| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9249089542748136, | |
| "eval_steps": 500, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4e-05, | |
| "loss": 1.0185, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-05, | |
| "loss": 0.8649, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00012, | |
| "loss": 0.7784, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00016, | |
| "loss": 0.7386, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 0.7037, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001990530303030303, | |
| "loss": 0.7019, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001981060606060606, | |
| "loss": 0.7117, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019715909090909094, | |
| "loss": 0.672, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019621212121212123, | |
| "loss": 0.664, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019526515151515152, | |
| "loss": 0.6666, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001943181818181818, | |
| "loss": 0.6685, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019337121212121213, | |
| "loss": 0.6788, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019242424242424245, | |
| "loss": 0.6673, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019147727272727274, | |
| "loss": 0.6628, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019053030303030303, | |
| "loss": 0.6643, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00018958333333333332, | |
| "loss": 0.6607, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00018863636363636364, | |
| "loss": 0.6706, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00018768939393939396, | |
| "loss": 0.6709, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00018674242424242425, | |
| "loss": 0.6616, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00018579545454545454, | |
| "loss": 0.6566, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00018484848484848484, | |
| "loss": 0.6513, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00018390151515151518, | |
| "loss": 0.6797, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00018295454545454547, | |
| "loss": 0.6599, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00018200757575757577, | |
| "loss": 0.6561, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00018106060606060606, | |
| "loss": 0.662, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00018011363636363638, | |
| "loss": 0.6629, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0001791666666666667, | |
| "loss": 0.6475, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00017821969696969699, | |
| "loss": 0.6607, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00017727272727272728, | |
| "loss": 0.6512, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017632575757575757, | |
| "loss": 0.6484, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001753787878787879, | |
| "loss": 0.6403, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001744318181818182, | |
| "loss": 0.6537, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001734848484848485, | |
| "loss": 0.6516, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001725378787878788, | |
| "loss": 0.6577, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00017159090909090908, | |
| "loss": 0.6374, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001706439393939394, | |
| "loss": 0.6551, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00016969696969696972, | |
| "loss": 0.6388, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00016875, | |
| "loss": 0.64, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001678030303030303, | |
| "loss": 0.6579, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001668560606060606, | |
| "loss": 0.6525, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00016590909090909094, | |
| "loss": 0.6261, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00016496212121212123, | |
| "loss": 0.6351, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00016401515151515152, | |
| "loss": 0.6537, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0001630681818181818, | |
| "loss": 0.6448, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00016212121212121213, | |
| "loss": 0.638, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00016117424242424245, | |
| "loss": 0.6503, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00016022727272727274, | |
| "loss": 0.6378, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00015928030303030303, | |
| "loss": 0.643, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00015833333333333332, | |
| "loss": 0.6235, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00015738636363636364, | |
| "loss": 0.647, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00015643939393939396, | |
| "loss": 0.6408, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00015549242424242425, | |
| "loss": 0.6391, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00015454545454545454, | |
| "loss": 0.6356, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00015359848484848484, | |
| "loss": 0.6317, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00015265151515151515, | |
| "loss": 0.6413, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00015170454545454547, | |
| "loss": 0.6338, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00015075757575757576, | |
| "loss": 0.6422, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00014981060606060606, | |
| "loss": 0.6442, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00014886363636363635, | |
| "loss": 0.6523, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0001479166666666667, | |
| "loss": 0.6349, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00014696969696969698, | |
| "loss": 0.6389, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014602272727272728, | |
| "loss": 0.6468, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014507575757575757, | |
| "loss": 0.6431, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00014412878787878789, | |
| "loss": 0.6287, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001431818181818182, | |
| "loss": 0.6438, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0001422348484848485, | |
| "loss": 0.6274, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0001412878787878788, | |
| "loss": 0.6286, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00014034090909090908, | |
| "loss": 0.6401, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001393939393939394, | |
| "loss": 0.6472, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00013844696969696972, | |
| "loss": 0.6458, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001375, | |
| "loss": 0.6117, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001365530303030303, | |
| "loss": 0.6271, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0001356060606060606, | |
| "loss": 0.6287, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00013465909090909094, | |
| "loss": 0.6336, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00013371212121212123, | |
| "loss": 0.6404, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00013276515151515152, | |
| "loss": 0.6312, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0001318181818181818, | |
| "loss": 0.6168, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00013087121212121213, | |
| "loss": 0.6272, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012992424242424245, | |
| "loss": 0.6477, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012897727272727274, | |
| "loss": 0.6477, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012803030303030303, | |
| "loss": 0.6227, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00012708333333333332, | |
| "loss": 0.6224, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00012613636363636364, | |
| "loss": 0.6315, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00012518939393939396, | |
| "loss": 0.631, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00012424242424242425, | |
| "loss": 0.6285, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00012329545454545454, | |
| "loss": 0.6359, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00012234848484848484, | |
| "loss": 0.6282, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00012140151515151517, | |
| "loss": 0.6196, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00012045454545454546, | |
| "loss": 0.6346, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00011950757575757576, | |
| "loss": 0.6323, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00011856060606060606, | |
| "loss": 0.6108, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00011761363636363636, | |
| "loss": 0.6324, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00011666666666666668, | |
| "loss": 0.618, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00011571969696969698, | |
| "loss": 0.6099, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00011477272727272728, | |
| "loss": 0.6251, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00011382575757575758, | |
| "loss": 0.6209, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0001128787878787879, | |
| "loss": 0.6218, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00011193181818181819, | |
| "loss": 0.6299, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0001109848484848485, | |
| "loss": 0.6211, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00011003787878787879, | |
| "loss": 0.6072, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00010909090909090909, | |
| "loss": 0.6264, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00010814393939393941, | |
| "loss": 0.6248, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0001071969696969697, | |
| "loss": 0.6125, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00010625000000000001, | |
| "loss": 0.6294, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0001053030303030303, | |
| "loss": 0.6193, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0001043560606060606, | |
| "loss": 0.6293, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010340909090909092, | |
| "loss": 0.629, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00010246212121212121, | |
| "loss": 0.6353, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00010151515151515152, | |
| "loss": 0.6268, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00010056818181818181, | |
| "loss": 0.6256, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.962121212121213e-05, | |
| "loss": 0.6404, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.867424242424242e-05, | |
| "loss": 0.6347, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.772727272727274e-05, | |
| "loss": 0.6317, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.678030303030303e-05, | |
| "loss": 0.6335, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.583333333333334e-05, | |
| "loss": 0.6239, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.488636363636364e-05, | |
| "loss": 0.6302, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.393939393939395e-05, | |
| "loss": 0.6182, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.299242424242425e-05, | |
| "loss": 0.6219, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.204545454545454e-05, | |
| "loss": 0.6229, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.109848484848486e-05, | |
| "loss": 0.6413, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.015151515151515e-05, | |
| "loss": 0.6237, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.920454545454546e-05, | |
| "loss": 0.6397, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.825757575757576e-05, | |
| "loss": 0.6259, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.731060606060605e-05, | |
| "loss": 0.634, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.636363636363637e-05, | |
| "loss": 0.622, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.541666666666666e-05, | |
| "loss": 0.6279, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.446969696969697e-05, | |
| "loss": 0.6306, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.352272727272727e-05, | |
| "loss": 0.6288, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.257575757575758e-05, | |
| "loss": 0.6297, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.162878787878789e-05, | |
| "loss": 0.6119, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.068181818181818e-05, | |
| "loss": 0.6227, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.97348484848485e-05, | |
| "loss": 0.6317, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.878787878787879e-05, | |
| "loss": 0.619, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.784090909090909e-05, | |
| "loss": 0.6156, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.68939393939394e-05, | |
| "loss": 0.6116, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.59469696969697e-05, | |
| "loss": 0.6009, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.6226, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.40530303030303e-05, | |
| "loss": 0.6246, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.310606060606062e-05, | |
| "loss": 0.6035, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.215909090909091e-05, | |
| "loss": 0.6059, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.121212121212121e-05, | |
| "loss": 0.6387, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.026515151515152e-05, | |
| "loss": 0.6314, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.931818181818182e-05, | |
| "loss": 0.6246, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 6.837121212121213e-05, | |
| "loss": 0.6448, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 6.742424242424242e-05, | |
| "loss": 0.5997, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.647727272727274e-05, | |
| "loss": 0.6238, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.553030303030303e-05, | |
| "loss": 0.614, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.458333333333334e-05, | |
| "loss": 0.6162, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.363636363636364e-05, | |
| "loss": 0.6087, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.268939393939395e-05, | |
| "loss": 0.6215, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.174242424242425e-05, | |
| "loss": 0.6022, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.079545454545454e-05, | |
| "loss": 0.6294, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.9848484848484854e-05, | |
| "loss": 0.6187, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.890151515151515e-05, | |
| "loss": 0.6194, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.7954545454545464e-05, | |
| "loss": 0.6109, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.700757575757576e-05, | |
| "loss": 0.638, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.606060606060606e-05, | |
| "loss": 0.6156, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.5113636363636366e-05, | |
| "loss": 0.6209, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.4166666666666664e-05, | |
| "loss": 0.6068, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.3219696969696976e-05, | |
| "loss": 0.6161, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.2272727272727274e-05, | |
| "loss": 0.6354, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.132575757575758e-05, | |
| "loss": 0.6317, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.037878787878788e-05, | |
| "loss": 0.6227, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.943181818181818e-05, | |
| "loss": 0.6176, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.848484848484849e-05, | |
| "loss": 0.6274, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.753787878787879e-05, | |
| "loss": 0.6159, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.659090909090909e-05, | |
| "loss": 0.6093, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.5643939393939396e-05, | |
| "loss": 0.6166, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.46969696969697e-05, | |
| "loss": 0.6046, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.6297, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.2803030303030305e-05, | |
| "loss": 0.6106, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.185606060606061e-05, | |
| "loss": 0.6064, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.0909090909090915e-05, | |
| "loss": 0.6138, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.996212121212121e-05, | |
| "loss": 0.6102, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.901515151515152e-05, | |
| "loss": 0.61, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.8068181818181816e-05, | |
| "loss": 0.6107, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.712121212121212e-05, | |
| "loss": 0.6086, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.6174242424242427e-05, | |
| "loss": 0.6163, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.522727272727273e-05, | |
| "loss": 0.636, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.428030303030303e-05, | |
| "loss": 0.6237, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 0.6187, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.238636363636364e-05, | |
| "loss": 0.6195, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.143939393939394e-05, | |
| "loss": 0.6333, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.0492424242424243e-05, | |
| "loss": 0.613, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.954545454545455e-05, | |
| "loss": 0.6452, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8598484848484853e-05, | |
| "loss": 0.6202, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.7651515151515152e-05, | |
| "loss": 0.6159, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.6704545454545453e-05, | |
| "loss": 0.6231, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.575757575757576e-05, | |
| "loss": 0.6287, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.4810606060606064e-05, | |
| "loss": 0.6168, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3863636363636365e-05, | |
| "loss": 0.6125, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.2916666666666667e-05, | |
| "loss": 0.6122, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.1969696969696972e-05, | |
| "loss": 0.6087, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.1022727272727274e-05, | |
| "loss": 0.6207, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.0075757575757575e-05, | |
| "loss": 0.6175, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.912878787878788e-05, | |
| "loss": 0.6029, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 0.6183, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7234848484848487e-05, | |
| "loss": 0.6214, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.628787878787879e-05, | |
| "loss": 0.6314, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.534090909090909e-05, | |
| "loss": 0.6215, | |
| "step": 4000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 4324, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 6.358088323636838e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |