, + "step": 37230 + }, + { + "epoch": 9.74, + "learning_rate": 0.0001, + "loss": 0.1554, + "step": 37240 + }, + { + "epoch": 9.75, + "learning_rate": 0.0001, + "loss": 0.1601, + "step": 37250 + }, + { + "epoch": 9.75, + "learning_rate": 0.0001, + "loss": 0.1547, + "step": 37260 + }, + { + "epoch": 9.75, + "learning_rate": 0.0001, + "loss": 0.159, + "step": 37270 + }, + { + "epoch": 9.75, + "learning_rate": 0.0001, + "loss": 0.1536, + "step": 37280 + }, + { + "epoch": 9.76, + "learning_rate": 0.0001, + "loss": 0.1551, + "step": 37290 + }, + { + "epoch": 9.76, + "learning_rate": 0.0001, + "loss": 0.1645, + "step": 37300 + }, + { + "epoch": 9.76, + "learning_rate": 0.0001, + "loss": 0.1586, + "step": 37310 + }, + { + "epoch": 9.76, + "learning_rate": 0.0001, + "loss": 0.1537, + "step": 37320 + }, + { + "epoch": 9.77, + "learning_rate": 0.0001, + "loss": 0.1637, + "step": 37330 + }, + { + "epoch": 9.77, + "learning_rate": 0.0001, + "loss": 0.1546, + "step": 37340 + }, + { + "epoch": 9.77, + "learning_rate": 0.0001, + "loss": 0.1492, + "step": 37350 + }, + { + "epoch": 9.77, + "learning_rate": 0.0001, + "loss": 0.154, + "step": 37360 + }, + { + "epoch": 9.78, + "learning_rate": 0.0001, + "loss": 0.1564, + "step": 37370 + }, + { + "epoch": 9.78, + "learning_rate": 0.0001, + "loss": 0.1595, + "step": 37380 + }, + { + "epoch": 9.78, + "learning_rate": 0.0001, + "loss": 0.1515, + "step": 37390 + }, + { + "epoch": 9.79, + "learning_rate": 0.0001, + "loss": 0.1599, + "step": 37400 + }, + { + "epoch": 9.79, + "learning_rate": 0.0001, + "loss": 0.1548, + "step": 37410 + }, + { + "epoch": 9.79, + "learning_rate": 0.0001, + "loss": 0.1527, + "step": 37420 + }, + { + "epoch": 9.79, + "learning_rate": 0.0001, + "loss": 0.1583, + "step": 37430 + }, + { + "epoch": 9.8, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 37440 + }, + { + "epoch": 9.8, + "learning_rate": 0.0001, + "loss": 0.1588, + "step": 37450 + }, + { + "epoch": 9.8, + "learning_rate": 0.0001, + "loss": 0.1588, + "step": 37460 + }, + { + "epoch": 9.8, + "learning_rate": 0.0001, + "loss": 0.1489, + "step": 37470 + }, + { + "epoch": 9.81, + "learning_rate": 0.0001, + "loss": 0.1614, + "step": 37480 + }, + { + "epoch": 9.81, + "learning_rate": 0.0001, + "loss": 0.1502, + "step": 37490 + }, + { + "epoch": 9.81, + "learning_rate": 0.0001, + "loss": 0.1594, + "step": 37500 + }, + { + "epoch": 9.81, + "learning_rate": 0.0001, + "loss": 0.1559, + "step": 37510 + }, + { + "epoch": 9.82, + "learning_rate": 0.0001, + "loss": 0.1559, + "step": 37520 + }, + { + "epoch": 9.82, + "learning_rate": 0.0001, + "loss": 0.1584, + "step": 37530 + }, + { + "epoch": 9.82, + "learning_rate": 0.0001, + "loss": 0.1568, + "step": 37540 + }, + { + "epoch": 9.82, + "learning_rate": 0.0001, + "loss": 0.1645, + "step": 37550 + }, + { + "epoch": 9.83, + "learning_rate": 0.0001, + "loss": 0.1551, + "step": 37560 + }, + { + "epoch": 9.83, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 37570 + }, + { + "epoch": 9.83, + "learning_rate": 0.0001, + "loss": 0.156, + "step": 37580 + }, + { + "epoch": 9.84, + "learning_rate": 0.0001, + "loss": 0.1602, + "step": 37590 + }, + { + "epoch": 9.84, + "learning_rate": 0.0001, + "loss": 0.1576, + "step": 37600 + }, + { + "epoch": 9.84, + "learning_rate": 0.0001, + "loss": 0.1548, + "step": 37610 + }, + { + "epoch": 9.84, + "learning_rate": 0.0001, + "loss": 0.1588, + "step": 37620 + }, + { + "epoch": 9.85, + "learning_rate": 0.0001, + "loss": 0.1631, + "step": 37630 + }, + { + "epoch": 9.85, + "learning_rate": 0.0001, + "loss": 0.1547, + "step": 37640 + }, + { + "epoch": 9.85, + "learning_rate": 0.0001, + "loss": 0.155, + "step": 37650 + }, + { + "epoch": 9.85, + "learning_rate": 0.0001, + "loss": 0.1591, + "step": 37660 + }, + { + "epoch": 9.86, + "learning_rate": 0.0001, + "loss": 0.1677, + "step": 37670 + }, + { + "epoch": 9.86, + "learning_rate": 0.0001, + "loss": 0.1545, + "step": 37680 + }, + { + "epoch": 9.86, + "learning_rate": 0.0001, + "loss": 0.1604, + "step": 37690 + }, + { + "epoch": 9.86, + "learning_rate": 0.0001, + "loss": 0.1628, + "step": 37700 + }, + { + "epoch": 9.87, + "learning_rate": 0.0001, + "loss": 0.1636, + "step": 37710 + }, + { + "epoch": 9.87, + "learning_rate": 0.0001, + "loss": 0.1526, + "step": 37720 + }, + { + "epoch": 9.87, + "learning_rate": 0.0001, + "loss": 0.1531, + "step": 37730 + }, + { + "epoch": 9.87, + "learning_rate": 0.0001, + "loss": 0.1613, + "step": 37740 + }, + { + "epoch": 9.88, + "learning_rate": 0.0001, + "loss": 0.1555, + "step": 37750 + }, + { + "epoch": 9.88, + "learning_rate": 0.0001, + "loss": 0.163, + "step": 37760 + }, + { + "epoch": 9.88, + "learning_rate": 0.0001, + "loss": 0.1547, + "step": 37770 + }, + { + "epoch": 9.88, + "learning_rate": 0.0001, + "loss": 0.1628, + "step": 37780 + }, + { + "epoch": 9.89, + "learning_rate": 0.0001, + "loss": 0.155, + "step": 37790 + }, + { + "epoch": 9.89, + "learning_rate": 0.0001, + "loss": 0.1631, + "step": 37800 + }, + { + "epoch": 9.89, + "learning_rate": 0.0001, + "loss": 0.1618, + "step": 37810 + }, + { + "epoch": 9.9, + "learning_rate": 0.0001, + "loss": 0.1596, + "step": 37820 + }, + { + "epoch": 9.9, + "learning_rate": 0.0001, + "loss": 0.1611, + "step": 37830 + }, + { + "epoch": 9.9, + "learning_rate": 0.0001, + "loss": 0.1615, + "step": 37840 + }, + { + "epoch": 9.9, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 37850 + }, + { + "epoch": 9.91, + "learning_rate": 0.0001, + "loss": 0.1613, + "step": 37860 + }, + { + "epoch": 9.91, + "learning_rate": 0.0001, + "loss": 0.1562, + "step": 37870 + }, + { + "epoch": 9.91, + "learning_rate": 0.0001, + "loss": 0.1547, + "step": 37880 + }, + { + "epoch": 9.91, + "learning_rate": 0.0001, + "loss": 0.1507, + "step": 37890 + }, + { + "epoch": 9.92, + "learning_rate": 0.0001, + "loss": 0.1608, + "step": 37900 + }, + { + "epoch": 9.92, + "learning_rate": 0.0001, + "loss": 0.1607, + "step": 37910 + }, + { + "epoch": 9.92, + "learning_rate": 0.0001, + "loss": 0.1607, + "step": 37920 + }, + { + "epoch": 9.92, + "learning_rate": 0.0001, + "loss": 0.1594, + "step": 37930 + }, + { + "epoch": 9.93, + "learning_rate": 0.0001, + "loss": 0.1646, + "step": 37940 + }, + { + "epoch": 9.93, + "learning_rate": 0.0001, + "loss": 0.1574, + "step": 37950 + }, + { + "epoch": 9.93, + "learning_rate": 0.0001, + "loss": 0.1565, + "step": 37960 + }, + { + "epoch": 9.93, + "learning_rate": 0.0001, + "loss": 0.1629, + "step": 37970 + }, + { + "epoch": 9.94, + "learning_rate": 0.0001, + "loss": 0.1534, + "step": 37980 + }, + { + "epoch": 9.94, + "learning_rate": 0.0001, + "loss": 0.157, + "step": 37990 + }, + { + "epoch": 9.94, + "learning_rate": 0.0001, + "loss": 0.1617, + "step": 38000 + }, + { + "epoch": 9.95, + "learning_rate": 0.0001, + "loss": 0.1586, + "step": 38010 + }, + { + "epoch": 9.95, + "learning_rate": 0.0001, + "loss": 0.1652, + "step": 38020 + }, + { + "epoch": 9.95, + "learning_rate": 0.0001, + "loss": 0.1593, + "step": 38030 + }, + { + "epoch": 9.95, + "learning_rate": 0.0001, + "loss": 0.1585, + "step": 38040 + }, + { + "epoch": 9.96, + "learning_rate": 0.0001, + "loss": 0.1597, + "step": 38050 + }, + { + "epoch": 9.96, + "learning_rate": 0.0001, + "loss": 0.1524, + "step": 38060 + }, + { + "epoch": 9.96, + "learning_rate": 0.0001, + "loss": 0.1663, + "step": 38070 + }, + { + "epoch": 9.96, + "learning_rate": 0.0001, + "loss": 0.1654, + "step": 38080 + }, + { + "epoch": 9.97, + "learning_rate": 0.0001, + "loss": 0.1529, + "step": 38090 + }, + { + "epoch": 9.97, + "learning_rate": 0.0001, + "loss": 0.1573, + "step": 38100 + }, + { + "epoch": 9.97, + "learning_rate": 0.0001, + "loss": 0.1563, + "step": 38110 + }, + { + "epoch": 9.97, + "learning_rate": 0.0001, + "loss": 0.1637, + "step": 38120 + }, + { + "epoch": 9.98, + "learning_rate": 0.0001, + "loss": 0.1591, + "step": 38130 + }, + { + "epoch": 9.98, + "learning_rate": 0.0001, + "loss": 0.1593, + "step": 38140 + }, + { + "epoch": 9.98, + "learning_rate": 0.0001, + "loss": 0.1595, + "step": 38150 + }, + { + "epoch": 9.98, + "learning_rate": 0.0001, + "loss": 0.1617, + "step": 38160 + }, + { + "epoch": 9.99, + "learning_rate": 0.0001, + "loss": 0.1534, + "step": 38170 + }, + { + "epoch": 9.99, + "learning_rate": 0.0001, + "loss": 0.1524, + "step": 38180 + }, + { + "epoch": 9.99, + "learning_rate": 0.0001, + "loss": 0.1547, + "step": 38190 + }, + { + "epoch": 9.99, + "learning_rate": 0.0001, + "loss": 0.16, + "step": 38200 + }, + { + "epoch": 10.0, + "learning_rate": 0.0001, + "loss": 0.1575, + "step": 38210 + }, + { + "epoch": 10.0, + "learning_rate": 0.0001, + "loss": 0.151, + "step": 38220 + }, + { + "epoch": 10.0, + "eval_gen_len": 61.8111, + "eval_loss": 1.3773375749588013, + "eval_rouge1": 57.0069, + "eval_rouge2": 34.8854, + "eval_rougeL": 46.5402, + "eval_rougeLsum": 52.6352, + "eval_runtime": 38.376, + "eval_samples_per_second": 22.905, + "eval_steps_per_second": 0.365, + "step": 38220 + }, + { + "epoch": 10.0, + "step": 38220, + "total_flos": 1.1269919104682689e+19, + "train_loss": 0.28210093728663843, + "train_runtime": 21334.6616, + "train_samples_per_second": 114.625, + "train_steps_per_second": 1.791 + } + ], + "max_steps": 38220, + "num_train_epochs": 10, + "total_flos": 1.1269919104682689e+19, + "trial_name": null, + "trial_params": null +}