{ "best_metric": 1.3524357080459595, "best_model_checkpoint": "models/EyeBERTv10/weights/checkpoint-100000", "epoch": 0.13923969556632962, "global_step": 100000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.99999375e-05, "loss": 10.4215, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.9999375000000006e-05, "loss": 9.7265, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.999875e-05, "loss": 9.2712, "step": 20 }, { "epoch": 0.0, "learning_rate": 4.9998125e-05, "loss": 8.7031, "step": 30 }, { "epoch": 0.0, "learning_rate": 4.9997500000000004e-05, "loss": 8.4228, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.999687500000001e-05, "loss": 8.0989, "step": 50 }, { "epoch": 0.0, "learning_rate": 4.999625e-05, "loss": 7.8738, "step": 60 }, { "epoch": 0.0, "learning_rate": 4.9995625e-05, "loss": 7.6055, "step": 70 }, { "epoch": 0.0, "learning_rate": 4.9995000000000005e-05, "loss": 7.2637, "step": 80 }, { "epoch": 0.0, "learning_rate": 4.9994375e-05, "loss": 7.2769, "step": 90 }, { "epoch": 0.0, "learning_rate": 4.999375e-05, "loss": 7.1161, "step": 100 }, { "epoch": 0.0, "learning_rate": 4.9993125e-05, "loss": 7.072, "step": 110 }, { "epoch": 0.0, "learning_rate": 4.9992500000000006e-05, "loss": 6.9431, "step": 120 }, { "epoch": 0.0, "learning_rate": 4.9991875e-05, "loss": 6.9566, "step": 130 }, { "epoch": 0.0, "learning_rate": 4.999125e-05, "loss": 7.0397, "step": 140 }, { "epoch": 0.0, "learning_rate": 4.9990625000000004e-05, "loss": 7.1445, "step": 150 }, { "epoch": 0.0, "learning_rate": 4.999e-05, "loss": 6.8521, "step": 160 }, { "epoch": 0.0, "learning_rate": 4.9989375000000005e-05, "loss": 6.8662, "step": 170 }, { "epoch": 0.0, "learning_rate": 4.998875e-05, "loss": 6.9422, "step": 180 }, { "epoch": 0.0, "learning_rate": 4.9988125000000005e-05, "loss": 6.8326, "step": 190 }, { "epoch": 0.0, "learning_rate": 4.99875e-05, "loss": 6.8945, "step": 200 }, { "epoch": 0.0, "learning_rate": 4.9986875e-05, "loss": 6.7354, "step": 210 }, { "epoch": 0.0, "learning_rate": 4.998625e-05, "loss": 6.8377, "step": 220 }, { "epoch": 0.0, "learning_rate": 4.9985625e-05, "loss": 6.8651, "step": 230 }, { "epoch": 0.0, "learning_rate": 4.9985e-05, "loss": 6.6876, "step": 240 }, { "epoch": 0.0, "learning_rate": 4.9984375e-05, "loss": 6.7303, "step": 250 }, { "epoch": 0.0, "learning_rate": 4.9983750000000004e-05, "loss": 6.7041, "step": 260 }, { "epoch": 0.0, "learning_rate": 4.9983125e-05, "loss": 6.807, "step": 270 }, { "epoch": 0.0, "learning_rate": 4.9982500000000005e-05, "loss": 6.7255, "step": 280 }, { "epoch": 0.0, "learning_rate": 4.9981875e-05, "loss": 7.1604, "step": 290 }, { "epoch": 0.0, "learning_rate": 4.998125e-05, "loss": 6.7561, "step": 300 }, { "epoch": 0.0, "learning_rate": 4.9980625e-05, "loss": 6.7432, "step": 310 }, { "epoch": 0.0, "learning_rate": 4.9980000000000006e-05, "loss": 6.7885, "step": 320 }, { "epoch": 0.0, "learning_rate": 4.9979375e-05, "loss": 7.2084, "step": 330 }, { "epoch": 0.0, "learning_rate": 4.997875e-05, "loss": 6.7919, "step": 340 }, { "epoch": 0.0, "learning_rate": 4.9978125000000004e-05, "loss": 6.7852, "step": 350 }, { "epoch": 0.0, "learning_rate": 4.997750000000001e-05, "loss": 6.8428, "step": 360 }, { "epoch": 0.0, "learning_rate": 4.9976875e-05, "loss": 6.7765, "step": 370 }, { "epoch": 0.0, "learning_rate": 4.997625e-05, "loss": 6.8604, "step": 380 }, { "epoch": 0.0, "learning_rate": 4.9975625000000005e-05, "loss": 6.8293, "step": 390 }, { "epoch": 0.0, "learning_rate": 4.9975e-05, "loss": 6.7377, "step": 400 }, { "epoch": 0.0, "learning_rate": 4.9974375e-05, "loss": 6.77, "step": 410 }, { "epoch": 0.0, "learning_rate": 4.997375e-05, "loss": 6.7213, "step": 420 }, { "epoch": 0.0, "learning_rate": 4.9973125000000006e-05, "loss": 6.7387, "step": 430 }, { "epoch": 0.0, "learning_rate": 4.99725e-05, "loss": 6.5909, "step": 440 }, { "epoch": 0.0, "learning_rate": 4.9971875e-05, "loss": 6.8196, "step": 450 }, { "epoch": 0.0, "learning_rate": 4.9971250000000004e-05, "loss": 6.7757, "step": 460 }, { "epoch": 0.0, "learning_rate": 4.997062500000001e-05, "loss": 6.7628, "step": 470 }, { "epoch": 0.0, "learning_rate": 4.997e-05, "loss": 6.7206, "step": 480 }, { "epoch": 0.0, "learning_rate": 4.9969375e-05, "loss": 6.591, "step": 490 }, { "epoch": 0.0, "learning_rate": 4.9968750000000005e-05, "loss": 6.6129, "step": 500 }, { "epoch": 0.0, "learning_rate": 4.9968125e-05, "loss": 6.7242, "step": 510 }, { "epoch": 0.0, "learning_rate": 4.99675e-05, "loss": 6.6734, "step": 520 }, { "epoch": 0.0, "learning_rate": 4.9966875e-05, "loss": 6.6125, "step": 530 }, { "epoch": 0.0, "learning_rate": 4.9966250000000006e-05, "loss": 6.6411, "step": 540 }, { "epoch": 0.0, "learning_rate": 4.9965625e-05, "loss": 6.4995, "step": 550 }, { "epoch": 0.0, "learning_rate": 4.9965e-05, "loss": 6.675, "step": 560 }, { "epoch": 0.0, "learning_rate": 4.9964375000000004e-05, "loss": 6.3114, "step": 570 }, { "epoch": 0.0, "learning_rate": 4.996375e-05, "loss": 6.4305, "step": 580 }, { "epoch": 0.0, "learning_rate": 4.9963125000000005e-05, "loss": 6.405, "step": 590 }, { "epoch": 0.0, "learning_rate": 4.99625e-05, "loss": 6.4082, "step": 600 }, { "epoch": 0.0, "learning_rate": 4.9961875000000005e-05, "loss": 6.5629, "step": 610 }, { "epoch": 0.0, "learning_rate": 4.996125e-05, "loss": 6.3581, "step": 620 }, { "epoch": 0.0, "learning_rate": 4.9960625e-05, "loss": 6.323, "step": 630 }, { "epoch": 0.0, "learning_rate": 4.996e-05, "loss": 6.2987, "step": 640 }, { "epoch": 0.0, "learning_rate": 4.9959375e-05, "loss": 6.3087, "step": 650 }, { "epoch": 0.0, "learning_rate": 4.9958750000000003e-05, "loss": 6.5717, "step": 660 }, { "epoch": 0.0, "learning_rate": 4.9958125e-05, "loss": 6.1976, "step": 670 }, { "epoch": 0.0, "learning_rate": 4.9957500000000004e-05, "loss": 6.375, "step": 680 }, { "epoch": 0.0, "learning_rate": 4.9956875e-05, "loss": 6.263, "step": 690 }, { "epoch": 0.0, "learning_rate": 4.9956250000000005e-05, "loss": 6.4501, "step": 700 }, { "epoch": 0.0, "learning_rate": 4.9955625e-05, "loss": 6.0482, "step": 710 }, { "epoch": 0.0, "learning_rate": 4.9955e-05, "loss": 6.4436, "step": 720 }, { "epoch": 0.0, "learning_rate": 4.9954375e-05, "loss": 6.2933, "step": 730 }, { "epoch": 0.0, "learning_rate": 4.9953750000000006e-05, "loss": 5.9614, "step": 740 }, { "epoch": 0.0, "learning_rate": 4.9953125e-05, "loss": 6.3687, "step": 750 }, { "epoch": 0.0, "learning_rate": 4.99525e-05, "loss": 6.2533, "step": 760 }, { "epoch": 0.0, "learning_rate": 4.9951875000000004e-05, "loss": 6.3574, "step": 770 }, { "epoch": 0.0, "learning_rate": 4.995125000000001e-05, "loss": 6.1889, "step": 780 }, { "epoch": 0.0, "learning_rate": 4.9950625000000004e-05, "loss": 6.2137, "step": 790 }, { "epoch": 0.0, "learning_rate": 4.995e-05, "loss": 6.3349, "step": 800 }, { "epoch": 0.0, "learning_rate": 4.9949375000000005e-05, "loss": 6.0768, "step": 810 }, { "epoch": 0.0, "learning_rate": 4.994875e-05, "loss": 6.2237, "step": 820 }, { "epoch": 0.0, "learning_rate": 4.9948125e-05, "loss": 6.3743, "step": 830 }, { "epoch": 0.0, "learning_rate": 4.99475e-05, "loss": 6.1115, "step": 840 }, { "epoch": 0.0, "learning_rate": 4.9946875000000006e-05, "loss": 6.2327, "step": 850 }, { "epoch": 0.0, "learning_rate": 4.994625e-05, "loss": 6.1463, "step": 860 }, { "epoch": 0.0, "learning_rate": 4.9945625e-05, "loss": 6.1507, "step": 870 }, { "epoch": 0.0, "learning_rate": 4.9945000000000004e-05, "loss": 6.1983, "step": 880 }, { "epoch": 0.0, "learning_rate": 4.994437500000001e-05, "loss": 6.04, "step": 890 }, { "epoch": 0.0, "learning_rate": 4.994375e-05, "loss": 6.1951, "step": 900 }, { "epoch": 0.0, "learning_rate": 4.9943125e-05, "loss": 6.1546, "step": 910 }, { "epoch": 0.0, "learning_rate": 4.9942500000000005e-05, "loss": 5.9664, "step": 920 }, { "epoch": 0.0, "learning_rate": 4.9941875e-05, "loss": 6.1142, "step": 930 }, { "epoch": 0.0, "learning_rate": 4.994125e-05, "loss": 5.982, "step": 940 }, { "epoch": 0.0, "learning_rate": 4.9940625e-05, "loss": 6.1539, "step": 950 }, { "epoch": 0.0, "learning_rate": 4.9940000000000006e-05, "loss": 6.1975, "step": 960 }, { "epoch": 0.0, "learning_rate": 4.9939375e-05, "loss": 6.2161, "step": 970 }, { "epoch": 0.0, "learning_rate": 4.993875e-05, "loss": 6.2377, "step": 980 }, { "epoch": 0.0, "learning_rate": 4.9938125000000004e-05, "loss": 5.9643, "step": 990 }, { "epoch": 0.0, "learning_rate": 4.99375e-05, "loss": 5.8701, "step": 1000 }, { "epoch": 0.0, "eval_loss": 6.044287204742432, "eval_runtime": 1290.7405, "eval_samples_per_second": 44.963, "eval_steps_per_second": 5.621, "step": 1000 }, { "epoch": 0.0, "learning_rate": 4.9936875e-05, "loss": 6.1357, "step": 1010 }, { "epoch": 0.0, "learning_rate": 4.993625e-05, "loss": 6.0856, "step": 1020 }, { "epoch": 0.0, "learning_rate": 4.9935625000000005e-05, "loss": 6.1778, "step": 1030 }, { "epoch": 0.0, "learning_rate": 4.9935e-05, "loss": 6.042, "step": 1040 }, { "epoch": 0.0, "learning_rate": 4.9934375e-05, "loss": 6.0345, "step": 1050 }, { "epoch": 0.0, "learning_rate": 4.993375e-05, "loss": 6.1273, "step": 1060 }, { "epoch": 0.0, "learning_rate": 4.9933125e-05, "loss": 6.1635, "step": 1070 }, { "epoch": 0.0, "learning_rate": 4.9932500000000003e-05, "loss": 6.1199, "step": 1080 }, { "epoch": 0.0, "learning_rate": 4.9931875e-05, "loss": 6.1156, "step": 1090 }, { "epoch": 0.0, "learning_rate": 4.9931250000000004e-05, "loss": 5.9087, "step": 1100 }, { "epoch": 0.0, "learning_rate": 4.9930625e-05, "loss": 5.7586, "step": 1110 }, { "epoch": 0.0, "learning_rate": 4.9930000000000005e-05, "loss": 5.8263, "step": 1120 }, { "epoch": 0.0, "learning_rate": 4.9929375e-05, "loss": 5.7533, "step": 1130 }, { "epoch": 0.0, "learning_rate": 4.992875e-05, "loss": 6.0566, "step": 1140 }, { "epoch": 0.0, "learning_rate": 4.9928125e-05, "loss": 5.9131, "step": 1150 }, { "epoch": 0.0, "learning_rate": 4.9927500000000006e-05, "loss": 6.1455, "step": 1160 }, { "epoch": 0.0, "learning_rate": 4.9926875e-05, "loss": 5.9283, "step": 1170 }, { "epoch": 0.0, "learning_rate": 4.992625e-05, "loss": 6.0183, "step": 1180 }, { "epoch": 0.0, "learning_rate": 4.9925625000000004e-05, "loss": 5.9118, "step": 1190 }, { "epoch": 0.0, "learning_rate": 4.992500000000001e-05, "loss": 6.0648, "step": 1200 }, { "epoch": 0.0, "learning_rate": 4.9924375000000004e-05, "loss": 6.003, "step": 1210 }, { "epoch": 0.0, "learning_rate": 4.992375e-05, "loss": 5.8928, "step": 1220 }, { "epoch": 0.0, "learning_rate": 4.9923125000000005e-05, "loss": 6.041, "step": 1230 }, { "epoch": 0.0, "learning_rate": 4.99225e-05, "loss": 6.1297, "step": 1240 }, { "epoch": 0.0, "learning_rate": 4.9921875e-05, "loss": 5.8835, "step": 1250 }, { "epoch": 0.0, "learning_rate": 4.992125e-05, "loss": 5.805, "step": 1260 }, { "epoch": 0.0, "learning_rate": 4.9920625000000006e-05, "loss": 6.0591, "step": 1270 }, { "epoch": 0.0, "learning_rate": 4.992e-05, "loss": 5.8117, "step": 1280 }, { "epoch": 0.0, "learning_rate": 4.9919375e-05, "loss": 6.0531, "step": 1290 }, { "epoch": 0.0, "learning_rate": 4.9918750000000004e-05, "loss": 5.8621, "step": 1300 }, { "epoch": 0.0, "learning_rate": 4.991812500000001e-05, "loss": 5.9835, "step": 1310 }, { "epoch": 0.0, "learning_rate": 4.99175e-05, "loss": 6.0515, "step": 1320 }, { "epoch": 0.0, "learning_rate": 4.9916875e-05, "loss": 5.7792, "step": 1330 }, { "epoch": 0.0, "learning_rate": 4.9916250000000005e-05, "loss": 5.9157, "step": 1340 }, { "epoch": 0.0, "learning_rate": 4.9915625e-05, "loss": 5.9257, "step": 1350 }, { "epoch": 0.0, "learning_rate": 4.9915e-05, "loss": 6.091, "step": 1360 }, { "epoch": 0.0, "learning_rate": 4.9914375e-05, "loss": 5.7793, "step": 1370 }, { "epoch": 0.0, "learning_rate": 4.9913750000000006e-05, "loss": 5.8409, "step": 1380 }, { "epoch": 0.0, "learning_rate": 4.9913125e-05, "loss": 5.5957, "step": 1390 }, { "epoch": 0.0, "learning_rate": 4.99125e-05, "loss": 5.5886, "step": 1400 }, { "epoch": 0.0, "learning_rate": 4.9911875000000004e-05, "loss": 5.759, "step": 1410 }, { "epoch": 0.0, "learning_rate": 4.991125e-05, "loss": 5.8165, "step": 1420 }, { "epoch": 0.0, "learning_rate": 4.9910625e-05, "loss": 5.8275, "step": 1430 }, { "epoch": 0.0, "learning_rate": 4.991e-05, "loss": 5.9068, "step": 1440 }, { "epoch": 0.0, "learning_rate": 4.9909375000000005e-05, "loss": 5.6743, "step": 1450 }, { "epoch": 0.0, "learning_rate": 4.990875e-05, "loss": 5.6995, "step": 1460 }, { "epoch": 0.0, "learning_rate": 4.9908125e-05, "loss": 5.7515, "step": 1470 }, { "epoch": 0.0, "learning_rate": 4.99075e-05, "loss": 5.8694, "step": 1480 }, { "epoch": 0.0, "learning_rate": 4.9906875e-05, "loss": 5.7796, "step": 1490 }, { "epoch": 0.0, "learning_rate": 4.9906250000000004e-05, "loss": 5.6033, "step": 1500 }, { "epoch": 0.0, "learning_rate": 4.9905625e-05, "loss": 5.9326, "step": 1510 }, { "epoch": 0.0, "learning_rate": 4.9905000000000004e-05, "loss": 5.8005, "step": 1520 }, { "epoch": 0.0, "learning_rate": 4.9904375e-05, "loss": 5.4725, "step": 1530 }, { "epoch": 0.0, "learning_rate": 4.9903750000000005e-05, "loss": 5.8301, "step": 1540 }, { "epoch": 0.0, "learning_rate": 4.9903125e-05, "loss": 5.5641, "step": 1550 }, { "epoch": 0.0, "learning_rate": 4.99025e-05, "loss": 5.8324, "step": 1560 }, { "epoch": 0.0, "learning_rate": 4.9901875e-05, "loss": 5.7555, "step": 1570 }, { "epoch": 0.0, "learning_rate": 4.9901250000000006e-05, "loss": 5.6391, "step": 1580 }, { "epoch": 0.0, "learning_rate": 4.9900625e-05, "loss": 5.9191, "step": 1590 }, { "epoch": 0.0, "learning_rate": 4.99e-05, "loss": 5.6807, "step": 1600 }, { "epoch": 0.0, "learning_rate": 4.9899375000000004e-05, "loss": 5.9131, "step": 1610 }, { "epoch": 0.0, "learning_rate": 4.989875e-05, "loss": 5.6132, "step": 1620 }, { "epoch": 0.0, "learning_rate": 4.9898125000000004e-05, "loss": 5.867, "step": 1630 }, { "epoch": 0.0, "learning_rate": 4.98975e-05, "loss": 5.8107, "step": 1640 }, { "epoch": 0.0, "learning_rate": 4.9896875000000005e-05, "loss": 5.7436, "step": 1650 }, { "epoch": 0.0, "learning_rate": 4.989625e-05, "loss": 5.8651, "step": 1660 }, { "epoch": 0.0, "learning_rate": 4.9895625e-05, "loss": 5.6806, "step": 1670 }, { "epoch": 0.0, "learning_rate": 4.9895e-05, "loss": 5.7277, "step": 1680 }, { "epoch": 0.0, "learning_rate": 4.9894375000000006e-05, "loss": 5.7747, "step": 1690 }, { "epoch": 0.0, "learning_rate": 4.989375e-05, "loss": 5.7684, "step": 1700 }, { "epoch": 0.0, "learning_rate": 4.9893125e-05, "loss": 5.8705, "step": 1710 }, { "epoch": 0.0, "learning_rate": 4.9892500000000004e-05, "loss": 5.7197, "step": 1720 }, { "epoch": 0.0, "learning_rate": 4.989187500000001e-05, "loss": 5.7807, "step": 1730 }, { "epoch": 0.0, "learning_rate": 4.989125e-05, "loss": 5.734, "step": 1740 }, { "epoch": 0.0, "learning_rate": 4.9890625e-05, "loss": 5.7511, "step": 1750 }, { "epoch": 0.0, "learning_rate": 4.9890000000000005e-05, "loss": 5.8212, "step": 1760 }, { "epoch": 0.0, "learning_rate": 4.9889375e-05, "loss": 5.6603, "step": 1770 }, { "epoch": 0.0, "learning_rate": 4.988875e-05, "loss": 5.5919, "step": 1780 }, { "epoch": 0.0, "learning_rate": 4.9888125e-05, "loss": 5.5897, "step": 1790 }, { "epoch": 0.0, "learning_rate": 4.9887500000000006e-05, "loss": 5.7401, "step": 1800 }, { "epoch": 0.0, "learning_rate": 4.9886875e-05, "loss": 5.5912, "step": 1810 }, { "epoch": 0.0, "learning_rate": 4.988625e-05, "loss": 5.8376, "step": 1820 }, { "epoch": 0.0, "learning_rate": 4.9885625000000004e-05, "loss": 5.7624, "step": 1830 }, { "epoch": 0.0, "learning_rate": 4.9885e-05, "loss": 5.7179, "step": 1840 }, { "epoch": 0.0, "learning_rate": 4.9884375e-05, "loss": 5.6135, "step": 1850 }, { "epoch": 0.0, "learning_rate": 4.988375e-05, "loss": 5.8143, "step": 1860 }, { "epoch": 0.0, "learning_rate": 4.9883125000000005e-05, "loss": 5.5802, "step": 1870 }, { "epoch": 0.0, "learning_rate": 4.98825e-05, "loss": 5.7082, "step": 1880 }, { "epoch": 0.0, "learning_rate": 4.9881875e-05, "loss": 5.718, "step": 1890 }, { "epoch": 0.0, "learning_rate": 4.988125e-05, "loss": 5.6469, "step": 1900 }, { "epoch": 0.0, "learning_rate": 4.9880625e-05, "loss": 5.8176, "step": 1910 }, { "epoch": 0.0, "learning_rate": 4.9880000000000004e-05, "loss": 5.6331, "step": 1920 }, { "epoch": 0.0, "learning_rate": 4.9879375e-05, "loss": 5.7142, "step": 1930 }, { "epoch": 0.0, "learning_rate": 4.9878750000000004e-05, "loss": 5.6422, "step": 1940 }, { "epoch": 0.0, "learning_rate": 4.9878125e-05, "loss": 5.6615, "step": 1950 }, { "epoch": 0.0, "learning_rate": 4.9877500000000005e-05, "loss": 5.5616, "step": 1960 }, { "epoch": 0.0, "learning_rate": 4.9876875e-05, "loss": 5.5871, "step": 1970 }, { "epoch": 0.0, "learning_rate": 4.9876250000000005e-05, "loss": 5.5977, "step": 1980 }, { "epoch": 0.0, "learning_rate": 4.9875625e-05, "loss": 5.6369, "step": 1990 }, { "epoch": 0.0, "learning_rate": 4.9875000000000006e-05, "loss": 5.529, "step": 2000 }, { "epoch": 0.0, "eval_loss": 5.546490669250488, "eval_runtime": 1290.4872, "eval_samples_per_second": 44.972, "eval_steps_per_second": 5.622, "step": 2000 }, { "epoch": 0.0, "learning_rate": 4.9874375e-05, "loss": 5.6742, "step": 2010 }, { "epoch": 0.0, "learning_rate": 4.987375e-05, "loss": 5.4999, "step": 2020 }, { "epoch": 0.0, "learning_rate": 4.9873125000000004e-05, "loss": 5.7466, "step": 2030 }, { "epoch": 0.0, "learning_rate": 4.98725e-05, "loss": 5.4872, "step": 2040 }, { "epoch": 0.0, "learning_rate": 4.9871875000000004e-05, "loss": 5.735, "step": 2050 }, { "epoch": 0.0, "learning_rate": 4.987125e-05, "loss": 5.5832, "step": 2060 }, { "epoch": 0.0, "learning_rate": 4.9870625000000005e-05, "loss": 5.735, "step": 2070 }, { "epoch": 0.0, "learning_rate": 4.987e-05, "loss": 5.6389, "step": 2080 }, { "epoch": 0.0, "learning_rate": 4.9869375e-05, "loss": 5.7397, "step": 2090 }, { "epoch": 0.0, "learning_rate": 4.986875e-05, "loss": 5.558, "step": 2100 }, { "epoch": 0.0, "learning_rate": 4.9868125000000006e-05, "loss": 5.4038, "step": 2110 }, { "epoch": 0.0, "learning_rate": 4.98675e-05, "loss": 5.6118, "step": 2120 }, { "epoch": 0.0, "learning_rate": 4.9866875e-05, "loss": 5.6965, "step": 2130 }, { "epoch": 0.0, "learning_rate": 4.9866250000000004e-05, "loss": 5.6473, "step": 2140 }, { "epoch": 0.0, "learning_rate": 4.986562500000001e-05, "loss": 5.382, "step": 2150 }, { "epoch": 0.0, "learning_rate": 4.9865e-05, "loss": 5.5608, "step": 2160 }, { "epoch": 0.0, "learning_rate": 4.9864375e-05, "loss": 5.575, "step": 2170 }, { "epoch": 0.0, "learning_rate": 4.9863750000000005e-05, "loss": 5.6916, "step": 2180 }, { "epoch": 0.0, "learning_rate": 4.9863125e-05, "loss": 5.6779, "step": 2190 }, { "epoch": 0.0, "learning_rate": 4.98625e-05, "loss": 5.5786, "step": 2200 }, { "epoch": 0.0, "learning_rate": 4.9861875e-05, "loss": 5.5532, "step": 2210 }, { "epoch": 0.0, "learning_rate": 4.9861250000000006e-05, "loss": 5.3066, "step": 2220 }, { "epoch": 0.0, "learning_rate": 4.9860625e-05, "loss": 5.5132, "step": 2230 }, { "epoch": 0.0, "learning_rate": 4.986e-05, "loss": 5.508, "step": 2240 }, { "epoch": 0.0, "learning_rate": 4.9859375000000004e-05, "loss": 5.4825, "step": 2250 }, { "epoch": 0.0, "learning_rate": 4.985875e-05, "loss": 5.501, "step": 2260 }, { "epoch": 0.0, "learning_rate": 4.9858125e-05, "loss": 5.4365, "step": 2270 }, { "epoch": 0.0, "learning_rate": 4.98575e-05, "loss": 5.4862, "step": 2280 }, { "epoch": 0.0, "learning_rate": 4.9856875000000005e-05, "loss": 5.5737, "step": 2290 }, { "epoch": 0.0, "learning_rate": 4.985625e-05, "loss": 5.4607, "step": 2300 }, { "epoch": 0.0, "learning_rate": 4.9855625e-05, "loss": 5.4364, "step": 2310 }, { "epoch": 0.0, "learning_rate": 4.9855e-05, "loss": 5.4407, "step": 2320 }, { "epoch": 0.0, "learning_rate": 4.9854375e-05, "loss": 5.3583, "step": 2330 }, { "epoch": 0.0, "learning_rate": 4.9853750000000004e-05, "loss": 5.3946, "step": 2340 }, { "epoch": 0.0, "learning_rate": 4.9853125e-05, "loss": 5.438, "step": 2350 }, { "epoch": 0.0, "learning_rate": 4.9852500000000004e-05, "loss": 5.3303, "step": 2360 }, { "epoch": 0.0, "learning_rate": 4.9851875e-05, "loss": 5.5209, "step": 2370 }, { "epoch": 0.0, "learning_rate": 4.9851250000000005e-05, "loss": 5.6679, "step": 2380 }, { "epoch": 0.0, "learning_rate": 4.9850625e-05, "loss": 5.3585, "step": 2390 }, { "epoch": 0.0, "learning_rate": 4.9850000000000006e-05, "loss": 5.5441, "step": 2400 }, { "epoch": 0.0, "learning_rate": 4.9849375e-05, "loss": 5.1516, "step": 2410 }, { "epoch": 0.0, "learning_rate": 4.984875e-05, "loss": 5.372, "step": 2420 }, { "epoch": 0.0, "learning_rate": 4.9848125e-05, "loss": 5.3733, "step": 2430 }, { "epoch": 0.0, "learning_rate": 4.98475e-05, "loss": 5.4744, "step": 2440 }, { "epoch": 0.0, "learning_rate": 4.9846875000000004e-05, "loss": 5.3295, "step": 2450 }, { "epoch": 0.0, "learning_rate": 4.984625e-05, "loss": 5.3578, "step": 2460 }, { "epoch": 0.0, "learning_rate": 4.9845625000000004e-05, "loss": 5.4026, "step": 2470 }, { "epoch": 0.0, "learning_rate": 4.9845e-05, "loss": 5.4387, "step": 2480 }, { "epoch": 0.0, "learning_rate": 4.9844375000000005e-05, "loss": 5.3882, "step": 2490 }, { "epoch": 0.0, "learning_rate": 4.984375e-05, "loss": 5.471, "step": 2500 }, { "epoch": 0.0, "learning_rate": 4.9843125e-05, "loss": 5.504, "step": 2510 }, { "epoch": 0.0, "learning_rate": 4.98425e-05, "loss": 5.3762, "step": 2520 }, { "epoch": 0.0, "learning_rate": 4.9841875000000006e-05, "loss": 5.5108, "step": 2530 }, { "epoch": 0.0, "learning_rate": 4.984125e-05, "loss": 5.4382, "step": 2540 }, { "epoch": 0.0, "learning_rate": 4.9840625e-05, "loss": 5.4066, "step": 2550 }, { "epoch": 0.0, "learning_rate": 4.9840000000000004e-05, "loss": 5.355, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.983937500000001e-05, "loss": 5.2863, "step": 2570 }, { "epoch": 0.0, "learning_rate": 4.983875e-05, "loss": 5.0904, "step": 2580 }, { "epoch": 0.0, "learning_rate": 4.9838125e-05, "loss": 5.1451, "step": 2590 }, { "epoch": 0.0, "learning_rate": 4.9837500000000005e-05, "loss": 5.2127, "step": 2600 }, { "epoch": 0.0, "learning_rate": 4.9836875e-05, "loss": 5.4163, "step": 2610 }, { "epoch": 0.0, "learning_rate": 4.983625e-05, "loss": 5.5231, "step": 2620 }, { "epoch": 0.0, "learning_rate": 4.9835625e-05, "loss": 5.1872, "step": 2630 }, { "epoch": 0.0, "learning_rate": 4.9835000000000007e-05, "loss": 5.3788, "step": 2640 }, { "epoch": 0.0, "learning_rate": 4.9834375e-05, "loss": 5.2673, "step": 2650 }, { "epoch": 0.0, "learning_rate": 4.983375e-05, "loss": 5.458, "step": 2660 }, { "epoch": 0.0, "learning_rate": 4.9833125000000004e-05, "loss": 5.4682, "step": 2670 }, { "epoch": 0.0, "learning_rate": 4.98325e-05, "loss": 5.3808, "step": 2680 }, { "epoch": 0.0, "learning_rate": 4.9831875e-05, "loss": 5.1942, "step": 2690 }, { "epoch": 0.0, "learning_rate": 4.983125e-05, "loss": 5.4653, "step": 2700 }, { "epoch": 0.0, "learning_rate": 4.9830625000000005e-05, "loss": 5.4359, "step": 2710 }, { "epoch": 0.0, "learning_rate": 4.983e-05, "loss": 5.2673, "step": 2720 }, { "epoch": 0.0, "learning_rate": 4.9829375e-05, "loss": 5.0108, "step": 2730 }, { "epoch": 0.0, "learning_rate": 4.982875e-05, "loss": 5.1554, "step": 2740 }, { "epoch": 0.0, "learning_rate": 4.9828125e-05, "loss": 5.469, "step": 2750 }, { "epoch": 0.0, "learning_rate": 4.9827500000000004e-05, "loss": 5.4264, "step": 2760 }, { "epoch": 0.0, "learning_rate": 4.9826875e-05, "loss": 5.2732, "step": 2770 }, { "epoch": 0.0, "learning_rate": 4.9826250000000004e-05, "loss": 5.4984, "step": 2780 }, { "epoch": 0.0, "learning_rate": 4.9825625e-05, "loss": 5.1537, "step": 2790 }, { "epoch": 0.0, "learning_rate": 4.9825000000000005e-05, "loss": 5.2632, "step": 2800 }, { "epoch": 0.0, "learning_rate": 4.9824375e-05, "loss": 5.2513, "step": 2810 }, { "epoch": 0.0, "learning_rate": 4.9823750000000006e-05, "loss": 5.6471, "step": 2820 }, { "epoch": 0.0, "learning_rate": 4.9823125e-05, "loss": 5.4319, "step": 2830 }, { "epoch": 0.0, "learning_rate": 4.98225e-05, "loss": 5.631, "step": 2840 }, { "epoch": 0.0, "learning_rate": 4.9821875e-05, "loss": 5.518, "step": 2850 }, { "epoch": 0.0, "learning_rate": 4.982125e-05, "loss": 5.3744, "step": 2860 }, { "epoch": 0.0, "learning_rate": 4.9820625000000004e-05, "loss": 5.3745, "step": 2870 }, { "epoch": 0.0, "learning_rate": 4.982e-05, "loss": 5.1186, "step": 2880 }, { "epoch": 0.0, "learning_rate": 4.9819375000000004e-05, "loss": 5.1986, "step": 2890 }, { "epoch": 0.0, "learning_rate": 4.981875e-05, "loss": 5.4569, "step": 2900 }, { "epoch": 0.0, "learning_rate": 4.9818125000000005e-05, "loss": 5.387, "step": 2910 }, { "epoch": 0.0, "learning_rate": 4.98175e-05, "loss": 5.3456, "step": 2920 }, { "epoch": 0.0, "learning_rate": 4.9816875e-05, "loss": 5.3023, "step": 2930 }, { "epoch": 0.0, "learning_rate": 4.981625e-05, "loss": 5.3567, "step": 2940 }, { "epoch": 0.0, "learning_rate": 4.9815625000000006e-05, "loss": 5.2196, "step": 2950 }, { "epoch": 0.0, "learning_rate": 4.9815e-05, "loss": 5.4705, "step": 2960 }, { "epoch": 0.0, "learning_rate": 4.9814375e-05, "loss": 5.4674, "step": 2970 }, { "epoch": 0.0, "learning_rate": 4.9813750000000004e-05, "loss": 5.3274, "step": 2980 }, { "epoch": 0.0, "learning_rate": 4.981312500000001e-05, "loss": 5.5175, "step": 2990 }, { "epoch": 0.0, "learning_rate": 4.98125e-05, "loss": 5.1719, "step": 3000 }, { "epoch": 0.0, "eval_loss": 5.23661994934082, "eval_runtime": 1290.3728, "eval_samples_per_second": 44.976, "eval_steps_per_second": 5.622, "step": 3000 }, { "epoch": 0.0, "learning_rate": 4.9811875e-05, "loss": 5.1404, "step": 3010 }, { "epoch": 0.0, "learning_rate": 4.9811250000000005e-05, "loss": 5.2816, "step": 3020 }, { "epoch": 0.0, "learning_rate": 4.9810625e-05, "loss": 5.2866, "step": 3030 }, { "epoch": 0.0, "learning_rate": 4.981e-05, "loss": 5.2371, "step": 3040 }, { "epoch": 0.0, "learning_rate": 4.9809375e-05, "loss": 5.2764, "step": 3050 }, { "epoch": 0.0, "learning_rate": 4.9808750000000007e-05, "loss": 5.2766, "step": 3060 }, { "epoch": 0.0, "learning_rate": 4.9808125e-05, "loss": 5.323, "step": 3070 }, { "epoch": 0.0, "learning_rate": 4.98075e-05, "loss": 5.1826, "step": 3080 }, { "epoch": 0.0, "learning_rate": 4.9806875000000004e-05, "loss": 5.0908, "step": 3090 }, { "epoch": 0.0, "learning_rate": 4.980625e-05, "loss": 5.0548, "step": 3100 }, { "epoch": 0.0, "learning_rate": 4.9805625e-05, "loss": 5.1214, "step": 3110 }, { "epoch": 0.0, "learning_rate": 4.9805e-05, "loss": 5.2052, "step": 3120 }, { "epoch": 0.0, "learning_rate": 4.9804375000000005e-05, "loss": 5.3539, "step": 3130 }, { "epoch": 0.0, "learning_rate": 4.980375e-05, "loss": 5.3156, "step": 3140 }, { "epoch": 0.0, "learning_rate": 4.9803125e-05, "loss": 5.3648, "step": 3150 }, { "epoch": 0.0, "learning_rate": 4.98025e-05, "loss": 5.0443, "step": 3160 }, { "epoch": 0.0, "learning_rate": 4.980187500000001e-05, "loss": 5.1141, "step": 3170 }, { "epoch": 0.0, "learning_rate": 4.9801250000000004e-05, "loss": 5.1947, "step": 3180 }, { "epoch": 0.0, "learning_rate": 4.9800625e-05, "loss": 5.363, "step": 3190 }, { "epoch": 0.0, "learning_rate": 4.9800000000000004e-05, "loss": 5.3531, "step": 3200 }, { "epoch": 0.0, "learning_rate": 4.9799375e-05, "loss": 5.1716, "step": 3210 }, { "epoch": 0.0, "learning_rate": 4.9798750000000005e-05, "loss": 5.3822, "step": 3220 }, { "epoch": 0.0, "learning_rate": 4.9798125e-05, "loss": 5.2315, "step": 3230 }, { "epoch": 0.0, "learning_rate": 4.9797500000000006e-05, "loss": 5.0566, "step": 3240 }, { "epoch": 0.0, "learning_rate": 4.9796875e-05, "loss": 5.1457, "step": 3250 }, { "epoch": 0.0, "learning_rate": 4.979625e-05, "loss": 5.2059, "step": 3260 }, { "epoch": 0.0, "learning_rate": 4.9795625e-05, "loss": 5.3245, "step": 3270 }, { "epoch": 0.0, "learning_rate": 4.9795e-05, "loss": 5.1982, "step": 3280 }, { "epoch": 0.0, "learning_rate": 4.9794375000000004e-05, "loss": 5.2997, "step": 3290 }, { "epoch": 0.0, "learning_rate": 4.979375e-05, "loss": 4.8939, "step": 3300 }, { "epoch": 0.0, "learning_rate": 4.9793125000000004e-05, "loss": 5.1763, "step": 3310 }, { "epoch": 0.0, "learning_rate": 4.97925e-05, "loss": 5.3171, "step": 3320 }, { "epoch": 0.0, "learning_rate": 4.9791875000000005e-05, "loss": 4.8775, "step": 3330 }, { "epoch": 0.0, "learning_rate": 4.979125e-05, "loss": 5.1351, "step": 3340 }, { "epoch": 0.0, "learning_rate": 4.9790625e-05, "loss": 5.3803, "step": 3350 }, { "epoch": 0.0, "learning_rate": 4.979e-05, "loss": 5.1125, "step": 3360 }, { "epoch": 0.0, "learning_rate": 4.9789375000000006e-05, "loss": 4.877, "step": 3370 }, { "epoch": 0.0, "learning_rate": 4.978875e-05, "loss": 5.2229, "step": 3380 }, { "epoch": 0.0, "learning_rate": 4.9788125e-05, "loss": 5.1194, "step": 3390 }, { "epoch": 0.0, "learning_rate": 4.9787500000000004e-05, "loss": 5.205, "step": 3400 }, { "epoch": 0.0, "learning_rate": 4.978687500000001e-05, "loss": 5.0868, "step": 3410 }, { "epoch": 0.0, "learning_rate": 4.978625e-05, "loss": 5.0289, "step": 3420 }, { "epoch": 0.0, "learning_rate": 4.9785625e-05, "loss": 5.0166, "step": 3430 }, { "epoch": 0.0, "learning_rate": 4.9785000000000005e-05, "loss": 5.0617, "step": 3440 }, { "epoch": 0.0, "learning_rate": 4.9784375e-05, "loss": 5.384, "step": 3450 }, { "epoch": 0.0, "learning_rate": 4.978375e-05, "loss": 5.0531, "step": 3460 }, { "epoch": 0.0, "learning_rate": 4.9783125e-05, "loss": 5.0491, "step": 3470 }, { "epoch": 0.0, "learning_rate": 4.9782500000000007e-05, "loss": 5.1168, "step": 3480 }, { "epoch": 0.0, "learning_rate": 4.9781875e-05, "loss": 5.2709, "step": 3490 }, { "epoch": 0.0, "learning_rate": 4.978125e-05, "loss": 5.1075, "step": 3500 }, { "epoch": 0.0, "learning_rate": 4.9780625000000004e-05, "loss": 4.9556, "step": 3510 }, { "epoch": 0.0, "learning_rate": 4.978e-05, "loss": 5.1925, "step": 3520 }, { "epoch": 0.0, "learning_rate": 4.9779375e-05, "loss": 5.1607, "step": 3530 }, { "epoch": 0.0, "learning_rate": 4.977875e-05, "loss": 5.1624, "step": 3540 }, { "epoch": 0.0, "learning_rate": 4.9778125000000005e-05, "loss": 5.2167, "step": 3550 }, { "epoch": 0.0, "learning_rate": 4.97775e-05, "loss": 5.1407, "step": 3560 }, { "epoch": 0.0, "learning_rate": 4.9776875e-05, "loss": 5.3289, "step": 3570 }, { "epoch": 0.0, "learning_rate": 4.977625e-05, "loss": 5.2374, "step": 3580 }, { "epoch": 0.0, "learning_rate": 4.977562500000001e-05, "loss": 5.2668, "step": 3590 }, { "epoch": 0.01, "learning_rate": 4.9775000000000004e-05, "loss": 5.1647, "step": 3600 }, { "epoch": 0.01, "learning_rate": 4.9774375e-05, "loss": 5.3235, "step": 3610 }, { "epoch": 0.01, "learning_rate": 4.9773750000000004e-05, "loss": 5.0411, "step": 3620 }, { "epoch": 0.01, "learning_rate": 4.9773125e-05, "loss": 5.2167, "step": 3630 }, { "epoch": 0.01, "learning_rate": 4.97725e-05, "loss": 5.0623, "step": 3640 }, { "epoch": 0.01, "learning_rate": 4.9771875e-05, "loss": 5.2098, "step": 3650 }, { "epoch": 0.01, "learning_rate": 4.9771250000000006e-05, "loss": 5.1417, "step": 3660 }, { "epoch": 0.01, "learning_rate": 4.9770625e-05, "loss": 5.1985, "step": 3670 }, { "epoch": 0.01, "learning_rate": 4.977e-05, "loss": 5.1262, "step": 3680 }, { "epoch": 0.01, "learning_rate": 4.9769375e-05, "loss": 4.9582, "step": 3690 }, { "epoch": 0.01, "learning_rate": 4.976875e-05, "loss": 4.9633, "step": 3700 }, { "epoch": 0.01, "learning_rate": 4.9768125000000004e-05, "loss": 5.316, "step": 3710 }, { "epoch": 0.01, "learning_rate": 4.97675e-05, "loss": 5.0506, "step": 3720 }, { "epoch": 0.01, "learning_rate": 4.9766875000000005e-05, "loss": 5.0421, "step": 3730 }, { "epoch": 0.01, "learning_rate": 4.976625e-05, "loss": 5.0589, "step": 3740 }, { "epoch": 0.01, "learning_rate": 4.9765625000000005e-05, "loss": 5.0365, "step": 3750 }, { "epoch": 0.01, "learning_rate": 4.9765e-05, "loss": 5.0121, "step": 3760 }, { "epoch": 0.01, "learning_rate": 4.9764375e-05, "loss": 5.0514, "step": 3770 }, { "epoch": 0.01, "learning_rate": 4.976375e-05, "loss": 5.1168, "step": 3780 }, { "epoch": 0.01, "learning_rate": 4.9763125000000006e-05, "loss": 5.0953, "step": 3790 }, { "epoch": 0.01, "learning_rate": 4.9762500000000003e-05, "loss": 4.9919, "step": 3800 }, { "epoch": 0.01, "learning_rate": 4.9761875e-05, "loss": 5.2, "step": 3810 }, { "epoch": 0.01, "learning_rate": 4.9761250000000004e-05, "loss": 5.24, "step": 3820 }, { "epoch": 0.01, "learning_rate": 4.976062500000001e-05, "loss": 5.1549, "step": 3830 }, { "epoch": 0.01, "learning_rate": 4.976e-05, "loss": 5.1695, "step": 3840 }, { "epoch": 0.01, "learning_rate": 4.9759375e-05, "loss": 5.0373, "step": 3850 }, { "epoch": 0.01, "learning_rate": 4.9758750000000005e-05, "loss": 4.9885, "step": 3860 }, { "epoch": 0.01, "learning_rate": 4.9758125e-05, "loss": 5.2995, "step": 3870 }, { "epoch": 0.01, "learning_rate": 4.97575e-05, "loss": 4.9755, "step": 3880 }, { "epoch": 0.01, "learning_rate": 4.9756875e-05, "loss": 4.9689, "step": 3890 }, { "epoch": 0.01, "learning_rate": 4.975625000000001e-05, "loss": 5.3017, "step": 3900 }, { "epoch": 0.01, "learning_rate": 4.9755625e-05, "loss": 5.1364, "step": 3910 }, { "epoch": 0.01, "learning_rate": 4.9755e-05, "loss": 4.8806, "step": 3920 }, { "epoch": 0.01, "learning_rate": 4.9754375000000004e-05, "loss": 4.9147, "step": 3930 }, { "epoch": 0.01, "learning_rate": 4.975375e-05, "loss": 4.9342, "step": 3940 }, { "epoch": 0.01, "learning_rate": 4.9753125e-05, "loss": 5.0386, "step": 3950 }, { "epoch": 0.01, "learning_rate": 4.97525e-05, "loss": 5.1696, "step": 3960 }, { "epoch": 0.01, "learning_rate": 4.9751875000000006e-05, "loss": 4.9817, "step": 3970 }, { "epoch": 0.01, "learning_rate": 4.975125e-05, "loss": 5.3313, "step": 3980 }, { "epoch": 0.01, "learning_rate": 4.9750625e-05, "loss": 5.1854, "step": 3990 }, { "epoch": 0.01, "learning_rate": 4.975e-05, "loss": 4.9209, "step": 4000 }, { "epoch": 0.01, "eval_loss": 4.9565348625183105, "eval_runtime": 1291.1452, "eval_samples_per_second": 44.949, "eval_steps_per_second": 5.619, "step": 4000 }, { "epoch": 0.01, "learning_rate": 4.974937500000001e-05, "loss": 4.9402, "step": 4010 }, { "epoch": 0.01, "learning_rate": 4.9748750000000004e-05, "loss": 5.1335, "step": 4020 }, { "epoch": 0.01, "learning_rate": 4.9748125e-05, "loss": 5.0239, "step": 4030 }, { "epoch": 0.01, "learning_rate": 4.9747500000000004e-05, "loss": 4.9243, "step": 4040 }, { "epoch": 0.01, "learning_rate": 4.9746875e-05, "loss": 4.885, "step": 4050 }, { "epoch": 0.01, "learning_rate": 4.974625e-05, "loss": 4.8526, "step": 4060 }, { "epoch": 0.01, "learning_rate": 4.9745625e-05, "loss": 4.9724, "step": 4070 }, { "epoch": 0.01, "learning_rate": 4.9745000000000006e-05, "loss": 5.2256, "step": 4080 }, { "epoch": 0.01, "learning_rate": 4.9744375e-05, "loss": 4.6898, "step": 4090 }, { "epoch": 0.01, "learning_rate": 4.974375e-05, "loss": 4.9986, "step": 4100 }, { "epoch": 0.01, "learning_rate": 4.9743125e-05, "loss": 4.9575, "step": 4110 }, { "epoch": 0.01, "learning_rate": 4.97425e-05, "loss": 4.8754, "step": 4120 }, { "epoch": 0.01, "learning_rate": 4.9741875000000004e-05, "loss": 4.828, "step": 4130 }, { "epoch": 0.01, "learning_rate": 4.974125e-05, "loss": 4.9701, "step": 4140 }, { "epoch": 0.01, "learning_rate": 4.9740625000000005e-05, "loss": 5.0385, "step": 4150 }, { "epoch": 0.01, "learning_rate": 4.974e-05, "loss": 4.938, "step": 4160 }, { "epoch": 0.01, "learning_rate": 4.9739375000000005e-05, "loss": 4.646, "step": 4170 }, { "epoch": 0.01, "learning_rate": 4.973875e-05, "loss": 4.9187, "step": 4180 }, { "epoch": 0.01, "learning_rate": 4.9738125e-05, "loss": 5.1335, "step": 4190 }, { "epoch": 0.01, "learning_rate": 4.97375e-05, "loss": 5.2601, "step": 4200 }, { "epoch": 0.01, "learning_rate": 4.9736875000000007e-05, "loss": 4.9061, "step": 4210 }, { "epoch": 0.01, "learning_rate": 4.9736250000000003e-05, "loss": 4.8484, "step": 4220 }, { "epoch": 0.01, "learning_rate": 4.9735625e-05, "loss": 5.0929, "step": 4230 }, { "epoch": 0.01, "learning_rate": 4.9735000000000004e-05, "loss": 5.1923, "step": 4240 }, { "epoch": 0.01, "learning_rate": 4.9734375e-05, "loss": 5.0951, "step": 4250 }, { "epoch": 0.01, "learning_rate": 4.973375e-05, "loss": 4.8763, "step": 4260 }, { "epoch": 0.01, "learning_rate": 4.9733125e-05, "loss": 4.934, "step": 4270 }, { "epoch": 0.01, "learning_rate": 4.9732500000000005e-05, "loss": 5.1082, "step": 4280 }, { "epoch": 0.01, "learning_rate": 4.9731875e-05, "loss": 5.0792, "step": 4290 }, { "epoch": 0.01, "learning_rate": 4.973125e-05, "loss": 4.935, "step": 4300 }, { "epoch": 0.01, "learning_rate": 4.9730625e-05, "loss": 4.9378, "step": 4310 }, { "epoch": 0.01, "learning_rate": 4.973000000000001e-05, "loss": 4.9178, "step": 4320 }, { "epoch": 0.01, "learning_rate": 4.9729375000000004e-05, "loss": 4.7575, "step": 4330 }, { "epoch": 0.01, "learning_rate": 4.972875e-05, "loss": 4.9522, "step": 4340 }, { "epoch": 0.01, "learning_rate": 4.9728125000000004e-05, "loss": 4.8881, "step": 4350 }, { "epoch": 0.01, "learning_rate": 4.972750000000001e-05, "loss": 5.0173, "step": 4360 }, { "epoch": 0.01, "learning_rate": 4.9726875e-05, "loss": 4.9399, "step": 4370 }, { "epoch": 0.01, "learning_rate": 4.972625e-05, "loss": 4.8249, "step": 4380 }, { "epoch": 0.01, "learning_rate": 4.9725625000000006e-05, "loss": 4.8757, "step": 4390 }, { "epoch": 0.01, "learning_rate": 4.9725e-05, "loss": 5.185, "step": 4400 }, { "epoch": 0.01, "learning_rate": 4.9724375e-05, "loss": 5.0086, "step": 4410 }, { "epoch": 0.01, "learning_rate": 4.972375e-05, "loss": 4.8594, "step": 4420 }, { "epoch": 0.01, "learning_rate": 4.972312500000001e-05, "loss": 4.5972, "step": 4430 }, { "epoch": 0.01, "learning_rate": 4.97225e-05, "loss": 4.946, "step": 4440 }, { "epoch": 0.01, "learning_rate": 4.9721875e-05, "loss": 5.1036, "step": 4450 }, { "epoch": 0.01, "learning_rate": 4.9721250000000004e-05, "loss": 4.727, "step": 4460 }, { "epoch": 0.01, "learning_rate": 4.9720625e-05, "loss": 4.9311, "step": 4470 }, { "epoch": 0.01, "learning_rate": 4.972e-05, "loss": 5.0765, "step": 4480 }, { "epoch": 0.01, "learning_rate": 4.9719375e-05, "loss": 4.8082, "step": 4490 }, { "epoch": 0.01, "learning_rate": 4.9718750000000006e-05, "loss": 4.6347, "step": 4500 }, { "epoch": 0.01, "learning_rate": 4.9718125e-05, "loss": 4.9208, "step": 4510 }, { "epoch": 0.01, "learning_rate": 4.97175e-05, "loss": 4.8189, "step": 4520 }, { "epoch": 0.01, "learning_rate": 4.9716875e-05, "loss": 5.0238, "step": 4530 }, { "epoch": 0.01, "learning_rate": 4.971625e-05, "loss": 4.9812, "step": 4540 }, { "epoch": 0.01, "learning_rate": 4.9715625000000004e-05, "loss": 5.0075, "step": 4550 }, { "epoch": 0.01, "learning_rate": 4.9715e-05, "loss": 4.6772, "step": 4560 }, { "epoch": 0.01, "learning_rate": 4.9714375000000005e-05, "loss": 4.8833, "step": 4570 }, { "epoch": 0.01, "learning_rate": 4.971375e-05, "loss": 4.9346, "step": 4580 }, { "epoch": 0.01, "learning_rate": 4.9713125000000005e-05, "loss": 4.9001, "step": 4590 }, { "epoch": 0.01, "learning_rate": 4.97125e-05, "loss": 4.6977, "step": 4600 }, { "epoch": 0.01, "learning_rate": 4.9711875e-05, "loss": 4.72, "step": 4610 }, { "epoch": 0.01, "learning_rate": 4.971125e-05, "loss": 4.9938, "step": 4620 }, { "epoch": 0.01, "learning_rate": 4.9710625000000007e-05, "loss": 4.9463, "step": 4630 }, { "epoch": 0.01, "learning_rate": 4.9710000000000003e-05, "loss": 4.8265, "step": 4640 }, { "epoch": 0.01, "learning_rate": 4.9709375e-05, "loss": 4.6999, "step": 4650 }, { "epoch": 0.01, "learning_rate": 4.9708750000000004e-05, "loss": 4.6282, "step": 4660 }, { "epoch": 0.01, "learning_rate": 4.9708125e-05, "loss": 4.8076, "step": 4670 }, { "epoch": 0.01, "learning_rate": 4.97075e-05, "loss": 4.7768, "step": 4680 }, { "epoch": 0.01, "learning_rate": 4.9706875e-05, "loss": 4.8861, "step": 4690 }, { "epoch": 0.01, "learning_rate": 4.9706250000000005e-05, "loss": 4.7596, "step": 4700 }, { "epoch": 0.01, "learning_rate": 4.9705625e-05, "loss": 4.8524, "step": 4710 }, { "epoch": 0.01, "learning_rate": 4.9705e-05, "loss": 4.7703, "step": 4720 }, { "epoch": 0.01, "learning_rate": 4.9704375e-05, "loss": 4.8896, "step": 4730 }, { "epoch": 0.01, "learning_rate": 4.970375000000001e-05, "loss": 4.602, "step": 4740 }, { "epoch": 0.01, "learning_rate": 4.9703125000000004e-05, "loss": 4.8619, "step": 4750 }, { "epoch": 0.01, "learning_rate": 4.97025e-05, "loss": 4.6292, "step": 4760 }, { "epoch": 0.01, "learning_rate": 4.9701875000000004e-05, "loss": 4.7711, "step": 4770 }, { "epoch": 0.01, "learning_rate": 4.970125000000001e-05, "loss": 4.5958, "step": 4780 }, { "epoch": 0.01, "learning_rate": 4.9700625e-05, "loss": 4.8532, "step": 4790 }, { "epoch": 0.01, "learning_rate": 4.97e-05, "loss": 4.758, "step": 4800 }, { "epoch": 0.01, "learning_rate": 4.9699375000000006e-05, "loss": 5.0698, "step": 4810 }, { "epoch": 0.01, "learning_rate": 4.969875e-05, "loss": 4.7093, "step": 4820 }, { "epoch": 0.01, "learning_rate": 4.9698125e-05, "loss": 4.5394, "step": 4830 }, { "epoch": 0.01, "learning_rate": 4.96975e-05, "loss": 4.8743, "step": 4840 }, { "epoch": 0.01, "learning_rate": 4.969687500000001e-05, "loss": 4.6109, "step": 4850 }, { "epoch": 0.01, "learning_rate": 4.969625e-05, "loss": 4.9002, "step": 4860 }, { "epoch": 0.01, "learning_rate": 4.9695625e-05, "loss": 4.8014, "step": 4870 }, { "epoch": 0.01, "learning_rate": 4.9695000000000004e-05, "loss": 4.7614, "step": 4880 }, { "epoch": 0.01, "learning_rate": 4.9694375e-05, "loss": 4.6379, "step": 4890 }, { "epoch": 0.01, "learning_rate": 4.969375e-05, "loss": 4.8297, "step": 4900 }, { "epoch": 0.01, "learning_rate": 4.9693125e-05, "loss": 4.8103, "step": 4910 }, { "epoch": 0.01, "learning_rate": 4.9692500000000006e-05, "loss": 4.6609, "step": 4920 }, { "epoch": 0.01, "learning_rate": 4.9691875e-05, "loss": 4.9071, "step": 4930 }, { "epoch": 0.01, "learning_rate": 4.969125e-05, "loss": 4.6381, "step": 4940 }, { "epoch": 0.01, "learning_rate": 4.9690625e-05, "loss": 4.9194, "step": 4950 }, { "epoch": 0.01, "learning_rate": 4.969e-05, "loss": 4.8628, "step": 4960 }, { "epoch": 0.01, "learning_rate": 4.9689375000000004e-05, "loss": 4.8936, "step": 4970 }, { "epoch": 0.01, "learning_rate": 4.968875e-05, "loss": 4.713, "step": 4980 }, { "epoch": 0.01, "learning_rate": 4.9688125000000005e-05, "loss": 4.4379, "step": 4990 }, { "epoch": 0.01, "learning_rate": 4.96875e-05, "loss": 4.443, "step": 5000 }, { "epoch": 0.01, "eval_loss": 4.665901184082031, "eval_runtime": 1291.5299, "eval_samples_per_second": 44.936, "eval_steps_per_second": 5.617, "step": 5000 }, { "epoch": 0.01, "learning_rate": 4.9686875000000005e-05, "loss": 4.9263, "step": 5010 }, { "epoch": 0.01, "learning_rate": 4.968625e-05, "loss": 4.8888, "step": 5020 }, { "epoch": 0.01, "learning_rate": 4.9685625e-05, "loss": 4.8267, "step": 5030 }, { "epoch": 0.01, "learning_rate": 4.9685e-05, "loss": 4.7469, "step": 5040 }, { "epoch": 0.01, "learning_rate": 4.9684375e-05, "loss": 4.8116, "step": 5050 }, { "epoch": 0.01, "learning_rate": 4.9683750000000004e-05, "loss": 4.8608, "step": 5060 }, { "epoch": 0.01, "learning_rate": 4.9683125e-05, "loss": 4.6568, "step": 5070 }, { "epoch": 0.01, "learning_rate": 4.9682500000000004e-05, "loss": 4.4885, "step": 5080 }, { "epoch": 0.01, "learning_rate": 4.9681875e-05, "loss": 4.8361, "step": 5090 }, { "epoch": 0.01, "learning_rate": 4.968125e-05, "loss": 4.6793, "step": 5100 }, { "epoch": 0.01, "learning_rate": 4.9680625e-05, "loss": 4.3473, "step": 5110 }, { "epoch": 0.01, "learning_rate": 4.9680000000000005e-05, "loss": 4.9485, "step": 5120 }, { "epoch": 0.01, "learning_rate": 4.9679375e-05, "loss": 4.7565, "step": 5130 }, { "epoch": 0.01, "learning_rate": 4.967875e-05, "loss": 5.1085, "step": 5140 }, { "epoch": 0.01, "learning_rate": 4.9678125e-05, "loss": 4.6892, "step": 5150 }, { "epoch": 0.01, "learning_rate": 4.967750000000001e-05, "loss": 4.5416, "step": 5160 }, { "epoch": 0.01, "learning_rate": 4.9676875000000004e-05, "loss": 4.8231, "step": 5170 }, { "epoch": 0.01, "learning_rate": 4.967625e-05, "loss": 4.4526, "step": 5180 }, { "epoch": 0.01, "learning_rate": 4.9675625000000004e-05, "loss": 4.5484, "step": 5190 }, { "epoch": 0.01, "learning_rate": 4.967500000000001e-05, "loss": 4.4713, "step": 5200 }, { "epoch": 0.01, "learning_rate": 4.9674375e-05, "loss": 4.6772, "step": 5210 }, { "epoch": 0.01, "learning_rate": 4.967375e-05, "loss": 4.7286, "step": 5220 }, { "epoch": 0.01, "learning_rate": 4.9673125000000006e-05, "loss": 4.7338, "step": 5230 }, { "epoch": 0.01, "learning_rate": 4.96725e-05, "loss": 4.3111, "step": 5240 }, { "epoch": 0.01, "learning_rate": 4.9671875e-05, "loss": 4.567, "step": 5250 }, { "epoch": 0.01, "learning_rate": 4.967125e-05, "loss": 4.5364, "step": 5260 }, { "epoch": 0.01, "learning_rate": 4.967062500000001e-05, "loss": 4.7666, "step": 5270 }, { "epoch": 0.01, "learning_rate": 4.967e-05, "loss": 4.6751, "step": 5280 }, { "epoch": 0.01, "learning_rate": 4.9669375e-05, "loss": 4.9996, "step": 5290 }, { "epoch": 0.01, "learning_rate": 4.9668750000000005e-05, "loss": 4.769, "step": 5300 }, { "epoch": 0.01, "learning_rate": 4.9668125e-05, "loss": 4.7779, "step": 5310 }, { "epoch": 0.01, "learning_rate": 4.96675e-05, "loss": 4.6935, "step": 5320 }, { "epoch": 0.01, "learning_rate": 4.9666875e-05, "loss": 4.8173, "step": 5330 }, { "epoch": 0.01, "learning_rate": 4.9666250000000006e-05, "loss": 4.6378, "step": 5340 }, { "epoch": 0.01, "learning_rate": 4.9665625e-05, "loss": 4.6975, "step": 5350 }, { "epoch": 0.01, "learning_rate": 4.9665e-05, "loss": 4.6202, "step": 5360 }, { "epoch": 0.01, "learning_rate": 4.9664375e-05, "loss": 4.7363, "step": 5370 }, { "epoch": 0.01, "learning_rate": 4.966375e-05, "loss": 4.703, "step": 5380 }, { "epoch": 0.01, "learning_rate": 4.9663125000000004e-05, "loss": 4.4604, "step": 5390 }, { "epoch": 0.01, "learning_rate": 4.96625e-05, "loss": 4.351, "step": 5400 }, { "epoch": 0.01, "learning_rate": 4.9661875000000005e-05, "loss": 4.6252, "step": 5410 }, { "epoch": 0.01, "learning_rate": 4.966125e-05, "loss": 4.7222, "step": 5420 }, { "epoch": 0.01, "learning_rate": 4.9660625000000005e-05, "loss": 4.753, "step": 5430 }, { "epoch": 0.01, "learning_rate": 4.966e-05, "loss": 4.536, "step": 5440 }, { "epoch": 0.01, "learning_rate": 4.9659375e-05, "loss": 4.5324, "step": 5450 }, { "epoch": 0.01, "learning_rate": 4.965875e-05, "loss": 4.4998, "step": 5460 }, { "epoch": 0.01, "learning_rate": 4.9658125e-05, "loss": 4.6248, "step": 5470 }, { "epoch": 0.01, "learning_rate": 4.9657500000000004e-05, "loss": 4.5996, "step": 5480 }, { "epoch": 0.01, "learning_rate": 4.9656875e-05, "loss": 4.2794, "step": 5490 }, { "epoch": 0.01, "learning_rate": 4.9656250000000004e-05, "loss": 4.5031, "step": 5500 }, { "epoch": 0.01, "learning_rate": 4.9655625e-05, "loss": 4.5028, "step": 5510 }, { "epoch": 0.01, "learning_rate": 4.9655000000000005e-05, "loss": 4.6702, "step": 5520 }, { "epoch": 0.01, "learning_rate": 4.9654375e-05, "loss": 4.5889, "step": 5530 }, { "epoch": 0.01, "learning_rate": 4.9653750000000005e-05, "loss": 4.6203, "step": 5540 }, { "epoch": 0.01, "learning_rate": 4.9653125e-05, "loss": 4.3109, "step": 5550 }, { "epoch": 0.01, "learning_rate": 4.96525e-05, "loss": 4.6682, "step": 5560 }, { "epoch": 0.01, "learning_rate": 4.9651875e-05, "loss": 4.4218, "step": 5570 }, { "epoch": 0.01, "learning_rate": 4.965125000000001e-05, "loss": 4.628, "step": 5580 }, { "epoch": 0.01, "learning_rate": 4.9650625000000004e-05, "loss": 4.3221, "step": 5590 }, { "epoch": 0.01, "learning_rate": 4.965e-05, "loss": 4.3831, "step": 5600 }, { "epoch": 0.01, "learning_rate": 4.9649375000000004e-05, "loss": 4.6072, "step": 5610 }, { "epoch": 0.01, "learning_rate": 4.964875000000001e-05, "loss": 4.5748, "step": 5620 }, { "epoch": 0.01, "learning_rate": 4.9648125e-05, "loss": 4.3094, "step": 5630 }, { "epoch": 0.01, "learning_rate": 4.96475e-05, "loss": 4.6253, "step": 5640 }, { "epoch": 0.01, "learning_rate": 4.9646875000000006e-05, "loss": 4.3336, "step": 5650 }, { "epoch": 0.01, "learning_rate": 4.964625e-05, "loss": 4.5384, "step": 5660 }, { "epoch": 0.01, "learning_rate": 4.9645625e-05, "loss": 4.6564, "step": 5670 }, { "epoch": 0.01, "learning_rate": 4.9645e-05, "loss": 4.5371, "step": 5680 }, { "epoch": 0.01, "learning_rate": 4.964437500000001e-05, "loss": 4.3877, "step": 5690 }, { "epoch": 0.01, "learning_rate": 4.964375e-05, "loss": 4.1628, "step": 5700 }, { "epoch": 0.01, "learning_rate": 4.9643125e-05, "loss": 4.2434, "step": 5710 }, { "epoch": 0.01, "learning_rate": 4.9642500000000005e-05, "loss": 4.5242, "step": 5720 }, { "epoch": 0.01, "learning_rate": 4.9641875e-05, "loss": 4.6608, "step": 5730 }, { "epoch": 0.01, "learning_rate": 4.964125e-05, "loss": 4.0727, "step": 5740 }, { "epoch": 0.01, "learning_rate": 4.9640625e-05, "loss": 4.4966, "step": 5750 }, { "epoch": 0.01, "learning_rate": 4.9640000000000006e-05, "loss": 4.356, "step": 5760 }, { "epoch": 0.01, "learning_rate": 4.9639375e-05, "loss": 4.4053, "step": 5770 }, { "epoch": 0.01, "learning_rate": 4.963875e-05, "loss": 4.2831, "step": 5780 }, { "epoch": 0.01, "learning_rate": 4.9638125000000003e-05, "loss": 4.2383, "step": 5790 }, { "epoch": 0.01, "learning_rate": 4.96375e-05, "loss": 4.5428, "step": 5800 }, { "epoch": 0.01, "learning_rate": 4.9636875000000004e-05, "loss": 4.1701, "step": 5810 }, { "epoch": 0.01, "learning_rate": 4.963625e-05, "loss": 4.3831, "step": 5820 }, { "epoch": 0.01, "learning_rate": 4.9635625000000005e-05, "loss": 4.3697, "step": 5830 }, { "epoch": 0.01, "learning_rate": 4.9635e-05, "loss": 4.1746, "step": 5840 }, { "epoch": 0.01, "learning_rate": 4.9634375e-05, "loss": 4.104, "step": 5850 }, { "epoch": 0.01, "learning_rate": 4.963375e-05, "loss": 4.3071, "step": 5860 }, { "epoch": 0.01, "learning_rate": 4.9633125e-05, "loss": 4.2418, "step": 5870 }, { "epoch": 0.01, "learning_rate": 4.96325e-05, "loss": 4.5622, "step": 5880 }, { "epoch": 0.01, "learning_rate": 4.9631875e-05, "loss": 4.4406, "step": 5890 }, { "epoch": 0.01, "learning_rate": 4.9631250000000004e-05, "loss": 4.2963, "step": 5900 }, { "epoch": 0.01, "learning_rate": 4.9630625e-05, "loss": 4.4012, "step": 5910 }, { "epoch": 0.01, "learning_rate": 4.9630000000000004e-05, "loss": 4.4318, "step": 5920 }, { "epoch": 0.01, "learning_rate": 4.9629375e-05, "loss": 4.1245, "step": 5930 }, { "epoch": 0.01, "learning_rate": 4.9628750000000005e-05, "loss": 3.9864, "step": 5940 }, { "epoch": 0.01, "learning_rate": 4.9628125e-05, "loss": 4.4461, "step": 5950 }, { "epoch": 0.01, "learning_rate": 4.9627500000000006e-05, "loss": 4.0817, "step": 5960 }, { "epoch": 0.01, "learning_rate": 4.9626875e-05, "loss": 4.3074, "step": 5970 }, { "epoch": 0.01, "learning_rate": 4.962625e-05, "loss": 4.2605, "step": 5980 }, { "epoch": 0.01, "learning_rate": 4.9625625e-05, "loss": 4.4763, "step": 5990 }, { "epoch": 0.01, "learning_rate": 4.962500000000001e-05, "loss": 4.2494, "step": 6000 }, { "epoch": 0.01, "eval_loss": 4.199699878692627, "eval_runtime": 1291.0726, "eval_samples_per_second": 44.952, "eval_steps_per_second": 5.619, "step": 6000 }, { "epoch": 0.01, "learning_rate": 4.9624375000000004e-05, "loss": 4.4983, "step": 6010 }, { "epoch": 0.01, "learning_rate": 4.962375e-05, "loss": 4.4506, "step": 6020 }, { "epoch": 0.01, "learning_rate": 4.9623125000000004e-05, "loss": 4.3629, "step": 6030 }, { "epoch": 0.01, "learning_rate": 4.962250000000001e-05, "loss": 4.3736, "step": 6040 }, { "epoch": 0.01, "learning_rate": 4.9621875e-05, "loss": 4.3252, "step": 6050 }, { "epoch": 0.01, "learning_rate": 4.962125e-05, "loss": 4.4878, "step": 6060 }, { "epoch": 0.01, "learning_rate": 4.9620625000000006e-05, "loss": 4.2264, "step": 6070 }, { "epoch": 0.01, "learning_rate": 4.962e-05, "loss": 4.2321, "step": 6080 }, { "epoch": 0.01, "learning_rate": 4.9619375e-05, "loss": 4.1035, "step": 6090 }, { "epoch": 0.01, "learning_rate": 4.961875e-05, "loss": 4.2984, "step": 6100 }, { "epoch": 0.01, "learning_rate": 4.961812500000001e-05, "loss": 4.3376, "step": 6110 }, { "epoch": 0.01, "learning_rate": 4.96175e-05, "loss": 4.2871, "step": 6120 }, { "epoch": 0.01, "learning_rate": 4.9616875e-05, "loss": 4.0909, "step": 6130 }, { "epoch": 0.01, "learning_rate": 4.9616250000000005e-05, "loss": 4.2851, "step": 6140 }, { "epoch": 0.01, "learning_rate": 4.9615625e-05, "loss": 4.2851, "step": 6150 }, { "epoch": 0.01, "learning_rate": 4.9615e-05, "loss": 4.1529, "step": 6160 }, { "epoch": 0.01, "learning_rate": 4.9614375e-05, "loss": 4.426, "step": 6170 }, { "epoch": 0.01, "learning_rate": 4.9613750000000006e-05, "loss": 4.0831, "step": 6180 }, { "epoch": 0.01, "learning_rate": 4.9613125e-05, "loss": 3.9893, "step": 6190 }, { "epoch": 0.01, "learning_rate": 4.96125e-05, "loss": 4.0959, "step": 6200 }, { "epoch": 0.01, "learning_rate": 4.9611875000000003e-05, "loss": 4.2922, "step": 6210 }, { "epoch": 0.01, "learning_rate": 4.961125e-05, "loss": 4.0622, "step": 6220 }, { "epoch": 0.01, "learning_rate": 4.9610625000000004e-05, "loss": 4.4173, "step": 6230 }, { "epoch": 0.01, "learning_rate": 4.961e-05, "loss": 4.2111, "step": 6240 }, { "epoch": 0.01, "learning_rate": 4.9609375000000005e-05, "loss": 4.0252, "step": 6250 }, { "epoch": 0.01, "learning_rate": 4.960875e-05, "loss": 4.0679, "step": 6260 }, { "epoch": 0.01, "learning_rate": 4.9608125e-05, "loss": 4.3444, "step": 6270 }, { "epoch": 0.01, "learning_rate": 4.96075e-05, "loss": 4.3168, "step": 6280 }, { "epoch": 0.01, "learning_rate": 4.9606875e-05, "loss": 4.2019, "step": 6290 }, { "epoch": 0.01, "learning_rate": 4.960625e-05, "loss": 4.1883, "step": 6300 }, { "epoch": 0.01, "learning_rate": 4.9605625e-05, "loss": 4.1526, "step": 6310 }, { "epoch": 0.01, "learning_rate": 4.9605000000000004e-05, "loss": 4.1274, "step": 6320 }, { "epoch": 0.01, "learning_rate": 4.9604375e-05, "loss": 4.0769, "step": 6330 }, { "epoch": 0.01, "learning_rate": 4.9603750000000004e-05, "loss": 4.1712, "step": 6340 }, { "epoch": 0.01, "learning_rate": 4.9603125e-05, "loss": 3.9084, "step": 6350 }, { "epoch": 0.01, "learning_rate": 4.9602500000000005e-05, "loss": 4.0311, "step": 6360 }, { "epoch": 0.01, "learning_rate": 4.9601875e-05, "loss": 3.8593, "step": 6370 }, { "epoch": 0.01, "learning_rate": 4.9601250000000006e-05, "loss": 4.495, "step": 6380 }, { "epoch": 0.01, "learning_rate": 4.9600625e-05, "loss": 4.1586, "step": 6390 }, { "epoch": 0.01, "learning_rate": 4.96e-05, "loss": 4.1642, "step": 6400 }, { "epoch": 0.01, "learning_rate": 4.9599375e-05, "loss": 4.1348, "step": 6410 }, { "epoch": 0.01, "learning_rate": 4.959875000000001e-05, "loss": 4.0144, "step": 6420 }, { "epoch": 0.01, "learning_rate": 4.9598125000000004e-05, "loss": 4.4614, "step": 6430 }, { "epoch": 0.01, "learning_rate": 4.95975e-05, "loss": 4.1228, "step": 6440 }, { "epoch": 0.01, "learning_rate": 4.9596875000000004e-05, "loss": 4.2363, "step": 6450 }, { "epoch": 0.01, "learning_rate": 4.959625e-05, "loss": 3.9664, "step": 6460 }, { "epoch": 0.01, "learning_rate": 4.9595625e-05, "loss": 4.3137, "step": 6470 }, { "epoch": 0.01, "learning_rate": 4.9595e-05, "loss": 4.1001, "step": 6480 }, { "epoch": 0.01, "learning_rate": 4.9594375000000006e-05, "loss": 4.1747, "step": 6490 }, { "epoch": 0.01, "learning_rate": 4.959375e-05, "loss": 4.1912, "step": 6500 }, { "epoch": 0.01, "learning_rate": 4.9593125e-05, "loss": 4.2483, "step": 6510 }, { "epoch": 0.01, "learning_rate": 4.95925e-05, "loss": 4.2366, "step": 6520 }, { "epoch": 0.01, "learning_rate": 4.959187500000001e-05, "loss": 4.1883, "step": 6530 }, { "epoch": 0.01, "learning_rate": 4.959125e-05, "loss": 4.0729, "step": 6540 }, { "epoch": 0.01, "learning_rate": 4.9590625e-05, "loss": 4.2181, "step": 6550 }, { "epoch": 0.01, "learning_rate": 4.9590000000000005e-05, "loss": 4.4574, "step": 6560 }, { "epoch": 0.01, "learning_rate": 4.9589375e-05, "loss": 3.9946, "step": 6570 }, { "epoch": 0.01, "learning_rate": 4.958875e-05, "loss": 4.3376, "step": 6580 }, { "epoch": 0.01, "learning_rate": 4.9588125e-05, "loss": 3.9525, "step": 6590 }, { "epoch": 0.01, "learning_rate": 4.9587500000000006e-05, "loss": 4.2535, "step": 6600 }, { "epoch": 0.01, "learning_rate": 4.9586875e-05, "loss": 4.2741, "step": 6610 }, { "epoch": 0.01, "learning_rate": 4.958625e-05, "loss": 4.3139, "step": 6620 }, { "epoch": 0.01, "learning_rate": 4.9585625000000004e-05, "loss": 3.8646, "step": 6630 }, { "epoch": 0.01, "learning_rate": 4.9585e-05, "loss": 4.2046, "step": 6640 }, { "epoch": 0.01, "learning_rate": 4.9584375000000004e-05, "loss": 3.8861, "step": 6650 }, { "epoch": 0.01, "learning_rate": 4.958375e-05, "loss": 3.9607, "step": 6660 }, { "epoch": 0.01, "learning_rate": 4.9583125000000005e-05, "loss": 3.8816, "step": 6670 }, { "epoch": 0.01, "learning_rate": 4.95825e-05, "loss": 4.2083, "step": 6680 }, { "epoch": 0.01, "learning_rate": 4.9581875e-05, "loss": 4.2578, "step": 6690 }, { "epoch": 0.01, "learning_rate": 4.958125e-05, "loss": 4.1309, "step": 6700 }, { "epoch": 0.01, "learning_rate": 4.9580625000000006e-05, "loss": 4.0814, "step": 6710 }, { "epoch": 0.01, "learning_rate": 4.958e-05, "loss": 4.1366, "step": 6720 }, { "epoch": 0.01, "learning_rate": 4.9579375e-05, "loss": 4.1373, "step": 6730 }, { "epoch": 0.01, "learning_rate": 4.9578750000000004e-05, "loss": 4.0746, "step": 6740 }, { "epoch": 0.01, "learning_rate": 4.9578125e-05, "loss": 3.932, "step": 6750 }, { "epoch": 0.01, "learning_rate": 4.9577500000000004e-05, "loss": 4.4765, "step": 6760 }, { "epoch": 0.01, "learning_rate": 4.9576875e-05, "loss": 4.3016, "step": 6770 }, { "epoch": 0.01, "learning_rate": 4.9576250000000005e-05, "loss": 4.0456, "step": 6780 }, { "epoch": 0.01, "learning_rate": 4.9575625e-05, "loss": 3.6256, "step": 6790 }, { "epoch": 0.01, "learning_rate": 4.9575000000000006e-05, "loss": 3.8826, "step": 6800 }, { "epoch": 0.01, "learning_rate": 4.9574375e-05, "loss": 4.0948, "step": 6810 }, { "epoch": 0.01, "learning_rate": 4.957375e-05, "loss": 4.1409, "step": 6820 }, { "epoch": 0.01, "learning_rate": 4.9573125e-05, "loss": 4.3881, "step": 6830 }, { "epoch": 0.01, "learning_rate": 4.957250000000001e-05, "loss": 4.1364, "step": 6840 }, { "epoch": 0.01, "learning_rate": 4.9571875000000004e-05, "loss": 4.3749, "step": 6850 }, { "epoch": 0.01, "learning_rate": 4.957125e-05, "loss": 4.0804, "step": 6860 }, { "epoch": 0.01, "learning_rate": 4.9570625000000004e-05, "loss": 4.1474, "step": 6870 }, { "epoch": 0.01, "learning_rate": 4.957e-05, "loss": 3.8757, "step": 6880 }, { "epoch": 0.01, "learning_rate": 4.9569375e-05, "loss": 4.0067, "step": 6890 }, { "epoch": 0.01, "learning_rate": 4.956875e-05, "loss": 3.8939, "step": 6900 }, { "epoch": 0.01, "learning_rate": 4.9568125000000006e-05, "loss": 4.0438, "step": 6910 }, { "epoch": 0.01, "learning_rate": 4.95675e-05, "loss": 4.1511, "step": 6920 }, { "epoch": 0.01, "learning_rate": 4.9566875e-05, "loss": 4.4667, "step": 6930 }, { "epoch": 0.01, "learning_rate": 4.956625e-05, "loss": 4.0703, "step": 6940 }, { "epoch": 0.01, "learning_rate": 4.956562500000001e-05, "loss": 4.3461, "step": 6950 }, { "epoch": 0.01, "learning_rate": 4.9565e-05, "loss": 3.9892, "step": 6960 }, { "epoch": 0.01, "learning_rate": 4.9564375e-05, "loss": 3.612, "step": 6970 }, { "epoch": 0.01, "learning_rate": 4.9563750000000005e-05, "loss": 4.1341, "step": 6980 }, { "epoch": 0.01, "learning_rate": 4.9563125e-05, "loss": 3.8654, "step": 6990 }, { "epoch": 0.01, "learning_rate": 4.95625e-05, "loss": 3.7124, "step": 7000 }, { "epoch": 0.01, "eval_loss": 3.8291847705841064, "eval_runtime": 1290.6573, "eval_samples_per_second": 44.966, "eval_steps_per_second": 5.621, "step": 7000 }, { "epoch": 0.01, "learning_rate": 4.9561875e-05, "loss": 4.0486, "step": 7010 }, { "epoch": 0.01, "learning_rate": 4.9561250000000006e-05, "loss": 3.6687, "step": 7020 }, { "epoch": 0.01, "learning_rate": 4.9560625e-05, "loss": 4.1076, "step": 7030 }, { "epoch": 0.01, "learning_rate": 4.956e-05, "loss": 3.9366, "step": 7040 }, { "epoch": 0.01, "learning_rate": 4.9559375000000004e-05, "loss": 3.9861, "step": 7050 }, { "epoch": 0.01, "learning_rate": 4.955875e-05, "loss": 3.6556, "step": 7060 }, { "epoch": 0.01, "learning_rate": 4.9558125e-05, "loss": 3.9875, "step": 7070 }, { "epoch": 0.01, "learning_rate": 4.95575e-05, "loss": 4.1649, "step": 7080 }, { "epoch": 0.01, "learning_rate": 4.9556875000000005e-05, "loss": 4.0953, "step": 7090 }, { "epoch": 0.01, "learning_rate": 4.955625e-05, "loss": 3.9598, "step": 7100 }, { "epoch": 0.01, "learning_rate": 4.9555625e-05, "loss": 4.0831, "step": 7110 }, { "epoch": 0.01, "learning_rate": 4.9555e-05, "loss": 4.1012, "step": 7120 }, { "epoch": 0.01, "learning_rate": 4.9554375000000006e-05, "loss": 3.9145, "step": 7130 }, { "epoch": 0.01, "learning_rate": 4.955375e-05, "loss": 3.9787, "step": 7140 }, { "epoch": 0.01, "learning_rate": 4.9553125e-05, "loss": 4.059, "step": 7150 }, { "epoch": 0.01, "learning_rate": 4.9552500000000004e-05, "loss": 4.0313, "step": 7160 }, { "epoch": 0.01, "learning_rate": 4.9551875e-05, "loss": 4.0148, "step": 7170 }, { "epoch": 0.01, "learning_rate": 4.9551250000000004e-05, "loss": 3.9957, "step": 7180 }, { "epoch": 0.01, "learning_rate": 4.9550625e-05, "loss": 3.9103, "step": 7190 }, { "epoch": 0.01, "learning_rate": 4.9550000000000005e-05, "loss": 3.7941, "step": 7200 }, { "epoch": 0.01, "learning_rate": 4.9549375e-05, "loss": 3.9534, "step": 7210 }, { "epoch": 0.01, "learning_rate": 4.9548750000000006e-05, "loss": 3.713, "step": 7220 }, { "epoch": 0.01, "learning_rate": 4.9548125e-05, "loss": 3.9206, "step": 7230 }, { "epoch": 0.01, "learning_rate": 4.95475e-05, "loss": 3.6663, "step": 7240 }, { "epoch": 0.01, "learning_rate": 4.9546875e-05, "loss": 4.0116, "step": 7250 }, { "epoch": 0.01, "learning_rate": 4.954625e-05, "loss": 4.1163, "step": 7260 }, { "epoch": 0.01, "learning_rate": 4.9545625000000004e-05, "loss": 3.6151, "step": 7270 }, { "epoch": 0.01, "learning_rate": 4.9545e-05, "loss": 3.8196, "step": 7280 }, { "epoch": 0.01, "learning_rate": 4.9544375000000005e-05, "loss": 3.8835, "step": 7290 }, { "epoch": 0.01, "learning_rate": 4.954375e-05, "loss": 3.5992, "step": 7300 }, { "epoch": 0.01, "learning_rate": 4.9543125e-05, "loss": 3.8251, "step": 7310 }, { "epoch": 0.01, "learning_rate": 4.95425e-05, "loss": 3.7277, "step": 7320 }, { "epoch": 0.01, "learning_rate": 4.9541875000000006e-05, "loss": 3.9705, "step": 7330 }, { "epoch": 0.01, "learning_rate": 4.954125e-05, "loss": 3.776, "step": 7340 }, { "epoch": 0.01, "learning_rate": 4.9540625e-05, "loss": 4.101, "step": 7350 }, { "epoch": 0.01, "learning_rate": 4.9540000000000003e-05, "loss": 3.9977, "step": 7360 }, { "epoch": 0.01, "learning_rate": 4.953937500000001e-05, "loss": 3.8368, "step": 7370 }, { "epoch": 0.01, "learning_rate": 4.953875e-05, "loss": 3.7347, "step": 7380 }, { "epoch": 0.01, "learning_rate": 4.9538125e-05, "loss": 3.9016, "step": 7390 }, { "epoch": 0.01, "learning_rate": 4.9537500000000005e-05, "loss": 3.7959, "step": 7400 }, { "epoch": 0.01, "learning_rate": 4.9536875e-05, "loss": 3.9565, "step": 7410 }, { "epoch": 0.01, "learning_rate": 4.953625e-05, "loss": 3.9758, "step": 7420 }, { "epoch": 0.01, "learning_rate": 4.9535625e-05, "loss": 4.0549, "step": 7430 }, { "epoch": 0.01, "learning_rate": 4.9535000000000006e-05, "loss": 3.7673, "step": 7440 }, { "epoch": 0.01, "learning_rate": 4.9534375e-05, "loss": 3.5792, "step": 7450 }, { "epoch": 0.01, "learning_rate": 4.953375e-05, "loss": 3.8941, "step": 7460 }, { "epoch": 0.01, "learning_rate": 4.9533125000000004e-05, "loss": 3.7686, "step": 7470 }, { "epoch": 0.01, "learning_rate": 4.95325e-05, "loss": 3.9049, "step": 7480 }, { "epoch": 0.01, "learning_rate": 4.9531875e-05, "loss": 3.8492, "step": 7490 }, { "epoch": 0.01, "learning_rate": 4.953125e-05, "loss": 4.0722, "step": 7500 }, { "epoch": 0.01, "learning_rate": 4.9530625000000005e-05, "loss": 3.5641, "step": 7510 }, { "epoch": 0.01, "learning_rate": 4.953e-05, "loss": 4.2266, "step": 7520 }, { "epoch": 0.01, "learning_rate": 4.9529375e-05, "loss": 3.9837, "step": 7530 }, { "epoch": 0.01, "learning_rate": 4.952875e-05, "loss": 3.6495, "step": 7540 }, { "epoch": 0.01, "learning_rate": 4.9528125000000006e-05, "loss": 3.5167, "step": 7550 }, { "epoch": 0.01, "learning_rate": 4.95275e-05, "loss": 3.8765, "step": 7560 }, { "epoch": 0.01, "learning_rate": 4.9526875e-05, "loss": 3.8782, "step": 7570 }, { "epoch": 0.01, "learning_rate": 4.9526250000000004e-05, "loss": 3.5327, "step": 7580 }, { "epoch": 0.01, "learning_rate": 4.9525625e-05, "loss": 3.7482, "step": 7590 }, { "epoch": 0.01, "learning_rate": 4.9525000000000004e-05, "loss": 3.947, "step": 7600 }, { "epoch": 0.01, "learning_rate": 4.9524375e-05, "loss": 3.9392, "step": 7610 }, { "epoch": 0.01, "learning_rate": 4.9523750000000005e-05, "loss": 4.0527, "step": 7620 }, { "epoch": 0.01, "learning_rate": 4.9523125e-05, "loss": 3.7612, "step": 7630 }, { "epoch": 0.01, "learning_rate": 4.9522500000000006e-05, "loss": 3.6594, "step": 7640 }, { "epoch": 0.01, "learning_rate": 4.9521875e-05, "loss": 3.7504, "step": 7650 }, { "epoch": 0.01, "learning_rate": 4.952125e-05, "loss": 3.5791, "step": 7660 }, { "epoch": 0.01, "learning_rate": 4.9520625e-05, "loss": 3.9749, "step": 7670 }, { "epoch": 0.01, "learning_rate": 4.952e-05, "loss": 3.9398, "step": 7680 }, { "epoch": 0.01, "learning_rate": 4.9519375000000004e-05, "loss": 3.4283, "step": 7690 }, { "epoch": 0.01, "learning_rate": 4.951875e-05, "loss": 3.7188, "step": 7700 }, { "epoch": 0.01, "learning_rate": 4.9518125000000005e-05, "loss": 4.0613, "step": 7710 }, { "epoch": 0.01, "learning_rate": 4.95175e-05, "loss": 3.7282, "step": 7720 }, { "epoch": 0.01, "learning_rate": 4.9516875e-05, "loss": 4.1714, "step": 7730 }, { "epoch": 0.01, "learning_rate": 4.951625e-05, "loss": 3.8217, "step": 7740 }, { "epoch": 0.01, "learning_rate": 4.9515625000000006e-05, "loss": 4.1396, "step": 7750 }, { "epoch": 0.01, "learning_rate": 4.9515e-05, "loss": 3.3593, "step": 7760 }, { "epoch": 0.01, "learning_rate": 4.9514375e-05, "loss": 3.7897, "step": 7770 }, { "epoch": 0.01, "learning_rate": 4.9513750000000003e-05, "loss": 3.8672, "step": 7780 }, { "epoch": 0.01, "learning_rate": 4.951312500000001e-05, "loss": 3.4582, "step": 7790 }, { "epoch": 0.01, "learning_rate": 4.95125e-05, "loss": 3.8057, "step": 7800 }, { "epoch": 0.01, "learning_rate": 4.9511875e-05, "loss": 3.5574, "step": 7810 }, { "epoch": 0.01, "learning_rate": 4.9511250000000005e-05, "loss": 3.6635, "step": 7820 }, { "epoch": 0.01, "learning_rate": 4.9510625e-05, "loss": 3.9062, "step": 7830 }, { "epoch": 0.01, "learning_rate": 4.951e-05, "loss": 3.9078, "step": 7840 }, { "epoch": 0.01, "learning_rate": 4.9509375e-05, "loss": 3.8434, "step": 7850 }, { "epoch": 0.01, "learning_rate": 4.9508750000000006e-05, "loss": 3.5484, "step": 7860 }, { "epoch": 0.01, "learning_rate": 4.9508125e-05, "loss": 3.8928, "step": 7870 }, { "epoch": 0.01, "learning_rate": 4.95075e-05, "loss": 3.6295, "step": 7880 }, { "epoch": 0.01, "learning_rate": 4.9506875000000004e-05, "loss": 3.7458, "step": 7890 }, { "epoch": 0.01, "learning_rate": 4.950625000000001e-05, "loss": 3.7509, "step": 7900 }, { "epoch": 0.01, "learning_rate": 4.9505625e-05, "loss": 4.1396, "step": 7910 }, { "epoch": 0.01, "learning_rate": 4.9505e-05, "loss": 3.5691, "step": 7920 }, { "epoch": 0.01, "learning_rate": 4.9504375000000005e-05, "loss": 3.6733, "step": 7930 }, { "epoch": 0.01, "learning_rate": 4.950375e-05, "loss": 3.7956, "step": 7940 }, { "epoch": 0.01, "learning_rate": 4.9503125e-05, "loss": 4.2452, "step": 7950 }, { "epoch": 0.01, "learning_rate": 4.95025e-05, "loss": 4.0007, "step": 7960 }, { "epoch": 0.01, "learning_rate": 4.9501875000000006e-05, "loss": 3.6776, "step": 7970 }, { "epoch": 0.01, "learning_rate": 4.950125e-05, "loss": 3.9383, "step": 7980 }, { "epoch": 0.01, "learning_rate": 4.9500625e-05, "loss": 3.4481, "step": 7990 }, { "epoch": 0.01, "learning_rate": 4.9500000000000004e-05, "loss": 3.9361, "step": 8000 }, { "epoch": 0.01, "eval_loss": 3.578571319580078, "eval_runtime": 1291.0219, "eval_samples_per_second": 44.954, "eval_steps_per_second": 5.62, "step": 8000 }, { "epoch": 0.01, "learning_rate": 4.9499375e-05, "loss": 3.4723, "step": 8010 }, { "epoch": 0.01, "learning_rate": 4.9498750000000004e-05, "loss": 3.4799, "step": 8020 }, { "epoch": 0.01, "learning_rate": 4.9498125e-05, "loss": 3.3268, "step": 8030 }, { "epoch": 0.01, "learning_rate": 4.9497500000000005e-05, "loss": 3.2724, "step": 8040 }, { "epoch": 0.01, "learning_rate": 4.9496875e-05, "loss": 3.8901, "step": 8050 }, { "epoch": 0.01, "learning_rate": 4.9496250000000006e-05, "loss": 3.9391, "step": 8060 }, { "epoch": 0.01, "learning_rate": 4.9495625e-05, "loss": 3.645, "step": 8070 }, { "epoch": 0.01, "learning_rate": 4.9495e-05, "loss": 3.8628, "step": 8080 }, { "epoch": 0.01, "learning_rate": 4.9494375e-05, "loss": 3.7129, "step": 8090 }, { "epoch": 0.01, "learning_rate": 4.949375e-05, "loss": 3.5434, "step": 8100 }, { "epoch": 0.01, "learning_rate": 4.9493125000000004e-05, "loss": 3.7123, "step": 8110 }, { "epoch": 0.01, "learning_rate": 4.94925e-05, "loss": 3.6897, "step": 8120 }, { "epoch": 0.01, "learning_rate": 4.9491875000000005e-05, "loss": 3.3288, "step": 8130 }, { "epoch": 0.01, "learning_rate": 4.949125e-05, "loss": 3.4082, "step": 8140 }, { "epoch": 0.01, "learning_rate": 4.9490625e-05, "loss": 3.7781, "step": 8150 }, { "epoch": 0.01, "learning_rate": 4.949e-05, "loss": 3.4184, "step": 8160 }, { "epoch": 0.01, "learning_rate": 4.9489375000000006e-05, "loss": 3.84, "step": 8170 }, { "epoch": 0.01, "learning_rate": 4.948875e-05, "loss": 3.9866, "step": 8180 }, { "epoch": 0.01, "learning_rate": 4.9488125e-05, "loss": 3.7153, "step": 8190 }, { "epoch": 0.01, "learning_rate": 4.9487500000000003e-05, "loss": 3.5568, "step": 8200 }, { "epoch": 0.01, "learning_rate": 4.948687500000001e-05, "loss": 3.8304, "step": 8210 }, { "epoch": 0.01, "learning_rate": 4.948625e-05, "loss": 3.5526, "step": 8220 }, { "epoch": 0.01, "learning_rate": 4.9485625e-05, "loss": 3.7829, "step": 8230 }, { "epoch": 0.01, "learning_rate": 4.9485000000000005e-05, "loss": 3.667, "step": 8240 }, { "epoch": 0.01, "learning_rate": 4.9484375e-05, "loss": 3.75, "step": 8250 }, { "epoch": 0.01, "learning_rate": 4.948375e-05, "loss": 3.5902, "step": 8260 }, { "epoch": 0.01, "learning_rate": 4.9483125e-05, "loss": 3.5043, "step": 8270 }, { "epoch": 0.01, "learning_rate": 4.9482500000000006e-05, "loss": 3.9359, "step": 8280 }, { "epoch": 0.01, "learning_rate": 4.9481875e-05, "loss": 3.7477, "step": 8290 }, { "epoch": 0.01, "learning_rate": 4.948125e-05, "loss": 3.6164, "step": 8300 }, { "epoch": 0.01, "learning_rate": 4.9480625000000004e-05, "loss": 3.5659, "step": 8310 }, { "epoch": 0.01, "learning_rate": 4.948000000000001e-05, "loss": 3.3963, "step": 8320 }, { "epoch": 0.01, "learning_rate": 4.9479375e-05, "loss": 3.7214, "step": 8330 }, { "epoch": 0.01, "learning_rate": 4.947875e-05, "loss": 3.7314, "step": 8340 }, { "epoch": 0.01, "learning_rate": 4.9478125000000005e-05, "loss": 3.7615, "step": 8350 }, { "epoch": 0.01, "learning_rate": 4.94775e-05, "loss": 3.7091, "step": 8360 }, { "epoch": 0.01, "learning_rate": 4.9476875e-05, "loss": 3.4318, "step": 8370 }, { "epoch": 0.01, "learning_rate": 4.947625e-05, "loss": 3.3764, "step": 8380 }, { "epoch": 0.01, "learning_rate": 4.9475625000000006e-05, "loss": 4.0291, "step": 8390 }, { "epoch": 0.01, "learning_rate": 4.9475e-05, "loss": 3.2857, "step": 8400 }, { "epoch": 0.01, "learning_rate": 4.9474375e-05, "loss": 3.931, "step": 8410 }, { "epoch": 0.01, "learning_rate": 4.9473750000000004e-05, "loss": 3.7972, "step": 8420 }, { "epoch": 0.01, "learning_rate": 4.9473125e-05, "loss": 3.8959, "step": 8430 }, { "epoch": 0.01, "learning_rate": 4.9472500000000004e-05, "loss": 3.7975, "step": 8440 }, { "epoch": 0.01, "learning_rate": 4.9471875e-05, "loss": 3.6078, "step": 8450 }, { "epoch": 0.01, "learning_rate": 4.9471250000000005e-05, "loss": 3.7358, "step": 8460 }, { "epoch": 0.01, "learning_rate": 4.9470625e-05, "loss": 3.2634, "step": 8470 }, { "epoch": 0.01, "learning_rate": 4.947e-05, "loss": 3.6984, "step": 8480 }, { "epoch": 0.01, "learning_rate": 4.9469375e-05, "loss": 3.6459, "step": 8490 }, { "epoch": 0.01, "learning_rate": 4.946875e-05, "loss": 3.4683, "step": 8500 }, { "epoch": 0.01, "learning_rate": 4.9468125e-05, "loss": 3.6039, "step": 8510 }, { "epoch": 0.01, "learning_rate": 4.94675e-05, "loss": 3.5672, "step": 8520 }, { "epoch": 0.01, "learning_rate": 4.9466875000000004e-05, "loss": 3.6584, "step": 8530 }, { "epoch": 0.01, "learning_rate": 4.946625e-05, "loss": 3.7361, "step": 8540 }, { "epoch": 0.01, "learning_rate": 4.9465625000000005e-05, "loss": 3.3914, "step": 8550 }, { "epoch": 0.01, "learning_rate": 4.9465e-05, "loss": 3.8755, "step": 8560 }, { "epoch": 0.01, "learning_rate": 4.9464375e-05, "loss": 3.4521, "step": 8570 }, { "epoch": 0.01, "learning_rate": 4.946375e-05, "loss": 3.7667, "step": 8580 }, { "epoch": 0.01, "learning_rate": 4.9463125000000006e-05, "loss": 3.7986, "step": 8590 }, { "epoch": 0.01, "learning_rate": 4.94625e-05, "loss": 3.3521, "step": 8600 }, { "epoch": 0.01, "learning_rate": 4.9461875e-05, "loss": 3.6678, "step": 8610 }, { "epoch": 0.01, "learning_rate": 4.9461250000000004e-05, "loss": 3.8021, "step": 8620 }, { "epoch": 0.01, "learning_rate": 4.946062500000001e-05, "loss": 3.8331, "step": 8630 }, { "epoch": 0.01, "learning_rate": 4.946e-05, "loss": 3.7464, "step": 8640 }, { "epoch": 0.01, "learning_rate": 4.9459375e-05, "loss": 3.9039, "step": 8650 }, { "epoch": 0.01, "learning_rate": 4.9458750000000005e-05, "loss": 3.5446, "step": 8660 }, { "epoch": 0.01, "learning_rate": 4.9458125e-05, "loss": 3.6315, "step": 8670 }, { "epoch": 0.01, "learning_rate": 4.94575e-05, "loss": 3.5182, "step": 8680 }, { "epoch": 0.01, "learning_rate": 4.9456875e-05, "loss": 3.8995, "step": 8690 }, { "epoch": 0.01, "learning_rate": 4.9456250000000006e-05, "loss": 3.9399, "step": 8700 }, { "epoch": 0.01, "learning_rate": 4.9455625e-05, "loss": 3.5354, "step": 8710 }, { "epoch": 0.01, "learning_rate": 4.9455e-05, "loss": 3.6385, "step": 8720 }, { "epoch": 0.01, "learning_rate": 4.9454375000000004e-05, "loss": 3.4879, "step": 8730 }, { "epoch": 0.01, "learning_rate": 4.945375000000001e-05, "loss": 3.3559, "step": 8740 }, { "epoch": 0.01, "learning_rate": 4.9453125e-05, "loss": 3.5404, "step": 8750 }, { "epoch": 0.01, "learning_rate": 4.94525e-05, "loss": 3.6796, "step": 8760 }, { "epoch": 0.01, "learning_rate": 4.9451875000000005e-05, "loss": 3.6906, "step": 8770 }, { "epoch": 0.01, "learning_rate": 4.945125e-05, "loss": 3.8954, "step": 8780 }, { "epoch": 0.01, "learning_rate": 4.9450625e-05, "loss": 3.6951, "step": 8790 }, { "epoch": 0.01, "learning_rate": 4.945e-05, "loss": 3.8207, "step": 8800 }, { "epoch": 0.01, "learning_rate": 4.9449375000000006e-05, "loss": 3.4884, "step": 8810 }, { "epoch": 0.01, "learning_rate": 4.944875e-05, "loss": 3.786, "step": 8820 }, { "epoch": 0.01, "learning_rate": 4.9448125e-05, "loss": 3.704, "step": 8830 }, { "epoch": 0.01, "learning_rate": 4.9447500000000004e-05, "loss": 4.0115, "step": 8840 }, { "epoch": 0.01, "learning_rate": 4.9446875e-05, "loss": 3.3402, "step": 8850 }, { "epoch": 0.01, "learning_rate": 4.9446250000000005e-05, "loss": 3.6847, "step": 8860 }, { "epoch": 0.01, "learning_rate": 4.9445625e-05, "loss": 3.6482, "step": 8870 }, { "epoch": 0.01, "learning_rate": 4.9445000000000005e-05, "loss": 3.4033, "step": 8880 }, { "epoch": 0.01, "learning_rate": 4.9444375e-05, "loss": 3.5919, "step": 8890 }, { "epoch": 0.01, "learning_rate": 4.944375e-05, "loss": 3.5481, "step": 8900 }, { "epoch": 0.01, "learning_rate": 4.9443125e-05, "loss": 3.3364, "step": 8910 }, { "epoch": 0.01, "learning_rate": 4.94425e-05, "loss": 3.3548, "step": 8920 }, { "epoch": 0.01, "learning_rate": 4.9441875000000003e-05, "loss": 3.5279, "step": 8930 }, { "epoch": 0.01, "learning_rate": 4.944125e-05, "loss": 3.5439, "step": 8940 }, { "epoch": 0.01, "learning_rate": 4.9440625000000004e-05, "loss": 3.6632, "step": 8950 }, { "epoch": 0.01, "learning_rate": 4.944e-05, "loss": 3.2497, "step": 8960 }, { "epoch": 0.01, "learning_rate": 4.9439375000000005e-05, "loss": 3.4721, "step": 8970 }, { "epoch": 0.01, "learning_rate": 4.943875e-05, "loss": 3.3256, "step": 8980 }, { "epoch": 0.01, "learning_rate": 4.9438125e-05, "loss": 3.5589, "step": 8990 }, { "epoch": 0.01, "learning_rate": 4.94375e-05, "loss": 3.6398, "step": 9000 }, { "epoch": 0.01, "eval_loss": 3.4082789421081543, "eval_runtime": 1290.6956, "eval_samples_per_second": 44.965, "eval_steps_per_second": 5.621, "step": 9000 }, { "epoch": 0.01, "learning_rate": 4.9436875000000006e-05, "loss": 3.5698, "step": 9010 }, { "epoch": 0.01, "learning_rate": 4.943625e-05, "loss": 3.4002, "step": 9020 }, { "epoch": 0.01, "learning_rate": 4.9435625e-05, "loss": 3.3791, "step": 9030 }, { "epoch": 0.01, "learning_rate": 4.9435000000000004e-05, "loss": 3.6341, "step": 9040 }, { "epoch": 0.01, "learning_rate": 4.943437500000001e-05, "loss": 3.6095, "step": 9050 }, { "epoch": 0.01, "learning_rate": 4.9433750000000004e-05, "loss": 3.1007, "step": 9060 }, { "epoch": 0.01, "learning_rate": 4.9433125e-05, "loss": 3.5754, "step": 9070 }, { "epoch": 0.01, "learning_rate": 4.9432500000000005e-05, "loss": 3.8251, "step": 9080 }, { "epoch": 0.01, "learning_rate": 4.9431875e-05, "loss": 3.3762, "step": 9090 }, { "epoch": 0.01, "learning_rate": 4.943125e-05, "loss": 3.6238, "step": 9100 }, { "epoch": 0.01, "learning_rate": 4.9430625e-05, "loss": 3.5951, "step": 9110 }, { "epoch": 0.01, "learning_rate": 4.9430000000000006e-05, "loss": 3.5522, "step": 9120 }, { "epoch": 0.01, "learning_rate": 4.9429375e-05, "loss": 3.294, "step": 9130 }, { "epoch": 0.01, "learning_rate": 4.942875e-05, "loss": 3.1976, "step": 9140 }, { "epoch": 0.01, "learning_rate": 4.9428125000000004e-05, "loss": 3.4196, "step": 9150 }, { "epoch": 0.01, "learning_rate": 4.942750000000001e-05, "loss": 3.6929, "step": 9160 }, { "epoch": 0.01, "learning_rate": 4.9426875e-05, "loss": 3.3984, "step": 9170 }, { "epoch": 0.01, "learning_rate": 4.942625e-05, "loss": 3.2055, "step": 9180 }, { "epoch": 0.01, "learning_rate": 4.9425625000000005e-05, "loss": 3.6268, "step": 9190 }, { "epoch": 0.01, "learning_rate": 4.9425e-05, "loss": 3.4288, "step": 9200 }, { "epoch": 0.01, "learning_rate": 4.9424375e-05, "loss": 3.2284, "step": 9210 }, { "epoch": 0.01, "learning_rate": 4.942375e-05, "loss": 3.5418, "step": 9220 }, { "epoch": 0.01, "learning_rate": 4.9423125000000006e-05, "loss": 3.4934, "step": 9230 }, { "epoch": 0.01, "learning_rate": 4.94225e-05, "loss": 3.0759, "step": 9240 }, { "epoch": 0.01, "learning_rate": 4.9421875e-05, "loss": 3.2583, "step": 9250 }, { "epoch": 0.01, "learning_rate": 4.9421250000000004e-05, "loss": 3.4111, "step": 9260 }, { "epoch": 0.01, "learning_rate": 4.9420625e-05, "loss": 3.4947, "step": 9270 }, { "epoch": 0.01, "learning_rate": 4.942e-05, "loss": 3.591, "step": 9280 }, { "epoch": 0.01, "learning_rate": 4.9419375e-05, "loss": 3.3657, "step": 9290 }, { "epoch": 0.01, "learning_rate": 4.9418750000000005e-05, "loss": 3.5002, "step": 9300 }, { "epoch": 0.01, "learning_rate": 4.9418125e-05, "loss": 3.4835, "step": 9310 }, { "epoch": 0.01, "learning_rate": 4.94175e-05, "loss": 3.5944, "step": 9320 }, { "epoch": 0.01, "learning_rate": 4.9416875e-05, "loss": 3.3964, "step": 9330 }, { "epoch": 0.01, "learning_rate": 4.941625e-05, "loss": 3.3253, "step": 9340 }, { "epoch": 0.01, "learning_rate": 4.9415625000000003e-05, "loss": 3.5001, "step": 9350 }, { "epoch": 0.01, "learning_rate": 4.9415e-05, "loss": 3.4858, "step": 9360 }, { "epoch": 0.01, "learning_rate": 4.9414375000000004e-05, "loss": 3.4862, "step": 9370 }, { "epoch": 0.01, "learning_rate": 4.941375e-05, "loss": 3.8342, "step": 9380 }, { "epoch": 0.01, "learning_rate": 4.9413125000000005e-05, "loss": 3.5336, "step": 9390 }, { "epoch": 0.01, "learning_rate": 4.94125e-05, "loss": 3.373, "step": 9400 }, { "epoch": 0.01, "learning_rate": 4.9411875e-05, "loss": 3.5508, "step": 9410 }, { "epoch": 0.01, "learning_rate": 4.941125e-05, "loss": 3.3746, "step": 9420 }, { "epoch": 0.01, "learning_rate": 4.9410625000000006e-05, "loss": 3.5696, "step": 9430 }, { "epoch": 0.01, "learning_rate": 4.941e-05, "loss": 3.486, "step": 9440 }, { "epoch": 0.01, "learning_rate": 4.9409375e-05, "loss": 3.4811, "step": 9450 }, { "epoch": 0.01, "learning_rate": 4.9408750000000004e-05, "loss": 3.4771, "step": 9460 }, { "epoch": 0.01, "learning_rate": 4.940812500000001e-05, "loss": 3.233, "step": 9470 }, { "epoch": 0.01, "learning_rate": 4.9407500000000004e-05, "loss": 3.2796, "step": 9480 }, { "epoch": 0.01, "learning_rate": 4.9406875e-05, "loss": 3.097, "step": 9490 }, { "epoch": 0.01, "learning_rate": 4.9406250000000005e-05, "loss": 3.4439, "step": 9500 }, { "epoch": 0.01, "learning_rate": 4.9405625e-05, "loss": 3.3275, "step": 9510 }, { "epoch": 0.01, "learning_rate": 4.9405e-05, "loss": 3.6515, "step": 9520 }, { "epoch": 0.01, "learning_rate": 4.9404375e-05, "loss": 3.4626, "step": 9530 }, { "epoch": 0.01, "learning_rate": 4.9403750000000006e-05, "loss": 3.668, "step": 9540 }, { "epoch": 0.01, "learning_rate": 4.9403125e-05, "loss": 3.2561, "step": 9550 }, { "epoch": 0.01, "learning_rate": 4.94025e-05, "loss": 3.6071, "step": 9560 }, { "epoch": 0.01, "learning_rate": 4.9401875000000004e-05, "loss": 3.4366, "step": 9570 }, { "epoch": 0.01, "learning_rate": 4.940125000000001e-05, "loss": 3.3348, "step": 9580 }, { "epoch": 0.01, "learning_rate": 4.9400625e-05, "loss": 3.2542, "step": 9590 }, { "epoch": 0.01, "learning_rate": 4.94e-05, "loss": 3.3632, "step": 9600 }, { "epoch": 0.01, "learning_rate": 4.9399375000000005e-05, "loss": 3.3642, "step": 9610 }, { "epoch": 0.01, "learning_rate": 4.939875e-05, "loss": 3.5836, "step": 9620 }, { "epoch": 0.01, "learning_rate": 4.9398125e-05, "loss": 3.2301, "step": 9630 }, { "epoch": 0.01, "learning_rate": 4.93975e-05, "loss": 3.1996, "step": 9640 }, { "epoch": 0.01, "learning_rate": 4.9396875000000006e-05, "loss": 3.5631, "step": 9650 }, { "epoch": 0.01, "learning_rate": 4.939625e-05, "loss": 3.7293, "step": 9660 }, { "epoch": 0.01, "learning_rate": 4.9395625e-05, "loss": 3.4949, "step": 9670 }, { "epoch": 0.01, "learning_rate": 4.9395000000000004e-05, "loss": 3.1382, "step": 9680 }, { "epoch": 0.01, "learning_rate": 4.9394375e-05, "loss": 3.5623, "step": 9690 }, { "epoch": 0.01, "learning_rate": 4.939375e-05, "loss": 3.409, "step": 9700 }, { "epoch": 0.01, "learning_rate": 4.9393125e-05, "loss": 3.293, "step": 9710 }, { "epoch": 0.01, "learning_rate": 4.9392500000000005e-05, "loss": 3.4939, "step": 9720 }, { "epoch": 0.01, "learning_rate": 4.9391875e-05, "loss": 3.8907, "step": 9730 }, { "epoch": 0.01, "learning_rate": 4.939125e-05, "loss": 3.3106, "step": 9740 }, { "epoch": 0.01, "learning_rate": 4.9390625e-05, "loss": 3.4932, "step": 9750 }, { "epoch": 0.01, "learning_rate": 4.939e-05, "loss": 3.1819, "step": 9760 }, { "epoch": 0.01, "learning_rate": 4.9389375000000003e-05, "loss": 3.6377, "step": 9770 }, { "epoch": 0.01, "learning_rate": 4.938875e-05, "loss": 3.5221, "step": 9780 }, { "epoch": 0.01, "learning_rate": 4.9388125000000004e-05, "loss": 3.4369, "step": 9790 }, { "epoch": 0.01, "learning_rate": 4.93875e-05, "loss": 3.7737, "step": 9800 }, { "epoch": 0.01, "learning_rate": 4.9386875000000005e-05, "loss": 3.4624, "step": 9810 }, { "epoch": 0.01, "learning_rate": 4.938625e-05, "loss": 3.5221, "step": 9820 }, { "epoch": 0.01, "learning_rate": 4.9385625e-05, "loss": 3.0842, "step": 9830 }, { "epoch": 0.01, "learning_rate": 4.9385e-05, "loss": 3.4306, "step": 9840 }, { "epoch": 0.01, "learning_rate": 4.9384375000000006e-05, "loss": 3.2496, "step": 9850 }, { "epoch": 0.01, "learning_rate": 4.938375e-05, "loss": 3.2743, "step": 9860 }, { "epoch": 0.01, "learning_rate": 4.9383125e-05, "loss": 3.0378, "step": 9870 }, { "epoch": 0.01, "learning_rate": 4.9382500000000004e-05, "loss": 3.5076, "step": 9880 }, { "epoch": 0.01, "learning_rate": 4.9381875e-05, "loss": 3.2568, "step": 9890 }, { "epoch": 0.01, "learning_rate": 4.9381250000000004e-05, "loss": 3.3286, "step": 9900 }, { "epoch": 0.01, "learning_rate": 4.9380625e-05, "loss": 3.2772, "step": 9910 }, { "epoch": 0.01, "learning_rate": 4.9380000000000005e-05, "loss": 3.179, "step": 9920 }, { "epoch": 0.01, "learning_rate": 4.9379375e-05, "loss": 3.3218, "step": 9930 }, { "epoch": 0.01, "learning_rate": 4.937875e-05, "loss": 3.2904, "step": 9940 }, { "epoch": 0.01, "learning_rate": 4.9378125e-05, "loss": 2.9548, "step": 9950 }, { "epoch": 0.01, "learning_rate": 4.9377500000000006e-05, "loss": 3.7028, "step": 9960 }, { "epoch": 0.01, "learning_rate": 4.9376875e-05, "loss": 3.4251, "step": 9970 }, { "epoch": 0.01, "learning_rate": 4.937625e-05, "loss": 3.3478, "step": 9980 }, { "epoch": 0.01, "learning_rate": 4.9375625000000004e-05, "loss": 3.2976, "step": 9990 }, { "epoch": 0.01, "learning_rate": 4.937500000000001e-05, "loss": 3.4461, "step": 10000 }, { "epoch": 0.01, "eval_loss": 3.2043068408966064, "eval_runtime": 1290.7027, "eval_samples_per_second": 44.965, "eval_steps_per_second": 5.621, "step": 10000 }, { "epoch": 0.01, "learning_rate": 4.9374375e-05, "loss": 3.3673, "step": 10010 }, { "epoch": 0.01, "learning_rate": 4.937375e-05, "loss": 3.4854, "step": 10020 }, { "epoch": 0.01, "learning_rate": 4.9373125000000005e-05, "loss": 3.4896, "step": 10030 }, { "epoch": 0.01, "learning_rate": 4.93725e-05, "loss": 3.3921, "step": 10040 }, { "epoch": 0.01, "learning_rate": 4.9371875e-05, "loss": 3.5224, "step": 10050 }, { "epoch": 0.01, "learning_rate": 4.937125e-05, "loss": 3.2742, "step": 10060 }, { "epoch": 0.01, "learning_rate": 4.9370625000000006e-05, "loss": 3.5311, "step": 10070 }, { "epoch": 0.01, "learning_rate": 4.937e-05, "loss": 3.4085, "step": 10080 }, { "epoch": 0.01, "learning_rate": 4.9369375e-05, "loss": 3.1062, "step": 10090 }, { "epoch": 0.01, "learning_rate": 4.9368750000000004e-05, "loss": 3.5506, "step": 10100 }, { "epoch": 0.01, "learning_rate": 4.9368125e-05, "loss": 3.5411, "step": 10110 }, { "epoch": 0.01, "learning_rate": 4.93675e-05, "loss": 3.699, "step": 10120 }, { "epoch": 0.01, "learning_rate": 4.9366875e-05, "loss": 3.4206, "step": 10130 }, { "epoch": 0.01, "learning_rate": 4.9366250000000005e-05, "loss": 3.4676, "step": 10140 }, { "epoch": 0.01, "learning_rate": 4.9365625e-05, "loss": 3.0131, "step": 10150 }, { "epoch": 0.01, "learning_rate": 4.9365e-05, "loss": 3.2674, "step": 10160 }, { "epoch": 0.01, "learning_rate": 4.9364375e-05, "loss": 3.186, "step": 10170 }, { "epoch": 0.01, "learning_rate": 4.936375e-05, "loss": 3.1644, "step": 10180 }, { "epoch": 0.01, "learning_rate": 4.9363125000000004e-05, "loss": 3.3497, "step": 10190 }, { "epoch": 0.01, "learning_rate": 4.93625e-05, "loss": 3.2108, "step": 10200 }, { "epoch": 0.01, "learning_rate": 4.9361875000000004e-05, "loss": 3.2988, "step": 10210 }, { "epoch": 0.01, "learning_rate": 4.936125e-05, "loss": 3.5928, "step": 10220 }, { "epoch": 0.01, "learning_rate": 4.9360625000000005e-05, "loss": 3.2536, "step": 10230 }, { "epoch": 0.01, "learning_rate": 4.936e-05, "loss": 3.4038, "step": 10240 }, { "epoch": 0.01, "learning_rate": 4.9359375000000005e-05, "loss": 3.1755, "step": 10250 }, { "epoch": 0.01, "learning_rate": 4.935875e-05, "loss": 3.2453, "step": 10260 }, { "epoch": 0.01, "learning_rate": 4.9358125000000006e-05, "loss": 3.2187, "step": 10270 }, { "epoch": 0.01, "learning_rate": 4.93575e-05, "loss": 3.2603, "step": 10280 }, { "epoch": 0.01, "learning_rate": 4.9356875e-05, "loss": 3.3355, "step": 10290 }, { "epoch": 0.01, "learning_rate": 4.9356250000000004e-05, "loss": 3.2602, "step": 10300 }, { "epoch": 0.01, "learning_rate": 4.9355625e-05, "loss": 3.2615, "step": 10310 }, { "epoch": 0.01, "learning_rate": 4.9355000000000004e-05, "loss": 3.2973, "step": 10320 }, { "epoch": 0.01, "learning_rate": 4.9354375e-05, "loss": 3.2866, "step": 10330 }, { "epoch": 0.01, "learning_rate": 4.9353750000000005e-05, "loss": 3.4226, "step": 10340 }, { "epoch": 0.01, "learning_rate": 4.9353125e-05, "loss": 3.1798, "step": 10350 }, { "epoch": 0.01, "learning_rate": 4.93525e-05, "loss": 3.146, "step": 10360 }, { "epoch": 0.01, "learning_rate": 4.9351875e-05, "loss": 3.2464, "step": 10370 }, { "epoch": 0.01, "learning_rate": 4.9351250000000006e-05, "loss": 3.1374, "step": 10380 }, { "epoch": 0.01, "learning_rate": 4.9350625e-05, "loss": 3.433, "step": 10390 }, { "epoch": 0.01, "learning_rate": 4.935e-05, "loss": 3.3449, "step": 10400 }, { "epoch": 0.01, "learning_rate": 4.9349375000000004e-05, "loss": 2.9269, "step": 10410 }, { "epoch": 0.01, "learning_rate": 4.934875000000001e-05, "loss": 3.1669, "step": 10420 }, { "epoch": 0.01, "learning_rate": 4.9348125e-05, "loss": 3.4554, "step": 10430 }, { "epoch": 0.01, "learning_rate": 4.93475e-05, "loss": 3.4824, "step": 10440 }, { "epoch": 0.01, "learning_rate": 4.9346875000000005e-05, "loss": 3.6861, "step": 10450 }, { "epoch": 0.01, "learning_rate": 4.934625e-05, "loss": 3.4263, "step": 10460 }, { "epoch": 0.01, "learning_rate": 4.9345625e-05, "loss": 3.3093, "step": 10470 }, { "epoch": 0.01, "learning_rate": 4.9345e-05, "loss": 3.2426, "step": 10480 }, { "epoch": 0.01, "learning_rate": 4.9344375000000006e-05, "loss": 3.2701, "step": 10490 }, { "epoch": 0.01, "learning_rate": 4.9343749999999997e-05, "loss": 3.5368, "step": 10500 }, { "epoch": 0.01, "learning_rate": 4.9343125e-05, "loss": 3.5442, "step": 10510 }, { "epoch": 0.01, "learning_rate": 4.9342500000000004e-05, "loss": 2.9055, "step": 10520 }, { "epoch": 0.01, "learning_rate": 4.9341875e-05, "loss": 3.1482, "step": 10530 }, { "epoch": 0.01, "learning_rate": 4.934125e-05, "loss": 3.5763, "step": 10540 }, { "epoch": 0.01, "learning_rate": 4.9340625e-05, "loss": 3.1986, "step": 10550 }, { "epoch": 0.01, "learning_rate": 4.9340000000000005e-05, "loss": 2.969, "step": 10560 }, { "epoch": 0.01, "learning_rate": 4.9339375e-05, "loss": 3.0896, "step": 10570 }, { "epoch": 0.01, "learning_rate": 4.933875e-05, "loss": 3.5655, "step": 10580 }, { "epoch": 0.01, "learning_rate": 4.9338125e-05, "loss": 3.3841, "step": 10590 }, { "epoch": 0.01, "learning_rate": 4.93375e-05, "loss": 3.1081, "step": 10600 }, { "epoch": 0.01, "learning_rate": 4.9336875000000004e-05, "loss": 3.2995, "step": 10610 }, { "epoch": 0.01, "learning_rate": 4.933625e-05, "loss": 3.3516, "step": 10620 }, { "epoch": 0.01, "learning_rate": 4.9335625000000004e-05, "loss": 3.2118, "step": 10630 }, { "epoch": 0.01, "learning_rate": 4.9335e-05, "loss": 3.0774, "step": 10640 }, { "epoch": 0.01, "learning_rate": 4.9334375000000005e-05, "loss": 3.2535, "step": 10650 }, { "epoch": 0.01, "learning_rate": 4.933375e-05, "loss": 3.3348, "step": 10660 }, { "epoch": 0.01, "learning_rate": 4.9333125000000005e-05, "loss": 2.9281, "step": 10670 }, { "epoch": 0.01, "learning_rate": 4.93325e-05, "loss": 3.1008, "step": 10680 }, { "epoch": 0.01, "learning_rate": 4.9331875e-05, "loss": 3.1371, "step": 10690 }, { "epoch": 0.01, "learning_rate": 4.933125e-05, "loss": 3.1068, "step": 10700 }, { "epoch": 0.01, "learning_rate": 4.9330625e-05, "loss": 3.0763, "step": 10710 }, { "epoch": 0.01, "learning_rate": 4.9330000000000004e-05, "loss": 2.9777, "step": 10720 }, { "epoch": 0.01, "learning_rate": 4.9329375e-05, "loss": 3.1927, "step": 10730 }, { "epoch": 0.01, "learning_rate": 4.9328750000000004e-05, "loss": 2.9179, "step": 10740 }, { "epoch": 0.01, "learning_rate": 4.9328125e-05, "loss": 3.1902, "step": 10750 }, { "epoch": 0.01, "learning_rate": 4.9327500000000005e-05, "loss": 2.7897, "step": 10760 }, { "epoch": 0.01, "learning_rate": 4.9326875e-05, "loss": 3.1677, "step": 10770 }, { "epoch": 0.02, "learning_rate": 4.932625e-05, "loss": 3.2404, "step": 10780 }, { "epoch": 0.02, "learning_rate": 4.9325625e-05, "loss": 3.2591, "step": 10790 }, { "epoch": 0.02, "learning_rate": 4.9325000000000006e-05, "loss": 3.0062, "step": 10800 }, { "epoch": 0.02, "learning_rate": 4.9324375e-05, "loss": 3.1786, "step": 10810 }, { "epoch": 0.02, "learning_rate": 4.932375e-05, "loss": 3.179, "step": 10820 }, { "epoch": 0.02, "learning_rate": 4.9323125000000004e-05, "loss": 2.9489, "step": 10830 }, { "epoch": 0.02, "learning_rate": 4.932250000000001e-05, "loss": 3.2355, "step": 10840 }, { "epoch": 0.02, "learning_rate": 4.9321875e-05, "loss": 3.2281, "step": 10850 }, { "epoch": 0.02, "learning_rate": 4.932125e-05, "loss": 3.2089, "step": 10860 }, { "epoch": 0.02, "learning_rate": 4.9320625000000005e-05, "loss": 3.5328, "step": 10870 }, { "epoch": 0.02, "learning_rate": 4.932e-05, "loss": 3.1328, "step": 10880 }, { "epoch": 0.02, "learning_rate": 4.9319375e-05, "loss": 3.3662, "step": 10890 }, { "epoch": 0.02, "learning_rate": 4.931875e-05, "loss": 3.2203, "step": 10900 }, { "epoch": 0.02, "learning_rate": 4.9318125000000006e-05, "loss": 3.0367, "step": 10910 }, { "epoch": 0.02, "learning_rate": 4.93175e-05, "loss": 3.2762, "step": 10920 }, { "epoch": 0.02, "learning_rate": 4.9316875e-05, "loss": 3.22, "step": 10930 }, { "epoch": 0.02, "learning_rate": 4.9316250000000004e-05, "loss": 3.221, "step": 10940 }, { "epoch": 0.02, "learning_rate": 4.9315625e-05, "loss": 3.2589, "step": 10950 }, { "epoch": 0.02, "learning_rate": 4.9315e-05, "loss": 3.1118, "step": 10960 }, { "epoch": 0.02, "learning_rate": 4.9314375e-05, "loss": 3.2071, "step": 10970 }, { "epoch": 0.02, "learning_rate": 4.9313750000000005e-05, "loss": 3.1683, "step": 10980 }, { "epoch": 0.02, "learning_rate": 4.9313125e-05, "loss": 3.1918, "step": 10990 }, { "epoch": 0.02, "learning_rate": 4.93125e-05, "loss": 3.2806, "step": 11000 }, { "epoch": 0.02, "eval_loss": 3.0368127822875977, "eval_runtime": 1291.3468, "eval_samples_per_second": 44.942, "eval_steps_per_second": 5.618, "step": 11000 }, { "epoch": 0.02, "learning_rate": 4.9311875e-05, "loss": 3.0215, "step": 11010 }, { "epoch": 0.02, "learning_rate": 4.931125000000001e-05, "loss": 2.9273, "step": 11020 }, { "epoch": 0.02, "learning_rate": 4.9310625000000004e-05, "loss": 3.322, "step": 11030 }, { "epoch": 0.02, "learning_rate": 4.931e-05, "loss": 3.18, "step": 11040 }, { "epoch": 0.02, "learning_rate": 4.9309375000000004e-05, "loss": 2.9707, "step": 11050 }, { "epoch": 0.02, "learning_rate": 4.930875e-05, "loss": 3.2583, "step": 11060 }, { "epoch": 0.02, "learning_rate": 4.9308125000000005e-05, "loss": 3.1384, "step": 11070 }, { "epoch": 0.02, "learning_rate": 4.93075e-05, "loss": 3.2307, "step": 11080 }, { "epoch": 0.02, "learning_rate": 4.9306875000000006e-05, "loss": 3.2889, "step": 11090 }, { "epoch": 0.02, "learning_rate": 4.930625e-05, "loss": 3.1536, "step": 11100 }, { "epoch": 0.02, "learning_rate": 4.9305625e-05, "loss": 3.1033, "step": 11110 }, { "epoch": 0.02, "learning_rate": 4.9305e-05, "loss": 3.114, "step": 11120 }, { "epoch": 0.02, "learning_rate": 4.9304375e-05, "loss": 2.7169, "step": 11130 }, { "epoch": 0.02, "learning_rate": 4.9303750000000004e-05, "loss": 3.2093, "step": 11140 }, { "epoch": 0.02, "learning_rate": 4.9303125e-05, "loss": 3.4982, "step": 11150 }, { "epoch": 0.02, "learning_rate": 4.9302500000000004e-05, "loss": 3.0432, "step": 11160 }, { "epoch": 0.02, "learning_rate": 4.9301875e-05, "loss": 2.9971, "step": 11170 }, { "epoch": 0.02, "learning_rate": 4.9301250000000005e-05, "loss": 3.2066, "step": 11180 }, { "epoch": 0.02, "learning_rate": 4.9300625e-05, "loss": 2.7408, "step": 11190 }, { "epoch": 0.02, "learning_rate": 4.93e-05, "loss": 2.9796, "step": 11200 }, { "epoch": 0.02, "learning_rate": 4.9299375e-05, "loss": 3.3378, "step": 11210 }, { "epoch": 0.02, "learning_rate": 4.9298750000000006e-05, "loss": 2.9784, "step": 11220 }, { "epoch": 0.02, "learning_rate": 4.9298125e-05, "loss": 2.7574, "step": 11230 }, { "epoch": 0.02, "learning_rate": 4.92975e-05, "loss": 3.2446, "step": 11240 }, { "epoch": 0.02, "learning_rate": 4.9296875000000004e-05, "loss": 3.0218, "step": 11250 }, { "epoch": 0.02, "learning_rate": 4.929625000000001e-05, "loss": 3.0277, "step": 11260 }, { "epoch": 0.02, "learning_rate": 4.9295625e-05, "loss": 3.0225, "step": 11270 }, { "epoch": 0.02, "learning_rate": 4.9295e-05, "loss": 3.1752, "step": 11280 }, { "epoch": 0.02, "learning_rate": 4.9294375000000005e-05, "loss": 3.3066, "step": 11290 }, { "epoch": 0.02, "learning_rate": 4.929375e-05, "loss": 3.0175, "step": 11300 }, { "epoch": 0.02, "learning_rate": 4.9293125e-05, "loss": 3.3492, "step": 11310 }, { "epoch": 0.02, "learning_rate": 4.92925e-05, "loss": 3.1667, "step": 11320 }, { "epoch": 0.02, "learning_rate": 4.9291875000000007e-05, "loss": 3.0802, "step": 11330 }, { "epoch": 0.02, "learning_rate": 4.929125e-05, "loss": 2.9216, "step": 11340 }, { "epoch": 0.02, "learning_rate": 4.9290625e-05, "loss": 3.0988, "step": 11350 }, { "epoch": 0.02, "learning_rate": 4.9290000000000004e-05, "loss": 3.2656, "step": 11360 }, { "epoch": 0.02, "learning_rate": 4.9289375e-05, "loss": 3.1209, "step": 11370 }, { "epoch": 0.02, "learning_rate": 4.928875e-05, "loss": 3.2368, "step": 11380 }, { "epoch": 0.02, "learning_rate": 4.9288125e-05, "loss": 3.1238, "step": 11390 }, { "epoch": 0.02, "learning_rate": 4.9287500000000005e-05, "loss": 3.3314, "step": 11400 }, { "epoch": 0.02, "learning_rate": 4.9286875e-05, "loss": 3.2883, "step": 11410 }, { "epoch": 0.02, "learning_rate": 4.928625e-05, "loss": 3.3284, "step": 11420 }, { "epoch": 0.02, "learning_rate": 4.9285625e-05, "loss": 3.0403, "step": 11430 }, { "epoch": 0.02, "learning_rate": 4.928500000000001e-05, "loss": 2.8881, "step": 11440 }, { "epoch": 0.02, "learning_rate": 4.9284375000000004e-05, "loss": 2.9968, "step": 11450 }, { "epoch": 0.02, "learning_rate": 4.928375e-05, "loss": 3.4168, "step": 11460 }, { "epoch": 0.02, "learning_rate": 4.9283125000000004e-05, "loss": 3.1828, "step": 11470 }, { "epoch": 0.02, "learning_rate": 4.92825e-05, "loss": 2.8552, "step": 11480 }, { "epoch": 0.02, "learning_rate": 4.9281875000000005e-05, "loss": 3.3487, "step": 11490 }, { "epoch": 0.02, "learning_rate": 4.928125e-05, "loss": 3.3312, "step": 11500 }, { "epoch": 0.02, "learning_rate": 4.9280625000000006e-05, "loss": 3.0766, "step": 11510 }, { "epoch": 0.02, "learning_rate": 4.928e-05, "loss": 2.8558, "step": 11520 }, { "epoch": 0.02, "learning_rate": 4.9279375e-05, "loss": 3.2239, "step": 11530 }, { "epoch": 0.02, "learning_rate": 4.927875e-05, "loss": 3.256, "step": 11540 }, { "epoch": 0.02, "learning_rate": 4.9278125e-05, "loss": 3.3188, "step": 11550 }, { "epoch": 0.02, "learning_rate": 4.9277500000000004e-05, "loss": 3.0546, "step": 11560 }, { "epoch": 0.02, "learning_rate": 4.9276875e-05, "loss": 3.2326, "step": 11570 }, { "epoch": 0.02, "learning_rate": 4.9276250000000004e-05, "loss": 3.1798, "step": 11580 }, { "epoch": 0.02, "learning_rate": 4.9275625e-05, "loss": 2.9517, "step": 11590 }, { "epoch": 0.02, "learning_rate": 4.9275000000000005e-05, "loss": 2.8993, "step": 11600 }, { "epoch": 0.02, "learning_rate": 4.9274375e-05, "loss": 3.0196, "step": 11610 }, { "epoch": 0.02, "learning_rate": 4.927375e-05, "loss": 3.1123, "step": 11620 }, { "epoch": 0.02, "learning_rate": 4.9273125e-05, "loss": 2.9998, "step": 11630 }, { "epoch": 0.02, "learning_rate": 4.9272500000000006e-05, "loss": 3.0828, "step": 11640 }, { "epoch": 0.02, "learning_rate": 4.9271875e-05, "loss": 3.294, "step": 11650 }, { "epoch": 0.02, "learning_rate": 4.927125e-05, "loss": 3.3501, "step": 11660 }, { "epoch": 0.02, "learning_rate": 4.9270625000000004e-05, "loss": 2.9809, "step": 11670 }, { "epoch": 0.02, "learning_rate": 4.927000000000001e-05, "loss": 3.3666, "step": 11680 }, { "epoch": 0.02, "learning_rate": 4.9269375e-05, "loss": 3.4348, "step": 11690 }, { "epoch": 0.02, "learning_rate": 4.926875e-05, "loss": 3.2334, "step": 11700 }, { "epoch": 0.02, "learning_rate": 4.9268125000000005e-05, "loss": 3.1802, "step": 11710 }, { "epoch": 0.02, "learning_rate": 4.92675e-05, "loss": 3.1418, "step": 11720 }, { "epoch": 0.02, "learning_rate": 4.9266875e-05, "loss": 2.8603, "step": 11730 }, { "epoch": 0.02, "learning_rate": 4.926625e-05, "loss": 3.2243, "step": 11740 }, { "epoch": 0.02, "learning_rate": 4.9265625000000007e-05, "loss": 3.0931, "step": 11750 }, { "epoch": 0.02, "learning_rate": 4.9265e-05, "loss": 3.2331, "step": 11760 }, { "epoch": 0.02, "learning_rate": 4.9264375e-05, "loss": 2.9789, "step": 11770 }, { "epoch": 0.02, "learning_rate": 4.9263750000000004e-05, "loss": 3.2558, "step": 11780 }, { "epoch": 0.02, "learning_rate": 4.9263125e-05, "loss": 3.1451, "step": 11790 }, { "epoch": 0.02, "learning_rate": 4.92625e-05, "loss": 3.1114, "step": 11800 }, { "epoch": 0.02, "learning_rate": 4.9261875e-05, "loss": 2.9598, "step": 11810 }, { "epoch": 0.02, "learning_rate": 4.9261250000000005e-05, "loss": 2.7883, "step": 11820 }, { "epoch": 0.02, "learning_rate": 4.9260625e-05, "loss": 3.1869, "step": 11830 }, { "epoch": 0.02, "learning_rate": 4.926e-05, "loss": 3.035, "step": 11840 }, { "epoch": 0.02, "learning_rate": 4.9259375e-05, "loss": 3.3383, "step": 11850 }, { "epoch": 0.02, "learning_rate": 4.925875000000001e-05, "loss": 2.9423, "step": 11860 }, { "epoch": 0.02, "learning_rate": 4.9258125000000004e-05, "loss": 2.8412, "step": 11870 }, { "epoch": 0.02, "learning_rate": 4.92575e-05, "loss": 2.9428, "step": 11880 }, { "epoch": 0.02, "learning_rate": 4.9256875000000004e-05, "loss": 3.006, "step": 11890 }, { "epoch": 0.02, "learning_rate": 4.925625e-05, "loss": 2.8168, "step": 11900 }, { "epoch": 0.02, "learning_rate": 4.9255625e-05, "loss": 3.0569, "step": 11910 }, { "epoch": 0.02, "learning_rate": 4.9255e-05, "loss": 2.9725, "step": 11920 }, { "epoch": 0.02, "learning_rate": 4.9254375000000006e-05, "loss": 2.8952, "step": 11930 }, { "epoch": 0.02, "learning_rate": 4.925375e-05, "loss": 3.0806, "step": 11940 }, { "epoch": 0.02, "learning_rate": 4.9253125e-05, "loss": 2.8342, "step": 11950 }, { "epoch": 0.02, "learning_rate": 4.92525e-05, "loss": 2.9804, "step": 11960 }, { "epoch": 0.02, "learning_rate": 4.9251875e-05, "loss": 2.904, "step": 11970 }, { "epoch": 0.02, "learning_rate": 4.9251250000000004e-05, "loss": 3.2092, "step": 11980 }, { "epoch": 0.02, "learning_rate": 4.9250625e-05, "loss": 3.0802, "step": 11990 }, { "epoch": 0.02, "learning_rate": 4.9250000000000004e-05, "loss": 3.2538, "step": 12000 }, { "epoch": 0.02, "eval_loss": 2.9008471965789795, "eval_runtime": 1290.8169, "eval_samples_per_second": 44.961, "eval_steps_per_second": 5.62, "step": 12000 }, { "epoch": 0.02, "learning_rate": 4.9249375e-05, "loss": 2.9914, "step": 12010 }, { "epoch": 0.02, "learning_rate": 4.9248750000000005e-05, "loss": 3.1978, "step": 12020 }, { "epoch": 0.02, "learning_rate": 4.9248125e-05, "loss": 3.168, "step": 12030 }, { "epoch": 0.02, "learning_rate": 4.92475e-05, "loss": 3.0325, "step": 12040 }, { "epoch": 0.02, "learning_rate": 4.9246875e-05, "loss": 2.949, "step": 12050 }, { "epoch": 0.02, "learning_rate": 4.9246250000000006e-05, "loss": 3.3464, "step": 12060 }, { "epoch": 0.02, "learning_rate": 4.9245625e-05, "loss": 2.821, "step": 12070 }, { "epoch": 0.02, "learning_rate": 4.9245e-05, "loss": 2.5214, "step": 12080 }, { "epoch": 0.02, "learning_rate": 4.9244375000000004e-05, "loss": 2.9258, "step": 12090 }, { "epoch": 0.02, "learning_rate": 4.924375e-05, "loss": 3.0793, "step": 12100 }, { "epoch": 0.02, "learning_rate": 4.9243125e-05, "loss": 3.083, "step": 12110 }, { "epoch": 0.02, "learning_rate": 4.92425e-05, "loss": 3.1146, "step": 12120 }, { "epoch": 0.02, "learning_rate": 4.9241875000000005e-05, "loss": 3.173, "step": 12130 }, { "epoch": 0.02, "learning_rate": 4.924125e-05, "loss": 2.7969, "step": 12140 }, { "epoch": 0.02, "learning_rate": 4.9240625e-05, "loss": 3.1594, "step": 12150 }, { "epoch": 0.02, "learning_rate": 4.924e-05, "loss": 3.2232, "step": 12160 }, { "epoch": 0.02, "learning_rate": 4.9239375000000007e-05, "loss": 2.8916, "step": 12170 }, { "epoch": 0.02, "learning_rate": 4.9238750000000004e-05, "loss": 2.8619, "step": 12180 }, { "epoch": 0.02, "learning_rate": 4.9238125e-05, "loss": 2.9829, "step": 12190 }, { "epoch": 0.02, "learning_rate": 4.9237500000000004e-05, "loss": 2.7603, "step": 12200 }, { "epoch": 0.02, "learning_rate": 4.923687500000001e-05, "loss": 3.033, "step": 12210 }, { "epoch": 0.02, "learning_rate": 4.923625e-05, "loss": 3.4367, "step": 12220 }, { "epoch": 0.02, "learning_rate": 4.9235625e-05, "loss": 3.173, "step": 12230 }, { "epoch": 0.02, "learning_rate": 4.9235000000000005e-05, "loss": 3.1659, "step": 12240 }, { "epoch": 0.02, "learning_rate": 4.9234375e-05, "loss": 3.4029, "step": 12250 }, { "epoch": 0.02, "learning_rate": 4.923375e-05, "loss": 2.9323, "step": 12260 }, { "epoch": 0.02, "learning_rate": 4.9233125e-05, "loss": 3.0574, "step": 12270 }, { "epoch": 0.02, "learning_rate": 4.923250000000001e-05, "loss": 3.068, "step": 12280 }, { "epoch": 0.02, "learning_rate": 4.9231875000000004e-05, "loss": 3.1348, "step": 12290 }, { "epoch": 0.02, "learning_rate": 4.923125e-05, "loss": 2.8839, "step": 12300 }, { "epoch": 0.02, "learning_rate": 4.9230625000000004e-05, "loss": 2.6386, "step": 12310 }, { "epoch": 0.02, "learning_rate": 4.923e-05, "loss": 2.9799, "step": 12320 }, { "epoch": 0.02, "learning_rate": 4.9229375e-05, "loss": 2.8461, "step": 12330 }, { "epoch": 0.02, "learning_rate": 4.922875e-05, "loss": 2.9998, "step": 12340 }, { "epoch": 0.02, "learning_rate": 4.9228125000000006e-05, "loss": 2.7169, "step": 12350 }, { "epoch": 0.02, "learning_rate": 4.92275e-05, "loss": 3.4619, "step": 12360 }, { "epoch": 0.02, "learning_rate": 4.9226875e-05, "loss": 3.0961, "step": 12370 }, { "epoch": 0.02, "learning_rate": 4.922625e-05, "loss": 3.1378, "step": 12380 }, { "epoch": 0.02, "learning_rate": 4.9225625e-05, "loss": 3.0483, "step": 12390 }, { "epoch": 0.02, "learning_rate": 4.9225000000000004e-05, "loss": 3.2622, "step": 12400 }, { "epoch": 0.02, "learning_rate": 4.9224375e-05, "loss": 3.1196, "step": 12410 }, { "epoch": 0.02, "learning_rate": 4.9223750000000005e-05, "loss": 2.901, "step": 12420 }, { "epoch": 0.02, "learning_rate": 4.9223125e-05, "loss": 3.0162, "step": 12430 }, { "epoch": 0.02, "learning_rate": 4.9222500000000005e-05, "loss": 2.6859, "step": 12440 }, { "epoch": 0.02, "learning_rate": 4.9221875e-05, "loss": 2.7737, "step": 12450 }, { "epoch": 0.02, "learning_rate": 4.922125e-05, "loss": 2.8407, "step": 12460 }, { "epoch": 0.02, "learning_rate": 4.9220625e-05, "loss": 3.1173, "step": 12470 }, { "epoch": 0.02, "learning_rate": 4.9220000000000006e-05, "loss": 2.7749, "step": 12480 }, { "epoch": 0.02, "learning_rate": 4.9219375000000003e-05, "loss": 3.0709, "step": 12490 }, { "epoch": 0.02, "learning_rate": 4.921875e-05, "loss": 2.7529, "step": 12500 }, { "epoch": 0.02, "learning_rate": 4.9218125000000004e-05, "loss": 3.1316, "step": 12510 }, { "epoch": 0.02, "learning_rate": 4.92175e-05, "loss": 2.7336, "step": 12520 }, { "epoch": 0.02, "learning_rate": 4.9216875e-05, "loss": 2.7769, "step": 12530 }, { "epoch": 0.02, "learning_rate": 4.921625e-05, "loss": 2.8616, "step": 12540 }, { "epoch": 0.02, "learning_rate": 4.9215625000000005e-05, "loss": 3.1076, "step": 12550 }, { "epoch": 0.02, "learning_rate": 4.9215e-05, "loss": 2.6112, "step": 12560 }, { "epoch": 0.02, "learning_rate": 4.9214375e-05, "loss": 2.9286, "step": 12570 }, { "epoch": 0.02, "learning_rate": 4.921375e-05, "loss": 2.9151, "step": 12580 }, { "epoch": 0.02, "learning_rate": 4.921312500000001e-05, "loss": 2.9439, "step": 12590 }, { "epoch": 0.02, "learning_rate": 4.9212500000000004e-05, "loss": 2.9318, "step": 12600 }, { "epoch": 0.02, "learning_rate": 4.9211875e-05, "loss": 3.1139, "step": 12610 }, { "epoch": 0.02, "learning_rate": 4.9211250000000004e-05, "loss": 3.0182, "step": 12620 }, { "epoch": 0.02, "learning_rate": 4.921062500000001e-05, "loss": 3.1018, "step": 12630 }, { "epoch": 0.02, "learning_rate": 4.921e-05, "loss": 2.8633, "step": 12640 }, { "epoch": 0.02, "learning_rate": 4.9209375e-05, "loss": 2.9768, "step": 12650 }, { "epoch": 0.02, "learning_rate": 4.9208750000000006e-05, "loss": 3.0282, "step": 12660 }, { "epoch": 0.02, "learning_rate": 4.9208125e-05, "loss": 2.5559, "step": 12670 }, { "epoch": 0.02, "learning_rate": 4.92075e-05, "loss": 3.3861, "step": 12680 }, { "epoch": 0.02, "learning_rate": 4.9206875e-05, "loss": 2.9903, "step": 12690 }, { "epoch": 0.02, "learning_rate": 4.920625000000001e-05, "loss": 2.7813, "step": 12700 }, { "epoch": 0.02, "learning_rate": 4.9205625e-05, "loss": 2.9504, "step": 12710 }, { "epoch": 0.02, "learning_rate": 4.9205e-05, "loss": 2.5905, "step": 12720 }, { "epoch": 0.02, "learning_rate": 4.9204375000000004e-05, "loss": 2.8423, "step": 12730 }, { "epoch": 0.02, "learning_rate": 4.920375e-05, "loss": 3.1319, "step": 12740 }, { "epoch": 0.02, "learning_rate": 4.9203125e-05, "loss": 3.0152, "step": 12750 }, { "epoch": 0.02, "learning_rate": 4.92025e-05, "loss": 3.1268, "step": 12760 }, { "epoch": 0.02, "learning_rate": 4.9201875000000006e-05, "loss": 2.7621, "step": 12770 }, { "epoch": 0.02, "learning_rate": 4.920125e-05, "loss": 3.17, "step": 12780 }, { "epoch": 0.02, "learning_rate": 4.9200625e-05, "loss": 3.044, "step": 12790 }, { "epoch": 0.02, "learning_rate": 4.92e-05, "loss": 3.2553, "step": 12800 }, { "epoch": 0.02, "learning_rate": 4.9199375e-05, "loss": 2.7475, "step": 12810 }, { "epoch": 0.02, "learning_rate": 4.9198750000000004e-05, "loss": 2.9328, "step": 12820 }, { "epoch": 0.02, "learning_rate": 4.9198125e-05, "loss": 2.7254, "step": 12830 }, { "epoch": 0.02, "learning_rate": 4.9197500000000005e-05, "loss": 2.8976, "step": 12840 }, { "epoch": 0.02, "learning_rate": 4.9196875e-05, "loss": 3.0479, "step": 12850 }, { "epoch": 0.02, "learning_rate": 4.9196250000000005e-05, "loss": 2.8499, "step": 12860 }, { "epoch": 0.02, "learning_rate": 4.9195625e-05, "loss": 3.0119, "step": 12870 }, { "epoch": 0.02, "learning_rate": 4.9195e-05, "loss": 2.6405, "step": 12880 }, { "epoch": 0.02, "learning_rate": 4.9194375e-05, "loss": 2.6932, "step": 12890 }, { "epoch": 0.02, "learning_rate": 4.9193750000000007e-05, "loss": 3.0618, "step": 12900 }, { "epoch": 0.02, "learning_rate": 4.9193125000000003e-05, "loss": 3.067, "step": 12910 }, { "epoch": 0.02, "learning_rate": 4.91925e-05, "loss": 2.7462, "step": 12920 }, { "epoch": 0.02, "learning_rate": 4.9191875000000004e-05, "loss": 2.8699, "step": 12930 }, { "epoch": 0.02, "learning_rate": 4.919125e-05, "loss": 3.0433, "step": 12940 }, { "epoch": 0.02, "learning_rate": 4.9190625e-05, "loss": 3.2487, "step": 12950 }, { "epoch": 0.02, "learning_rate": 4.919e-05, "loss": 3.2311, "step": 12960 }, { "epoch": 0.02, "learning_rate": 4.9189375000000005e-05, "loss": 2.5745, "step": 12970 }, { "epoch": 0.02, "learning_rate": 4.918875e-05, "loss": 2.9283, "step": 12980 }, { "epoch": 0.02, "learning_rate": 4.9188125e-05, "loss": 3.0513, "step": 12990 }, { "epoch": 0.02, "learning_rate": 4.91875e-05, "loss": 3.0708, "step": 13000 }, { "epoch": 0.02, "eval_loss": 2.7813737392425537, "eval_runtime": 1291.0921, "eval_samples_per_second": 44.951, "eval_steps_per_second": 5.619, "step": 13000 }, { "epoch": 0.02, "learning_rate": 4.918687500000001e-05, "loss": 2.8165, "step": 13010 }, { "epoch": 0.02, "learning_rate": 4.9186250000000004e-05, "loss": 2.8699, "step": 13020 }, { "epoch": 0.02, "learning_rate": 4.9185625e-05, "loss": 2.8334, "step": 13030 }, { "epoch": 0.02, "learning_rate": 4.9185000000000004e-05, "loss": 2.8677, "step": 13040 }, { "epoch": 0.02, "learning_rate": 4.918437500000001e-05, "loss": 3.0485, "step": 13050 }, { "epoch": 0.02, "learning_rate": 4.918375e-05, "loss": 2.937, "step": 13060 }, { "epoch": 0.02, "learning_rate": 4.9183125e-05, "loss": 2.7071, "step": 13070 }, { "epoch": 0.02, "learning_rate": 4.9182500000000006e-05, "loss": 2.6373, "step": 13080 }, { "epoch": 0.02, "learning_rate": 4.9181875e-05, "loss": 3.0923, "step": 13090 }, { "epoch": 0.02, "learning_rate": 4.918125e-05, "loss": 2.921, "step": 13100 }, { "epoch": 0.02, "learning_rate": 4.9180625e-05, "loss": 3.1064, "step": 13110 }, { "epoch": 0.02, "learning_rate": 4.918000000000001e-05, "loss": 2.8849, "step": 13120 }, { "epoch": 0.02, "learning_rate": 4.9179375e-05, "loss": 2.7596, "step": 13130 }, { "epoch": 0.02, "learning_rate": 4.917875e-05, "loss": 2.7495, "step": 13140 }, { "epoch": 0.02, "learning_rate": 4.9178125000000004e-05, "loss": 2.9096, "step": 13150 }, { "epoch": 0.02, "learning_rate": 4.91775e-05, "loss": 3.0813, "step": 13160 }, { "epoch": 0.02, "learning_rate": 4.9176875e-05, "loss": 2.9572, "step": 13170 }, { "epoch": 0.02, "learning_rate": 4.917625e-05, "loss": 2.8779, "step": 13180 }, { "epoch": 0.02, "learning_rate": 4.9175625000000006e-05, "loss": 3.0485, "step": 13190 }, { "epoch": 0.02, "learning_rate": 4.9175e-05, "loss": 2.9377, "step": 13200 }, { "epoch": 0.02, "learning_rate": 4.9174375e-05, "loss": 3.0486, "step": 13210 }, { "epoch": 0.02, "learning_rate": 4.917375e-05, "loss": 2.7736, "step": 13220 }, { "epoch": 0.02, "learning_rate": 4.9173125e-05, "loss": 3.0231, "step": 13230 }, { "epoch": 0.02, "learning_rate": 4.9172500000000004e-05, "loss": 3.0176, "step": 13240 }, { "epoch": 0.02, "learning_rate": 4.9171875e-05, "loss": 2.9529, "step": 13250 }, { "epoch": 0.02, "learning_rate": 4.9171250000000005e-05, "loss": 3.1179, "step": 13260 }, { "epoch": 0.02, "learning_rate": 4.9170625e-05, "loss": 2.7672, "step": 13270 }, { "epoch": 0.02, "learning_rate": 4.9170000000000005e-05, "loss": 2.9213, "step": 13280 }, { "epoch": 0.02, "learning_rate": 4.9169375e-05, "loss": 2.7064, "step": 13290 }, { "epoch": 0.02, "learning_rate": 4.916875e-05, "loss": 2.7942, "step": 13300 }, { "epoch": 0.02, "learning_rate": 4.9168125e-05, "loss": 2.7852, "step": 13310 }, { "epoch": 0.02, "learning_rate": 4.91675e-05, "loss": 2.9432, "step": 13320 }, { "epoch": 0.02, "learning_rate": 4.9166875000000003e-05, "loss": 2.9927, "step": 13330 }, { "epoch": 0.02, "learning_rate": 4.916625e-05, "loss": 2.7665, "step": 13340 }, { "epoch": 0.02, "learning_rate": 4.9165625000000004e-05, "loss": 2.8335, "step": 13350 }, { "epoch": 0.02, "learning_rate": 4.9165e-05, "loss": 3.0379, "step": 13360 }, { "epoch": 0.02, "learning_rate": 4.9164375000000005e-05, "loss": 2.9696, "step": 13370 }, { "epoch": 0.02, "learning_rate": 4.916375e-05, "loss": 2.8452, "step": 13380 }, { "epoch": 0.02, "learning_rate": 4.9163125000000005e-05, "loss": 2.9542, "step": 13390 }, { "epoch": 0.02, "learning_rate": 4.91625e-05, "loss": 2.8811, "step": 13400 }, { "epoch": 0.02, "learning_rate": 4.9161875e-05, "loss": 2.8721, "step": 13410 }, { "epoch": 0.02, "learning_rate": 4.916125e-05, "loss": 2.9939, "step": 13420 }, { "epoch": 0.02, "learning_rate": 4.916062500000001e-05, "loss": 2.9803, "step": 13430 }, { "epoch": 0.02, "learning_rate": 4.9160000000000004e-05, "loss": 2.9277, "step": 13440 }, { "epoch": 0.02, "learning_rate": 4.9159375e-05, "loss": 3.1057, "step": 13450 }, { "epoch": 0.02, "learning_rate": 4.9158750000000004e-05, "loss": 2.635, "step": 13460 }, { "epoch": 0.02, "learning_rate": 4.915812500000001e-05, "loss": 3.0945, "step": 13470 }, { "epoch": 0.02, "learning_rate": 4.91575e-05, "loss": 2.9596, "step": 13480 }, { "epoch": 0.02, "learning_rate": 4.9156875e-05, "loss": 2.8895, "step": 13490 }, { "epoch": 0.02, "learning_rate": 4.9156250000000006e-05, "loss": 2.8866, "step": 13500 }, { "epoch": 0.02, "learning_rate": 4.9155625e-05, "loss": 2.9176, "step": 13510 }, { "epoch": 0.02, "learning_rate": 4.9155e-05, "loss": 2.5653, "step": 13520 }, { "epoch": 0.02, "learning_rate": 4.9154375e-05, "loss": 2.8112, "step": 13530 }, { "epoch": 0.02, "learning_rate": 4.915375000000001e-05, "loss": 2.6158, "step": 13540 }, { "epoch": 0.02, "learning_rate": 4.9153125e-05, "loss": 3.0132, "step": 13550 }, { "epoch": 0.02, "learning_rate": 4.91525e-05, "loss": 2.9344, "step": 13560 }, { "epoch": 0.02, "learning_rate": 4.9151875000000004e-05, "loss": 2.6639, "step": 13570 }, { "epoch": 0.02, "learning_rate": 4.915125e-05, "loss": 2.9821, "step": 13580 }, { "epoch": 0.02, "learning_rate": 4.9150625e-05, "loss": 2.9943, "step": 13590 }, { "epoch": 0.02, "learning_rate": 4.915e-05, "loss": 3.1098, "step": 13600 }, { "epoch": 0.02, "learning_rate": 4.9149375000000006e-05, "loss": 3.0714, "step": 13610 }, { "epoch": 0.02, "learning_rate": 4.914875e-05, "loss": 2.814, "step": 13620 }, { "epoch": 0.02, "learning_rate": 4.9148125e-05, "loss": 2.8466, "step": 13630 }, { "epoch": 0.02, "learning_rate": 4.91475e-05, "loss": 3.0146, "step": 13640 }, { "epoch": 0.02, "learning_rate": 4.9146875e-05, "loss": 2.6098, "step": 13650 }, { "epoch": 0.02, "learning_rate": 4.9146250000000004e-05, "loss": 2.8639, "step": 13660 }, { "epoch": 0.02, "learning_rate": 4.9145625e-05, "loss": 2.4896, "step": 13670 }, { "epoch": 0.02, "learning_rate": 4.9145000000000005e-05, "loss": 3.0011, "step": 13680 }, { "epoch": 0.02, "learning_rate": 4.9144375e-05, "loss": 2.803, "step": 13690 }, { "epoch": 0.02, "learning_rate": 4.9143750000000005e-05, "loss": 2.7534, "step": 13700 }, { "epoch": 0.02, "learning_rate": 4.9143125e-05, "loss": 2.7145, "step": 13710 }, { "epoch": 0.02, "learning_rate": 4.91425e-05, "loss": 2.892, "step": 13720 }, { "epoch": 0.02, "learning_rate": 4.9141875e-05, "loss": 2.9333, "step": 13730 }, { "epoch": 0.02, "learning_rate": 4.914125e-05, "loss": 3.1241, "step": 13740 }, { "epoch": 0.02, "learning_rate": 4.9140625000000004e-05, "loss": 2.75, "step": 13750 }, { "epoch": 0.02, "learning_rate": 4.914e-05, "loss": 2.8514, "step": 13760 }, { "epoch": 0.02, "learning_rate": 4.9139375000000004e-05, "loss": 2.9492, "step": 13770 }, { "epoch": 0.02, "learning_rate": 4.913875e-05, "loss": 2.7484, "step": 13780 }, { "epoch": 0.02, "learning_rate": 4.9138125000000005e-05, "loss": 2.7673, "step": 13790 }, { "epoch": 0.02, "learning_rate": 4.91375e-05, "loss": 2.9044, "step": 13800 }, { "epoch": 0.02, "learning_rate": 4.9136875000000005e-05, "loss": 2.9115, "step": 13810 }, { "epoch": 0.02, "learning_rate": 4.913625e-05, "loss": 2.6977, "step": 13820 }, { "epoch": 0.02, "learning_rate": 4.9135625e-05, "loss": 2.6948, "step": 13830 }, { "epoch": 0.02, "learning_rate": 4.9135e-05, "loss": 2.9551, "step": 13840 }, { "epoch": 0.02, "learning_rate": 4.913437500000001e-05, "loss": 2.8382, "step": 13850 }, { "epoch": 0.02, "learning_rate": 4.9133750000000004e-05, "loss": 2.5074, "step": 13860 }, { "epoch": 0.02, "learning_rate": 4.9133125e-05, "loss": 3.0042, "step": 13870 }, { "epoch": 0.02, "learning_rate": 4.9132500000000004e-05, "loss": 2.7412, "step": 13880 }, { "epoch": 0.02, "learning_rate": 4.913187500000001e-05, "loss": 2.5494, "step": 13890 }, { "epoch": 0.02, "learning_rate": 4.913125e-05, "loss": 2.7924, "step": 13900 }, { "epoch": 0.02, "learning_rate": 4.9130625e-05, "loss": 2.9895, "step": 13910 }, { "epoch": 0.02, "learning_rate": 4.9130000000000006e-05, "loss": 2.9434, "step": 13920 }, { "epoch": 0.02, "learning_rate": 4.9129375e-05, "loss": 2.7507, "step": 13930 }, { "epoch": 0.02, "learning_rate": 4.912875e-05, "loss": 2.5789, "step": 13940 }, { "epoch": 0.02, "learning_rate": 4.9128125e-05, "loss": 2.6028, "step": 13950 }, { "epoch": 0.02, "learning_rate": 4.912750000000001e-05, "loss": 2.5901, "step": 13960 }, { "epoch": 0.02, "learning_rate": 4.9126875e-05, "loss": 2.4629, "step": 13970 }, { "epoch": 0.02, "learning_rate": 4.912625e-05, "loss": 2.7467, "step": 13980 }, { "epoch": 0.02, "learning_rate": 4.9125625000000005e-05, "loss": 3.2225, "step": 13990 }, { "epoch": 0.02, "learning_rate": 4.9125e-05, "loss": 2.9054, "step": 14000 }, { "epoch": 0.02, "eval_loss": 2.6749837398529053, "eval_runtime": 1290.9638, "eval_samples_per_second": 44.956, "eval_steps_per_second": 5.62, "step": 14000 }, { "epoch": 0.02, "learning_rate": 4.9124375e-05, "loss": 2.5837, "step": 14010 }, { "epoch": 0.02, "learning_rate": 4.912375e-05, "loss": 2.5908, "step": 14020 }, { "epoch": 0.02, "learning_rate": 4.9123125000000006e-05, "loss": 2.5649, "step": 14030 }, { "epoch": 0.02, "learning_rate": 4.91225e-05, "loss": 2.6649, "step": 14040 }, { "epoch": 0.02, "learning_rate": 4.9121875e-05, "loss": 2.7297, "step": 14050 }, { "epoch": 0.02, "learning_rate": 4.9121250000000003e-05, "loss": 2.7665, "step": 14060 }, { "epoch": 0.02, "learning_rate": 4.9120625e-05, "loss": 2.991, "step": 14070 }, { "epoch": 0.02, "learning_rate": 4.9120000000000004e-05, "loss": 3.0699, "step": 14080 }, { "epoch": 0.02, "learning_rate": 4.9119375e-05, "loss": 2.6899, "step": 14090 }, { "epoch": 0.02, "learning_rate": 4.9118750000000005e-05, "loss": 2.3959, "step": 14100 }, { "epoch": 0.02, "learning_rate": 4.9118125e-05, "loss": 2.8429, "step": 14110 }, { "epoch": 0.02, "learning_rate": 4.91175e-05, "loss": 2.8712, "step": 14120 }, { "epoch": 0.02, "learning_rate": 4.9116875e-05, "loss": 2.6129, "step": 14130 }, { "epoch": 0.02, "learning_rate": 4.911625e-05, "loss": 2.7031, "step": 14140 }, { "epoch": 0.02, "learning_rate": 4.9115625e-05, "loss": 2.6171, "step": 14150 }, { "epoch": 0.02, "learning_rate": 4.9115e-05, "loss": 2.6806, "step": 14160 }, { "epoch": 0.02, "learning_rate": 4.9114375000000004e-05, "loss": 2.589, "step": 14170 }, { "epoch": 0.02, "learning_rate": 4.911375e-05, "loss": 2.7601, "step": 14180 }, { "epoch": 0.02, "learning_rate": 4.9113125000000004e-05, "loss": 2.6848, "step": 14190 }, { "epoch": 0.02, "learning_rate": 4.91125e-05, "loss": 2.6525, "step": 14200 }, { "epoch": 0.02, "learning_rate": 4.9111875000000005e-05, "loss": 2.8716, "step": 14210 }, { "epoch": 0.02, "learning_rate": 4.911125e-05, "loss": 2.582, "step": 14220 }, { "epoch": 0.02, "learning_rate": 4.9110625000000006e-05, "loss": 2.8345, "step": 14230 }, { "epoch": 0.02, "learning_rate": 4.911e-05, "loss": 2.7354, "step": 14240 }, { "epoch": 0.02, "learning_rate": 4.9109375e-05, "loss": 2.7099, "step": 14250 }, { "epoch": 0.02, "learning_rate": 4.910875e-05, "loss": 2.5057, "step": 14260 }, { "epoch": 0.02, "learning_rate": 4.910812500000001e-05, "loss": 2.7597, "step": 14270 }, { "epoch": 0.02, "learning_rate": 4.9107500000000004e-05, "loss": 2.7338, "step": 14280 }, { "epoch": 0.02, "learning_rate": 4.9106875e-05, "loss": 2.4919, "step": 14290 }, { "epoch": 0.02, "learning_rate": 4.9106250000000004e-05, "loss": 2.5683, "step": 14300 }, { "epoch": 0.02, "learning_rate": 4.910562500000001e-05, "loss": 2.5857, "step": 14310 }, { "epoch": 0.02, "learning_rate": 4.9105e-05, "loss": 2.5945, "step": 14320 }, { "epoch": 0.02, "learning_rate": 4.9104375e-05, "loss": 2.8768, "step": 14330 }, { "epoch": 0.02, "learning_rate": 4.9103750000000006e-05, "loss": 2.7584, "step": 14340 }, { "epoch": 0.02, "learning_rate": 4.9103125e-05, "loss": 2.8377, "step": 14350 }, { "epoch": 0.02, "learning_rate": 4.91025e-05, "loss": 2.8573, "step": 14360 }, { "epoch": 0.02, "learning_rate": 4.9101875e-05, "loss": 2.7623, "step": 14370 }, { "epoch": 0.02, "learning_rate": 4.910125000000001e-05, "loss": 2.8397, "step": 14380 }, { "epoch": 0.02, "learning_rate": 4.9100625e-05, "loss": 2.6243, "step": 14390 }, { "epoch": 0.02, "learning_rate": 4.91e-05, "loss": 2.6289, "step": 14400 }, { "epoch": 0.02, "learning_rate": 4.9099375000000005e-05, "loss": 2.8688, "step": 14410 }, { "epoch": 0.02, "learning_rate": 4.909875e-05, "loss": 2.6879, "step": 14420 }, { "epoch": 0.02, "learning_rate": 4.9098125e-05, "loss": 2.7816, "step": 14430 }, { "epoch": 0.02, "learning_rate": 4.90975e-05, "loss": 2.7272, "step": 14440 }, { "epoch": 0.02, "learning_rate": 4.9096875000000006e-05, "loss": 3.0019, "step": 14450 }, { "epoch": 0.02, "learning_rate": 4.909625e-05, "loss": 2.5757, "step": 14460 }, { "epoch": 0.02, "learning_rate": 4.9095625e-05, "loss": 2.687, "step": 14470 }, { "epoch": 0.02, "learning_rate": 4.9095000000000003e-05, "loss": 2.9618, "step": 14480 }, { "epoch": 0.02, "learning_rate": 4.9094375e-05, "loss": 2.7559, "step": 14490 }, { "epoch": 0.02, "learning_rate": 4.9093750000000004e-05, "loss": 2.5938, "step": 14500 }, { "epoch": 0.02, "learning_rate": 4.9093125e-05, "loss": 2.6396, "step": 14510 }, { "epoch": 0.02, "learning_rate": 4.9092500000000005e-05, "loss": 2.4885, "step": 14520 }, { "epoch": 0.02, "learning_rate": 4.9091875e-05, "loss": 2.3722, "step": 14530 }, { "epoch": 0.02, "learning_rate": 4.909125e-05, "loss": 2.5715, "step": 14540 }, { "epoch": 0.02, "learning_rate": 4.9090625e-05, "loss": 2.5744, "step": 14550 }, { "epoch": 0.02, "learning_rate": 4.9090000000000006e-05, "loss": 2.7874, "step": 14560 }, { "epoch": 0.02, "learning_rate": 4.9089375e-05, "loss": 2.571, "step": 14570 }, { "epoch": 0.02, "learning_rate": 4.908875e-05, "loss": 2.6225, "step": 14580 }, { "epoch": 0.02, "learning_rate": 4.9088125000000004e-05, "loss": 2.5392, "step": 14590 }, { "epoch": 0.02, "learning_rate": 4.90875e-05, "loss": 2.825, "step": 14600 }, { "epoch": 0.02, "learning_rate": 4.9086875000000004e-05, "loss": 2.4171, "step": 14610 }, { "epoch": 0.02, "learning_rate": 4.908625e-05, "loss": 2.7003, "step": 14620 }, { "epoch": 0.02, "learning_rate": 4.9085625000000005e-05, "loss": 2.5579, "step": 14630 }, { "epoch": 0.02, "learning_rate": 4.9085e-05, "loss": 2.8341, "step": 14640 }, { "epoch": 0.02, "learning_rate": 4.9084375000000006e-05, "loss": 2.6007, "step": 14650 }, { "epoch": 0.02, "learning_rate": 4.908375e-05, "loss": 2.5834, "step": 14660 }, { "epoch": 0.02, "learning_rate": 4.9083125e-05, "loss": 2.6363, "step": 14670 }, { "epoch": 0.02, "learning_rate": 4.90825e-05, "loss": 2.1591, "step": 14680 }, { "epoch": 0.02, "learning_rate": 4.908187500000001e-05, "loss": 2.5829, "step": 14690 }, { "epoch": 0.02, "learning_rate": 4.9081250000000004e-05, "loss": 2.568, "step": 14700 }, { "epoch": 0.02, "learning_rate": 4.9080625e-05, "loss": 2.5674, "step": 14710 }, { "epoch": 0.02, "learning_rate": 4.9080000000000004e-05, "loss": 3.0467, "step": 14720 }, { "epoch": 0.02, "learning_rate": 4.9079375e-05, "loss": 2.7763, "step": 14730 }, { "epoch": 0.02, "learning_rate": 4.907875e-05, "loss": 2.504, "step": 14740 }, { "epoch": 0.02, "learning_rate": 4.9078125e-05, "loss": 2.4532, "step": 14750 }, { "epoch": 0.02, "learning_rate": 4.9077500000000006e-05, "loss": 2.5822, "step": 14760 }, { "epoch": 0.02, "learning_rate": 4.9076875e-05, "loss": 2.7954, "step": 14770 }, { "epoch": 0.02, "learning_rate": 4.907625e-05, "loss": 2.6088, "step": 14780 }, { "epoch": 0.02, "learning_rate": 4.9075625e-05, "loss": 2.6456, "step": 14790 }, { "epoch": 0.02, "learning_rate": 4.907500000000001e-05, "loss": 2.9884, "step": 14800 }, { "epoch": 0.02, "learning_rate": 4.9074375e-05, "loss": 2.7033, "step": 14810 }, { "epoch": 0.02, "learning_rate": 4.907375e-05, "loss": 2.6741, "step": 14820 }, { "epoch": 0.02, "learning_rate": 4.9073125000000005e-05, "loss": 2.7656, "step": 14830 }, { "epoch": 0.02, "learning_rate": 4.90725e-05, "loss": 2.6418, "step": 14840 }, { "epoch": 0.02, "learning_rate": 4.9071875e-05, "loss": 2.6544, "step": 14850 }, { "epoch": 0.02, "learning_rate": 4.907125e-05, "loss": 2.7269, "step": 14860 }, { "epoch": 0.02, "learning_rate": 4.9070625000000006e-05, "loss": 2.7914, "step": 14870 }, { "epoch": 0.02, "learning_rate": 4.907e-05, "loss": 2.7932, "step": 14880 }, { "epoch": 0.02, "learning_rate": 4.9069375e-05, "loss": 3.0636, "step": 14890 }, { "epoch": 0.02, "learning_rate": 4.9068750000000003e-05, "loss": 2.5489, "step": 14900 }, { "epoch": 0.02, "learning_rate": 4.9068125e-05, "loss": 2.7673, "step": 14910 }, { "epoch": 0.02, "learning_rate": 4.90675e-05, "loss": 2.5083, "step": 14920 }, { "epoch": 0.02, "learning_rate": 4.9066875e-05, "loss": 2.6733, "step": 14930 }, { "epoch": 0.02, "learning_rate": 4.9066250000000005e-05, "loss": 2.8056, "step": 14940 }, { "epoch": 0.02, "learning_rate": 4.9065625e-05, "loss": 2.3809, "step": 14950 }, { "epoch": 0.02, "learning_rate": 4.9065e-05, "loss": 2.6266, "step": 14960 }, { "epoch": 0.02, "learning_rate": 4.9064375e-05, "loss": 2.9067, "step": 14970 }, { "epoch": 0.02, "learning_rate": 4.9063750000000006e-05, "loss": 2.7121, "step": 14980 }, { "epoch": 0.02, "learning_rate": 4.9063125e-05, "loss": 2.9192, "step": 14990 }, { "epoch": 0.02, "learning_rate": 4.90625e-05, "loss": 2.8296, "step": 15000 }, { "epoch": 0.02, "eval_loss": 2.582259178161621, "eval_runtime": 1290.2582, "eval_samples_per_second": 44.98, "eval_steps_per_second": 5.623, "step": 15000 }, { "epoch": 0.02, "learning_rate": 4.9061875000000004e-05, "loss": 2.7399, "step": 15010 }, { "epoch": 0.02, "learning_rate": 4.906125e-05, "loss": 2.768, "step": 15020 }, { "epoch": 0.02, "learning_rate": 4.9060625000000004e-05, "loss": 2.8903, "step": 15030 }, { "epoch": 0.02, "learning_rate": 4.906e-05, "loss": 2.6387, "step": 15040 }, { "epoch": 0.02, "learning_rate": 4.9059375000000005e-05, "loss": 2.7289, "step": 15050 }, { "epoch": 0.02, "learning_rate": 4.905875e-05, "loss": 2.6187, "step": 15060 }, { "epoch": 0.02, "learning_rate": 4.9058125000000006e-05, "loss": 2.8434, "step": 15070 }, { "epoch": 0.02, "learning_rate": 4.90575e-05, "loss": 2.6991, "step": 15080 }, { "epoch": 0.02, "learning_rate": 4.9056875e-05, "loss": 2.4814, "step": 15090 }, { "epoch": 0.02, "learning_rate": 4.905625e-05, "loss": 2.9566, "step": 15100 }, { "epoch": 0.02, "learning_rate": 4.905562500000001e-05, "loss": 2.6882, "step": 15110 }, { "epoch": 0.02, "learning_rate": 4.9055000000000004e-05, "loss": 2.8896, "step": 15120 }, { "epoch": 0.02, "learning_rate": 4.9054375e-05, "loss": 2.7369, "step": 15130 }, { "epoch": 0.02, "learning_rate": 4.9053750000000004e-05, "loss": 2.5184, "step": 15140 }, { "epoch": 0.02, "learning_rate": 4.9053125e-05, "loss": 2.4872, "step": 15150 }, { "epoch": 0.02, "learning_rate": 4.90525e-05, "loss": 2.5547, "step": 15160 }, { "epoch": 0.02, "learning_rate": 4.9051875e-05, "loss": 2.5908, "step": 15170 }, { "epoch": 0.02, "learning_rate": 4.9051250000000006e-05, "loss": 2.5078, "step": 15180 }, { "epoch": 0.02, "learning_rate": 4.9050625e-05, "loss": 2.5902, "step": 15190 }, { "epoch": 0.02, "learning_rate": 4.905e-05, "loss": 2.854, "step": 15200 }, { "epoch": 0.02, "learning_rate": 4.9049375e-05, "loss": 2.5185, "step": 15210 }, { "epoch": 0.02, "learning_rate": 4.904875000000001e-05, "loss": 2.5728, "step": 15220 }, { "epoch": 0.02, "learning_rate": 4.9048125e-05, "loss": 2.8556, "step": 15230 }, { "epoch": 0.02, "learning_rate": 4.90475e-05, "loss": 2.695, "step": 15240 }, { "epoch": 0.02, "learning_rate": 4.9046875000000005e-05, "loss": 2.6546, "step": 15250 }, { "epoch": 0.02, "learning_rate": 4.904625e-05, "loss": 2.6413, "step": 15260 }, { "epoch": 0.02, "learning_rate": 4.9045625e-05, "loss": 2.7465, "step": 15270 }, { "epoch": 0.02, "learning_rate": 4.9045e-05, "loss": 2.9357, "step": 15280 }, { "epoch": 0.02, "learning_rate": 4.9044375000000006e-05, "loss": 2.8877, "step": 15290 }, { "epoch": 0.02, "learning_rate": 4.904375e-05, "loss": 2.7875, "step": 15300 }, { "epoch": 0.02, "learning_rate": 4.9043125e-05, "loss": 2.7323, "step": 15310 }, { "epoch": 0.02, "learning_rate": 4.9042500000000004e-05, "loss": 2.546, "step": 15320 }, { "epoch": 0.02, "learning_rate": 4.9041875e-05, "loss": 2.5875, "step": 15330 }, { "epoch": 0.02, "learning_rate": 4.904125e-05, "loss": 2.7816, "step": 15340 }, { "epoch": 0.02, "learning_rate": 4.9040625e-05, "loss": 2.203, "step": 15350 }, { "epoch": 0.02, "learning_rate": 4.9040000000000005e-05, "loss": 2.8168, "step": 15360 }, { "epoch": 0.02, "learning_rate": 4.9039375e-05, "loss": 2.4811, "step": 15370 }, { "epoch": 0.02, "learning_rate": 4.903875e-05, "loss": 2.8464, "step": 15380 }, { "epoch": 0.02, "learning_rate": 4.9038125e-05, "loss": 2.5762, "step": 15390 }, { "epoch": 0.02, "learning_rate": 4.9037500000000006e-05, "loss": 2.7452, "step": 15400 }, { "epoch": 0.02, "learning_rate": 4.9036875e-05, "loss": 2.619, "step": 15410 }, { "epoch": 0.02, "learning_rate": 4.903625e-05, "loss": 2.6376, "step": 15420 }, { "epoch": 0.02, "learning_rate": 4.9035625000000004e-05, "loss": 2.607, "step": 15430 }, { "epoch": 0.02, "learning_rate": 4.9035e-05, "loss": 2.6564, "step": 15440 }, { "epoch": 0.02, "learning_rate": 4.9034375000000004e-05, "loss": 3.0532, "step": 15450 }, { "epoch": 0.02, "learning_rate": 4.903375e-05, "loss": 2.8436, "step": 15460 }, { "epoch": 0.02, "learning_rate": 4.9033125000000005e-05, "loss": 2.7677, "step": 15470 }, { "epoch": 0.02, "learning_rate": 4.90325e-05, "loss": 2.6006, "step": 15480 }, { "epoch": 0.02, "learning_rate": 4.9031875000000006e-05, "loss": 2.4156, "step": 15490 }, { "epoch": 0.02, "learning_rate": 4.903125e-05, "loss": 2.8186, "step": 15500 }, { "epoch": 0.02, "learning_rate": 4.9030625e-05, "loss": 2.8003, "step": 15510 }, { "epoch": 0.02, "learning_rate": 4.903e-05, "loss": 2.591, "step": 15520 }, { "epoch": 0.02, "learning_rate": 4.9029375e-05, "loss": 2.5875, "step": 15530 }, { "epoch": 0.02, "learning_rate": 4.9028750000000004e-05, "loss": 2.6418, "step": 15540 }, { "epoch": 0.02, "learning_rate": 4.9028125e-05, "loss": 2.4225, "step": 15550 }, { "epoch": 0.02, "learning_rate": 4.9027500000000005e-05, "loss": 2.4539, "step": 15560 }, { "epoch": 0.02, "learning_rate": 4.9026875e-05, "loss": 2.772, "step": 15570 }, { "epoch": 0.02, "learning_rate": 4.902625e-05, "loss": 2.9025, "step": 15580 }, { "epoch": 0.02, "learning_rate": 4.9025625e-05, "loss": 2.4599, "step": 15590 }, { "epoch": 0.02, "learning_rate": 4.9025000000000006e-05, "loss": 2.5949, "step": 15600 }, { "epoch": 0.02, "learning_rate": 4.9024375e-05, "loss": 2.719, "step": 15610 }, { "epoch": 0.02, "learning_rate": 4.902375e-05, "loss": 2.905, "step": 15620 }, { "epoch": 0.02, "learning_rate": 4.9023125e-05, "loss": 2.8562, "step": 15630 }, { "epoch": 0.02, "learning_rate": 4.902250000000001e-05, "loss": 2.9112, "step": 15640 }, { "epoch": 0.02, "learning_rate": 4.9021875e-05, "loss": 2.5921, "step": 15650 }, { "epoch": 0.02, "learning_rate": 4.902125e-05, "loss": 2.648, "step": 15660 }, { "epoch": 0.02, "learning_rate": 4.9020625000000005e-05, "loss": 2.8241, "step": 15670 }, { "epoch": 0.02, "learning_rate": 4.902e-05, "loss": 2.8667, "step": 15680 }, { "epoch": 0.02, "learning_rate": 4.9019375e-05, "loss": 2.1954, "step": 15690 }, { "epoch": 0.02, "learning_rate": 4.901875e-05, "loss": 2.5605, "step": 15700 }, { "epoch": 0.02, "learning_rate": 4.9018125000000006e-05, "loss": 2.3522, "step": 15710 }, { "epoch": 0.02, "learning_rate": 4.90175e-05, "loss": 2.5578, "step": 15720 }, { "epoch": 0.02, "learning_rate": 4.9016875e-05, "loss": 2.2441, "step": 15730 }, { "epoch": 0.02, "learning_rate": 4.9016250000000004e-05, "loss": 2.4577, "step": 15740 }, { "epoch": 0.02, "learning_rate": 4.901562500000001e-05, "loss": 2.5527, "step": 15750 }, { "epoch": 0.02, "learning_rate": 4.9015e-05, "loss": 2.8992, "step": 15760 }, { "epoch": 0.02, "learning_rate": 4.9014375e-05, "loss": 2.7292, "step": 15770 }, { "epoch": 0.02, "learning_rate": 4.9013750000000005e-05, "loss": 3.1055, "step": 15780 }, { "epoch": 0.02, "learning_rate": 4.9013125e-05, "loss": 2.964, "step": 15790 }, { "epoch": 0.02, "learning_rate": 4.90125e-05, "loss": 2.6776, "step": 15800 }, { "epoch": 0.02, "learning_rate": 4.9011875e-05, "loss": 2.5582, "step": 15810 }, { "epoch": 0.02, "learning_rate": 4.9011250000000006e-05, "loss": 2.5314, "step": 15820 }, { "epoch": 0.02, "learning_rate": 4.9010625e-05, "loss": 2.6508, "step": 15830 }, { "epoch": 0.02, "learning_rate": 4.901e-05, "loss": 2.3967, "step": 15840 }, { "epoch": 0.02, "learning_rate": 4.9009375000000004e-05, "loss": 3.0418, "step": 15850 }, { "epoch": 0.02, "learning_rate": 4.900875e-05, "loss": 2.4721, "step": 15860 }, { "epoch": 0.02, "learning_rate": 4.9008125000000004e-05, "loss": 2.7166, "step": 15870 }, { "epoch": 0.02, "learning_rate": 4.90075e-05, "loss": 2.6688, "step": 15880 }, { "epoch": 0.02, "learning_rate": 4.9006875000000005e-05, "loss": 3.0, "step": 15890 }, { "epoch": 0.02, "learning_rate": 4.900625e-05, "loss": 2.7725, "step": 15900 }, { "epoch": 0.02, "learning_rate": 4.9005625000000006e-05, "loss": 2.5319, "step": 15910 }, { "epoch": 0.02, "learning_rate": 4.9005e-05, "loss": 2.4069, "step": 15920 }, { "epoch": 0.02, "learning_rate": 4.9004375e-05, "loss": 2.6238, "step": 15930 }, { "epoch": 0.02, "learning_rate": 4.900375e-05, "loss": 2.8392, "step": 15940 }, { "epoch": 0.02, "learning_rate": 4.9003125e-05, "loss": 2.6517, "step": 15950 }, { "epoch": 0.02, "learning_rate": 4.9002500000000004e-05, "loss": 2.85, "step": 15960 }, { "epoch": 0.02, "learning_rate": 4.9001875e-05, "loss": 2.6566, "step": 15970 }, { "epoch": 0.02, "learning_rate": 4.9001250000000005e-05, "loss": 2.356, "step": 15980 }, { "epoch": 0.02, "learning_rate": 4.9000625e-05, "loss": 2.8514, "step": 15990 }, { "epoch": 0.02, "learning_rate": 4.9e-05, "loss": 2.2976, "step": 16000 }, { "epoch": 0.02, "eval_loss": 2.4929113388061523, "eval_runtime": 1290.3377, "eval_samples_per_second": 44.977, "eval_steps_per_second": 5.623, "step": 16000 }, { "epoch": 0.02, "learning_rate": 4.8999375e-05, "loss": 2.7762, "step": 16010 }, { "epoch": 0.02, "learning_rate": 4.8998750000000006e-05, "loss": 2.5577, "step": 16020 }, { "epoch": 0.02, "learning_rate": 4.8998125e-05, "loss": 2.4271, "step": 16030 }, { "epoch": 0.02, "learning_rate": 4.89975e-05, "loss": 2.4648, "step": 16040 }, { "epoch": 0.02, "learning_rate": 4.8996875000000003e-05, "loss": 2.5327, "step": 16050 }, { "epoch": 0.02, "learning_rate": 4.899625000000001e-05, "loss": 2.3696, "step": 16060 }, { "epoch": 0.02, "learning_rate": 4.8995625e-05, "loss": 2.9112, "step": 16070 }, { "epoch": 0.02, "learning_rate": 4.8995e-05, "loss": 2.7018, "step": 16080 }, { "epoch": 0.02, "learning_rate": 4.8994375000000005e-05, "loss": 2.7662, "step": 16090 }, { "epoch": 0.02, "learning_rate": 4.899375e-05, "loss": 2.4624, "step": 16100 }, { "epoch": 0.02, "learning_rate": 4.8993125e-05, "loss": 2.6158, "step": 16110 }, { "epoch": 0.02, "learning_rate": 4.89925e-05, "loss": 2.5821, "step": 16120 }, { "epoch": 0.02, "learning_rate": 4.8991875000000006e-05, "loss": 2.9413, "step": 16130 }, { "epoch": 0.02, "learning_rate": 4.899125e-05, "loss": 2.7114, "step": 16140 }, { "epoch": 0.02, "learning_rate": 4.8990625e-05, "loss": 2.3002, "step": 16150 }, { "epoch": 0.02, "learning_rate": 4.8990000000000004e-05, "loss": 2.6734, "step": 16160 }, { "epoch": 0.02, "learning_rate": 4.898937500000001e-05, "loss": 2.4721, "step": 16170 }, { "epoch": 0.02, "learning_rate": 4.898875e-05, "loss": 2.5045, "step": 16180 }, { "epoch": 0.02, "learning_rate": 4.8988125e-05, "loss": 2.5806, "step": 16190 }, { "epoch": 0.02, "learning_rate": 4.8987500000000005e-05, "loss": 2.5661, "step": 16200 }, { "epoch": 0.02, "learning_rate": 4.8986875e-05, "loss": 2.5768, "step": 16210 }, { "epoch": 0.02, "learning_rate": 4.898625e-05, "loss": 2.7179, "step": 16220 }, { "epoch": 0.02, "learning_rate": 4.8985625e-05, "loss": 2.4915, "step": 16230 }, { "epoch": 0.02, "learning_rate": 4.8985000000000006e-05, "loss": 2.4805, "step": 16240 }, { "epoch": 0.02, "learning_rate": 4.8984375e-05, "loss": 2.3566, "step": 16250 }, { "epoch": 0.02, "learning_rate": 4.898375e-05, "loss": 2.5632, "step": 16260 }, { "epoch": 0.02, "learning_rate": 4.8983125000000004e-05, "loss": 2.889, "step": 16270 }, { "epoch": 0.02, "learning_rate": 4.89825e-05, "loss": 2.6335, "step": 16280 }, { "epoch": 0.02, "learning_rate": 4.8981875000000004e-05, "loss": 2.5953, "step": 16290 }, { "epoch": 0.02, "learning_rate": 4.898125e-05, "loss": 2.5337, "step": 16300 }, { "epoch": 0.02, "learning_rate": 4.8980625000000005e-05, "loss": 2.5967, "step": 16310 }, { "epoch": 0.02, "learning_rate": 4.898e-05, "loss": 2.703, "step": 16320 }, { "epoch": 0.02, "learning_rate": 4.8979375000000006e-05, "loss": 2.5311, "step": 16330 }, { "epoch": 0.02, "learning_rate": 4.897875e-05, "loss": 2.8072, "step": 16340 }, { "epoch": 0.02, "learning_rate": 4.8978125e-05, "loss": 2.5001, "step": 16350 }, { "epoch": 0.02, "learning_rate": 4.89775e-05, "loss": 2.7506, "step": 16360 }, { "epoch": 0.02, "learning_rate": 4.8976875e-05, "loss": 2.7103, "step": 16370 }, { "epoch": 0.02, "learning_rate": 4.8976250000000004e-05, "loss": 2.5521, "step": 16380 }, { "epoch": 0.02, "learning_rate": 4.8975625e-05, "loss": 2.6595, "step": 16390 }, { "epoch": 0.02, "learning_rate": 4.8975000000000005e-05, "loss": 2.4046, "step": 16400 }, { "epoch": 0.02, "learning_rate": 4.8974375e-05, "loss": 2.7818, "step": 16410 }, { "epoch": 0.02, "learning_rate": 4.897375e-05, "loss": 2.4486, "step": 16420 }, { "epoch": 0.02, "learning_rate": 4.8973125e-05, "loss": 2.2736, "step": 16430 }, { "epoch": 0.02, "learning_rate": 4.8972500000000006e-05, "loss": 2.9559, "step": 16440 }, { "epoch": 0.02, "learning_rate": 4.8971875e-05, "loss": 2.4986, "step": 16450 }, { "epoch": 0.02, "learning_rate": 4.897125e-05, "loss": 2.4672, "step": 16460 }, { "epoch": 0.02, "learning_rate": 4.8970625000000003e-05, "loss": 2.6911, "step": 16470 }, { "epoch": 0.02, "learning_rate": 4.897000000000001e-05, "loss": 2.6333, "step": 16480 }, { "epoch": 0.02, "learning_rate": 4.8969375e-05, "loss": 2.6808, "step": 16490 }, { "epoch": 0.02, "learning_rate": 4.896875e-05, "loss": 2.7001, "step": 16500 }, { "epoch": 0.02, "learning_rate": 4.8968125000000005e-05, "loss": 2.4086, "step": 16510 }, { "epoch": 0.02, "learning_rate": 4.89675e-05, "loss": 2.3836, "step": 16520 }, { "epoch": 0.02, "learning_rate": 4.8966875e-05, "loss": 2.6516, "step": 16530 }, { "epoch": 0.02, "learning_rate": 4.896625e-05, "loss": 2.8111, "step": 16540 }, { "epoch": 0.02, "learning_rate": 4.8965625000000006e-05, "loss": 2.6961, "step": 16550 }, { "epoch": 0.02, "learning_rate": 4.8965e-05, "loss": 2.2921, "step": 16560 }, { "epoch": 0.02, "learning_rate": 4.8964375e-05, "loss": 2.4377, "step": 16570 }, { "epoch": 0.02, "learning_rate": 4.8963750000000004e-05, "loss": 2.3989, "step": 16580 }, { "epoch": 0.02, "learning_rate": 4.896312500000001e-05, "loss": 2.5814, "step": 16590 }, { "epoch": 0.02, "learning_rate": 4.89625e-05, "loss": 2.4593, "step": 16600 }, { "epoch": 0.02, "learning_rate": 4.8961875e-05, "loss": 2.4566, "step": 16610 }, { "epoch": 0.02, "learning_rate": 4.8961250000000005e-05, "loss": 2.7457, "step": 16620 }, { "epoch": 0.02, "learning_rate": 4.8960625e-05, "loss": 2.4088, "step": 16630 }, { "epoch": 0.02, "learning_rate": 4.896e-05, "loss": 2.7596, "step": 16640 }, { "epoch": 0.02, "learning_rate": 4.8959375e-05, "loss": 2.4464, "step": 16650 }, { "epoch": 0.02, "learning_rate": 4.8958750000000006e-05, "loss": 2.3342, "step": 16660 }, { "epoch": 0.02, "learning_rate": 4.8958125e-05, "loss": 2.7267, "step": 16670 }, { "epoch": 0.02, "learning_rate": 4.89575e-05, "loss": 2.5061, "step": 16680 }, { "epoch": 0.02, "learning_rate": 4.8956875000000004e-05, "loss": 2.5773, "step": 16690 }, { "epoch": 0.02, "learning_rate": 4.895625e-05, "loss": 2.7869, "step": 16700 }, { "epoch": 0.02, "learning_rate": 4.8955625000000004e-05, "loss": 2.6725, "step": 16710 }, { "epoch": 0.02, "learning_rate": 4.8955e-05, "loss": 2.3466, "step": 16720 }, { "epoch": 0.02, "learning_rate": 4.8954375000000005e-05, "loss": 2.4134, "step": 16730 }, { "epoch": 0.02, "learning_rate": 4.895375e-05, "loss": 2.6744, "step": 16740 }, { "epoch": 0.02, "learning_rate": 4.8953125e-05, "loss": 2.7119, "step": 16750 }, { "epoch": 0.02, "learning_rate": 4.89525e-05, "loss": 2.8039, "step": 16760 }, { "epoch": 0.02, "learning_rate": 4.8951875e-05, "loss": 2.6431, "step": 16770 }, { "epoch": 0.02, "learning_rate": 4.895125e-05, "loss": 2.5865, "step": 16780 }, { "epoch": 0.02, "learning_rate": 4.8950625e-05, "loss": 2.7621, "step": 16790 }, { "epoch": 0.02, "learning_rate": 4.8950000000000004e-05, "loss": 2.5456, "step": 16800 }, { "epoch": 0.02, "learning_rate": 4.8949375e-05, "loss": 2.3872, "step": 16810 }, { "epoch": 0.02, "learning_rate": 4.8948750000000005e-05, "loss": 2.3969, "step": 16820 }, { "epoch": 0.02, "learning_rate": 4.8948125e-05, "loss": 2.4773, "step": 16830 }, { "epoch": 0.02, "learning_rate": 4.89475e-05, "loss": 2.6536, "step": 16840 }, { "epoch": 0.02, "learning_rate": 4.8946875e-05, "loss": 2.339, "step": 16850 }, { "epoch": 0.02, "learning_rate": 4.8946250000000006e-05, "loss": 2.4016, "step": 16860 }, { "epoch": 0.02, "learning_rate": 4.8945625e-05, "loss": 2.322, "step": 16870 }, { "epoch": 0.02, "learning_rate": 4.8945e-05, "loss": 2.2405, "step": 16880 }, { "epoch": 0.02, "learning_rate": 4.8944375000000004e-05, "loss": 2.1527, "step": 16890 }, { "epoch": 0.02, "learning_rate": 4.894375000000001e-05, "loss": 2.6174, "step": 16900 }, { "epoch": 0.02, "learning_rate": 4.8943125000000004e-05, "loss": 2.5548, "step": 16910 }, { "epoch": 0.02, "learning_rate": 4.89425e-05, "loss": 2.7806, "step": 16920 }, { "epoch": 0.02, "learning_rate": 4.8941875000000005e-05, "loss": 2.7504, "step": 16930 }, { "epoch": 0.02, "learning_rate": 4.894125e-05, "loss": 2.4435, "step": 16940 }, { "epoch": 0.02, "learning_rate": 4.8940625e-05, "loss": 2.5278, "step": 16950 }, { "epoch": 0.02, "learning_rate": 4.894e-05, "loss": 2.8752, "step": 16960 }, { "epoch": 0.02, "learning_rate": 4.8939375000000006e-05, "loss": 2.6004, "step": 16970 }, { "epoch": 0.02, "learning_rate": 4.893875e-05, "loss": 2.535, "step": 16980 }, { "epoch": 0.02, "learning_rate": 4.8938125e-05, "loss": 2.3233, "step": 16990 }, { "epoch": 0.02, "learning_rate": 4.8937500000000004e-05, "loss": 2.4564, "step": 17000 }, { "epoch": 0.02, "eval_loss": 2.423708200454712, "eval_runtime": 1290.3674, "eval_samples_per_second": 44.976, "eval_steps_per_second": 5.622, "step": 17000 }, { "epoch": 0.02, "learning_rate": 4.893687500000001e-05, "loss": 2.4155, "step": 17010 }, { "epoch": 0.02, "learning_rate": 4.893625e-05, "loss": 2.5934, "step": 17020 }, { "epoch": 0.02, "learning_rate": 4.8935625e-05, "loss": 2.6323, "step": 17030 }, { "epoch": 0.02, "learning_rate": 4.8935000000000005e-05, "loss": 2.9814, "step": 17040 }, { "epoch": 0.02, "learning_rate": 4.8934375e-05, "loss": 2.2281, "step": 17050 }, { "epoch": 0.02, "learning_rate": 4.893375e-05, "loss": 2.8294, "step": 17060 }, { "epoch": 0.02, "learning_rate": 4.8933125e-05, "loss": 2.5084, "step": 17070 }, { "epoch": 0.02, "learning_rate": 4.8932500000000006e-05, "loss": 2.724, "step": 17080 }, { "epoch": 0.02, "learning_rate": 4.8931875e-05, "loss": 2.2164, "step": 17090 }, { "epoch": 0.02, "learning_rate": 4.893125e-05, "loss": 2.1859, "step": 17100 }, { "epoch": 0.02, "learning_rate": 4.8930625000000004e-05, "loss": 2.6202, "step": 17110 }, { "epoch": 0.02, "learning_rate": 4.893e-05, "loss": 2.5255, "step": 17120 }, { "epoch": 0.02, "learning_rate": 4.8929375000000004e-05, "loss": 2.8065, "step": 17130 }, { "epoch": 0.02, "learning_rate": 4.892875e-05, "loss": 2.7457, "step": 17140 }, { "epoch": 0.02, "learning_rate": 4.8928125000000005e-05, "loss": 2.6274, "step": 17150 }, { "epoch": 0.02, "learning_rate": 4.89275e-05, "loss": 2.552, "step": 17160 }, { "epoch": 0.02, "learning_rate": 4.8926875e-05, "loss": 2.4959, "step": 17170 }, { "epoch": 0.02, "learning_rate": 4.892625e-05, "loss": 2.6067, "step": 17180 }, { "epoch": 0.02, "learning_rate": 4.8925625e-05, "loss": 2.73, "step": 17190 }, { "epoch": 0.02, "learning_rate": 4.8925e-05, "loss": 2.4185, "step": 17200 }, { "epoch": 0.02, "learning_rate": 4.8924375e-05, "loss": 2.358, "step": 17210 }, { "epoch": 0.02, "learning_rate": 4.8923750000000004e-05, "loss": 2.5773, "step": 17220 }, { "epoch": 0.02, "learning_rate": 4.8923125e-05, "loss": 2.5016, "step": 17230 }, { "epoch": 0.02, "learning_rate": 4.8922500000000005e-05, "loss": 2.6724, "step": 17240 }, { "epoch": 0.02, "learning_rate": 4.8921875e-05, "loss": 2.4934, "step": 17250 }, { "epoch": 0.02, "learning_rate": 4.892125e-05, "loss": 2.4501, "step": 17260 }, { "epoch": 0.02, "learning_rate": 4.8920625e-05, "loss": 2.6738, "step": 17270 }, { "epoch": 0.02, "learning_rate": 4.8920000000000006e-05, "loss": 2.3536, "step": 17280 }, { "epoch": 0.02, "learning_rate": 4.8919375e-05, "loss": 2.645, "step": 17290 }, { "epoch": 0.02, "learning_rate": 4.891875e-05, "loss": 2.5596, "step": 17300 }, { "epoch": 0.02, "learning_rate": 4.8918125000000004e-05, "loss": 2.4132, "step": 17310 }, { "epoch": 0.02, "learning_rate": 4.891750000000001e-05, "loss": 2.489, "step": 17320 }, { "epoch": 0.02, "learning_rate": 4.8916875000000004e-05, "loss": 2.5053, "step": 17330 }, { "epoch": 0.02, "learning_rate": 4.891625e-05, "loss": 2.4435, "step": 17340 }, { "epoch": 0.02, "learning_rate": 4.8915625000000005e-05, "loss": 2.6929, "step": 17350 }, { "epoch": 0.02, "learning_rate": 4.8915e-05, "loss": 2.5605, "step": 17360 }, { "epoch": 0.02, "learning_rate": 4.8914375e-05, "loss": 2.3623, "step": 17370 }, { "epoch": 0.02, "learning_rate": 4.891375e-05, "loss": 2.6638, "step": 17380 }, { "epoch": 0.02, "learning_rate": 4.8913125000000006e-05, "loss": 2.5111, "step": 17390 }, { "epoch": 0.02, "learning_rate": 4.89125e-05, "loss": 2.5772, "step": 17400 }, { "epoch": 0.02, "learning_rate": 4.8911875e-05, "loss": 2.7349, "step": 17410 }, { "epoch": 0.02, "learning_rate": 4.8911250000000004e-05, "loss": 2.8456, "step": 17420 }, { "epoch": 0.02, "learning_rate": 4.891062500000001e-05, "loss": 2.6385, "step": 17430 }, { "epoch": 0.02, "learning_rate": 4.891e-05, "loss": 2.2741, "step": 17440 }, { "epoch": 0.02, "learning_rate": 4.8909375e-05, "loss": 2.6875, "step": 17450 }, { "epoch": 0.02, "learning_rate": 4.8908750000000005e-05, "loss": 2.5092, "step": 17460 }, { "epoch": 0.02, "learning_rate": 4.8908125e-05, "loss": 2.5131, "step": 17470 }, { "epoch": 0.02, "learning_rate": 4.89075e-05, "loss": 2.6311, "step": 17480 }, { "epoch": 0.02, "learning_rate": 4.8906875e-05, "loss": 2.3845, "step": 17490 }, { "epoch": 0.02, "learning_rate": 4.8906250000000006e-05, "loss": 2.6637, "step": 17500 }, { "epoch": 0.02, "learning_rate": 4.8905625e-05, "loss": 2.2941, "step": 17510 }, { "epoch": 0.02, "learning_rate": 4.8905e-05, "loss": 2.4365, "step": 17520 }, { "epoch": 0.02, "learning_rate": 4.8904375000000004e-05, "loss": 2.7494, "step": 17530 }, { "epoch": 0.02, "learning_rate": 4.890375e-05, "loss": 2.5332, "step": 17540 }, { "epoch": 0.02, "learning_rate": 4.8903125e-05, "loss": 2.5088, "step": 17550 }, { "epoch": 0.02, "learning_rate": 4.89025e-05, "loss": 2.6672, "step": 17560 }, { "epoch": 0.02, "learning_rate": 4.8901875000000005e-05, "loss": 2.4189, "step": 17570 }, { "epoch": 0.02, "learning_rate": 4.890125e-05, "loss": 2.4737, "step": 17580 }, { "epoch": 0.02, "learning_rate": 4.8900625e-05, "loss": 2.7027, "step": 17590 }, { "epoch": 0.02, "learning_rate": 4.89e-05, "loss": 2.6053, "step": 17600 }, { "epoch": 0.02, "learning_rate": 4.8899375e-05, "loss": 2.5795, "step": 17610 }, { "epoch": 0.02, "learning_rate": 4.8898750000000003e-05, "loss": 2.445, "step": 17620 }, { "epoch": 0.02, "learning_rate": 4.8898125e-05, "loss": 2.4485, "step": 17630 }, { "epoch": 0.02, "learning_rate": 4.8897500000000004e-05, "loss": 2.5361, "step": 17640 }, { "epoch": 0.02, "learning_rate": 4.8896875e-05, "loss": 2.687, "step": 17650 }, { "epoch": 0.02, "learning_rate": 4.8896250000000005e-05, "loss": 2.6723, "step": 17660 }, { "epoch": 0.02, "learning_rate": 4.8895625e-05, "loss": 2.5734, "step": 17670 }, { "epoch": 0.02, "learning_rate": 4.8895e-05, "loss": 2.5413, "step": 17680 }, { "epoch": 0.02, "learning_rate": 4.8894375e-05, "loss": 2.4104, "step": 17690 }, { "epoch": 0.02, "learning_rate": 4.8893750000000006e-05, "loss": 2.095, "step": 17700 }, { "epoch": 0.02, "learning_rate": 4.8893125e-05, "loss": 2.4454, "step": 17710 }, { "epoch": 0.02, "learning_rate": 4.88925e-05, "loss": 2.3517, "step": 17720 }, { "epoch": 0.02, "learning_rate": 4.8891875000000004e-05, "loss": 2.4443, "step": 17730 }, { "epoch": 0.02, "learning_rate": 4.889125000000001e-05, "loss": 2.8372, "step": 17740 }, { "epoch": 0.02, "learning_rate": 4.8890625000000004e-05, "loss": 2.5336, "step": 17750 }, { "epoch": 0.02, "learning_rate": 4.889e-05, "loss": 2.2857, "step": 17760 }, { "epoch": 0.02, "learning_rate": 4.8889375000000005e-05, "loss": 2.3698, "step": 17770 }, { "epoch": 0.02, "learning_rate": 4.888875e-05, "loss": 2.6875, "step": 17780 }, { "epoch": 0.02, "learning_rate": 4.8888125e-05, "loss": 2.5567, "step": 17790 }, { "epoch": 0.02, "learning_rate": 4.88875e-05, "loss": 2.6278, "step": 17800 }, { "epoch": 0.02, "learning_rate": 4.8886875000000006e-05, "loss": 2.4606, "step": 17810 }, { "epoch": 0.02, "learning_rate": 4.888625e-05, "loss": 2.6622, "step": 17820 }, { "epoch": 0.02, "learning_rate": 4.8885625e-05, "loss": 2.5298, "step": 17830 }, { "epoch": 0.02, "learning_rate": 4.8885000000000004e-05, "loss": 2.577, "step": 17840 }, { "epoch": 0.02, "learning_rate": 4.888437500000001e-05, "loss": 2.0774, "step": 17850 }, { "epoch": 0.02, "learning_rate": 4.888375e-05, "loss": 2.3797, "step": 17860 }, { "epoch": 0.02, "learning_rate": 4.8883125e-05, "loss": 2.4047, "step": 17870 }, { "epoch": 0.02, "learning_rate": 4.8882500000000005e-05, "loss": 2.3207, "step": 17880 }, { "epoch": 0.02, "learning_rate": 4.8881875e-05, "loss": 2.1892, "step": 17890 }, { "epoch": 0.02, "learning_rate": 4.888125e-05, "loss": 2.627, "step": 17900 }, { "epoch": 0.02, "learning_rate": 4.8880625e-05, "loss": 2.5002, "step": 17910 }, { "epoch": 0.02, "learning_rate": 4.8880000000000006e-05, "loss": 2.6301, "step": 17920 }, { "epoch": 0.02, "learning_rate": 4.8879375e-05, "loss": 2.1983, "step": 17930 }, { "epoch": 0.02, "learning_rate": 4.887875e-05, "loss": 2.3258, "step": 17940 }, { "epoch": 0.02, "learning_rate": 4.8878125000000004e-05, "loss": 2.5379, "step": 17950 }, { "epoch": 0.03, "learning_rate": 4.88775e-05, "loss": 2.1678, "step": 17960 }, { "epoch": 0.03, "learning_rate": 4.8876875e-05, "loss": 2.3281, "step": 17970 }, { "epoch": 0.03, "learning_rate": 4.887625e-05, "loss": 2.379, "step": 17980 }, { "epoch": 0.03, "learning_rate": 4.8875625000000005e-05, "loss": 2.3988, "step": 17990 }, { "epoch": 0.03, "learning_rate": 4.8875e-05, "loss": 2.76, "step": 18000 }, { "epoch": 0.03, "eval_loss": 2.3622002601623535, "eval_runtime": 1290.8431, "eval_samples_per_second": 44.96, "eval_steps_per_second": 5.62, "step": 18000 }, { "epoch": 0.03, "learning_rate": 4.8874375e-05, "loss": 2.5218, "step": 18010 }, { "epoch": 0.03, "learning_rate": 4.887375e-05, "loss": 2.7261, "step": 18020 }, { "epoch": 0.03, "learning_rate": 4.8873125e-05, "loss": 2.5675, "step": 18030 }, { "epoch": 0.03, "learning_rate": 4.8872500000000003e-05, "loss": 2.2968, "step": 18040 }, { "epoch": 0.03, "learning_rate": 4.8871875e-05, "loss": 2.2868, "step": 18050 }, { "epoch": 0.03, "learning_rate": 4.8871250000000004e-05, "loss": 2.7499, "step": 18060 }, { "epoch": 0.03, "learning_rate": 4.8870625e-05, "loss": 2.7739, "step": 18070 }, { "epoch": 0.03, "learning_rate": 4.8870000000000005e-05, "loss": 2.4248, "step": 18080 }, { "epoch": 0.03, "learning_rate": 4.8869375e-05, "loss": 2.3956, "step": 18090 }, { "epoch": 0.03, "learning_rate": 4.8868750000000005e-05, "loss": 2.2925, "step": 18100 }, { "epoch": 0.03, "learning_rate": 4.8868125e-05, "loss": 2.5896, "step": 18110 }, { "epoch": 0.03, "learning_rate": 4.8867500000000006e-05, "loss": 2.2441, "step": 18120 }, { "epoch": 0.03, "learning_rate": 4.8866875e-05, "loss": 2.523, "step": 18130 }, { "epoch": 0.03, "learning_rate": 4.886625e-05, "loss": 2.211, "step": 18140 }, { "epoch": 0.03, "learning_rate": 4.8865625000000004e-05, "loss": 2.6031, "step": 18150 }, { "epoch": 0.03, "learning_rate": 4.8865e-05, "loss": 2.5006, "step": 18160 }, { "epoch": 0.03, "learning_rate": 4.8864375000000004e-05, "loss": 2.6601, "step": 18170 }, { "epoch": 0.03, "learning_rate": 4.886375e-05, "loss": 2.5799, "step": 18180 }, { "epoch": 0.03, "learning_rate": 4.8863125000000005e-05, "loss": 2.4199, "step": 18190 }, { "epoch": 0.03, "learning_rate": 4.88625e-05, "loss": 2.4569, "step": 18200 }, { "epoch": 0.03, "learning_rate": 4.8861875e-05, "loss": 2.524, "step": 18210 }, { "epoch": 0.03, "learning_rate": 4.886125e-05, "loss": 2.2731, "step": 18220 }, { "epoch": 0.03, "learning_rate": 4.8860625000000006e-05, "loss": 2.6649, "step": 18230 }, { "epoch": 0.03, "learning_rate": 4.886e-05, "loss": 2.4574, "step": 18240 }, { "epoch": 0.03, "learning_rate": 4.8859375e-05, "loss": 2.4873, "step": 18250 }, { "epoch": 0.03, "learning_rate": 4.8858750000000004e-05, "loss": 2.4466, "step": 18260 }, { "epoch": 0.03, "learning_rate": 4.885812500000001e-05, "loss": 2.5025, "step": 18270 }, { "epoch": 0.03, "learning_rate": 4.88575e-05, "loss": 2.4212, "step": 18280 }, { "epoch": 0.03, "learning_rate": 4.8856875e-05, "loss": 2.2701, "step": 18290 }, { "epoch": 0.03, "learning_rate": 4.8856250000000005e-05, "loss": 2.3432, "step": 18300 }, { "epoch": 0.03, "learning_rate": 4.8855625e-05, "loss": 2.4973, "step": 18310 }, { "epoch": 0.03, "learning_rate": 4.8855e-05, "loss": 2.5243, "step": 18320 }, { "epoch": 0.03, "learning_rate": 4.8854375e-05, "loss": 2.5176, "step": 18330 }, { "epoch": 0.03, "learning_rate": 4.8853750000000006e-05, "loss": 2.4191, "step": 18340 }, { "epoch": 0.03, "learning_rate": 4.8853124999999997e-05, "loss": 2.274, "step": 18350 }, { "epoch": 0.03, "learning_rate": 4.88525e-05, "loss": 2.6769, "step": 18360 }, { "epoch": 0.03, "learning_rate": 4.8851875000000004e-05, "loss": 2.2587, "step": 18370 }, { "epoch": 0.03, "learning_rate": 4.885125e-05, "loss": 2.3508, "step": 18380 }, { "epoch": 0.03, "learning_rate": 4.8850625e-05, "loss": 2.5559, "step": 18390 }, { "epoch": 0.03, "learning_rate": 4.885e-05, "loss": 2.2076, "step": 18400 }, { "epoch": 0.03, "learning_rate": 4.8849375000000005e-05, "loss": 2.4783, "step": 18410 }, { "epoch": 0.03, "learning_rate": 4.884875e-05, "loss": 2.4439, "step": 18420 }, { "epoch": 0.03, "learning_rate": 4.8848125e-05, "loss": 2.5777, "step": 18430 }, { "epoch": 0.03, "learning_rate": 4.88475e-05, "loss": 2.582, "step": 18440 }, { "epoch": 0.03, "learning_rate": 4.8846875e-05, "loss": 2.5681, "step": 18450 }, { "epoch": 0.03, "learning_rate": 4.8846250000000003e-05, "loss": 2.2192, "step": 18460 }, { "epoch": 0.03, "learning_rate": 4.8845625e-05, "loss": 2.3074, "step": 18470 }, { "epoch": 0.03, "learning_rate": 4.8845000000000004e-05, "loss": 2.3684, "step": 18480 }, { "epoch": 0.03, "learning_rate": 4.8844375e-05, "loss": 2.3934, "step": 18490 }, { "epoch": 0.03, "learning_rate": 4.8843750000000005e-05, "loss": 2.4016, "step": 18500 }, { "epoch": 0.03, "learning_rate": 4.8843125e-05, "loss": 2.5863, "step": 18510 }, { "epoch": 0.03, "learning_rate": 4.8842500000000005e-05, "loss": 2.4957, "step": 18520 }, { "epoch": 0.03, "learning_rate": 4.8841875e-05, "loss": 2.3788, "step": 18530 }, { "epoch": 0.03, "learning_rate": 4.8841250000000006e-05, "loss": 2.5726, "step": 18540 }, { "epoch": 0.03, "learning_rate": 4.8840625e-05, "loss": 2.3077, "step": 18550 }, { "epoch": 0.03, "learning_rate": 4.884e-05, "loss": 2.4859, "step": 18560 }, { "epoch": 0.03, "learning_rate": 4.8839375000000004e-05, "loss": 2.4671, "step": 18570 }, { "epoch": 0.03, "learning_rate": 4.883875e-05, "loss": 2.4291, "step": 18580 }, { "epoch": 0.03, "learning_rate": 4.8838125000000004e-05, "loss": 2.448, "step": 18590 }, { "epoch": 0.03, "learning_rate": 4.88375e-05, "loss": 2.4512, "step": 18600 }, { "epoch": 0.03, "learning_rate": 4.8836875000000005e-05, "loss": 1.9776, "step": 18610 }, { "epoch": 0.03, "learning_rate": 4.883625e-05, "loss": 2.6788, "step": 18620 }, { "epoch": 0.03, "learning_rate": 4.8835625e-05, "loss": 2.403, "step": 18630 }, { "epoch": 0.03, "learning_rate": 4.8835e-05, "loss": 2.2626, "step": 18640 }, { "epoch": 0.03, "learning_rate": 4.8834375000000006e-05, "loss": 2.5175, "step": 18650 }, { "epoch": 0.03, "learning_rate": 4.883375e-05, "loss": 2.3175, "step": 18660 }, { "epoch": 0.03, "learning_rate": 4.8833125e-05, "loss": 2.3108, "step": 18670 }, { "epoch": 0.03, "learning_rate": 4.8832500000000004e-05, "loss": 2.4121, "step": 18680 }, { "epoch": 0.03, "learning_rate": 4.883187500000001e-05, "loss": 2.6836, "step": 18690 }, { "epoch": 0.03, "learning_rate": 4.883125e-05, "loss": 2.4697, "step": 18700 }, { "epoch": 0.03, "learning_rate": 4.8830625e-05, "loss": 2.5081, "step": 18710 }, { "epoch": 0.03, "learning_rate": 4.8830000000000005e-05, "loss": 2.234, "step": 18720 }, { "epoch": 0.03, "learning_rate": 4.8829375e-05, "loss": 2.2968, "step": 18730 }, { "epoch": 0.03, "learning_rate": 4.882875e-05, "loss": 1.9723, "step": 18740 }, { "epoch": 0.03, "learning_rate": 4.8828125e-05, "loss": 2.4328, "step": 18750 }, { "epoch": 0.03, "learning_rate": 4.8827500000000006e-05, "loss": 2.45, "step": 18760 }, { "epoch": 0.03, "learning_rate": 4.8826874999999997e-05, "loss": 2.2668, "step": 18770 }, { "epoch": 0.03, "learning_rate": 4.882625e-05, "loss": 2.583, "step": 18780 }, { "epoch": 0.03, "learning_rate": 4.8825625000000004e-05, "loss": 2.3014, "step": 18790 }, { "epoch": 0.03, "learning_rate": 4.8825e-05, "loss": 2.3337, "step": 18800 }, { "epoch": 0.03, "learning_rate": 4.8824375e-05, "loss": 2.2075, "step": 18810 }, { "epoch": 0.03, "learning_rate": 4.882375e-05, "loss": 2.2268, "step": 18820 }, { "epoch": 0.03, "learning_rate": 4.8823125000000005e-05, "loss": 2.3656, "step": 18830 }, { "epoch": 0.03, "learning_rate": 4.88225e-05, "loss": 2.5013, "step": 18840 }, { "epoch": 0.03, "learning_rate": 4.8821875e-05, "loss": 2.3533, "step": 18850 }, { "epoch": 0.03, "learning_rate": 4.882125e-05, "loss": 2.3308, "step": 18860 }, { "epoch": 0.03, "learning_rate": 4.8820625e-05, "loss": 2.7023, "step": 18870 }, { "epoch": 0.03, "learning_rate": 4.8820000000000004e-05, "loss": 2.3365, "step": 18880 }, { "epoch": 0.03, "learning_rate": 4.8819375e-05, "loss": 2.3457, "step": 18890 }, { "epoch": 0.03, "learning_rate": 4.8818750000000004e-05, "loss": 2.3595, "step": 18900 }, { "epoch": 0.03, "learning_rate": 4.8818125e-05, "loss": 2.2147, "step": 18910 }, { "epoch": 0.03, "learning_rate": 4.8817500000000005e-05, "loss": 2.8365, "step": 18920 }, { "epoch": 0.03, "learning_rate": 4.8816875e-05, "loss": 2.3506, "step": 18930 }, { "epoch": 0.03, "learning_rate": 4.8816250000000005e-05, "loss": 2.4074, "step": 18940 }, { "epoch": 0.03, "learning_rate": 4.8815625e-05, "loss": 2.6878, "step": 18950 }, { "epoch": 0.03, "learning_rate": 4.8815e-05, "loss": 2.4812, "step": 18960 }, { "epoch": 0.03, "learning_rate": 4.8814375e-05, "loss": 2.3347, "step": 18970 }, { "epoch": 0.03, "learning_rate": 4.881375e-05, "loss": 2.4096, "step": 18980 }, { "epoch": 0.03, "learning_rate": 4.8813125000000004e-05, "loss": 2.5002, "step": 18990 }, { "epoch": 0.03, "learning_rate": 4.88125e-05, "loss": 2.3611, "step": 19000 }, { "epoch": 0.03, "eval_loss": 2.312469005584717, "eval_runtime": 1291.4426, "eval_samples_per_second": 44.939, "eval_steps_per_second": 5.618, "step": 19000 }, { "epoch": 0.03, "learning_rate": 4.8811875000000004e-05, "loss": 2.2837, "step": 19010 }, { "epoch": 0.03, "learning_rate": 4.881125e-05, "loss": 2.4096, "step": 19020 }, { "epoch": 0.03, "learning_rate": 4.8810625000000005e-05, "loss": 2.105, "step": 19030 }, { "epoch": 0.03, "learning_rate": 4.881e-05, "loss": 2.4291, "step": 19040 }, { "epoch": 0.03, "learning_rate": 4.8809375e-05, "loss": 2.5225, "step": 19050 }, { "epoch": 0.03, "learning_rate": 4.880875e-05, "loss": 2.7026, "step": 19060 }, { "epoch": 0.03, "learning_rate": 4.8808125000000006e-05, "loss": 2.2696, "step": 19070 }, { "epoch": 0.03, "learning_rate": 4.88075e-05, "loss": 2.4002, "step": 19080 }, { "epoch": 0.03, "learning_rate": 4.8806875e-05, "loss": 2.3987, "step": 19090 }, { "epoch": 0.03, "learning_rate": 4.8806250000000004e-05, "loss": 2.519, "step": 19100 }, { "epoch": 0.03, "learning_rate": 4.880562500000001e-05, "loss": 2.3886, "step": 19110 }, { "epoch": 0.03, "learning_rate": 4.8805e-05, "loss": 2.5665, "step": 19120 }, { "epoch": 0.03, "learning_rate": 4.8804375e-05, "loss": 2.6187, "step": 19130 }, { "epoch": 0.03, "learning_rate": 4.8803750000000005e-05, "loss": 2.6529, "step": 19140 }, { "epoch": 0.03, "learning_rate": 4.8803125e-05, "loss": 2.2526, "step": 19150 }, { "epoch": 0.03, "learning_rate": 4.88025e-05, "loss": 2.756, "step": 19160 }, { "epoch": 0.03, "learning_rate": 4.8801875e-05, "loss": 2.1681, "step": 19170 }, { "epoch": 0.03, "learning_rate": 4.8801250000000006e-05, "loss": 2.4062, "step": 19180 }, { "epoch": 0.03, "learning_rate": 4.8800625e-05, "loss": 2.2737, "step": 19190 }, { "epoch": 0.03, "learning_rate": 4.88e-05, "loss": 2.2085, "step": 19200 }, { "epoch": 0.03, "learning_rate": 4.8799375000000004e-05, "loss": 2.5044, "step": 19210 }, { "epoch": 0.03, "learning_rate": 4.879875e-05, "loss": 2.2498, "step": 19220 }, { "epoch": 0.03, "learning_rate": 4.8798125e-05, "loss": 2.2483, "step": 19230 }, { "epoch": 0.03, "learning_rate": 4.87975e-05, "loss": 2.1992, "step": 19240 }, { "epoch": 0.03, "learning_rate": 4.8796875000000005e-05, "loss": 2.5385, "step": 19250 }, { "epoch": 0.03, "learning_rate": 4.879625e-05, "loss": 2.5624, "step": 19260 }, { "epoch": 0.03, "learning_rate": 4.8795625e-05, "loss": 2.3742, "step": 19270 }, { "epoch": 0.03, "learning_rate": 4.8795e-05, "loss": 2.2878, "step": 19280 }, { "epoch": 0.03, "learning_rate": 4.879437500000001e-05, "loss": 2.4579, "step": 19290 }, { "epoch": 0.03, "learning_rate": 4.8793750000000004e-05, "loss": 2.1167, "step": 19300 }, { "epoch": 0.03, "learning_rate": 4.8793125e-05, "loss": 2.3384, "step": 19310 }, { "epoch": 0.03, "learning_rate": 4.8792500000000004e-05, "loss": 2.5749, "step": 19320 }, { "epoch": 0.03, "learning_rate": 4.8791875e-05, "loss": 2.498, "step": 19330 }, { "epoch": 0.03, "learning_rate": 4.8791250000000005e-05, "loss": 2.359, "step": 19340 }, { "epoch": 0.03, "learning_rate": 4.8790625e-05, "loss": 2.3567, "step": 19350 }, { "epoch": 0.03, "learning_rate": 4.8790000000000006e-05, "loss": 2.5637, "step": 19360 }, { "epoch": 0.03, "learning_rate": 4.8789375e-05, "loss": 2.4467, "step": 19370 }, { "epoch": 0.03, "learning_rate": 4.878875e-05, "loss": 2.5859, "step": 19380 }, { "epoch": 0.03, "learning_rate": 4.8788125e-05, "loss": 2.4288, "step": 19390 }, { "epoch": 0.03, "learning_rate": 4.87875e-05, "loss": 2.3486, "step": 19400 }, { "epoch": 0.03, "learning_rate": 4.8786875000000004e-05, "loss": 2.0325, "step": 19410 }, { "epoch": 0.03, "learning_rate": 4.878625e-05, "loss": 2.4623, "step": 19420 }, { "epoch": 0.03, "learning_rate": 4.8785625000000004e-05, "loss": 2.2059, "step": 19430 }, { "epoch": 0.03, "learning_rate": 4.8785e-05, "loss": 2.6933, "step": 19440 }, { "epoch": 0.03, "learning_rate": 4.8784375000000005e-05, "loss": 2.4454, "step": 19450 }, { "epoch": 0.03, "learning_rate": 4.878375e-05, "loss": 2.4795, "step": 19460 }, { "epoch": 0.03, "learning_rate": 4.8783125e-05, "loss": 2.3142, "step": 19470 }, { "epoch": 0.03, "learning_rate": 4.87825e-05, "loss": 2.4202, "step": 19480 }, { "epoch": 0.03, "learning_rate": 4.8781875000000006e-05, "loss": 2.7098, "step": 19490 }, { "epoch": 0.03, "learning_rate": 4.878125e-05, "loss": 2.3359, "step": 19500 }, { "epoch": 0.03, "learning_rate": 4.8780625e-05, "loss": 2.2847, "step": 19510 }, { "epoch": 0.03, "learning_rate": 4.8780000000000004e-05, "loss": 2.0908, "step": 19520 }, { "epoch": 0.03, "learning_rate": 4.877937500000001e-05, "loss": 2.4552, "step": 19530 }, { "epoch": 0.03, "learning_rate": 4.877875e-05, "loss": 2.2058, "step": 19540 }, { "epoch": 0.03, "learning_rate": 4.8778125e-05, "loss": 2.3579, "step": 19550 }, { "epoch": 0.03, "learning_rate": 4.8777500000000005e-05, "loss": 2.6536, "step": 19560 }, { "epoch": 0.03, "learning_rate": 4.8776875e-05, "loss": 2.3224, "step": 19570 }, { "epoch": 0.03, "learning_rate": 4.877625e-05, "loss": 1.9972, "step": 19580 }, { "epoch": 0.03, "learning_rate": 4.8775625e-05, "loss": 2.2876, "step": 19590 }, { "epoch": 0.03, "learning_rate": 4.8775000000000007e-05, "loss": 2.2425, "step": 19600 }, { "epoch": 0.03, "learning_rate": 4.8774375e-05, "loss": 2.4191, "step": 19610 }, { "epoch": 0.03, "learning_rate": 4.877375e-05, "loss": 2.4486, "step": 19620 }, { "epoch": 0.03, "learning_rate": 4.8773125000000004e-05, "loss": 2.3372, "step": 19630 }, { "epoch": 0.03, "learning_rate": 4.87725e-05, "loss": 2.084, "step": 19640 }, { "epoch": 0.03, "learning_rate": 4.8771875e-05, "loss": 2.491, "step": 19650 }, { "epoch": 0.03, "learning_rate": 4.877125e-05, "loss": 2.3352, "step": 19660 }, { "epoch": 0.03, "learning_rate": 4.8770625000000005e-05, "loss": 2.2872, "step": 19670 }, { "epoch": 0.03, "learning_rate": 4.877e-05, "loss": 2.3502, "step": 19680 }, { "epoch": 0.03, "learning_rate": 4.8769375e-05, "loss": 2.4298, "step": 19690 }, { "epoch": 0.03, "learning_rate": 4.876875e-05, "loss": 2.2554, "step": 19700 }, { "epoch": 0.03, "learning_rate": 4.876812500000001e-05, "loss": 2.0552, "step": 19710 }, { "epoch": 0.03, "learning_rate": 4.8767500000000004e-05, "loss": 2.2006, "step": 19720 }, { "epoch": 0.03, "learning_rate": 4.8766875e-05, "loss": 2.2591, "step": 19730 }, { "epoch": 0.03, "learning_rate": 4.8766250000000004e-05, "loss": 2.3346, "step": 19740 }, { "epoch": 0.03, "learning_rate": 4.8765625e-05, "loss": 2.0991, "step": 19750 }, { "epoch": 0.03, "learning_rate": 4.8765e-05, "loss": 2.4276, "step": 19760 }, { "epoch": 0.03, "learning_rate": 4.8764375e-05, "loss": 2.6467, "step": 19770 }, { "epoch": 0.03, "learning_rate": 4.8763750000000006e-05, "loss": 2.7596, "step": 19780 }, { "epoch": 0.03, "learning_rate": 4.8763125e-05, "loss": 2.4871, "step": 19790 }, { "epoch": 0.03, "learning_rate": 4.87625e-05, "loss": 2.2693, "step": 19800 }, { "epoch": 0.03, "learning_rate": 4.8761875e-05, "loss": 2.4479, "step": 19810 }, { "epoch": 0.03, "learning_rate": 4.876125e-05, "loss": 2.2479, "step": 19820 }, { "epoch": 0.03, "learning_rate": 4.8760625000000004e-05, "loss": 2.3955, "step": 19830 }, { "epoch": 0.03, "learning_rate": 4.876e-05, "loss": 2.3623, "step": 19840 }, { "epoch": 0.03, "learning_rate": 4.8759375000000004e-05, "loss": 2.5029, "step": 19850 }, { "epoch": 0.03, "learning_rate": 4.875875e-05, "loss": 2.6117, "step": 19860 }, { "epoch": 0.03, "learning_rate": 4.8758125000000005e-05, "loss": 2.1472, "step": 19870 }, { "epoch": 0.03, "learning_rate": 4.87575e-05, "loss": 2.246, "step": 19880 }, { "epoch": 0.03, "learning_rate": 4.8756875e-05, "loss": 2.3558, "step": 19890 }, { "epoch": 0.03, "learning_rate": 4.875625e-05, "loss": 2.2117, "step": 19900 }, { "epoch": 0.03, "learning_rate": 4.8755625000000006e-05, "loss": 2.5146, "step": 19910 }, { "epoch": 0.03, "learning_rate": 4.8755e-05, "loss": 2.2097, "step": 19920 }, { "epoch": 0.03, "learning_rate": 4.8754375e-05, "loss": 2.3328, "step": 19930 }, { "epoch": 0.03, "learning_rate": 4.8753750000000004e-05, "loss": 2.1629, "step": 19940 }, { "epoch": 0.03, "learning_rate": 4.875312500000001e-05, "loss": 2.1864, "step": 19950 }, { "epoch": 0.03, "learning_rate": 4.87525e-05, "loss": 2.2445, "step": 19960 }, { "epoch": 0.03, "learning_rate": 4.8751875e-05, "loss": 2.5792, "step": 19970 }, { "epoch": 0.03, "learning_rate": 4.8751250000000005e-05, "loss": 2.319, "step": 19980 }, { "epoch": 0.03, "learning_rate": 4.8750625e-05, "loss": 2.2246, "step": 19990 }, { "epoch": 0.03, "learning_rate": 4.875e-05, "loss": 2.5503, "step": 20000 }, { "epoch": 0.03, "eval_loss": 2.260953426361084, "eval_runtime": 1290.4084, "eval_samples_per_second": 44.975, "eval_steps_per_second": 5.622, "step": 20000 }, { "epoch": 0.03, "learning_rate": 4.8749375e-05, "loss": 2.5879, "step": 20010 }, { "epoch": 0.03, "learning_rate": 4.8748750000000007e-05, "loss": 2.308, "step": 20020 }, { "epoch": 0.03, "learning_rate": 4.8748125e-05, "loss": 2.5936, "step": 20030 }, { "epoch": 0.03, "learning_rate": 4.87475e-05, "loss": 2.3681, "step": 20040 }, { "epoch": 0.03, "learning_rate": 4.8746875000000004e-05, "loss": 2.2918, "step": 20050 }, { "epoch": 0.03, "learning_rate": 4.874625e-05, "loss": 2.3842, "step": 20060 }, { "epoch": 0.03, "learning_rate": 4.8745625e-05, "loss": 2.1826, "step": 20070 }, { "epoch": 0.03, "learning_rate": 4.8745e-05, "loss": 2.2453, "step": 20080 }, { "epoch": 0.03, "learning_rate": 4.8744375000000005e-05, "loss": 2.3223, "step": 20090 }, { "epoch": 0.03, "learning_rate": 4.874375e-05, "loss": 2.2807, "step": 20100 }, { "epoch": 0.03, "learning_rate": 4.8743125e-05, "loss": 2.4205, "step": 20110 }, { "epoch": 0.03, "learning_rate": 4.87425e-05, "loss": 2.3504, "step": 20120 }, { "epoch": 0.03, "learning_rate": 4.874187500000001e-05, "loss": 2.5611, "step": 20130 }, { "epoch": 0.03, "learning_rate": 4.8741250000000004e-05, "loss": 2.3078, "step": 20140 }, { "epoch": 0.03, "learning_rate": 4.8740625e-05, "loss": 2.2333, "step": 20150 }, { "epoch": 0.03, "learning_rate": 4.8740000000000004e-05, "loss": 2.3508, "step": 20160 }, { "epoch": 0.03, "learning_rate": 4.8739375e-05, "loss": 2.1648, "step": 20170 }, { "epoch": 0.03, "learning_rate": 4.873875e-05, "loss": 2.2923, "step": 20180 }, { "epoch": 0.03, "learning_rate": 4.8738125e-05, "loss": 2.2962, "step": 20190 }, { "epoch": 0.03, "learning_rate": 4.8737500000000006e-05, "loss": 2.2935, "step": 20200 }, { "epoch": 0.03, "learning_rate": 4.8736875e-05, "loss": 2.1905, "step": 20210 }, { "epoch": 0.03, "learning_rate": 4.873625e-05, "loss": 2.2183, "step": 20220 }, { "epoch": 0.03, "learning_rate": 4.8735625e-05, "loss": 2.2322, "step": 20230 }, { "epoch": 0.03, "learning_rate": 4.8735e-05, "loss": 2.4208, "step": 20240 }, { "epoch": 0.03, "learning_rate": 4.8734375000000004e-05, "loss": 2.2594, "step": 20250 }, { "epoch": 0.03, "learning_rate": 4.873375e-05, "loss": 2.1274, "step": 20260 }, { "epoch": 0.03, "learning_rate": 4.8733125000000004e-05, "loss": 2.2988, "step": 20270 }, { "epoch": 0.03, "learning_rate": 4.87325e-05, "loss": 2.3847, "step": 20280 }, { "epoch": 0.03, "learning_rate": 4.8731875000000005e-05, "loss": 2.2657, "step": 20290 }, { "epoch": 0.03, "learning_rate": 4.873125e-05, "loss": 2.0483, "step": 20300 }, { "epoch": 0.03, "learning_rate": 4.8730625e-05, "loss": 2.5517, "step": 20310 }, { "epoch": 0.03, "learning_rate": 4.873e-05, "loss": 2.3813, "step": 20320 }, { "epoch": 0.03, "learning_rate": 4.8729375000000006e-05, "loss": 2.3064, "step": 20330 }, { "epoch": 0.03, "learning_rate": 4.872875e-05, "loss": 2.3169, "step": 20340 }, { "epoch": 0.03, "learning_rate": 4.8728125e-05, "loss": 2.2973, "step": 20350 }, { "epoch": 0.03, "learning_rate": 4.8727500000000004e-05, "loss": 2.7124, "step": 20360 }, { "epoch": 0.03, "learning_rate": 4.8726875e-05, "loss": 2.1093, "step": 20370 }, { "epoch": 0.03, "learning_rate": 4.872625e-05, "loss": 2.4657, "step": 20380 }, { "epoch": 0.03, "learning_rate": 4.8725625e-05, "loss": 2.361, "step": 20390 }, { "epoch": 0.03, "learning_rate": 4.8725000000000005e-05, "loss": 2.388, "step": 20400 }, { "epoch": 0.03, "learning_rate": 4.8724375e-05, "loss": 2.1121, "step": 20410 }, { "epoch": 0.03, "learning_rate": 4.872375e-05, "loss": 2.2672, "step": 20420 }, { "epoch": 0.03, "learning_rate": 4.8723125e-05, "loss": 2.1018, "step": 20430 }, { "epoch": 0.03, "learning_rate": 4.8722500000000007e-05, "loss": 2.1251, "step": 20440 }, { "epoch": 0.03, "learning_rate": 4.8721875000000004e-05, "loss": 2.4337, "step": 20450 }, { "epoch": 0.03, "learning_rate": 4.872125e-05, "loss": 2.1513, "step": 20460 }, { "epoch": 0.03, "learning_rate": 4.8720625000000004e-05, "loss": 2.3374, "step": 20470 }, { "epoch": 0.03, "learning_rate": 4.872000000000001e-05, "loss": 2.4692, "step": 20480 }, { "epoch": 0.03, "learning_rate": 4.8719375e-05, "loss": 2.3393, "step": 20490 }, { "epoch": 0.03, "learning_rate": 4.871875e-05, "loss": 2.3246, "step": 20500 }, { "epoch": 0.03, "learning_rate": 4.8718125000000005e-05, "loss": 2.3359, "step": 20510 }, { "epoch": 0.03, "learning_rate": 4.87175e-05, "loss": 2.2084, "step": 20520 }, { "epoch": 0.03, "learning_rate": 4.8716875e-05, "loss": 2.2088, "step": 20530 }, { "epoch": 0.03, "learning_rate": 4.871625e-05, "loss": 2.1942, "step": 20540 }, { "epoch": 0.03, "learning_rate": 4.871562500000001e-05, "loss": 2.5257, "step": 20550 }, { "epoch": 0.03, "learning_rate": 4.8715000000000004e-05, "loss": 2.4644, "step": 20560 }, { "epoch": 0.03, "learning_rate": 4.8714375e-05, "loss": 2.1786, "step": 20570 }, { "epoch": 0.03, "learning_rate": 4.8713750000000004e-05, "loss": 2.1099, "step": 20580 }, { "epoch": 0.03, "learning_rate": 4.8713125e-05, "loss": 2.3376, "step": 20590 }, { "epoch": 0.03, "learning_rate": 4.87125e-05, "loss": 2.1029, "step": 20600 }, { "epoch": 0.03, "learning_rate": 4.8711875e-05, "loss": 2.1818, "step": 20610 }, { "epoch": 0.03, "learning_rate": 4.8711250000000006e-05, "loss": 2.2186, "step": 20620 }, { "epoch": 0.03, "learning_rate": 4.8710625e-05, "loss": 2.3552, "step": 20630 }, { "epoch": 0.03, "learning_rate": 4.871e-05, "loss": 2.4427, "step": 20640 }, { "epoch": 0.03, "learning_rate": 4.8709375e-05, "loss": 2.3728, "step": 20650 }, { "epoch": 0.03, "learning_rate": 4.870875e-05, "loss": 2.5456, "step": 20660 }, { "epoch": 0.03, "learning_rate": 4.8708125000000004e-05, "loss": 2.3819, "step": 20670 }, { "epoch": 0.03, "learning_rate": 4.87075e-05, "loss": 2.3262, "step": 20680 }, { "epoch": 0.03, "learning_rate": 4.8706875000000005e-05, "loss": 2.3659, "step": 20690 }, { "epoch": 0.03, "learning_rate": 4.870625e-05, "loss": 2.5074, "step": 20700 }, { "epoch": 0.03, "learning_rate": 4.8705625000000005e-05, "loss": 2.3355, "step": 20710 }, { "epoch": 0.03, "learning_rate": 4.8705e-05, "loss": 2.538, "step": 20720 }, { "epoch": 0.03, "learning_rate": 4.8704375e-05, "loss": 2.3069, "step": 20730 }, { "epoch": 0.03, "learning_rate": 4.870375e-05, "loss": 2.144, "step": 20740 }, { "epoch": 0.03, "learning_rate": 4.8703125000000006e-05, "loss": 2.1511, "step": 20750 }, { "epoch": 0.03, "learning_rate": 4.87025e-05, "loss": 2.1981, "step": 20760 }, { "epoch": 0.03, "learning_rate": 4.8701875e-05, "loss": 2.3216, "step": 20770 }, { "epoch": 0.03, "learning_rate": 4.8701250000000004e-05, "loss": 2.4406, "step": 20780 }, { "epoch": 0.03, "learning_rate": 4.8700625e-05, "loss": 2.1347, "step": 20790 }, { "epoch": 0.03, "learning_rate": 4.87e-05, "loss": 2.5337, "step": 20800 }, { "epoch": 0.03, "learning_rate": 4.8699375e-05, "loss": 2.4846, "step": 20810 }, { "epoch": 0.03, "learning_rate": 4.8698750000000005e-05, "loss": 1.9419, "step": 20820 }, { "epoch": 0.03, "learning_rate": 4.8698125e-05, "loss": 2.1395, "step": 20830 }, { "epoch": 0.03, "learning_rate": 4.86975e-05, "loss": 2.3405, "step": 20840 }, { "epoch": 0.03, "learning_rate": 4.8696875e-05, "loss": 2.3276, "step": 20850 }, { "epoch": 0.03, "learning_rate": 4.869625000000001e-05, "loss": 2.4374, "step": 20860 }, { "epoch": 0.03, "learning_rate": 4.8695625000000004e-05, "loss": 2.3864, "step": 20870 }, { "epoch": 0.03, "learning_rate": 4.8695e-05, "loss": 2.3882, "step": 20880 }, { "epoch": 0.03, "learning_rate": 4.8694375000000004e-05, "loss": 2.2665, "step": 20890 }, { "epoch": 0.03, "learning_rate": 4.869375000000001e-05, "loss": 2.1009, "step": 20900 }, { "epoch": 0.03, "learning_rate": 4.8693125e-05, "loss": 2.4513, "step": 20910 }, { "epoch": 0.03, "learning_rate": 4.86925e-05, "loss": 2.4314, "step": 20920 }, { "epoch": 0.03, "learning_rate": 4.8691875000000005e-05, "loss": 2.2587, "step": 20930 }, { "epoch": 0.03, "learning_rate": 4.869125e-05, "loss": 2.291, "step": 20940 }, { "epoch": 0.03, "learning_rate": 4.8690625e-05, "loss": 1.9941, "step": 20950 }, { "epoch": 0.03, "learning_rate": 4.869e-05, "loss": 2.4618, "step": 20960 }, { "epoch": 0.03, "learning_rate": 4.868937500000001e-05, "loss": 2.2369, "step": 20970 }, { "epoch": 0.03, "learning_rate": 4.868875e-05, "loss": 2.613, "step": 20980 }, { "epoch": 0.03, "learning_rate": 4.8688125e-05, "loss": 2.2336, "step": 20990 }, { "epoch": 0.03, "learning_rate": 4.8687500000000004e-05, "loss": 2.4342, "step": 21000 }, { "epoch": 0.03, "eval_loss": 2.2093613147735596, "eval_runtime": 1290.3187, "eval_samples_per_second": 44.978, "eval_steps_per_second": 5.623, "step": 21000 }, { "epoch": 0.03, "learning_rate": 4.8686875e-05, "loss": 2.233, "step": 21010 }, { "epoch": 0.03, "learning_rate": 4.868625e-05, "loss": 2.487, "step": 21020 }, { "epoch": 0.03, "learning_rate": 4.8685625e-05, "loss": 2.4683, "step": 21030 }, { "epoch": 0.03, "learning_rate": 4.8685000000000006e-05, "loss": 2.3568, "step": 21040 }, { "epoch": 0.03, "learning_rate": 4.8684375e-05, "loss": 2.3289, "step": 21050 }, { "epoch": 0.03, "learning_rate": 4.868375e-05, "loss": 2.3722, "step": 21060 }, { "epoch": 0.03, "learning_rate": 4.8683125e-05, "loss": 2.3702, "step": 21070 }, { "epoch": 0.03, "learning_rate": 4.86825e-05, "loss": 2.472, "step": 21080 }, { "epoch": 0.03, "learning_rate": 4.8681875000000004e-05, "loss": 2.5465, "step": 21090 }, { "epoch": 0.03, "learning_rate": 4.868125e-05, "loss": 2.3161, "step": 21100 }, { "epoch": 0.03, "learning_rate": 4.8680625000000005e-05, "loss": 2.13, "step": 21110 }, { "epoch": 0.03, "learning_rate": 4.868e-05, "loss": 2.4498, "step": 21120 }, { "epoch": 0.03, "learning_rate": 4.8679375000000005e-05, "loss": 2.5581, "step": 21130 }, { "epoch": 0.03, "learning_rate": 4.867875e-05, "loss": 2.415, "step": 21140 }, { "epoch": 0.03, "learning_rate": 4.8678125e-05, "loss": 2.2081, "step": 21150 }, { "epoch": 0.03, "learning_rate": 4.86775e-05, "loss": 2.4024, "step": 21160 }, { "epoch": 0.03, "learning_rate": 4.8676875e-05, "loss": 2.4359, "step": 21170 }, { "epoch": 0.03, "learning_rate": 4.8676250000000003e-05, "loss": 2.3017, "step": 21180 }, { "epoch": 0.03, "learning_rate": 4.8675625e-05, "loss": 2.4693, "step": 21190 }, { "epoch": 0.03, "learning_rate": 4.8675000000000004e-05, "loss": 2.3517, "step": 21200 }, { "epoch": 0.03, "learning_rate": 4.8674375e-05, "loss": 1.9974, "step": 21210 }, { "epoch": 0.03, "learning_rate": 4.867375e-05, "loss": 2.3993, "step": 21220 }, { "epoch": 0.03, "learning_rate": 4.8673125e-05, "loss": 2.283, "step": 21230 }, { "epoch": 0.03, "learning_rate": 4.8672500000000005e-05, "loss": 2.1603, "step": 21240 }, { "epoch": 0.03, "learning_rate": 4.8671875e-05, "loss": 2.1959, "step": 21250 }, { "epoch": 0.03, "learning_rate": 4.867125e-05, "loss": 2.2641, "step": 21260 }, { "epoch": 0.03, "learning_rate": 4.8670625e-05, "loss": 2.424, "step": 21270 }, { "epoch": 0.03, "learning_rate": 4.867000000000001e-05, "loss": 2.371, "step": 21280 }, { "epoch": 0.03, "learning_rate": 4.8669375000000004e-05, "loss": 2.549, "step": 21290 }, { "epoch": 0.03, "learning_rate": 4.866875e-05, "loss": 2.3798, "step": 21300 }, { "epoch": 0.03, "learning_rate": 4.8668125000000004e-05, "loss": 2.3908, "step": 21310 }, { "epoch": 0.03, "learning_rate": 4.866750000000001e-05, "loss": 2.4423, "step": 21320 }, { "epoch": 0.03, "learning_rate": 4.8666875e-05, "loss": 2.4231, "step": 21330 }, { "epoch": 0.03, "learning_rate": 4.866625e-05, "loss": 2.2167, "step": 21340 }, { "epoch": 0.03, "learning_rate": 4.8665625000000006e-05, "loss": 2.506, "step": 21350 }, { "epoch": 0.03, "learning_rate": 4.8665e-05, "loss": 2.7232, "step": 21360 }, { "epoch": 0.03, "learning_rate": 4.8664375e-05, "loss": 2.3675, "step": 21370 }, { "epoch": 0.03, "learning_rate": 4.866375e-05, "loss": 2.3471, "step": 21380 }, { "epoch": 0.03, "learning_rate": 4.866312500000001e-05, "loss": 1.9485, "step": 21390 }, { "epoch": 0.03, "learning_rate": 4.86625e-05, "loss": 2.0076, "step": 21400 }, { "epoch": 0.03, "learning_rate": 4.8661875e-05, "loss": 2.3931, "step": 21410 }, { "epoch": 0.03, "learning_rate": 4.8661250000000004e-05, "loss": 2.3656, "step": 21420 }, { "epoch": 0.03, "learning_rate": 4.8660625e-05, "loss": 2.0859, "step": 21430 }, { "epoch": 0.03, "learning_rate": 4.866e-05, "loss": 2.1523, "step": 21440 }, { "epoch": 0.03, "learning_rate": 4.8659375e-05, "loss": 2.3771, "step": 21450 }, { "epoch": 0.03, "learning_rate": 4.8658750000000006e-05, "loss": 2.1812, "step": 21460 }, { "epoch": 0.03, "learning_rate": 4.8658125e-05, "loss": 2.5232, "step": 21470 }, { "epoch": 0.03, "learning_rate": 4.86575e-05, "loss": 1.9514, "step": 21480 }, { "epoch": 0.03, "learning_rate": 4.8656875e-05, "loss": 2.3563, "step": 21490 }, { "epoch": 0.03, "learning_rate": 4.865625e-05, "loss": 2.0763, "step": 21500 }, { "epoch": 0.03, "learning_rate": 4.8655625000000004e-05, "loss": 2.2868, "step": 21510 }, { "epoch": 0.03, "learning_rate": 4.8655e-05, "loss": 2.2824, "step": 21520 }, { "epoch": 0.03, "learning_rate": 4.8654375000000005e-05, "loss": 2.2368, "step": 21530 }, { "epoch": 0.03, "learning_rate": 4.865375e-05, "loss": 2.1187, "step": 21540 }, { "epoch": 0.03, "learning_rate": 4.8653125000000005e-05, "loss": 2.4042, "step": 21550 }, { "epoch": 0.03, "learning_rate": 4.86525e-05, "loss": 2.3941, "step": 21560 }, { "epoch": 0.03, "learning_rate": 4.8651875e-05, "loss": 2.413, "step": 21570 }, { "epoch": 0.03, "learning_rate": 4.865125e-05, "loss": 2.2152, "step": 21580 }, { "epoch": 0.03, "learning_rate": 4.8650625e-05, "loss": 2.4349, "step": 21590 }, { "epoch": 0.03, "learning_rate": 4.8650000000000003e-05, "loss": 2.3654, "step": 21600 }, { "epoch": 0.03, "learning_rate": 4.8649375e-05, "loss": 2.0259, "step": 21610 }, { "epoch": 0.03, "learning_rate": 4.8648750000000004e-05, "loss": 2.367, "step": 21620 }, { "epoch": 0.03, "learning_rate": 4.8648125e-05, "loss": 2.2524, "step": 21630 }, { "epoch": 0.03, "learning_rate": 4.8647500000000005e-05, "loss": 2.084, "step": 21640 }, { "epoch": 0.03, "learning_rate": 4.8646875e-05, "loss": 2.265, "step": 21650 }, { "epoch": 0.03, "learning_rate": 4.8646250000000005e-05, "loss": 2.2742, "step": 21660 }, { "epoch": 0.03, "learning_rate": 4.8645625e-05, "loss": 2.4076, "step": 21670 }, { "epoch": 0.03, "learning_rate": 4.8645e-05, "loss": 2.0673, "step": 21680 }, { "epoch": 0.03, "learning_rate": 4.8644375e-05, "loss": 2.6222, "step": 21690 }, { "epoch": 0.03, "learning_rate": 4.864375000000001e-05, "loss": 2.0792, "step": 21700 }, { "epoch": 0.03, "learning_rate": 4.8643125000000004e-05, "loss": 2.2572, "step": 21710 }, { "epoch": 0.03, "learning_rate": 4.86425e-05, "loss": 2.3266, "step": 21720 }, { "epoch": 0.03, "learning_rate": 4.8641875000000004e-05, "loss": 2.2862, "step": 21730 }, { "epoch": 0.03, "learning_rate": 4.864125000000001e-05, "loss": 2.2136, "step": 21740 }, { "epoch": 0.03, "learning_rate": 4.8640625e-05, "loss": 2.1342, "step": 21750 }, { "epoch": 0.03, "learning_rate": 4.864e-05, "loss": 2.4094, "step": 21760 }, { "epoch": 0.03, "learning_rate": 4.8639375000000006e-05, "loss": 2.2372, "step": 21770 }, { "epoch": 0.03, "learning_rate": 4.863875e-05, "loss": 2.2077, "step": 21780 }, { "epoch": 0.03, "learning_rate": 4.8638125e-05, "loss": 2.4428, "step": 21790 }, { "epoch": 0.03, "learning_rate": 4.86375e-05, "loss": 2.2329, "step": 21800 }, { "epoch": 0.03, "learning_rate": 4.863687500000001e-05, "loss": 2.2749, "step": 21810 }, { "epoch": 0.03, "learning_rate": 4.863625e-05, "loss": 2.2721, "step": 21820 }, { "epoch": 0.03, "learning_rate": 4.8635625e-05, "loss": 2.2965, "step": 21830 }, { "epoch": 0.03, "learning_rate": 4.8635000000000004e-05, "loss": 2.2071, "step": 21840 }, { "epoch": 0.03, "learning_rate": 4.8634375e-05, "loss": 2.4893, "step": 21850 }, { "epoch": 0.03, "learning_rate": 4.863375e-05, "loss": 2.5977, "step": 21860 }, { "epoch": 0.03, "learning_rate": 4.8633125e-05, "loss": 2.263, "step": 21870 }, { "epoch": 0.03, "learning_rate": 4.8632500000000006e-05, "loss": 2.0125, "step": 21880 }, { "epoch": 0.03, "learning_rate": 4.8631875e-05, "loss": 2.4471, "step": 21890 }, { "epoch": 0.03, "learning_rate": 4.863125e-05, "loss": 2.057, "step": 21900 }, { "epoch": 0.03, "learning_rate": 4.8630625e-05, "loss": 2.3869, "step": 21910 }, { "epoch": 0.03, "learning_rate": 4.863e-05, "loss": 2.2916, "step": 21920 }, { "epoch": 0.03, "learning_rate": 4.8629375000000004e-05, "loss": 2.2401, "step": 21930 }, { "epoch": 0.03, "learning_rate": 4.862875e-05, "loss": 2.2263, "step": 21940 }, { "epoch": 0.03, "learning_rate": 4.8628125000000005e-05, "loss": 2.2489, "step": 21950 }, { "epoch": 0.03, "learning_rate": 4.86275e-05, "loss": 2.139, "step": 21960 }, { "epoch": 0.03, "learning_rate": 4.8626875000000005e-05, "loss": 2.4174, "step": 21970 }, { "epoch": 0.03, "learning_rate": 4.862625e-05, "loss": 2.1562, "step": 21980 }, { "epoch": 0.03, "learning_rate": 4.8625625e-05, "loss": 2.2062, "step": 21990 }, { "epoch": 0.03, "learning_rate": 4.8625e-05, "loss": 2.3405, "step": 22000 }, { "epoch": 0.03, "eval_loss": 2.1775107383728027, "eval_runtime": 1290.4549, "eval_samples_per_second": 44.973, "eval_steps_per_second": 5.622, "step": 22000 }, { "epoch": 0.03, "learning_rate": 4.8624375e-05, "loss": 2.0546, "step": 22010 }, { "epoch": 0.03, "learning_rate": 4.8623750000000004e-05, "loss": 2.0661, "step": 22020 }, { "epoch": 0.03, "learning_rate": 4.8623125e-05, "loss": 2.553, "step": 22030 }, { "epoch": 0.03, "learning_rate": 4.8622500000000004e-05, "loss": 2.1326, "step": 22040 }, { "epoch": 0.03, "learning_rate": 4.8621875e-05, "loss": 2.3638, "step": 22050 }, { "epoch": 0.03, "learning_rate": 4.8621250000000005e-05, "loss": 2.5354, "step": 22060 }, { "epoch": 0.03, "learning_rate": 4.8620625e-05, "loss": 2.3506, "step": 22070 }, { "epoch": 0.03, "learning_rate": 4.8620000000000005e-05, "loss": 2.3234, "step": 22080 }, { "epoch": 0.03, "learning_rate": 4.8619375e-05, "loss": 2.1528, "step": 22090 }, { "epoch": 0.03, "learning_rate": 4.861875e-05, "loss": 2.5783, "step": 22100 }, { "epoch": 0.03, "learning_rate": 4.8618125e-05, "loss": 2.2355, "step": 22110 }, { "epoch": 0.03, "learning_rate": 4.861750000000001e-05, "loss": 2.4917, "step": 22120 }, { "epoch": 0.03, "learning_rate": 4.8616875000000004e-05, "loss": 2.1479, "step": 22130 }, { "epoch": 0.03, "learning_rate": 4.861625e-05, "loss": 2.3621, "step": 22140 }, { "epoch": 0.03, "learning_rate": 4.8615625000000004e-05, "loss": 2.4706, "step": 22150 }, { "epoch": 0.03, "learning_rate": 4.861500000000001e-05, "loss": 2.1382, "step": 22160 }, { "epoch": 0.03, "learning_rate": 4.8614375e-05, "loss": 2.2217, "step": 22170 }, { "epoch": 0.03, "learning_rate": 4.861375e-05, "loss": 2.0158, "step": 22180 }, { "epoch": 0.03, "learning_rate": 4.8613125000000006e-05, "loss": 2.5063, "step": 22190 }, { "epoch": 0.03, "learning_rate": 4.86125e-05, "loss": 2.389, "step": 22200 }, { "epoch": 0.03, "learning_rate": 4.8611875e-05, "loss": 2.1654, "step": 22210 }, { "epoch": 0.03, "learning_rate": 4.861125e-05, "loss": 2.424, "step": 22220 }, { "epoch": 0.03, "learning_rate": 4.861062500000001e-05, "loss": 2.3322, "step": 22230 }, { "epoch": 0.03, "learning_rate": 4.861e-05, "loss": 2.2738, "step": 22240 }, { "epoch": 0.03, "learning_rate": 4.8609375e-05, "loss": 1.9843, "step": 22250 }, { "epoch": 0.03, "learning_rate": 4.8608750000000004e-05, "loss": 2.5145, "step": 22260 }, { "epoch": 0.03, "learning_rate": 4.8608125e-05, "loss": 1.8718, "step": 22270 }, { "epoch": 0.03, "learning_rate": 4.86075e-05, "loss": 2.1117, "step": 22280 }, { "epoch": 0.03, "learning_rate": 4.8606875e-05, "loss": 1.986, "step": 22290 }, { "epoch": 0.03, "learning_rate": 4.8606250000000006e-05, "loss": 1.8881, "step": 22300 }, { "epoch": 0.03, "learning_rate": 4.8605625e-05, "loss": 2.2781, "step": 22310 }, { "epoch": 0.03, "learning_rate": 4.8605e-05, "loss": 2.4261, "step": 22320 }, { "epoch": 0.03, "learning_rate": 4.8604375e-05, "loss": 2.4298, "step": 22330 }, { "epoch": 0.03, "learning_rate": 4.860375e-05, "loss": 1.9964, "step": 22340 }, { "epoch": 0.03, "learning_rate": 4.8603125000000004e-05, "loss": 2.5758, "step": 22350 }, { "epoch": 0.03, "learning_rate": 4.86025e-05, "loss": 1.9528, "step": 22360 }, { "epoch": 0.03, "learning_rate": 4.8601875000000005e-05, "loss": 2.3746, "step": 22370 }, { "epoch": 0.03, "learning_rate": 4.860125e-05, "loss": 2.0225, "step": 22380 }, { "epoch": 0.03, "learning_rate": 4.8600625e-05, "loss": 2.4913, "step": 22390 }, { "epoch": 0.03, "learning_rate": 4.86e-05, "loss": 2.2644, "step": 22400 }, { "epoch": 0.03, "learning_rate": 4.8599375e-05, "loss": 2.5146, "step": 22410 }, { "epoch": 0.03, "learning_rate": 4.859875e-05, "loss": 2.1299, "step": 22420 }, { "epoch": 0.03, "learning_rate": 4.8598125e-05, "loss": 2.0692, "step": 22430 }, { "epoch": 0.03, "learning_rate": 4.8597500000000004e-05, "loss": 2.0313, "step": 22440 }, { "epoch": 0.03, "learning_rate": 4.8596875e-05, "loss": 2.2982, "step": 22450 }, { "epoch": 0.03, "learning_rate": 4.8596250000000004e-05, "loss": 2.3507, "step": 22460 }, { "epoch": 0.03, "learning_rate": 4.8595625e-05, "loss": 1.8823, "step": 22470 }, { "epoch": 0.03, "learning_rate": 4.8595000000000005e-05, "loss": 2.0221, "step": 22480 }, { "epoch": 0.03, "learning_rate": 4.8594375e-05, "loss": 2.2352, "step": 22490 }, { "epoch": 0.03, "learning_rate": 4.8593750000000005e-05, "loss": 2.4219, "step": 22500 }, { "epoch": 0.03, "learning_rate": 4.8593125e-05, "loss": 2.0994, "step": 22510 }, { "epoch": 0.03, "learning_rate": 4.85925e-05, "loss": 2.275, "step": 22520 }, { "epoch": 0.03, "learning_rate": 4.8591875e-05, "loss": 2.3879, "step": 22530 }, { "epoch": 0.03, "learning_rate": 4.859125000000001e-05, "loss": 2.1025, "step": 22540 }, { "epoch": 0.03, "learning_rate": 4.8590625000000004e-05, "loss": 2.1003, "step": 22550 }, { "epoch": 0.03, "learning_rate": 4.859e-05, "loss": 2.3505, "step": 22560 }, { "epoch": 0.03, "learning_rate": 4.8589375000000004e-05, "loss": 2.0997, "step": 22570 }, { "epoch": 0.03, "learning_rate": 4.858875000000001e-05, "loss": 2.3203, "step": 22580 }, { "epoch": 0.03, "learning_rate": 4.8588125e-05, "loss": 2.1672, "step": 22590 }, { "epoch": 0.03, "learning_rate": 4.85875e-05, "loss": 2.2919, "step": 22600 }, { "epoch": 0.03, "learning_rate": 4.8586875000000006e-05, "loss": 2.3548, "step": 22610 }, { "epoch": 0.03, "learning_rate": 4.858625e-05, "loss": 2.0557, "step": 22620 }, { "epoch": 0.03, "learning_rate": 4.8585625e-05, "loss": 2.0826, "step": 22630 }, { "epoch": 0.03, "learning_rate": 4.8585e-05, "loss": 2.5048, "step": 22640 }, { "epoch": 0.03, "learning_rate": 4.858437500000001e-05, "loss": 2.3477, "step": 22650 }, { "epoch": 0.03, "learning_rate": 4.858375e-05, "loss": 2.239, "step": 22660 }, { "epoch": 0.03, "learning_rate": 4.8583125e-05, "loss": 2.1443, "step": 22670 }, { "epoch": 0.03, "learning_rate": 4.8582500000000005e-05, "loss": 2.4111, "step": 22680 }, { "epoch": 0.03, "learning_rate": 4.8581875e-05, "loss": 2.4013, "step": 22690 }, { "epoch": 0.03, "learning_rate": 4.858125e-05, "loss": 2.38, "step": 22700 }, { "epoch": 0.03, "learning_rate": 4.8580625e-05, "loss": 2.1957, "step": 22710 }, { "epoch": 0.03, "learning_rate": 4.8580000000000006e-05, "loss": 2.2254, "step": 22720 }, { "epoch": 0.03, "learning_rate": 4.8579375e-05, "loss": 2.252, "step": 22730 }, { "epoch": 0.03, "learning_rate": 4.857875e-05, "loss": 2.424, "step": 22740 }, { "epoch": 0.03, "learning_rate": 4.8578125000000003e-05, "loss": 2.2479, "step": 22750 }, { "epoch": 0.03, "learning_rate": 4.85775e-05, "loss": 2.2819, "step": 22760 }, { "epoch": 0.03, "learning_rate": 4.8576875000000004e-05, "loss": 2.0261, "step": 22770 }, { "epoch": 0.03, "learning_rate": 4.857625e-05, "loss": 2.0856, "step": 22780 }, { "epoch": 0.03, "learning_rate": 4.8575625000000005e-05, "loss": 2.3566, "step": 22790 }, { "epoch": 0.03, "learning_rate": 4.8575e-05, "loss": 2.1271, "step": 22800 }, { "epoch": 0.03, "learning_rate": 4.8574375e-05, "loss": 2.2451, "step": 22810 }, { "epoch": 0.03, "learning_rate": 4.857375e-05, "loss": 2.0079, "step": 22820 }, { "epoch": 0.03, "learning_rate": 4.8573125000000006e-05, "loss": 2.2944, "step": 22830 }, { "epoch": 0.03, "learning_rate": 4.85725e-05, "loss": 2.2508, "step": 22840 }, { "epoch": 0.03, "learning_rate": 4.8571875e-05, "loss": 2.4204, "step": 22850 }, { "epoch": 0.03, "learning_rate": 4.8571250000000004e-05, "loss": 2.1546, "step": 22860 }, { "epoch": 0.03, "learning_rate": 4.8570625e-05, "loss": 1.997, "step": 22870 }, { "epoch": 0.03, "learning_rate": 4.8570000000000004e-05, "loss": 2.2632, "step": 22880 }, { "epoch": 0.03, "learning_rate": 4.8569375e-05, "loss": 2.0139, "step": 22890 }, { "epoch": 0.03, "learning_rate": 4.8568750000000005e-05, "loss": 2.2349, "step": 22900 }, { "epoch": 0.03, "learning_rate": 4.8568125e-05, "loss": 2.3016, "step": 22910 }, { "epoch": 0.03, "learning_rate": 4.8567500000000006e-05, "loss": 2.076, "step": 22920 }, { "epoch": 0.03, "learning_rate": 4.8566875e-05, "loss": 2.2716, "step": 22930 }, { "epoch": 0.03, "learning_rate": 4.856625e-05, "loss": 2.3025, "step": 22940 }, { "epoch": 0.03, "learning_rate": 4.8565625e-05, "loss": 2.3286, "step": 22950 }, { "epoch": 0.03, "learning_rate": 4.856500000000001e-05, "loss": 2.2962, "step": 22960 }, { "epoch": 0.03, "learning_rate": 4.8564375000000004e-05, "loss": 2.5217, "step": 22970 }, { "epoch": 0.03, "learning_rate": 4.856375e-05, "loss": 2.5527, "step": 22980 }, { "epoch": 0.03, "learning_rate": 4.8563125000000004e-05, "loss": 2.3345, "step": 22990 }, { "epoch": 0.03, "learning_rate": 4.85625e-05, "loss": 2.2353, "step": 23000 }, { "epoch": 0.03, "eval_loss": 2.1283891201019287, "eval_runtime": 1290.8409, "eval_samples_per_second": 44.96, "eval_steps_per_second": 5.62, "step": 23000 }, { "epoch": 0.03, "learning_rate": 4.8561875e-05, "loss": 2.1624, "step": 23010 }, { "epoch": 0.03, "learning_rate": 4.856125e-05, "loss": 2.3163, "step": 23020 }, { "epoch": 0.03, "learning_rate": 4.8560625000000006e-05, "loss": 2.4803, "step": 23030 }, { "epoch": 0.03, "learning_rate": 4.856e-05, "loss": 2.3134, "step": 23040 }, { "epoch": 0.03, "learning_rate": 4.8559375e-05, "loss": 2.3932, "step": 23050 }, { "epoch": 0.03, "learning_rate": 4.855875e-05, "loss": 2.2112, "step": 23060 }, { "epoch": 0.03, "learning_rate": 4.855812500000001e-05, "loss": 1.9779, "step": 23070 }, { "epoch": 0.03, "learning_rate": 4.85575e-05, "loss": 2.0204, "step": 23080 }, { "epoch": 0.03, "learning_rate": 4.8556875e-05, "loss": 2.1485, "step": 23090 }, { "epoch": 0.03, "learning_rate": 4.8556250000000005e-05, "loss": 2.276, "step": 23100 }, { "epoch": 0.03, "learning_rate": 4.8555625e-05, "loss": 2.3492, "step": 23110 }, { "epoch": 0.03, "learning_rate": 4.8555e-05, "loss": 2.4147, "step": 23120 }, { "epoch": 0.03, "learning_rate": 4.8554375e-05, "loss": 2.0006, "step": 23130 }, { "epoch": 0.03, "learning_rate": 4.8553750000000006e-05, "loss": 2.1937, "step": 23140 }, { "epoch": 0.03, "learning_rate": 4.8553125e-05, "loss": 2.143, "step": 23150 }, { "epoch": 0.03, "learning_rate": 4.85525e-05, "loss": 2.0272, "step": 23160 }, { "epoch": 0.03, "learning_rate": 4.8551875000000003e-05, "loss": 2.3846, "step": 23170 }, { "epoch": 0.03, "learning_rate": 4.855125e-05, "loss": 1.887, "step": 23180 }, { "epoch": 0.03, "learning_rate": 4.8550625e-05, "loss": 2.18, "step": 23190 }, { "epoch": 0.03, "learning_rate": 4.855e-05, "loss": 2.1072, "step": 23200 }, { "epoch": 0.03, "learning_rate": 4.8549375000000005e-05, "loss": 2.1508, "step": 23210 }, { "epoch": 0.03, "learning_rate": 4.854875e-05, "loss": 2.3114, "step": 23220 }, { "epoch": 0.03, "learning_rate": 4.8548125e-05, "loss": 2.3585, "step": 23230 }, { "epoch": 0.03, "learning_rate": 4.85475e-05, "loss": 2.2399, "step": 23240 }, { "epoch": 0.03, "learning_rate": 4.8546875000000006e-05, "loss": 2.0401, "step": 23250 }, { "epoch": 0.03, "learning_rate": 4.854625e-05, "loss": 2.266, "step": 23260 }, { "epoch": 0.03, "learning_rate": 4.8545625e-05, "loss": 2.0457, "step": 23270 }, { "epoch": 0.03, "learning_rate": 4.8545000000000004e-05, "loss": 2.2388, "step": 23280 }, { "epoch": 0.03, "learning_rate": 4.8544375e-05, "loss": 2.2701, "step": 23290 }, { "epoch": 0.03, "learning_rate": 4.8543750000000004e-05, "loss": 2.272, "step": 23300 }, { "epoch": 0.03, "learning_rate": 4.8543125e-05, "loss": 1.8597, "step": 23310 }, { "epoch": 0.03, "learning_rate": 4.8542500000000005e-05, "loss": 2.3223, "step": 23320 }, { "epoch": 0.03, "learning_rate": 4.8541875e-05, "loss": 2.0875, "step": 23330 }, { "epoch": 0.03, "learning_rate": 4.8541250000000006e-05, "loss": 2.0865, "step": 23340 }, { "epoch": 0.03, "learning_rate": 4.8540625e-05, "loss": 2.1806, "step": 23350 }, { "epoch": 0.03, "learning_rate": 4.854e-05, "loss": 2.1524, "step": 23360 }, { "epoch": 0.03, "learning_rate": 4.8539375e-05, "loss": 2.5023, "step": 23370 }, { "epoch": 0.03, "learning_rate": 4.853875000000001e-05, "loss": 2.2017, "step": 23380 }, { "epoch": 0.03, "learning_rate": 4.8538125000000004e-05, "loss": 2.1163, "step": 23390 }, { "epoch": 0.03, "learning_rate": 4.85375e-05, "loss": 2.1587, "step": 23400 }, { "epoch": 0.03, "learning_rate": 4.8536875000000004e-05, "loss": 2.4826, "step": 23410 }, { "epoch": 0.03, "learning_rate": 4.853625e-05, "loss": 2.1435, "step": 23420 }, { "epoch": 0.03, "learning_rate": 4.8535625e-05, "loss": 1.921, "step": 23430 }, { "epoch": 0.03, "learning_rate": 4.8535e-05, "loss": 2.0129, "step": 23440 }, { "epoch": 0.03, "learning_rate": 4.8534375000000006e-05, "loss": 2.4135, "step": 23450 }, { "epoch": 0.03, "learning_rate": 4.853375e-05, "loss": 2.185, "step": 23460 }, { "epoch": 0.03, "learning_rate": 4.8533125e-05, "loss": 2.2389, "step": 23470 }, { "epoch": 0.03, "learning_rate": 4.85325e-05, "loss": 2.0673, "step": 23480 }, { "epoch": 0.03, "learning_rate": 4.853187500000001e-05, "loss": 2.2528, "step": 23490 }, { "epoch": 0.03, "learning_rate": 4.853125e-05, "loss": 2.01, "step": 23500 }, { "epoch": 0.03, "learning_rate": 4.8530625e-05, "loss": 2.3307, "step": 23510 }, { "epoch": 0.03, "learning_rate": 4.8530000000000005e-05, "loss": 2.248, "step": 23520 }, { "epoch": 0.03, "learning_rate": 4.8529375e-05, "loss": 2.1837, "step": 23530 }, { "epoch": 0.03, "learning_rate": 4.852875e-05, "loss": 2.2282, "step": 23540 }, { "epoch": 0.03, "learning_rate": 4.8528125e-05, "loss": 2.2113, "step": 23550 }, { "epoch": 0.03, "learning_rate": 4.8527500000000006e-05, "loss": 2.0676, "step": 23560 }, { "epoch": 0.03, "learning_rate": 4.8526875e-05, "loss": 1.8505, "step": 23570 }, { "epoch": 0.03, "learning_rate": 4.852625e-05, "loss": 1.9939, "step": 23580 }, { "epoch": 0.03, "learning_rate": 4.8525625000000003e-05, "loss": 2.2118, "step": 23590 }, { "epoch": 0.03, "learning_rate": 4.8525e-05, "loss": 2.0001, "step": 23600 }, { "epoch": 0.03, "learning_rate": 4.8524375e-05, "loss": 2.3407, "step": 23610 }, { "epoch": 0.03, "learning_rate": 4.852375e-05, "loss": 2.2332, "step": 23620 }, { "epoch": 0.03, "learning_rate": 4.8523125000000005e-05, "loss": 2.3604, "step": 23630 }, { "epoch": 0.03, "learning_rate": 4.85225e-05, "loss": 2.3163, "step": 23640 }, { "epoch": 0.03, "learning_rate": 4.8521875e-05, "loss": 2.0798, "step": 23650 }, { "epoch": 0.03, "learning_rate": 4.852125e-05, "loss": 1.9969, "step": 23660 }, { "epoch": 0.03, "learning_rate": 4.8520625000000006e-05, "loss": 2.1271, "step": 23670 }, { "epoch": 0.03, "learning_rate": 4.852e-05, "loss": 2.0535, "step": 23680 }, { "epoch": 0.03, "learning_rate": 4.8519375e-05, "loss": 2.2134, "step": 23690 }, { "epoch": 0.03, "learning_rate": 4.8518750000000004e-05, "loss": 2.3404, "step": 23700 }, { "epoch": 0.03, "learning_rate": 4.8518125e-05, "loss": 2.0766, "step": 23710 }, { "epoch": 0.03, "learning_rate": 4.8517500000000004e-05, "loss": 2.3125, "step": 23720 }, { "epoch": 0.03, "learning_rate": 4.8516875e-05, "loss": 2.3108, "step": 23730 }, { "epoch": 0.03, "learning_rate": 4.8516250000000005e-05, "loss": 2.1331, "step": 23740 }, { "epoch": 0.03, "learning_rate": 4.8515625e-05, "loss": 2.1771, "step": 23750 }, { "epoch": 0.03, "learning_rate": 4.8515000000000006e-05, "loss": 2.2325, "step": 23760 }, { "epoch": 0.03, "learning_rate": 4.8514375e-05, "loss": 2.0434, "step": 23770 }, { "epoch": 0.03, "learning_rate": 4.851375e-05, "loss": 2.206, "step": 23780 }, { "epoch": 0.03, "learning_rate": 4.8513125e-05, "loss": 2.5005, "step": 23790 }, { "epoch": 0.03, "learning_rate": 4.85125e-05, "loss": 2.3062, "step": 23800 }, { "epoch": 0.03, "learning_rate": 4.8511875000000004e-05, "loss": 2.3613, "step": 23810 }, { "epoch": 0.03, "learning_rate": 4.851125e-05, "loss": 2.0761, "step": 23820 }, { "epoch": 0.03, "learning_rate": 4.8510625000000004e-05, "loss": 1.9985, "step": 23830 }, { "epoch": 0.03, "learning_rate": 4.851e-05, "loss": 2.1598, "step": 23840 }, { "epoch": 0.03, "learning_rate": 4.8509375e-05, "loss": 2.3622, "step": 23850 }, { "epoch": 0.03, "learning_rate": 4.850875e-05, "loss": 2.1267, "step": 23860 }, { "epoch": 0.03, "learning_rate": 4.8508125000000006e-05, "loss": 2.4003, "step": 23870 }, { "epoch": 0.03, "learning_rate": 4.85075e-05, "loss": 2.0126, "step": 23880 }, { "epoch": 0.03, "learning_rate": 4.8506875e-05, "loss": 2.2016, "step": 23890 }, { "epoch": 0.03, "learning_rate": 4.850625e-05, "loss": 2.0727, "step": 23900 }, { "epoch": 0.03, "learning_rate": 4.850562500000001e-05, "loss": 1.804, "step": 23910 }, { "epoch": 0.03, "learning_rate": 4.8505e-05, "loss": 2.3309, "step": 23920 }, { "epoch": 0.03, "learning_rate": 4.8504375e-05, "loss": 2.3776, "step": 23930 }, { "epoch": 0.03, "learning_rate": 4.8503750000000005e-05, "loss": 2.5115, "step": 23940 }, { "epoch": 0.03, "learning_rate": 4.8503125e-05, "loss": 2.1987, "step": 23950 }, { "epoch": 0.03, "learning_rate": 4.85025e-05, "loss": 2.1401, "step": 23960 }, { "epoch": 0.03, "learning_rate": 4.8501875e-05, "loss": 2.3737, "step": 23970 }, { "epoch": 0.03, "learning_rate": 4.8501250000000006e-05, "loss": 1.9799, "step": 23980 }, { "epoch": 0.03, "learning_rate": 4.8500625e-05, "loss": 2.0784, "step": 23990 }, { "epoch": 0.03, "learning_rate": 4.85e-05, "loss": 2.1754, "step": 24000 }, { "epoch": 0.03, "eval_loss": 2.0910816192626953, "eval_runtime": 1290.4774, "eval_samples_per_second": 44.973, "eval_steps_per_second": 5.622, "step": 24000 }, { "epoch": 0.03, "learning_rate": 4.8499375000000004e-05, "loss": 2.2032, "step": 24010 }, { "epoch": 0.03, "learning_rate": 4.849875000000001e-05, "loss": 2.0431, "step": 24020 }, { "epoch": 0.03, "learning_rate": 4.8498125e-05, "loss": 2.2558, "step": 24030 }, { "epoch": 0.03, "learning_rate": 4.84975e-05, "loss": 2.2203, "step": 24040 }, { "epoch": 0.03, "learning_rate": 4.8496875000000005e-05, "loss": 2.0644, "step": 24050 }, { "epoch": 0.03, "learning_rate": 4.849625e-05, "loss": 2.0956, "step": 24060 }, { "epoch": 0.03, "learning_rate": 4.8495625e-05, "loss": 2.1308, "step": 24070 }, { "epoch": 0.03, "learning_rate": 4.8495e-05, "loss": 2.0963, "step": 24080 }, { "epoch": 0.03, "learning_rate": 4.8494375000000006e-05, "loss": 2.1207, "step": 24090 }, { "epoch": 0.03, "learning_rate": 4.849375e-05, "loss": 1.9186, "step": 24100 }, { "epoch": 0.03, "learning_rate": 4.8493125e-05, "loss": 1.9804, "step": 24110 }, { "epoch": 0.03, "learning_rate": 4.8492500000000004e-05, "loss": 2.2329, "step": 24120 }, { "epoch": 0.03, "learning_rate": 4.8491875e-05, "loss": 2.3294, "step": 24130 }, { "epoch": 0.03, "learning_rate": 4.8491250000000004e-05, "loss": 2.2558, "step": 24140 }, { "epoch": 0.03, "learning_rate": 4.8490625e-05, "loss": 2.0778, "step": 24150 }, { "epoch": 0.03, "learning_rate": 4.8490000000000005e-05, "loss": 1.9095, "step": 24160 }, { "epoch": 0.03, "learning_rate": 4.8489375e-05, "loss": 2.3339, "step": 24170 }, { "epoch": 0.03, "learning_rate": 4.8488750000000006e-05, "loss": 2.3047, "step": 24180 }, { "epoch": 0.03, "learning_rate": 4.8488125e-05, "loss": 2.1478, "step": 24190 }, { "epoch": 0.03, "learning_rate": 4.84875e-05, "loss": 2.006, "step": 24200 }, { "epoch": 0.03, "learning_rate": 4.8486875e-05, "loss": 2.2709, "step": 24210 }, { "epoch": 0.03, "learning_rate": 4.848625e-05, "loss": 2.2143, "step": 24220 }, { "epoch": 0.03, "learning_rate": 4.8485625000000004e-05, "loss": 2.3752, "step": 24230 }, { "epoch": 0.03, "learning_rate": 4.8485e-05, "loss": 2.216, "step": 24240 }, { "epoch": 0.03, "learning_rate": 4.8484375000000005e-05, "loss": 2.0153, "step": 24250 }, { "epoch": 0.03, "learning_rate": 4.848375e-05, "loss": 2.1824, "step": 24260 }, { "epoch": 0.03, "learning_rate": 4.8483125e-05, "loss": 1.9447, "step": 24270 }, { "epoch": 0.03, "learning_rate": 4.84825e-05, "loss": 2.3678, "step": 24280 }, { "epoch": 0.03, "learning_rate": 4.8481875000000006e-05, "loss": 2.4331, "step": 24290 }, { "epoch": 0.03, "learning_rate": 4.848125e-05, "loss": 1.9175, "step": 24300 }, { "epoch": 0.03, "learning_rate": 4.8480625e-05, "loss": 2.2649, "step": 24310 }, { "epoch": 0.03, "learning_rate": 4.8480000000000003e-05, "loss": 2.1411, "step": 24320 }, { "epoch": 0.03, "learning_rate": 4.847937500000001e-05, "loss": 2.3599, "step": 24330 }, { "epoch": 0.03, "learning_rate": 4.847875e-05, "loss": 2.0995, "step": 24340 }, { "epoch": 0.03, "learning_rate": 4.8478125e-05, "loss": 2.3872, "step": 24350 }, { "epoch": 0.03, "learning_rate": 4.8477500000000005e-05, "loss": 2.6487, "step": 24360 }, { "epoch": 0.03, "learning_rate": 4.8476875e-05, "loss": 2.1907, "step": 24370 }, { "epoch": 0.03, "learning_rate": 4.847625e-05, "loss": 2.1512, "step": 24380 }, { "epoch": 0.03, "learning_rate": 4.8475625e-05, "loss": 2.1877, "step": 24390 }, { "epoch": 0.03, "learning_rate": 4.8475000000000006e-05, "loss": 2.1098, "step": 24400 }, { "epoch": 0.03, "learning_rate": 4.8474375e-05, "loss": 2.3352, "step": 24410 }, { "epoch": 0.03, "learning_rate": 4.847375e-05, "loss": 2.2871, "step": 24420 }, { "epoch": 0.03, "learning_rate": 4.8473125000000004e-05, "loss": 2.1483, "step": 24430 }, { "epoch": 0.03, "learning_rate": 4.847250000000001e-05, "loss": 1.9422, "step": 24440 }, { "epoch": 0.03, "learning_rate": 4.8471875e-05, "loss": 2.2697, "step": 24450 }, { "epoch": 0.03, "learning_rate": 4.847125e-05, "loss": 2.082, "step": 24460 }, { "epoch": 0.03, "learning_rate": 4.8470625000000005e-05, "loss": 2.2036, "step": 24470 }, { "epoch": 0.03, "learning_rate": 4.847e-05, "loss": 2.2707, "step": 24480 }, { "epoch": 0.03, "learning_rate": 4.8469375e-05, "loss": 2.1747, "step": 24490 }, { "epoch": 0.03, "learning_rate": 4.846875e-05, "loss": 2.1371, "step": 24500 }, { "epoch": 0.03, "learning_rate": 4.8468125000000006e-05, "loss": 2.1486, "step": 24510 }, { "epoch": 0.03, "learning_rate": 4.84675e-05, "loss": 2.3334, "step": 24520 }, { "epoch": 0.03, "learning_rate": 4.8466875e-05, "loss": 2.023, "step": 24530 }, { "epoch": 0.03, "learning_rate": 4.8466250000000004e-05, "loss": 1.9952, "step": 24540 }, { "epoch": 0.03, "learning_rate": 4.8465625e-05, "loss": 2.2259, "step": 24550 }, { "epoch": 0.03, "learning_rate": 4.8465000000000004e-05, "loss": 2.2988, "step": 24560 }, { "epoch": 0.03, "learning_rate": 4.8464375e-05, "loss": 2.2461, "step": 24570 }, { "epoch": 0.03, "learning_rate": 4.8463750000000005e-05, "loss": 2.0803, "step": 24580 }, { "epoch": 0.03, "learning_rate": 4.8463125e-05, "loss": 2.0301, "step": 24590 }, { "epoch": 0.03, "learning_rate": 4.84625e-05, "loss": 1.9962, "step": 24600 }, { "epoch": 0.03, "learning_rate": 4.8461875e-05, "loss": 2.1506, "step": 24610 }, { "epoch": 0.03, "learning_rate": 4.846125e-05, "loss": 2.4494, "step": 24620 }, { "epoch": 0.03, "learning_rate": 4.8460625e-05, "loss": 2.1515, "step": 24630 }, { "epoch": 0.03, "learning_rate": 4.846e-05, "loss": 2.2661, "step": 24640 }, { "epoch": 0.03, "learning_rate": 4.8459375000000004e-05, "loss": 2.1587, "step": 24650 }, { "epoch": 0.03, "learning_rate": 4.845875e-05, "loss": 2.7005, "step": 24660 }, { "epoch": 0.03, "learning_rate": 4.8458125000000005e-05, "loss": 2.1295, "step": 24670 }, { "epoch": 0.03, "learning_rate": 4.84575e-05, "loss": 2.4525, "step": 24680 }, { "epoch": 0.03, "learning_rate": 4.8456875e-05, "loss": 2.2316, "step": 24690 }, { "epoch": 0.03, "learning_rate": 4.845625e-05, "loss": 2.1294, "step": 24700 }, { "epoch": 0.03, "learning_rate": 4.8455625000000006e-05, "loss": 1.9656, "step": 24710 }, { "epoch": 0.03, "learning_rate": 4.8455e-05, "loss": 2.1311, "step": 24720 }, { "epoch": 0.03, "learning_rate": 4.8454375e-05, "loss": 2.1624, "step": 24730 }, { "epoch": 0.03, "learning_rate": 4.8453750000000003e-05, "loss": 2.1424, "step": 24740 }, { "epoch": 0.03, "learning_rate": 4.845312500000001e-05, "loss": 2.0182, "step": 24750 }, { "epoch": 0.03, "learning_rate": 4.84525e-05, "loss": 2.055, "step": 24760 }, { "epoch": 0.03, "learning_rate": 4.8451875e-05, "loss": 1.996, "step": 24770 }, { "epoch": 0.03, "learning_rate": 4.8451250000000005e-05, "loss": 1.9652, "step": 24780 }, { "epoch": 0.03, "learning_rate": 4.8450625e-05, "loss": 2.3495, "step": 24790 }, { "epoch": 0.03, "learning_rate": 4.845e-05, "loss": 2.1342, "step": 24800 }, { "epoch": 0.03, "learning_rate": 4.8449375e-05, "loss": 2.1159, "step": 24810 }, { "epoch": 0.03, "learning_rate": 4.8448750000000006e-05, "loss": 2.0516, "step": 24820 }, { "epoch": 0.03, "learning_rate": 4.8448125e-05, "loss": 2.0878, "step": 24830 }, { "epoch": 0.03, "learning_rate": 4.84475e-05, "loss": 2.1444, "step": 24840 }, { "epoch": 0.03, "learning_rate": 4.8446875000000004e-05, "loss": 1.8928, "step": 24850 }, { "epoch": 0.03, "learning_rate": 4.844625000000001e-05, "loss": 1.9209, "step": 24860 }, { "epoch": 0.03, "learning_rate": 4.8445625e-05, "loss": 2.2369, "step": 24870 }, { "epoch": 0.03, "learning_rate": 4.8445e-05, "loss": 2.1974, "step": 24880 }, { "epoch": 0.03, "learning_rate": 4.8444375000000005e-05, "loss": 2.0318, "step": 24890 }, { "epoch": 0.03, "learning_rate": 4.844375e-05, "loss": 2.0992, "step": 24900 }, { "epoch": 0.03, "learning_rate": 4.8443125e-05, "loss": 2.3078, "step": 24910 }, { "epoch": 0.03, "learning_rate": 4.84425e-05, "loss": 2.1223, "step": 24920 }, { "epoch": 0.03, "learning_rate": 4.8441875000000006e-05, "loss": 2.166, "step": 24930 }, { "epoch": 0.03, "learning_rate": 4.844125e-05, "loss": 2.2943, "step": 24940 }, { "epoch": 0.03, "learning_rate": 4.8440625e-05, "loss": 2.1463, "step": 24950 }, { "epoch": 0.03, "learning_rate": 4.8440000000000004e-05, "loss": 1.9269, "step": 24960 }, { "epoch": 0.03, "learning_rate": 4.8439375e-05, "loss": 2.2841, "step": 24970 }, { "epoch": 0.03, "learning_rate": 4.8438750000000004e-05, "loss": 2.0479, "step": 24980 }, { "epoch": 0.03, "learning_rate": 4.8438125e-05, "loss": 1.9568, "step": 24990 }, { "epoch": 0.03, "learning_rate": 4.8437500000000005e-05, "loss": 2.398, "step": 25000 }, { "epoch": 0.03, "eval_loss": 2.0704500675201416, "eval_runtime": 1290.3981, "eval_samples_per_second": 44.975, "eval_steps_per_second": 5.622, "step": 25000 }, { "epoch": 0.03, "learning_rate": 4.8436875e-05, "loss": 2.1488, "step": 25010 }, { "epoch": 0.03, "learning_rate": 4.843625e-05, "loss": 2.3856, "step": 25020 }, { "epoch": 0.03, "learning_rate": 4.8435625e-05, "loss": 2.1246, "step": 25030 }, { "epoch": 0.03, "learning_rate": 4.8435e-05, "loss": 2.0593, "step": 25040 }, { "epoch": 0.03, "learning_rate": 4.8434375e-05, "loss": 2.1007, "step": 25050 }, { "epoch": 0.03, "learning_rate": 4.843375e-05, "loss": 2.0035, "step": 25060 }, { "epoch": 0.03, "learning_rate": 4.8433125000000004e-05, "loss": 2.3294, "step": 25070 }, { "epoch": 0.03, "learning_rate": 4.84325e-05, "loss": 2.0574, "step": 25080 }, { "epoch": 0.03, "learning_rate": 4.8431875000000005e-05, "loss": 2.0746, "step": 25090 }, { "epoch": 0.03, "learning_rate": 4.843125e-05, "loss": 2.0982, "step": 25100 }, { "epoch": 0.03, "learning_rate": 4.8430625e-05, "loss": 2.2372, "step": 25110 }, { "epoch": 0.03, "learning_rate": 4.843e-05, "loss": 2.0825, "step": 25120 }, { "epoch": 0.03, "learning_rate": 4.8429375000000006e-05, "loss": 2.1668, "step": 25130 }, { "epoch": 0.04, "learning_rate": 4.842875e-05, "loss": 2.0782, "step": 25140 }, { "epoch": 0.04, "learning_rate": 4.8428125e-05, "loss": 2.0843, "step": 25150 }, { "epoch": 0.04, "learning_rate": 4.8427500000000003e-05, "loss": 2.2413, "step": 25160 }, { "epoch": 0.04, "learning_rate": 4.842687500000001e-05, "loss": 2.3154, "step": 25170 }, { "epoch": 0.04, "learning_rate": 4.8426250000000004e-05, "loss": 1.934, "step": 25180 }, { "epoch": 0.04, "learning_rate": 4.8425625e-05, "loss": 2.3848, "step": 25190 }, { "epoch": 0.04, "learning_rate": 4.8425000000000005e-05, "loss": 2.0206, "step": 25200 }, { "epoch": 0.04, "learning_rate": 4.8424375e-05, "loss": 2.3679, "step": 25210 }, { "epoch": 0.04, "learning_rate": 4.842375e-05, "loss": 2.1794, "step": 25220 }, { "epoch": 0.04, "learning_rate": 4.8423125e-05, "loss": 2.2393, "step": 25230 }, { "epoch": 0.04, "learning_rate": 4.8422500000000006e-05, "loss": 2.0295, "step": 25240 }, { "epoch": 0.04, "learning_rate": 4.8421875e-05, "loss": 2.1925, "step": 25250 }, { "epoch": 0.04, "learning_rate": 4.842125e-05, "loss": 2.3698, "step": 25260 }, { "epoch": 0.04, "learning_rate": 4.8420625000000004e-05, "loss": 2.4296, "step": 25270 }, { "epoch": 0.04, "learning_rate": 4.842000000000001e-05, "loss": 1.944, "step": 25280 }, { "epoch": 0.04, "learning_rate": 4.8419375e-05, "loss": 2.4407, "step": 25290 }, { "epoch": 0.04, "learning_rate": 4.841875e-05, "loss": 2.3673, "step": 25300 }, { "epoch": 0.04, "learning_rate": 4.8418125000000005e-05, "loss": 2.1087, "step": 25310 }, { "epoch": 0.04, "learning_rate": 4.84175e-05, "loss": 2.1512, "step": 25320 }, { "epoch": 0.04, "learning_rate": 4.8416875e-05, "loss": 2.0707, "step": 25330 }, { "epoch": 0.04, "learning_rate": 4.841625e-05, "loss": 1.8862, "step": 25340 }, { "epoch": 0.04, "learning_rate": 4.8415625000000006e-05, "loss": 2.0357, "step": 25350 }, { "epoch": 0.04, "learning_rate": 4.8415e-05, "loss": 2.0618, "step": 25360 }, { "epoch": 0.04, "learning_rate": 4.8414375e-05, "loss": 2.1775, "step": 25370 }, { "epoch": 0.04, "learning_rate": 4.8413750000000004e-05, "loss": 2.1302, "step": 25380 }, { "epoch": 0.04, "learning_rate": 4.8413125e-05, "loss": 2.1194, "step": 25390 }, { "epoch": 0.04, "learning_rate": 4.8412500000000004e-05, "loss": 2.0783, "step": 25400 }, { "epoch": 0.04, "learning_rate": 4.8411875e-05, "loss": 1.7968, "step": 25410 }, { "epoch": 0.04, "learning_rate": 4.8411250000000005e-05, "loss": 1.9836, "step": 25420 }, { "epoch": 0.04, "learning_rate": 4.8410625e-05, "loss": 2.3382, "step": 25430 }, { "epoch": 0.04, "learning_rate": 4.841e-05, "loss": 2.1181, "step": 25440 }, { "epoch": 0.04, "learning_rate": 4.8409375e-05, "loss": 2.0561, "step": 25450 }, { "epoch": 0.04, "learning_rate": 4.840875e-05, "loss": 2.0395, "step": 25460 }, { "epoch": 0.04, "learning_rate": 4.8408125e-05, "loss": 2.0382, "step": 25470 }, { "epoch": 0.04, "learning_rate": 4.84075e-05, "loss": 2.1502, "step": 25480 }, { "epoch": 0.04, "learning_rate": 4.8406875000000004e-05, "loss": 1.8919, "step": 25490 }, { "epoch": 0.04, "learning_rate": 4.840625e-05, "loss": 2.2428, "step": 25500 }, { "epoch": 0.04, "learning_rate": 4.8405625000000005e-05, "loss": 2.2364, "step": 25510 }, { "epoch": 0.04, "learning_rate": 4.8405e-05, "loss": 2.0568, "step": 25520 }, { "epoch": 0.04, "learning_rate": 4.8404375e-05, "loss": 2.0387, "step": 25530 }, { "epoch": 0.04, "learning_rate": 4.840375e-05, "loss": 2.0826, "step": 25540 }, { "epoch": 0.04, "learning_rate": 4.8403125000000006e-05, "loss": 2.1976, "step": 25550 }, { "epoch": 0.04, "learning_rate": 4.84025e-05, "loss": 2.2033, "step": 25560 }, { "epoch": 0.04, "learning_rate": 4.8401875e-05, "loss": 2.534, "step": 25570 }, { "epoch": 0.04, "learning_rate": 4.8401250000000004e-05, "loss": 2.2347, "step": 25580 }, { "epoch": 0.04, "learning_rate": 4.840062500000001e-05, "loss": 2.2659, "step": 25590 }, { "epoch": 0.04, "learning_rate": 4.8400000000000004e-05, "loss": 2.4238, "step": 25600 }, { "epoch": 0.04, "learning_rate": 4.8399375e-05, "loss": 2.0243, "step": 25610 }, { "epoch": 0.04, "learning_rate": 4.8398750000000005e-05, "loss": 2.2816, "step": 25620 }, { "epoch": 0.04, "learning_rate": 4.8398125e-05, "loss": 2.0196, "step": 25630 }, { "epoch": 0.04, "learning_rate": 4.83975e-05, "loss": 2.2563, "step": 25640 }, { "epoch": 0.04, "learning_rate": 4.8396875e-05, "loss": 2.1405, "step": 25650 }, { "epoch": 0.04, "learning_rate": 4.8396250000000006e-05, "loss": 2.0902, "step": 25660 }, { "epoch": 0.04, "learning_rate": 4.8395625e-05, "loss": 1.9822, "step": 25670 }, { "epoch": 0.04, "learning_rate": 4.8395e-05, "loss": 2.1213, "step": 25680 }, { "epoch": 0.04, "learning_rate": 4.8394375000000004e-05, "loss": 2.4831, "step": 25690 }, { "epoch": 0.04, "learning_rate": 4.839375000000001e-05, "loss": 2.1003, "step": 25700 }, { "epoch": 0.04, "learning_rate": 4.8393125e-05, "loss": 2.0799, "step": 25710 }, { "epoch": 0.04, "learning_rate": 4.83925e-05, "loss": 2.1622, "step": 25720 }, { "epoch": 0.04, "learning_rate": 4.8391875000000005e-05, "loss": 2.2023, "step": 25730 }, { "epoch": 0.04, "learning_rate": 4.839125e-05, "loss": 2.1111, "step": 25740 }, { "epoch": 0.04, "learning_rate": 4.8390625e-05, "loss": 1.8776, "step": 25750 }, { "epoch": 0.04, "learning_rate": 4.839e-05, "loss": 2.1046, "step": 25760 }, { "epoch": 0.04, "learning_rate": 4.8389375000000006e-05, "loss": 2.295, "step": 25770 }, { "epoch": 0.04, "learning_rate": 4.838875e-05, "loss": 2.1418, "step": 25780 }, { "epoch": 0.04, "learning_rate": 4.8388125e-05, "loss": 1.7722, "step": 25790 }, { "epoch": 0.04, "learning_rate": 4.8387500000000004e-05, "loss": 1.824, "step": 25800 }, { "epoch": 0.04, "learning_rate": 4.8386875e-05, "loss": 2.2154, "step": 25810 }, { "epoch": 0.04, "learning_rate": 4.838625e-05, "loss": 2.2087, "step": 25820 }, { "epoch": 0.04, "learning_rate": 4.8385625e-05, "loss": 2.4251, "step": 25830 }, { "epoch": 0.04, "learning_rate": 4.8385000000000005e-05, "loss": 2.0256, "step": 25840 }, { "epoch": 0.04, "learning_rate": 4.8384375e-05, "loss": 2.2371, "step": 25850 }, { "epoch": 0.04, "learning_rate": 4.838375e-05, "loss": 2.2865, "step": 25860 }, { "epoch": 0.04, "learning_rate": 4.8383125e-05, "loss": 2.6304, "step": 25870 }, { "epoch": 0.04, "learning_rate": 4.83825e-05, "loss": 2.3445, "step": 25880 }, { "epoch": 0.04, "learning_rate": 4.8381875e-05, "loss": 2.2187, "step": 25890 }, { "epoch": 0.04, "learning_rate": 4.838125e-05, "loss": 2.2783, "step": 25900 }, { "epoch": 0.04, "learning_rate": 4.8380625000000004e-05, "loss": 1.9517, "step": 25910 }, { "epoch": 0.04, "learning_rate": 4.838e-05, "loss": 1.9827, "step": 25920 }, { "epoch": 0.04, "learning_rate": 4.8379375000000005e-05, "loss": 2.3513, "step": 25930 }, { "epoch": 0.04, "learning_rate": 4.837875e-05, "loss": 2.0523, "step": 25940 }, { "epoch": 0.04, "learning_rate": 4.8378125e-05, "loss": 2.292, "step": 25950 }, { "epoch": 0.04, "learning_rate": 4.83775e-05, "loss": 2.0755, "step": 25960 }, { "epoch": 0.04, "learning_rate": 4.8376875000000006e-05, "loss": 2.3464, "step": 25970 }, { "epoch": 0.04, "learning_rate": 4.837625e-05, "loss": 2.1098, "step": 25980 }, { "epoch": 0.04, "learning_rate": 4.8375625e-05, "loss": 1.9917, "step": 25990 }, { "epoch": 0.04, "learning_rate": 4.8375000000000004e-05, "loss": 2.2089, "step": 26000 }, { "epoch": 0.04, "eval_loss": 2.0299570560455322, "eval_runtime": 1290.3367, "eval_samples_per_second": 44.977, "eval_steps_per_second": 5.623, "step": 26000 }, { "epoch": 0.04, "learning_rate": 4.8374375e-05, "loss": 2.029, "step": 26010 }, { "epoch": 0.04, "learning_rate": 4.8373750000000004e-05, "loss": 1.7763, "step": 26020 }, { "epoch": 0.04, "learning_rate": 4.8373125e-05, "loss": 2.2313, "step": 26030 }, { "epoch": 0.04, "learning_rate": 4.8372500000000005e-05, "loss": 1.9788, "step": 26040 }, { "epoch": 0.04, "learning_rate": 4.8371875e-05, "loss": 1.9496, "step": 26050 }, { "epoch": 0.04, "learning_rate": 4.837125e-05, "loss": 2.1578, "step": 26060 }, { "epoch": 0.04, "learning_rate": 4.8370625e-05, "loss": 2.1128, "step": 26070 }, { "epoch": 0.04, "learning_rate": 4.8370000000000006e-05, "loss": 2.0001, "step": 26080 }, { "epoch": 0.04, "learning_rate": 4.8369375e-05, "loss": 2.1884, "step": 26090 }, { "epoch": 0.04, "learning_rate": 4.836875e-05, "loss": 1.8899, "step": 26100 }, { "epoch": 0.04, "learning_rate": 4.8368125000000004e-05, "loss": 2.224, "step": 26110 }, { "epoch": 0.04, "learning_rate": 4.836750000000001e-05, "loss": 2.2457, "step": 26120 }, { "epoch": 0.04, "learning_rate": 4.8366875e-05, "loss": 2.0378, "step": 26130 }, { "epoch": 0.04, "learning_rate": 4.836625e-05, "loss": 1.9536, "step": 26140 }, { "epoch": 0.04, "learning_rate": 4.8365625000000005e-05, "loss": 2.0248, "step": 26150 }, { "epoch": 0.04, "learning_rate": 4.8365e-05, "loss": 2.1962, "step": 26160 }, { "epoch": 0.04, "learning_rate": 4.8364375e-05, "loss": 2.2093, "step": 26170 }, { "epoch": 0.04, "learning_rate": 4.836375e-05, "loss": 1.9724, "step": 26180 }, { "epoch": 0.04, "learning_rate": 4.8363125000000006e-05, "loss": 2.1542, "step": 26190 }, { "epoch": 0.04, "learning_rate": 4.83625e-05, "loss": 2.0964, "step": 26200 }, { "epoch": 0.04, "learning_rate": 4.8361875e-05, "loss": 2.2044, "step": 26210 }, { "epoch": 0.04, "learning_rate": 4.8361250000000004e-05, "loss": 2.1002, "step": 26220 }, { "epoch": 0.04, "learning_rate": 4.8360625e-05, "loss": 2.0794, "step": 26230 }, { "epoch": 0.04, "learning_rate": 4.836e-05, "loss": 1.8756, "step": 26240 }, { "epoch": 0.04, "learning_rate": 4.8359375e-05, "loss": 2.1119, "step": 26250 }, { "epoch": 0.04, "learning_rate": 4.8358750000000005e-05, "loss": 2.0419, "step": 26260 }, { "epoch": 0.04, "learning_rate": 4.8358125e-05, "loss": 1.8843, "step": 26270 }, { "epoch": 0.04, "learning_rate": 4.83575e-05, "loss": 2.1111, "step": 26280 }, { "epoch": 0.04, "learning_rate": 4.8356875e-05, "loss": 2.1021, "step": 26290 }, { "epoch": 0.04, "learning_rate": 4.835625e-05, "loss": 1.9234, "step": 26300 }, { "epoch": 0.04, "learning_rate": 4.8355625000000003e-05, "loss": 2.0662, "step": 26310 }, { "epoch": 0.04, "learning_rate": 4.8355e-05, "loss": 2.4605, "step": 26320 }, { "epoch": 0.04, "learning_rate": 4.8354375000000004e-05, "loss": 2.0635, "step": 26330 }, { "epoch": 0.04, "learning_rate": 4.835375e-05, "loss": 2.4258, "step": 26340 }, { "epoch": 0.04, "learning_rate": 4.8353125000000005e-05, "loss": 2.1056, "step": 26350 }, { "epoch": 0.04, "learning_rate": 4.83525e-05, "loss": 2.0409, "step": 26360 }, { "epoch": 0.04, "learning_rate": 4.8351875000000005e-05, "loss": 2.1377, "step": 26370 }, { "epoch": 0.04, "learning_rate": 4.835125e-05, "loss": 2.2108, "step": 26380 }, { "epoch": 0.04, "learning_rate": 4.8350625000000006e-05, "loss": 1.9704, "step": 26390 }, { "epoch": 0.04, "learning_rate": 4.835e-05, "loss": 2.0885, "step": 26400 }, { "epoch": 0.04, "learning_rate": 4.8349375e-05, "loss": 2.0268, "step": 26410 }, { "epoch": 0.04, "learning_rate": 4.8348750000000004e-05, "loss": 2.3031, "step": 26420 }, { "epoch": 0.04, "learning_rate": 4.8348125e-05, "loss": 2.2482, "step": 26430 }, { "epoch": 0.04, "learning_rate": 4.8347500000000004e-05, "loss": 2.1348, "step": 26440 }, { "epoch": 0.04, "learning_rate": 4.8346875e-05, "loss": 2.0045, "step": 26450 }, { "epoch": 0.04, "learning_rate": 4.8346250000000005e-05, "loss": 2.0446, "step": 26460 }, { "epoch": 0.04, "learning_rate": 4.8345625e-05, "loss": 2.2899, "step": 26470 }, { "epoch": 0.04, "learning_rate": 4.8345e-05, "loss": 2.1404, "step": 26480 }, { "epoch": 0.04, "learning_rate": 4.8344375e-05, "loss": 2.0691, "step": 26490 }, { "epoch": 0.04, "learning_rate": 4.8343750000000006e-05, "loss": 2.1413, "step": 26500 }, { "epoch": 0.04, "learning_rate": 4.8343125e-05, "loss": 2.0294, "step": 26510 }, { "epoch": 0.04, "learning_rate": 4.83425e-05, "loss": 1.9558, "step": 26520 }, { "epoch": 0.04, "learning_rate": 4.8341875000000004e-05, "loss": 2.1701, "step": 26530 }, { "epoch": 0.04, "learning_rate": 4.834125000000001e-05, "loss": 2.1649, "step": 26540 }, { "epoch": 0.04, "learning_rate": 4.8340625e-05, "loss": 2.2133, "step": 26550 }, { "epoch": 0.04, "learning_rate": 4.834e-05, "loss": 1.9913, "step": 26560 }, { "epoch": 0.04, "learning_rate": 4.8339375000000005e-05, "loss": 2.2469, "step": 26570 }, { "epoch": 0.04, "learning_rate": 4.833875e-05, "loss": 2.0699, "step": 26580 }, { "epoch": 0.04, "learning_rate": 4.8338125e-05, "loss": 1.954, "step": 26590 }, { "epoch": 0.04, "learning_rate": 4.83375e-05, "loss": 1.9621, "step": 26600 }, { "epoch": 0.04, "learning_rate": 4.8336875000000006e-05, "loss": 1.7908, "step": 26610 }, { "epoch": 0.04, "learning_rate": 4.8336249999999997e-05, "loss": 2.1028, "step": 26620 }, { "epoch": 0.04, "learning_rate": 4.8335625e-05, "loss": 2.351, "step": 26630 }, { "epoch": 0.04, "learning_rate": 4.8335000000000004e-05, "loss": 2.1521, "step": 26640 }, { "epoch": 0.04, "learning_rate": 4.8334375e-05, "loss": 2.0023, "step": 26650 }, { "epoch": 0.04, "learning_rate": 4.833375e-05, "loss": 2.1717, "step": 26660 }, { "epoch": 0.04, "learning_rate": 4.8333125e-05, "loss": 1.9745, "step": 26670 }, { "epoch": 0.04, "learning_rate": 4.8332500000000005e-05, "loss": 1.9389, "step": 26680 }, { "epoch": 0.04, "learning_rate": 4.8331875e-05, "loss": 2.3902, "step": 26690 }, { "epoch": 0.04, "learning_rate": 4.833125e-05, "loss": 2.2038, "step": 26700 }, { "epoch": 0.04, "learning_rate": 4.8330625e-05, "loss": 2.1117, "step": 26710 }, { "epoch": 0.04, "learning_rate": 4.833e-05, "loss": 2.3854, "step": 26720 }, { "epoch": 0.04, "learning_rate": 4.8329375000000003e-05, "loss": 1.99, "step": 26730 }, { "epoch": 0.04, "learning_rate": 4.832875e-05, "loss": 2.3675, "step": 26740 }, { "epoch": 0.04, "learning_rate": 4.8328125000000004e-05, "loss": 2.0723, "step": 26750 }, { "epoch": 0.04, "learning_rate": 4.83275e-05, "loss": 2.1447, "step": 26760 }, { "epoch": 0.04, "learning_rate": 4.8326875000000005e-05, "loss": 2.1433, "step": 26770 }, { "epoch": 0.04, "learning_rate": 4.832625e-05, "loss": 2.2676, "step": 26780 }, { "epoch": 0.04, "learning_rate": 4.8325625000000005e-05, "loss": 2.361, "step": 26790 }, { "epoch": 0.04, "learning_rate": 4.8325e-05, "loss": 2.0793, "step": 26800 }, { "epoch": 0.04, "learning_rate": 4.8324375000000006e-05, "loss": 2.1739, "step": 26810 }, { "epoch": 0.04, "learning_rate": 4.832375e-05, "loss": 2.1037, "step": 26820 }, { "epoch": 0.04, "learning_rate": 4.8323125e-05, "loss": 2.0027, "step": 26830 }, { "epoch": 0.04, "learning_rate": 4.8322500000000004e-05, "loss": 2.1973, "step": 26840 }, { "epoch": 0.04, "learning_rate": 4.8321875e-05, "loss": 2.2897, "step": 26850 }, { "epoch": 0.04, "learning_rate": 4.8321250000000004e-05, "loss": 1.9546, "step": 26860 }, { "epoch": 0.04, "learning_rate": 4.8320625e-05, "loss": 1.8408, "step": 26870 }, { "epoch": 0.04, "learning_rate": 4.8320000000000005e-05, "loss": 2.2963, "step": 26880 }, { "epoch": 0.04, "learning_rate": 4.8319375e-05, "loss": 2.1764, "step": 26890 }, { "epoch": 0.04, "learning_rate": 4.831875e-05, "loss": 2.0603, "step": 26900 }, { "epoch": 0.04, "learning_rate": 4.8318125e-05, "loss": 2.0234, "step": 26910 }, { "epoch": 0.04, "learning_rate": 4.8317500000000006e-05, "loss": 2.2214, "step": 26920 }, { "epoch": 0.04, "learning_rate": 4.8316875e-05, "loss": 2.1911, "step": 26930 }, { "epoch": 0.04, "learning_rate": 4.831625e-05, "loss": 2.1855, "step": 26940 }, { "epoch": 0.04, "learning_rate": 4.8315625000000004e-05, "loss": 2.107, "step": 26950 }, { "epoch": 0.04, "learning_rate": 4.831500000000001e-05, "loss": 2.306, "step": 26960 }, { "epoch": 0.04, "learning_rate": 4.8314375e-05, "loss": 1.7988, "step": 26970 }, { "epoch": 0.04, "learning_rate": 4.831375e-05, "loss": 2.1987, "step": 26980 }, { "epoch": 0.04, "learning_rate": 4.8313125000000005e-05, "loss": 2.1617, "step": 26990 }, { "epoch": 0.04, "learning_rate": 4.83125e-05, "loss": 2.2107, "step": 27000 }, { "epoch": 0.04, "eval_loss": 2.0014498233795166, "eval_runtime": 1290.2572, "eval_samples_per_second": 44.98, "eval_steps_per_second": 5.623, "step": 27000 }, { "epoch": 0.04, "learning_rate": 4.8311875e-05, "loss": 2.2194, "step": 27010 }, { "epoch": 0.04, "learning_rate": 4.831125e-05, "loss": 2.2904, "step": 27020 }, { "epoch": 0.04, "learning_rate": 4.8310625000000006e-05, "loss": 1.8194, "step": 27030 }, { "epoch": 0.04, "learning_rate": 4.8309999999999997e-05, "loss": 2.2581, "step": 27040 }, { "epoch": 0.04, "learning_rate": 4.8309375e-05, "loss": 2.0355, "step": 27050 }, { "epoch": 0.04, "learning_rate": 4.8308750000000004e-05, "loss": 1.9424, "step": 27060 }, { "epoch": 0.04, "learning_rate": 4.8308125e-05, "loss": 2.1575, "step": 27070 }, { "epoch": 0.04, "learning_rate": 4.83075e-05, "loss": 2.1704, "step": 27080 }, { "epoch": 0.04, "learning_rate": 4.8306875e-05, "loss": 2.0273, "step": 27090 }, { "epoch": 0.04, "learning_rate": 4.8306250000000005e-05, "loss": 1.9334, "step": 27100 }, { "epoch": 0.04, "learning_rate": 4.8305625e-05, "loss": 1.9978, "step": 27110 }, { "epoch": 0.04, "learning_rate": 4.8305e-05, "loss": 2.2261, "step": 27120 }, { "epoch": 0.04, "learning_rate": 4.8304375e-05, "loss": 2.1861, "step": 27130 }, { "epoch": 0.04, "learning_rate": 4.830375e-05, "loss": 2.1176, "step": 27140 }, { "epoch": 0.04, "learning_rate": 4.8303125000000004e-05, "loss": 1.9438, "step": 27150 }, { "epoch": 0.04, "learning_rate": 4.83025e-05, "loss": 2.1124, "step": 27160 }, { "epoch": 0.04, "learning_rate": 4.8301875000000004e-05, "loss": 2.2473, "step": 27170 }, { "epoch": 0.04, "learning_rate": 4.830125e-05, "loss": 1.9801, "step": 27180 }, { "epoch": 0.04, "learning_rate": 4.8300625000000005e-05, "loss": 2.2046, "step": 27190 }, { "epoch": 0.04, "learning_rate": 4.83e-05, "loss": 2.072, "step": 27200 }, { "epoch": 0.04, "learning_rate": 4.8299375000000005e-05, "loss": 2.0096, "step": 27210 }, { "epoch": 0.04, "learning_rate": 4.829875e-05, "loss": 1.887, "step": 27220 }, { "epoch": 0.04, "learning_rate": 4.8298125e-05, "loss": 1.9204, "step": 27230 }, { "epoch": 0.04, "learning_rate": 4.82975e-05, "loss": 2.1874, "step": 27240 }, { "epoch": 0.04, "learning_rate": 4.8296875e-05, "loss": 2.0132, "step": 27250 }, { "epoch": 0.04, "learning_rate": 4.8296250000000004e-05, "loss": 2.1451, "step": 27260 }, { "epoch": 0.04, "learning_rate": 4.8295625e-05, "loss": 2.0153, "step": 27270 }, { "epoch": 0.04, "learning_rate": 4.8295000000000004e-05, "loss": 2.1206, "step": 27280 }, { "epoch": 0.04, "learning_rate": 4.8294375e-05, "loss": 2.207, "step": 27290 }, { "epoch": 0.04, "learning_rate": 4.8293750000000005e-05, "loss": 2.0042, "step": 27300 }, { "epoch": 0.04, "learning_rate": 4.8293125e-05, "loss": 2.0752, "step": 27310 }, { "epoch": 0.04, "learning_rate": 4.82925e-05, "loss": 2.2219, "step": 27320 }, { "epoch": 0.04, "learning_rate": 4.8291875e-05, "loss": 2.1253, "step": 27330 }, { "epoch": 0.04, "learning_rate": 4.8291250000000006e-05, "loss": 2.0112, "step": 27340 }, { "epoch": 0.04, "learning_rate": 4.8290625e-05, "loss": 2.0906, "step": 27350 }, { "epoch": 0.04, "learning_rate": 4.829e-05, "loss": 1.9428, "step": 27360 }, { "epoch": 0.04, "learning_rate": 4.8289375000000004e-05, "loss": 1.9481, "step": 27370 }, { "epoch": 0.04, "learning_rate": 4.828875000000001e-05, "loss": 2.081, "step": 27380 }, { "epoch": 0.04, "learning_rate": 4.8288125e-05, "loss": 2.0821, "step": 27390 }, { "epoch": 0.04, "learning_rate": 4.82875e-05, "loss": 2.2023, "step": 27400 }, { "epoch": 0.04, "learning_rate": 4.8286875000000005e-05, "loss": 2.0543, "step": 27410 }, { "epoch": 0.04, "learning_rate": 4.828625e-05, "loss": 2.276, "step": 27420 }, { "epoch": 0.04, "learning_rate": 4.8285625e-05, "loss": 2.128, "step": 27430 }, { "epoch": 0.04, "learning_rate": 4.8285e-05, "loss": 1.8186, "step": 27440 }, { "epoch": 0.04, "learning_rate": 4.8284375000000006e-05, "loss": 2.2131, "step": 27450 }, { "epoch": 0.04, "learning_rate": 4.8283749999999997e-05, "loss": 2.2319, "step": 27460 }, { "epoch": 0.04, "learning_rate": 4.8283125e-05, "loss": 1.9128, "step": 27470 }, { "epoch": 0.04, "learning_rate": 4.8282500000000004e-05, "loss": 2.0607, "step": 27480 }, { "epoch": 0.04, "learning_rate": 4.8281875e-05, "loss": 1.9549, "step": 27490 }, { "epoch": 0.04, "learning_rate": 4.828125e-05, "loss": 1.931, "step": 27500 }, { "epoch": 0.04, "learning_rate": 4.8280625e-05, "loss": 2.1652, "step": 27510 }, { "epoch": 0.04, "learning_rate": 4.8280000000000005e-05, "loss": 2.1745, "step": 27520 }, { "epoch": 0.04, "learning_rate": 4.8279375e-05, "loss": 1.9563, "step": 27530 }, { "epoch": 0.04, "learning_rate": 4.827875e-05, "loss": 2.2533, "step": 27540 }, { "epoch": 0.04, "learning_rate": 4.8278125e-05, "loss": 2.1899, "step": 27550 }, { "epoch": 0.04, "learning_rate": 4.8277500000000007e-05, "loss": 2.1126, "step": 27560 }, { "epoch": 0.04, "learning_rate": 4.8276875000000004e-05, "loss": 2.0748, "step": 27570 }, { "epoch": 0.04, "learning_rate": 4.827625e-05, "loss": 2.0789, "step": 27580 }, { "epoch": 0.04, "learning_rate": 4.8275625000000004e-05, "loss": 2.2142, "step": 27590 }, { "epoch": 0.04, "learning_rate": 4.8275e-05, "loss": 2.2224, "step": 27600 }, { "epoch": 0.04, "learning_rate": 4.8274375000000005e-05, "loss": 2.1074, "step": 27610 }, { "epoch": 0.04, "learning_rate": 4.827375e-05, "loss": 2.0272, "step": 27620 }, { "epoch": 0.04, "learning_rate": 4.8273125000000005e-05, "loss": 2.1184, "step": 27630 }, { "epoch": 0.04, "learning_rate": 4.82725e-05, "loss": 2.1237, "step": 27640 }, { "epoch": 0.04, "learning_rate": 4.8271875e-05, "loss": 1.8887, "step": 27650 }, { "epoch": 0.04, "learning_rate": 4.827125e-05, "loss": 2.122, "step": 27660 }, { "epoch": 0.04, "learning_rate": 4.8270625e-05, "loss": 1.9449, "step": 27670 }, { "epoch": 0.04, "learning_rate": 4.8270000000000004e-05, "loss": 2.065, "step": 27680 }, { "epoch": 0.04, "learning_rate": 4.8269375e-05, "loss": 2.2044, "step": 27690 }, { "epoch": 0.04, "learning_rate": 4.8268750000000004e-05, "loss": 2.0708, "step": 27700 }, { "epoch": 0.04, "learning_rate": 4.8268125e-05, "loss": 2.0013, "step": 27710 }, { "epoch": 0.04, "learning_rate": 4.8267500000000005e-05, "loss": 1.8444, "step": 27720 }, { "epoch": 0.04, "learning_rate": 4.8266875e-05, "loss": 2.0308, "step": 27730 }, { "epoch": 0.04, "learning_rate": 4.826625e-05, "loss": 2.0701, "step": 27740 }, { "epoch": 0.04, "learning_rate": 4.8265625e-05, "loss": 2.1301, "step": 27750 }, { "epoch": 0.04, "learning_rate": 4.8265000000000006e-05, "loss": 2.0634, "step": 27760 }, { "epoch": 0.04, "learning_rate": 4.8264375e-05, "loss": 1.8155, "step": 27770 }, { "epoch": 0.04, "learning_rate": 4.826375e-05, "loss": 2.2152, "step": 27780 }, { "epoch": 0.04, "learning_rate": 4.8263125000000004e-05, "loss": 2.14, "step": 27790 }, { "epoch": 0.04, "learning_rate": 4.826250000000001e-05, "loss": 2.0501, "step": 27800 }, { "epoch": 0.04, "learning_rate": 4.8261875e-05, "loss": 1.7405, "step": 27810 }, { "epoch": 0.04, "learning_rate": 4.826125e-05, "loss": 1.9663, "step": 27820 }, { "epoch": 0.04, "learning_rate": 4.8260625000000005e-05, "loss": 2.093, "step": 27830 }, { "epoch": 0.04, "learning_rate": 4.826e-05, "loss": 1.9528, "step": 27840 }, { "epoch": 0.04, "learning_rate": 4.8259375e-05, "loss": 2.1037, "step": 27850 }, { "epoch": 0.04, "learning_rate": 4.825875e-05, "loss": 2.0077, "step": 27860 }, { "epoch": 0.04, "learning_rate": 4.8258125000000006e-05, "loss": 2.1292, "step": 27870 }, { "epoch": 0.04, "learning_rate": 4.82575e-05, "loss": 2.1007, "step": 27880 }, { "epoch": 0.04, "learning_rate": 4.8256875e-05, "loss": 1.9038, "step": 27890 }, { "epoch": 0.04, "learning_rate": 4.8256250000000004e-05, "loss": 2.0702, "step": 27900 }, { "epoch": 0.04, "learning_rate": 4.8255625e-05, "loss": 2.056, "step": 27910 }, { "epoch": 0.04, "learning_rate": 4.8255e-05, "loss": 1.9052, "step": 27920 }, { "epoch": 0.04, "learning_rate": 4.8254375e-05, "loss": 2.1694, "step": 27930 }, { "epoch": 0.04, "learning_rate": 4.8253750000000005e-05, "loss": 2.0448, "step": 27940 }, { "epoch": 0.04, "learning_rate": 4.8253125e-05, "loss": 1.9845, "step": 27950 }, { "epoch": 0.04, "learning_rate": 4.82525e-05, "loss": 2.1513, "step": 27960 }, { "epoch": 0.04, "learning_rate": 4.8251875e-05, "loss": 2.1091, "step": 27970 }, { "epoch": 0.04, "learning_rate": 4.825125000000001e-05, "loss": 2.003, "step": 27980 }, { "epoch": 0.04, "learning_rate": 4.8250625000000004e-05, "loss": 2.264, "step": 27990 }, { "epoch": 0.04, "learning_rate": 4.825e-05, "loss": 1.8918, "step": 28000 }, { "epoch": 0.04, "eval_loss": 1.9836450815200806, "eval_runtime": 1290.315, "eval_samples_per_second": 44.978, "eval_steps_per_second": 5.623, "step": 28000 }, { "epoch": 0.04, "learning_rate": 4.8249375000000004e-05, "loss": 2.1978, "step": 28010 }, { "epoch": 0.04, "learning_rate": 4.824875e-05, "loss": 2.219, "step": 28020 }, { "epoch": 0.04, "learning_rate": 4.8248125e-05, "loss": 1.9407, "step": 28030 }, { "epoch": 0.04, "learning_rate": 4.82475e-05, "loss": 2.3442, "step": 28040 }, { "epoch": 0.04, "learning_rate": 4.8246875000000006e-05, "loss": 2.2179, "step": 28050 }, { "epoch": 0.04, "learning_rate": 4.824625e-05, "loss": 2.129, "step": 28060 }, { "epoch": 0.04, "learning_rate": 4.8245625e-05, "loss": 1.9251, "step": 28070 }, { "epoch": 0.04, "learning_rate": 4.8245e-05, "loss": 1.9422, "step": 28080 }, { "epoch": 0.04, "learning_rate": 4.8244375e-05, "loss": 1.9071, "step": 28090 }, { "epoch": 0.04, "learning_rate": 4.8243750000000004e-05, "loss": 2.2151, "step": 28100 }, { "epoch": 0.04, "learning_rate": 4.8243125e-05, "loss": 2.1714, "step": 28110 }, { "epoch": 0.04, "learning_rate": 4.8242500000000004e-05, "loss": 2.3387, "step": 28120 }, { "epoch": 0.04, "learning_rate": 4.8241875e-05, "loss": 2.1877, "step": 28130 }, { "epoch": 0.04, "learning_rate": 4.8241250000000005e-05, "loss": 1.8939, "step": 28140 }, { "epoch": 0.04, "learning_rate": 4.8240625e-05, "loss": 1.6398, "step": 28150 }, { "epoch": 0.04, "learning_rate": 4.824e-05, "loss": 2.0846, "step": 28160 }, { "epoch": 0.04, "learning_rate": 4.8239375e-05, "loss": 1.8661, "step": 28170 }, { "epoch": 0.04, "learning_rate": 4.8238750000000006e-05, "loss": 1.8297, "step": 28180 }, { "epoch": 0.04, "learning_rate": 4.8238125e-05, "loss": 1.9914, "step": 28190 }, { "epoch": 0.04, "learning_rate": 4.82375e-05, "loss": 2.2286, "step": 28200 }, { "epoch": 0.04, "learning_rate": 4.8236875000000004e-05, "loss": 2.1448, "step": 28210 }, { "epoch": 0.04, "learning_rate": 4.823625000000001e-05, "loss": 1.8125, "step": 28220 }, { "epoch": 0.04, "learning_rate": 4.8235625e-05, "loss": 2.0851, "step": 28230 }, { "epoch": 0.04, "learning_rate": 4.8235e-05, "loss": 2.0234, "step": 28240 }, { "epoch": 0.04, "learning_rate": 4.8234375000000005e-05, "loss": 1.9115, "step": 28250 }, { "epoch": 0.04, "learning_rate": 4.823375e-05, "loss": 2.006, "step": 28260 }, { "epoch": 0.04, "learning_rate": 4.8233125e-05, "loss": 1.9976, "step": 28270 }, { "epoch": 0.04, "learning_rate": 4.82325e-05, "loss": 2.0605, "step": 28280 }, { "epoch": 0.04, "learning_rate": 4.8231875000000007e-05, "loss": 2.1565, "step": 28290 }, { "epoch": 0.04, "learning_rate": 4.823125e-05, "loss": 1.8454, "step": 28300 }, { "epoch": 0.04, "learning_rate": 4.8230625e-05, "loss": 2.0686, "step": 28310 }, { "epoch": 0.04, "learning_rate": 4.8230000000000004e-05, "loss": 1.813, "step": 28320 }, { "epoch": 0.04, "learning_rate": 4.8229375e-05, "loss": 2.038, "step": 28330 }, { "epoch": 0.04, "learning_rate": 4.822875e-05, "loss": 1.9448, "step": 28340 }, { "epoch": 0.04, "learning_rate": 4.8228125e-05, "loss": 2.1471, "step": 28350 }, { "epoch": 0.04, "learning_rate": 4.8227500000000005e-05, "loss": 2.0449, "step": 28360 }, { "epoch": 0.04, "learning_rate": 4.8226875e-05, "loss": 2.4226, "step": 28370 }, { "epoch": 0.04, "learning_rate": 4.822625e-05, "loss": 2.2111, "step": 28380 }, { "epoch": 0.04, "learning_rate": 4.8225625e-05, "loss": 1.9957, "step": 28390 }, { "epoch": 0.04, "learning_rate": 4.822500000000001e-05, "loss": 2.3001, "step": 28400 }, { "epoch": 0.04, "learning_rate": 4.8224375000000004e-05, "loss": 2.2951, "step": 28410 }, { "epoch": 0.04, "learning_rate": 4.822375e-05, "loss": 1.9291, "step": 28420 }, { "epoch": 0.04, "learning_rate": 4.8223125000000004e-05, "loss": 1.8151, "step": 28430 }, { "epoch": 0.04, "learning_rate": 4.82225e-05, "loss": 1.8748, "step": 28440 }, { "epoch": 0.04, "learning_rate": 4.8221875e-05, "loss": 1.8736, "step": 28450 }, { "epoch": 0.04, "learning_rate": 4.822125e-05, "loss": 1.897, "step": 28460 }, { "epoch": 0.04, "learning_rate": 4.8220625000000006e-05, "loss": 2.0346, "step": 28470 }, { "epoch": 0.04, "learning_rate": 4.822e-05, "loss": 2.1231, "step": 28480 }, { "epoch": 0.04, "learning_rate": 4.8219375e-05, "loss": 1.8616, "step": 28490 }, { "epoch": 0.04, "learning_rate": 4.821875e-05, "loss": 1.9975, "step": 28500 }, { "epoch": 0.04, "learning_rate": 4.8218125e-05, "loss": 2.0076, "step": 28510 }, { "epoch": 0.04, "learning_rate": 4.8217500000000004e-05, "loss": 2.1149, "step": 28520 }, { "epoch": 0.04, "learning_rate": 4.8216875e-05, "loss": 1.9896, "step": 28530 }, { "epoch": 0.04, "learning_rate": 4.8216250000000004e-05, "loss": 1.8497, "step": 28540 }, { "epoch": 0.04, "learning_rate": 4.8215625e-05, "loss": 2.1096, "step": 28550 }, { "epoch": 0.04, "learning_rate": 4.8215000000000005e-05, "loss": 1.881, "step": 28560 }, { "epoch": 0.04, "learning_rate": 4.8214375e-05, "loss": 2.1627, "step": 28570 }, { "epoch": 0.04, "learning_rate": 4.821375e-05, "loss": 1.8871, "step": 28580 }, { "epoch": 0.04, "learning_rate": 4.8213125e-05, "loss": 2.1501, "step": 28590 }, { "epoch": 0.04, "learning_rate": 4.8212500000000006e-05, "loss": 2.0191, "step": 28600 }, { "epoch": 0.04, "learning_rate": 4.8211875e-05, "loss": 1.9087, "step": 28610 }, { "epoch": 0.04, "learning_rate": 4.821125e-05, "loss": 2.105, "step": 28620 }, { "epoch": 0.04, "learning_rate": 4.8210625000000004e-05, "loss": 1.7039, "step": 28630 }, { "epoch": 0.04, "learning_rate": 4.821e-05, "loss": 1.8955, "step": 28640 }, { "epoch": 0.04, "learning_rate": 4.8209375e-05, "loss": 2.172, "step": 28650 }, { "epoch": 0.04, "learning_rate": 4.820875e-05, "loss": 2.0833, "step": 28660 }, { "epoch": 0.04, "learning_rate": 4.8208125000000005e-05, "loss": 1.9392, "step": 28670 }, { "epoch": 0.04, "learning_rate": 4.82075e-05, "loss": 2.2573, "step": 28680 }, { "epoch": 0.04, "learning_rate": 4.8206875e-05, "loss": 1.7853, "step": 28690 }, { "epoch": 0.04, "learning_rate": 4.820625e-05, "loss": 2.0003, "step": 28700 }, { "epoch": 0.04, "learning_rate": 4.8205625000000007e-05, "loss": 1.8164, "step": 28710 }, { "epoch": 0.04, "learning_rate": 4.8205000000000003e-05, "loss": 1.9141, "step": 28720 }, { "epoch": 0.04, "learning_rate": 4.8204375e-05, "loss": 1.9123, "step": 28730 }, { "epoch": 0.04, "learning_rate": 4.8203750000000004e-05, "loss": 1.9963, "step": 28740 }, { "epoch": 0.04, "learning_rate": 4.820312500000001e-05, "loss": 2.052, "step": 28750 }, { "epoch": 0.04, "learning_rate": 4.82025e-05, "loss": 2.1516, "step": 28760 }, { "epoch": 0.04, "learning_rate": 4.8201875e-05, "loss": 2.0179, "step": 28770 }, { "epoch": 0.04, "learning_rate": 4.8201250000000005e-05, "loss": 2.1899, "step": 28780 }, { "epoch": 0.04, "learning_rate": 4.8200625e-05, "loss": 1.8441, "step": 28790 }, { "epoch": 0.04, "learning_rate": 4.82e-05, "loss": 2.1017, "step": 28800 }, { "epoch": 0.04, "learning_rate": 4.8199375e-05, "loss": 2.075, "step": 28810 }, { "epoch": 0.04, "learning_rate": 4.819875000000001e-05, "loss": 2.1203, "step": 28820 }, { "epoch": 0.04, "learning_rate": 4.8198125000000004e-05, "loss": 2.1422, "step": 28830 }, { "epoch": 0.04, "learning_rate": 4.81975e-05, "loss": 1.9385, "step": 28840 }, { "epoch": 0.04, "learning_rate": 4.8196875000000004e-05, "loss": 2.0186, "step": 28850 }, { "epoch": 0.04, "learning_rate": 4.819625e-05, "loss": 1.9263, "step": 28860 }, { "epoch": 0.04, "learning_rate": 4.8195625e-05, "loss": 1.8973, "step": 28870 }, { "epoch": 0.04, "learning_rate": 4.8195e-05, "loss": 2.3611, "step": 28880 }, { "epoch": 0.04, "learning_rate": 4.8194375000000006e-05, "loss": 2.1176, "step": 28890 }, { "epoch": 0.04, "learning_rate": 4.819375e-05, "loss": 2.0479, "step": 28900 }, { "epoch": 0.04, "learning_rate": 4.8193125e-05, "loss": 2.0097, "step": 28910 }, { "epoch": 0.04, "learning_rate": 4.81925e-05, "loss": 2.0337, "step": 28920 }, { "epoch": 0.04, "learning_rate": 4.8191875e-05, "loss": 1.9712, "step": 28930 }, { "epoch": 0.04, "learning_rate": 4.8191250000000004e-05, "loss": 2.1635, "step": 28940 }, { "epoch": 0.04, "learning_rate": 4.8190625e-05, "loss": 2.0562, "step": 28950 }, { "epoch": 0.04, "learning_rate": 4.8190000000000004e-05, "loss": 1.9076, "step": 28960 }, { "epoch": 0.04, "learning_rate": 4.8189375e-05, "loss": 1.9923, "step": 28970 }, { "epoch": 0.04, "learning_rate": 4.8188750000000005e-05, "loss": 1.7329, "step": 28980 }, { "epoch": 0.04, "learning_rate": 4.8188125e-05, "loss": 1.8226, "step": 28990 }, { "epoch": 0.04, "learning_rate": 4.81875e-05, "loss": 1.9146, "step": 29000 }, { "epoch": 0.04, "eval_loss": 1.959999918937683, "eval_runtime": 1290.3101, "eval_samples_per_second": 44.978, "eval_steps_per_second": 5.623, "step": 29000 }, { "epoch": 0.04, "learning_rate": 4.8186875e-05, "loss": 1.7794, "step": 29010 }, { "epoch": 0.04, "learning_rate": 4.8186250000000006e-05, "loss": 1.8434, "step": 29020 }, { "epoch": 0.04, "learning_rate": 4.8185625e-05, "loss": 2.1186, "step": 29030 }, { "epoch": 0.04, "learning_rate": 4.8185e-05, "loss": 1.7925, "step": 29040 }, { "epoch": 0.04, "learning_rate": 4.8184375000000004e-05, "loss": 2.0031, "step": 29050 }, { "epoch": 0.04, "learning_rate": 4.818375e-05, "loss": 1.9604, "step": 29060 }, { "epoch": 0.04, "learning_rate": 4.8183125e-05, "loss": 2.176, "step": 29070 }, { "epoch": 0.04, "learning_rate": 4.81825e-05, "loss": 2.2015, "step": 29080 }, { "epoch": 0.04, "learning_rate": 4.8181875000000005e-05, "loss": 2.085, "step": 29090 }, { "epoch": 0.04, "learning_rate": 4.818125e-05, "loss": 2.2051, "step": 29100 }, { "epoch": 0.04, "learning_rate": 4.8180625e-05, "loss": 1.8834, "step": 29110 }, { "epoch": 0.04, "learning_rate": 4.818e-05, "loss": 1.8241, "step": 29120 }, { "epoch": 0.04, "learning_rate": 4.8179375000000007e-05, "loss": 2.102, "step": 29130 }, { "epoch": 0.04, "learning_rate": 4.8178750000000004e-05, "loss": 2.0871, "step": 29140 }, { "epoch": 0.04, "learning_rate": 4.8178125e-05, "loss": 2.0063, "step": 29150 }, { "epoch": 0.04, "learning_rate": 4.8177500000000004e-05, "loss": 1.945, "step": 29160 }, { "epoch": 0.04, "learning_rate": 4.817687500000001e-05, "loss": 2.0224, "step": 29170 }, { "epoch": 0.04, "learning_rate": 4.817625e-05, "loss": 2.0589, "step": 29180 }, { "epoch": 0.04, "learning_rate": 4.8175625e-05, "loss": 2.0808, "step": 29190 }, { "epoch": 0.04, "learning_rate": 4.8175000000000005e-05, "loss": 1.9518, "step": 29200 }, { "epoch": 0.04, "learning_rate": 4.8174375e-05, "loss": 2.2992, "step": 29210 }, { "epoch": 0.04, "learning_rate": 4.817375e-05, "loss": 1.7493, "step": 29220 }, { "epoch": 0.04, "learning_rate": 4.8173125e-05, "loss": 1.8616, "step": 29230 }, { "epoch": 0.04, "learning_rate": 4.817250000000001e-05, "loss": 2.377, "step": 29240 }, { "epoch": 0.04, "learning_rate": 4.8171875e-05, "loss": 2.1647, "step": 29250 }, { "epoch": 0.04, "learning_rate": 4.817125e-05, "loss": 2.0983, "step": 29260 }, { "epoch": 0.04, "learning_rate": 4.8170625000000004e-05, "loss": 2.2536, "step": 29270 }, { "epoch": 0.04, "learning_rate": 4.817e-05, "loss": 2.0083, "step": 29280 }, { "epoch": 0.04, "learning_rate": 4.8169375e-05, "loss": 2.2281, "step": 29290 }, { "epoch": 0.04, "learning_rate": 4.816875e-05, "loss": 1.7837, "step": 29300 }, { "epoch": 0.04, "learning_rate": 4.8168125000000006e-05, "loss": 1.8267, "step": 29310 }, { "epoch": 0.04, "learning_rate": 4.81675e-05, "loss": 1.924, "step": 29320 }, { "epoch": 0.04, "learning_rate": 4.8166875e-05, "loss": 2.0108, "step": 29330 }, { "epoch": 0.04, "learning_rate": 4.816625e-05, "loss": 1.9294, "step": 29340 }, { "epoch": 0.04, "learning_rate": 4.8165625e-05, "loss": 1.8225, "step": 29350 }, { "epoch": 0.04, "learning_rate": 4.8165000000000004e-05, "loss": 1.9627, "step": 29360 }, { "epoch": 0.04, "learning_rate": 4.8164375e-05, "loss": 2.2068, "step": 29370 }, { "epoch": 0.04, "learning_rate": 4.8163750000000005e-05, "loss": 1.9752, "step": 29380 }, { "epoch": 0.04, "learning_rate": 4.8163125e-05, "loss": 1.9449, "step": 29390 }, { "epoch": 0.04, "learning_rate": 4.8162500000000005e-05, "loss": 2.0803, "step": 29400 }, { "epoch": 0.04, "learning_rate": 4.8161875e-05, "loss": 1.932, "step": 29410 }, { "epoch": 0.04, "learning_rate": 4.816125e-05, "loss": 2.0314, "step": 29420 }, { "epoch": 0.04, "learning_rate": 4.8160625e-05, "loss": 1.8833, "step": 29430 }, { "epoch": 0.04, "learning_rate": 4.816e-05, "loss": 1.9965, "step": 29440 }, { "epoch": 0.04, "learning_rate": 4.8159375000000003e-05, "loss": 1.9432, "step": 29450 }, { "epoch": 0.04, "learning_rate": 4.815875e-05, "loss": 1.9782, "step": 29460 }, { "epoch": 0.04, "learning_rate": 4.8158125000000004e-05, "loss": 1.916, "step": 29470 }, { "epoch": 0.04, "learning_rate": 4.81575e-05, "loss": 2.1925, "step": 29480 }, { "epoch": 0.04, "learning_rate": 4.8156875e-05, "loss": 1.9248, "step": 29490 }, { "epoch": 0.04, "learning_rate": 4.815625e-05, "loss": 2.0987, "step": 29500 }, { "epoch": 0.04, "learning_rate": 4.8155625000000005e-05, "loss": 1.9756, "step": 29510 }, { "epoch": 0.04, "learning_rate": 4.8155e-05, "loss": 2.1215, "step": 29520 }, { "epoch": 0.04, "learning_rate": 4.8154375e-05, "loss": 2.1974, "step": 29530 }, { "epoch": 0.04, "learning_rate": 4.815375e-05, "loss": 2.1762, "step": 29540 }, { "epoch": 0.04, "learning_rate": 4.815312500000001e-05, "loss": 2.2737, "step": 29550 }, { "epoch": 0.04, "learning_rate": 4.8152500000000004e-05, "loss": 2.0171, "step": 29560 }, { "epoch": 0.04, "learning_rate": 4.8151875e-05, "loss": 1.8255, "step": 29570 }, { "epoch": 0.04, "learning_rate": 4.8151250000000004e-05, "loss": 1.9749, "step": 29580 }, { "epoch": 0.04, "learning_rate": 4.815062500000001e-05, "loss": 2.1661, "step": 29590 }, { "epoch": 0.04, "learning_rate": 4.815e-05, "loss": 1.9108, "step": 29600 }, { "epoch": 0.04, "learning_rate": 4.8149375e-05, "loss": 1.8904, "step": 29610 }, { "epoch": 0.04, "learning_rate": 4.8148750000000006e-05, "loss": 2.1435, "step": 29620 }, { "epoch": 0.04, "learning_rate": 4.8148125e-05, "loss": 1.7751, "step": 29630 }, { "epoch": 0.04, "learning_rate": 4.81475e-05, "loss": 2.0366, "step": 29640 }, { "epoch": 0.04, "learning_rate": 4.8146875e-05, "loss": 2.0249, "step": 29650 }, { "epoch": 0.04, "learning_rate": 4.814625000000001e-05, "loss": 2.0307, "step": 29660 }, { "epoch": 0.04, "learning_rate": 4.8145625e-05, "loss": 2.0009, "step": 29670 }, { "epoch": 0.04, "learning_rate": 4.8145e-05, "loss": 1.9623, "step": 29680 }, { "epoch": 0.04, "learning_rate": 4.8144375000000004e-05, "loss": 1.9389, "step": 29690 }, { "epoch": 0.04, "learning_rate": 4.814375e-05, "loss": 2.071, "step": 29700 }, { "epoch": 0.04, "learning_rate": 4.8143125e-05, "loss": 1.8434, "step": 29710 }, { "epoch": 0.04, "learning_rate": 4.81425e-05, "loss": 2.0824, "step": 29720 }, { "epoch": 0.04, "learning_rate": 4.8141875000000006e-05, "loss": 2.0062, "step": 29730 }, { "epoch": 0.04, "learning_rate": 4.814125e-05, "loss": 2.1086, "step": 29740 }, { "epoch": 0.04, "learning_rate": 4.8140625e-05, "loss": 2.1686, "step": 29750 }, { "epoch": 0.04, "learning_rate": 4.814e-05, "loss": 2.1543, "step": 29760 }, { "epoch": 0.04, "learning_rate": 4.8139375e-05, "loss": 2.0437, "step": 29770 }, { "epoch": 0.04, "learning_rate": 4.8138750000000004e-05, "loss": 2.166, "step": 29780 }, { "epoch": 0.04, "learning_rate": 4.8138125e-05, "loss": 1.8333, "step": 29790 }, { "epoch": 0.04, "learning_rate": 4.8137500000000005e-05, "loss": 1.9048, "step": 29800 }, { "epoch": 0.04, "learning_rate": 4.8136875e-05, "loss": 2.1004, "step": 29810 }, { "epoch": 0.04, "learning_rate": 4.8136250000000005e-05, "loss": 2.0753, "step": 29820 }, { "epoch": 0.04, "learning_rate": 4.8135625e-05, "loss": 2.0462, "step": 29830 }, { "epoch": 0.04, "learning_rate": 4.8135e-05, "loss": 2.2577, "step": 29840 }, { "epoch": 0.04, "learning_rate": 4.8134375e-05, "loss": 1.9045, "step": 29850 }, { "epoch": 0.04, "learning_rate": 4.813375e-05, "loss": 2.0537, "step": 29860 }, { "epoch": 0.04, "learning_rate": 4.8133125000000003e-05, "loss": 1.7584, "step": 29870 }, { "epoch": 0.04, "learning_rate": 4.81325e-05, "loss": 1.7032, "step": 29880 }, { "epoch": 0.04, "learning_rate": 4.8131875000000004e-05, "loss": 2.1635, "step": 29890 }, { "epoch": 0.04, "learning_rate": 4.813125e-05, "loss": 1.9861, "step": 29900 }, { "epoch": 0.04, "learning_rate": 4.8130625000000005e-05, "loss": 2.2141, "step": 29910 }, { "epoch": 0.04, "learning_rate": 4.813e-05, "loss": 1.8392, "step": 29920 }, { "epoch": 0.04, "learning_rate": 4.8129375000000005e-05, "loss": 1.9941, "step": 29930 }, { "epoch": 0.04, "learning_rate": 4.812875e-05, "loss": 1.8697, "step": 29940 }, { "epoch": 0.04, "learning_rate": 4.8128125e-05, "loss": 2.2119, "step": 29950 }, { "epoch": 0.04, "learning_rate": 4.81275e-05, "loss": 1.8025, "step": 29960 }, { "epoch": 0.04, "learning_rate": 4.812687500000001e-05, "loss": 1.8693, "step": 29970 }, { "epoch": 0.04, "learning_rate": 4.8126250000000004e-05, "loss": 1.8772, "step": 29980 }, { "epoch": 0.04, "learning_rate": 4.8125625e-05, "loss": 1.9477, "step": 29990 }, { "epoch": 0.04, "learning_rate": 4.8125000000000004e-05, "loss": 2.0911, "step": 30000 }, { "epoch": 0.04, "eval_loss": 1.9397389888763428, "eval_runtime": 1290.3898, "eval_samples_per_second": 44.976, "eval_steps_per_second": 5.622, "step": 30000 }, { "epoch": 0.04, "learning_rate": 4.812437500000001e-05, "loss": 2.1901, "step": 30010 }, { "epoch": 0.04, "learning_rate": 4.812375e-05, "loss": 2.2051, "step": 30020 }, { "epoch": 0.04, "learning_rate": 4.8123125e-05, "loss": 2.0982, "step": 30030 }, { "epoch": 0.04, "learning_rate": 4.8122500000000006e-05, "loss": 1.9931, "step": 30040 }, { "epoch": 0.04, "learning_rate": 4.8121875e-05, "loss": 2.2086, "step": 30050 }, { "epoch": 0.04, "learning_rate": 4.812125e-05, "loss": 1.7544, "step": 30060 }, { "epoch": 0.04, "learning_rate": 4.8120625e-05, "loss": 1.9648, "step": 30070 }, { "epoch": 0.04, "learning_rate": 4.812000000000001e-05, "loss": 2.186, "step": 30080 }, { "epoch": 0.04, "learning_rate": 4.8119375e-05, "loss": 1.8112, "step": 30090 }, { "epoch": 0.04, "learning_rate": 4.811875e-05, "loss": 2.3526, "step": 30100 }, { "epoch": 0.04, "learning_rate": 4.8118125000000004e-05, "loss": 2.0326, "step": 30110 }, { "epoch": 0.04, "learning_rate": 4.81175e-05, "loss": 2.194, "step": 30120 }, { "epoch": 0.04, "learning_rate": 4.8116875e-05, "loss": 1.6909, "step": 30130 }, { "epoch": 0.04, "learning_rate": 4.811625e-05, "loss": 1.9955, "step": 30140 }, { "epoch": 0.04, "learning_rate": 4.8115625000000006e-05, "loss": 1.8421, "step": 30150 }, { "epoch": 0.04, "learning_rate": 4.8115e-05, "loss": 1.9767, "step": 30160 }, { "epoch": 0.04, "learning_rate": 4.8114375e-05, "loss": 2.0239, "step": 30170 }, { "epoch": 0.04, "learning_rate": 4.811375e-05, "loss": 1.9627, "step": 30180 }, { "epoch": 0.04, "learning_rate": 4.8113125e-05, "loss": 1.9324, "step": 30190 }, { "epoch": 0.04, "learning_rate": 4.8112500000000004e-05, "loss": 1.8233, "step": 30200 }, { "epoch": 0.04, "learning_rate": 4.8111875e-05, "loss": 2.0035, "step": 30210 }, { "epoch": 0.04, "learning_rate": 4.8111250000000005e-05, "loss": 2.1572, "step": 30220 }, { "epoch": 0.04, "learning_rate": 4.8110625e-05, "loss": 1.8671, "step": 30230 }, { "epoch": 0.04, "learning_rate": 4.8110000000000005e-05, "loss": 1.9568, "step": 30240 }, { "epoch": 0.04, "learning_rate": 4.8109375e-05, "loss": 2.0527, "step": 30250 }, { "epoch": 0.04, "learning_rate": 4.810875e-05, "loss": 1.7696, "step": 30260 }, { "epoch": 0.04, "learning_rate": 4.8108125e-05, "loss": 2.1956, "step": 30270 }, { "epoch": 0.04, "learning_rate": 4.81075e-05, "loss": 2.0798, "step": 30280 }, { "epoch": 0.04, "learning_rate": 4.8106875000000003e-05, "loss": 2.1359, "step": 30290 }, { "epoch": 0.04, "learning_rate": 4.810625e-05, "loss": 2.0406, "step": 30300 }, { "epoch": 0.04, "learning_rate": 4.8105625000000004e-05, "loss": 2.0583, "step": 30310 }, { "epoch": 0.04, "learning_rate": 4.8105e-05, "loss": 2.1847, "step": 30320 }, { "epoch": 0.04, "learning_rate": 4.8104375000000005e-05, "loss": 1.9304, "step": 30330 }, { "epoch": 0.04, "learning_rate": 4.810375e-05, "loss": 1.9614, "step": 30340 }, { "epoch": 0.04, "learning_rate": 4.8103125000000005e-05, "loss": 2.2225, "step": 30350 }, { "epoch": 0.04, "learning_rate": 4.81025e-05, "loss": 1.7541, "step": 30360 }, { "epoch": 0.04, "learning_rate": 4.8101875e-05, "loss": 2.0205, "step": 30370 }, { "epoch": 0.04, "learning_rate": 4.810125e-05, "loss": 1.9629, "step": 30380 }, { "epoch": 0.04, "learning_rate": 4.810062500000001e-05, "loss": 2.2828, "step": 30390 }, { "epoch": 0.04, "learning_rate": 4.8100000000000004e-05, "loss": 1.9521, "step": 30400 }, { "epoch": 0.04, "learning_rate": 4.8099375e-05, "loss": 2.0419, "step": 30410 }, { "epoch": 0.04, "learning_rate": 4.8098750000000004e-05, "loss": 2.1576, "step": 30420 }, { "epoch": 0.04, "learning_rate": 4.809812500000001e-05, "loss": 2.1168, "step": 30430 }, { "epoch": 0.04, "learning_rate": 4.80975e-05, "loss": 1.9339, "step": 30440 }, { "epoch": 0.04, "learning_rate": 4.8096875e-05, "loss": 2.1276, "step": 30450 }, { "epoch": 0.04, "learning_rate": 4.8096250000000006e-05, "loss": 2.0948, "step": 30460 }, { "epoch": 0.04, "learning_rate": 4.8095625e-05, "loss": 1.8733, "step": 30470 }, { "epoch": 0.04, "learning_rate": 4.8095e-05, "loss": 1.9538, "step": 30480 }, { "epoch": 0.04, "learning_rate": 4.8094375e-05, "loss": 2.0342, "step": 30490 }, { "epoch": 0.04, "learning_rate": 4.809375000000001e-05, "loss": 2.0777, "step": 30500 }, { "epoch": 0.04, "learning_rate": 4.8093125e-05, "loss": 1.8658, "step": 30510 }, { "epoch": 0.04, "learning_rate": 4.80925e-05, "loss": 1.8529, "step": 30520 }, { "epoch": 0.04, "learning_rate": 4.8091875000000004e-05, "loss": 2.0437, "step": 30530 }, { "epoch": 0.04, "learning_rate": 4.809125e-05, "loss": 1.6918, "step": 30540 }, { "epoch": 0.04, "learning_rate": 4.8090625e-05, "loss": 1.9349, "step": 30550 }, { "epoch": 0.04, "learning_rate": 4.809e-05, "loss": 1.9816, "step": 30560 }, { "epoch": 0.04, "learning_rate": 4.8089375000000006e-05, "loss": 2.1628, "step": 30570 }, { "epoch": 0.04, "learning_rate": 4.808875e-05, "loss": 1.9844, "step": 30580 }, { "epoch": 0.04, "learning_rate": 4.8088125e-05, "loss": 2.0711, "step": 30590 }, { "epoch": 0.04, "learning_rate": 4.80875e-05, "loss": 1.9103, "step": 30600 }, { "epoch": 0.04, "learning_rate": 4.8086875e-05, "loss": 2.1487, "step": 30610 }, { "epoch": 0.04, "learning_rate": 4.8086250000000004e-05, "loss": 1.9924, "step": 30620 }, { "epoch": 0.04, "learning_rate": 4.8085625e-05, "loss": 2.1826, "step": 30630 }, { "epoch": 0.04, "learning_rate": 4.8085000000000005e-05, "loss": 1.942, "step": 30640 }, { "epoch": 0.04, "learning_rate": 4.8084375e-05, "loss": 1.7862, "step": 30650 }, { "epoch": 0.04, "learning_rate": 4.808375e-05, "loss": 1.9048, "step": 30660 }, { "epoch": 0.04, "learning_rate": 4.8083125e-05, "loss": 2.2473, "step": 30670 }, { "epoch": 0.04, "learning_rate": 4.8082500000000006e-05, "loss": 2.0804, "step": 30680 }, { "epoch": 0.04, "learning_rate": 4.8081875e-05, "loss": 1.802, "step": 30690 }, { "epoch": 0.04, "learning_rate": 4.808125e-05, "loss": 2.0815, "step": 30700 }, { "epoch": 0.04, "learning_rate": 4.8080625000000004e-05, "loss": 2.1763, "step": 30710 }, { "epoch": 0.04, "learning_rate": 4.808e-05, "loss": 2.1716, "step": 30720 }, { "epoch": 0.04, "learning_rate": 4.8079375000000004e-05, "loss": 2.0968, "step": 30730 }, { "epoch": 0.04, "learning_rate": 4.807875e-05, "loss": 2.0494, "step": 30740 }, { "epoch": 0.04, "learning_rate": 4.8078125000000005e-05, "loss": 2.009, "step": 30750 }, { "epoch": 0.04, "learning_rate": 4.80775e-05, "loss": 1.9409, "step": 30760 }, { "epoch": 0.04, "learning_rate": 4.8076875000000005e-05, "loss": 1.9602, "step": 30770 }, { "epoch": 0.04, "learning_rate": 4.807625e-05, "loss": 2.0658, "step": 30780 }, { "epoch": 0.04, "learning_rate": 4.8075625e-05, "loss": 1.9256, "step": 30790 }, { "epoch": 0.04, "learning_rate": 4.8075e-05, "loss": 1.9112, "step": 30800 }, { "epoch": 0.04, "learning_rate": 4.807437500000001e-05, "loss": 2.1307, "step": 30810 }, { "epoch": 0.04, "learning_rate": 4.8073750000000004e-05, "loss": 1.9719, "step": 30820 }, { "epoch": 0.04, "learning_rate": 4.8073125e-05, "loss": 1.9884, "step": 30830 }, { "epoch": 0.04, "learning_rate": 4.8072500000000004e-05, "loss": 1.7495, "step": 30840 }, { "epoch": 0.04, "learning_rate": 4.8071875e-05, "loss": 1.9409, "step": 30850 }, { "epoch": 0.04, "learning_rate": 4.807125e-05, "loss": 2.058, "step": 30860 }, { "epoch": 0.04, "learning_rate": 4.8070625e-05, "loss": 1.9445, "step": 30870 }, { "epoch": 0.04, "learning_rate": 4.8070000000000006e-05, "loss": 2.1292, "step": 30880 }, { "epoch": 0.04, "learning_rate": 4.8069375e-05, "loss": 1.9434, "step": 30890 }, { "epoch": 0.04, "learning_rate": 4.806875e-05, "loss": 1.9923, "step": 30900 }, { "epoch": 0.04, "learning_rate": 4.8068125e-05, "loss": 2.2934, "step": 30910 }, { "epoch": 0.04, "learning_rate": 4.806750000000001e-05, "loss": 1.9916, "step": 30920 }, { "epoch": 0.04, "learning_rate": 4.8066875e-05, "loss": 1.9099, "step": 30930 }, { "epoch": 0.04, "learning_rate": 4.806625e-05, "loss": 2.0682, "step": 30940 }, { "epoch": 0.04, "learning_rate": 4.8065625000000005e-05, "loss": 1.9379, "step": 30950 }, { "epoch": 0.04, "learning_rate": 4.8065e-05, "loss": 2.0724, "step": 30960 }, { "epoch": 0.04, "learning_rate": 4.8064375e-05, "loss": 1.928, "step": 30970 }, { "epoch": 0.04, "learning_rate": 4.806375e-05, "loss": 2.1079, "step": 30980 }, { "epoch": 0.04, "learning_rate": 4.8063125000000006e-05, "loss": 2.055, "step": 30990 }, { "epoch": 0.04, "learning_rate": 4.80625e-05, "loss": 1.9533, "step": 31000 }, { "epoch": 0.04, "eval_loss": 1.9091051816940308, "eval_runtime": 1290.5722, "eval_samples_per_second": 44.969, "eval_steps_per_second": 5.622, "step": 31000 }, { "epoch": 0.04, "learning_rate": 4.8061875e-05, "loss": 2.0427, "step": 31010 }, { "epoch": 0.04, "learning_rate": 4.806125e-05, "loss": 1.8114, "step": 31020 }, { "epoch": 0.04, "learning_rate": 4.8060625e-05, "loss": 1.9195, "step": 31030 }, { "epoch": 0.04, "learning_rate": 4.8060000000000004e-05, "loss": 2.2795, "step": 31040 }, { "epoch": 0.04, "learning_rate": 4.8059375e-05, "loss": 1.8801, "step": 31050 }, { "epoch": 0.04, "learning_rate": 4.8058750000000005e-05, "loss": 1.8031, "step": 31060 }, { "epoch": 0.04, "learning_rate": 4.8058125e-05, "loss": 1.8374, "step": 31070 }, { "epoch": 0.04, "learning_rate": 4.80575e-05, "loss": 1.9196, "step": 31080 }, { "epoch": 0.04, "learning_rate": 4.8056875e-05, "loss": 2.0022, "step": 31090 }, { "epoch": 0.04, "learning_rate": 4.8056250000000006e-05, "loss": 2.1707, "step": 31100 }, { "epoch": 0.04, "learning_rate": 4.8055625e-05, "loss": 1.9823, "step": 31110 }, { "epoch": 0.04, "learning_rate": 4.8055e-05, "loss": 2.1564, "step": 31120 }, { "epoch": 0.04, "learning_rate": 4.8054375000000004e-05, "loss": 2.1539, "step": 31130 }, { "epoch": 0.04, "learning_rate": 4.805375e-05, "loss": 1.9788, "step": 31140 }, { "epoch": 0.04, "learning_rate": 4.8053125000000004e-05, "loss": 2.0388, "step": 31150 }, { "epoch": 0.04, "learning_rate": 4.80525e-05, "loss": 2.0657, "step": 31160 }, { "epoch": 0.04, "learning_rate": 4.8051875000000005e-05, "loss": 1.924, "step": 31170 }, { "epoch": 0.04, "learning_rate": 4.805125e-05, "loss": 1.9471, "step": 31180 }, { "epoch": 0.04, "learning_rate": 4.8050625000000005e-05, "loss": 2.0768, "step": 31190 }, { "epoch": 0.04, "learning_rate": 4.805e-05, "loss": 2.0211, "step": 31200 }, { "epoch": 0.04, "learning_rate": 4.8049375e-05, "loss": 1.8849, "step": 31210 }, { "epoch": 0.04, "learning_rate": 4.804875e-05, "loss": 1.7188, "step": 31220 }, { "epoch": 0.04, "learning_rate": 4.804812500000001e-05, "loss": 1.8788, "step": 31230 }, { "epoch": 0.04, "learning_rate": 4.8047500000000004e-05, "loss": 1.8801, "step": 31240 }, { "epoch": 0.04, "learning_rate": 4.8046875e-05, "loss": 2.1079, "step": 31250 }, { "epoch": 0.04, "learning_rate": 4.8046250000000004e-05, "loss": 1.8732, "step": 31260 }, { "epoch": 0.04, "learning_rate": 4.8045625e-05, "loss": 2.0202, "step": 31270 }, { "epoch": 0.04, "learning_rate": 4.8045e-05, "loss": 1.9268, "step": 31280 }, { "epoch": 0.04, "learning_rate": 4.8044375e-05, "loss": 1.7062, "step": 31290 }, { "epoch": 0.04, "learning_rate": 4.8043750000000006e-05, "loss": 1.7676, "step": 31300 }, { "epoch": 0.04, "learning_rate": 4.8043125e-05, "loss": 1.8567, "step": 31310 }, { "epoch": 0.04, "learning_rate": 4.80425e-05, "loss": 2.0435, "step": 31320 }, { "epoch": 0.04, "learning_rate": 4.8041875e-05, "loss": 1.7977, "step": 31330 }, { "epoch": 0.04, "learning_rate": 4.804125000000001e-05, "loss": 2.0101, "step": 31340 }, { "epoch": 0.04, "learning_rate": 4.8040625e-05, "loss": 2.0164, "step": 31350 }, { "epoch": 0.04, "learning_rate": 4.804e-05, "loss": 2.2874, "step": 31360 }, { "epoch": 0.04, "learning_rate": 4.8039375000000005e-05, "loss": 1.9167, "step": 31370 }, { "epoch": 0.04, "learning_rate": 4.803875e-05, "loss": 1.9634, "step": 31380 }, { "epoch": 0.04, "learning_rate": 4.8038125e-05, "loss": 1.8654, "step": 31390 }, { "epoch": 0.04, "learning_rate": 4.80375e-05, "loss": 1.7539, "step": 31400 }, { "epoch": 0.04, "learning_rate": 4.8036875000000006e-05, "loss": 1.9541, "step": 31410 }, { "epoch": 0.04, "learning_rate": 4.803625e-05, "loss": 2.0576, "step": 31420 }, { "epoch": 0.04, "learning_rate": 4.8035625e-05, "loss": 2.0936, "step": 31430 }, { "epoch": 0.04, "learning_rate": 4.8035000000000003e-05, "loss": 1.8849, "step": 31440 }, { "epoch": 0.04, "learning_rate": 4.8034375e-05, "loss": 2.0824, "step": 31450 }, { "epoch": 0.04, "learning_rate": 4.803375e-05, "loss": 2.0158, "step": 31460 }, { "epoch": 0.04, "learning_rate": 4.8033125e-05, "loss": 1.878, "step": 31470 }, { "epoch": 0.04, "learning_rate": 4.8032500000000005e-05, "loss": 1.9859, "step": 31480 }, { "epoch": 0.04, "learning_rate": 4.8031875e-05, "loss": 2.0151, "step": 31490 }, { "epoch": 0.04, "learning_rate": 4.803125e-05, "loss": 1.8069, "step": 31500 }, { "epoch": 0.04, "learning_rate": 4.8030625e-05, "loss": 1.8979, "step": 31510 }, { "epoch": 0.04, "learning_rate": 4.8030000000000006e-05, "loss": 2.1596, "step": 31520 }, { "epoch": 0.04, "learning_rate": 4.8029375e-05, "loss": 1.9877, "step": 31530 }, { "epoch": 0.04, "learning_rate": 4.802875e-05, "loss": 2.0208, "step": 31540 }, { "epoch": 0.04, "learning_rate": 4.8028125000000004e-05, "loss": 2.2829, "step": 31550 }, { "epoch": 0.04, "learning_rate": 4.80275e-05, "loss": 1.95, "step": 31560 }, { "epoch": 0.04, "learning_rate": 4.8026875000000004e-05, "loss": 1.9794, "step": 31570 }, { "epoch": 0.04, "learning_rate": 4.802625e-05, "loss": 1.992, "step": 31580 }, { "epoch": 0.04, "learning_rate": 4.8025625000000005e-05, "loss": 2.1951, "step": 31590 }, { "epoch": 0.04, "learning_rate": 4.8025e-05, "loss": 1.992, "step": 31600 }, { "epoch": 0.04, "learning_rate": 4.8024375000000006e-05, "loss": 2.2083, "step": 31610 }, { "epoch": 0.04, "learning_rate": 4.802375e-05, "loss": 1.7376, "step": 31620 }, { "epoch": 0.04, "learning_rate": 4.8023125e-05, "loss": 1.9413, "step": 31630 }, { "epoch": 0.04, "learning_rate": 4.80225e-05, "loss": 2.0436, "step": 31640 }, { "epoch": 0.04, "learning_rate": 4.802187500000001e-05, "loss": 2.0351, "step": 31650 }, { "epoch": 0.04, "learning_rate": 4.8021250000000004e-05, "loss": 1.9803, "step": 31660 }, { "epoch": 0.04, "learning_rate": 4.8020625e-05, "loss": 1.7056, "step": 31670 }, { "epoch": 0.04, "learning_rate": 4.8020000000000004e-05, "loss": 2.0073, "step": 31680 }, { "epoch": 0.04, "learning_rate": 4.8019375e-05, "loss": 1.9408, "step": 31690 }, { "epoch": 0.04, "learning_rate": 4.801875e-05, "loss": 1.9514, "step": 31700 }, { "epoch": 0.04, "learning_rate": 4.8018125e-05, "loss": 1.8757, "step": 31710 }, { "epoch": 0.04, "learning_rate": 4.8017500000000006e-05, "loss": 2.0443, "step": 31720 }, { "epoch": 0.04, "learning_rate": 4.8016875e-05, "loss": 2.156, "step": 31730 }, { "epoch": 0.04, "learning_rate": 4.801625e-05, "loss": 2.0972, "step": 31740 }, { "epoch": 0.04, "learning_rate": 4.8015625e-05, "loss": 2.1462, "step": 31750 }, { "epoch": 0.04, "learning_rate": 4.801500000000001e-05, "loss": 2.2103, "step": 31760 }, { "epoch": 0.04, "learning_rate": 4.8014375e-05, "loss": 1.9277, "step": 31770 }, { "epoch": 0.04, "learning_rate": 4.801375e-05, "loss": 1.8715, "step": 31780 }, { "epoch": 0.04, "learning_rate": 4.8013125000000005e-05, "loss": 2.094, "step": 31790 }, { "epoch": 0.04, "learning_rate": 4.80125e-05, "loss": 2.0775, "step": 31800 }, { "epoch": 0.04, "learning_rate": 4.8011875e-05, "loss": 1.9859, "step": 31810 }, { "epoch": 0.04, "learning_rate": 4.801125e-05, "loss": 1.8885, "step": 31820 }, { "epoch": 0.04, "learning_rate": 4.8010625000000006e-05, "loss": 2.0184, "step": 31830 }, { "epoch": 0.04, "learning_rate": 4.801e-05, "loss": 2.0263, "step": 31840 }, { "epoch": 0.04, "learning_rate": 4.8009375e-05, "loss": 2.0127, "step": 31850 }, { "epoch": 0.04, "learning_rate": 4.8008750000000003e-05, "loss": 2.1736, "step": 31860 }, { "epoch": 0.04, "learning_rate": 4.800812500000001e-05, "loss": 2.0841, "step": 31870 }, { "epoch": 0.04, "learning_rate": 4.80075e-05, "loss": 1.9769, "step": 31880 }, { "epoch": 0.04, "learning_rate": 4.8006875e-05, "loss": 2.0223, "step": 31890 }, { "epoch": 0.04, "learning_rate": 4.8006250000000005e-05, "loss": 1.9745, "step": 31900 }, { "epoch": 0.04, "learning_rate": 4.8005625e-05, "loss": 1.9311, "step": 31910 }, { "epoch": 0.04, "learning_rate": 4.8005e-05, "loss": 2.0813, "step": 31920 }, { "epoch": 0.04, "learning_rate": 4.8004375e-05, "loss": 1.9212, "step": 31930 }, { "epoch": 0.04, "learning_rate": 4.8003750000000006e-05, "loss": 2.1274, "step": 31940 }, { "epoch": 0.04, "learning_rate": 4.8003125e-05, "loss": 1.6295, "step": 31950 }, { "epoch": 0.04, "learning_rate": 4.80025e-05, "loss": 1.6795, "step": 31960 }, { "epoch": 0.04, "learning_rate": 4.8001875000000004e-05, "loss": 2.3385, "step": 31970 }, { "epoch": 0.04, "learning_rate": 4.800125e-05, "loss": 2.062, "step": 31980 }, { "epoch": 0.04, "learning_rate": 4.8000625000000004e-05, "loss": 2.0338, "step": 31990 }, { "epoch": 0.04, "learning_rate": 4.8e-05, "loss": 2.1641, "step": 32000 }, { "epoch": 0.04, "eval_loss": 1.886854887008667, "eval_runtime": 1290.2784, "eval_samples_per_second": 44.979, "eval_steps_per_second": 5.623, "step": 32000 }, { "epoch": 0.04, "learning_rate": 4.7999375000000005e-05, "loss": 1.9275, "step": 32010 }, { "epoch": 0.04, "learning_rate": 4.799875e-05, "loss": 1.8181, "step": 32020 }, { "epoch": 0.04, "learning_rate": 4.7998125000000006e-05, "loss": 2.0185, "step": 32030 }, { "epoch": 0.04, "learning_rate": 4.79975e-05, "loss": 2.0486, "step": 32040 }, { "epoch": 0.04, "learning_rate": 4.7996875e-05, "loss": 1.8978, "step": 32050 }, { "epoch": 0.04, "learning_rate": 4.799625e-05, "loss": 2.1936, "step": 32060 }, { "epoch": 0.04, "learning_rate": 4.7995625e-05, "loss": 1.933, "step": 32070 }, { "epoch": 0.04, "learning_rate": 4.7995000000000004e-05, "loss": 2.0015, "step": 32080 }, { "epoch": 0.04, "learning_rate": 4.7994375e-05, "loss": 2.0511, "step": 32090 }, { "epoch": 0.04, "learning_rate": 4.7993750000000004e-05, "loss": 1.9728, "step": 32100 }, { "epoch": 0.04, "learning_rate": 4.7993125e-05, "loss": 1.8317, "step": 32110 }, { "epoch": 0.04, "learning_rate": 4.79925e-05, "loss": 2.243, "step": 32120 }, { "epoch": 0.04, "learning_rate": 4.7991875e-05, "loss": 2.0621, "step": 32130 }, { "epoch": 0.04, "learning_rate": 4.7991250000000006e-05, "loss": 1.8305, "step": 32140 }, { "epoch": 0.04, "learning_rate": 4.7990625e-05, "loss": 1.9827, "step": 32150 }, { "epoch": 0.04, "learning_rate": 4.799e-05, "loss": 1.8698, "step": 32160 }, { "epoch": 0.04, "learning_rate": 4.7989375e-05, "loss": 2.1303, "step": 32170 }, { "epoch": 0.04, "learning_rate": 4.798875000000001e-05, "loss": 1.9222, "step": 32180 }, { "epoch": 0.04, "learning_rate": 4.7988125e-05, "loss": 2.1299, "step": 32190 }, { "epoch": 0.04, "learning_rate": 4.79875e-05, "loss": 2.0037, "step": 32200 }, { "epoch": 0.04, "learning_rate": 4.7986875000000005e-05, "loss": 1.8975, "step": 32210 }, { "epoch": 0.04, "learning_rate": 4.798625e-05, "loss": 2.0975, "step": 32220 }, { "epoch": 0.04, "learning_rate": 4.7985625e-05, "loss": 2.0376, "step": 32230 }, { "epoch": 0.04, "learning_rate": 4.7985e-05, "loss": 2.1313, "step": 32240 }, { "epoch": 0.04, "learning_rate": 4.7984375000000006e-05, "loss": 2.0928, "step": 32250 }, { "epoch": 0.04, "learning_rate": 4.798375e-05, "loss": 2.0726, "step": 32260 }, { "epoch": 0.04, "learning_rate": 4.7983125e-05, "loss": 2.0643, "step": 32270 }, { "epoch": 0.04, "learning_rate": 4.7982500000000004e-05, "loss": 2.0688, "step": 32280 }, { "epoch": 0.04, "learning_rate": 4.798187500000001e-05, "loss": 1.9567, "step": 32290 }, { "epoch": 0.04, "learning_rate": 4.798125e-05, "loss": 1.9867, "step": 32300 }, { "epoch": 0.04, "learning_rate": 4.7980625e-05, "loss": 1.8441, "step": 32310 }, { "epoch": 0.05, "learning_rate": 4.7980000000000005e-05, "loss": 1.9684, "step": 32320 }, { "epoch": 0.05, "learning_rate": 4.7979375e-05, "loss": 2.1002, "step": 32330 }, { "epoch": 0.05, "learning_rate": 4.797875e-05, "loss": 1.951, "step": 32340 }, { "epoch": 0.05, "learning_rate": 4.7978125e-05, "loss": 1.7813, "step": 32350 }, { "epoch": 0.05, "learning_rate": 4.7977500000000006e-05, "loss": 1.8741, "step": 32360 }, { "epoch": 0.05, "learning_rate": 4.7976875e-05, "loss": 1.7686, "step": 32370 }, { "epoch": 0.05, "learning_rate": 4.797625e-05, "loss": 2.0978, "step": 32380 }, { "epoch": 0.05, "learning_rate": 4.7975625000000004e-05, "loss": 1.9024, "step": 32390 }, { "epoch": 0.05, "learning_rate": 4.7975e-05, "loss": 2.0667, "step": 32400 }, { "epoch": 0.05, "learning_rate": 4.7974375000000004e-05, "loss": 1.8983, "step": 32410 }, { "epoch": 0.05, "learning_rate": 4.797375e-05, "loss": 2.0845, "step": 32420 }, { "epoch": 0.05, "learning_rate": 4.7973125000000005e-05, "loss": 1.8862, "step": 32430 }, { "epoch": 0.05, "learning_rate": 4.79725e-05, "loss": 2.0764, "step": 32440 }, { "epoch": 0.05, "learning_rate": 4.7971875000000006e-05, "loss": 1.9386, "step": 32450 }, { "epoch": 0.05, "learning_rate": 4.797125e-05, "loss": 2.0498, "step": 32460 }, { "epoch": 0.05, "learning_rate": 4.7970625e-05, "loss": 1.6615, "step": 32470 }, { "epoch": 0.05, "learning_rate": 4.797e-05, "loss": 2.1689, "step": 32480 }, { "epoch": 0.05, "learning_rate": 4.7969375e-05, "loss": 2.1588, "step": 32490 }, { "epoch": 0.05, "learning_rate": 4.7968750000000004e-05, "loss": 1.8158, "step": 32500 }, { "epoch": 0.05, "learning_rate": 4.7968125e-05, "loss": 1.9571, "step": 32510 }, { "epoch": 0.05, "learning_rate": 4.7967500000000004e-05, "loss": 2.0218, "step": 32520 }, { "epoch": 0.05, "learning_rate": 4.7966875e-05, "loss": 1.9709, "step": 32530 }, { "epoch": 0.05, "learning_rate": 4.796625e-05, "loss": 1.7704, "step": 32540 }, { "epoch": 0.05, "learning_rate": 4.7965625e-05, "loss": 2.1043, "step": 32550 }, { "epoch": 0.05, "learning_rate": 4.7965000000000006e-05, "loss": 2.2567, "step": 32560 }, { "epoch": 0.05, "learning_rate": 4.7964375e-05, "loss": 1.756, "step": 32570 }, { "epoch": 0.05, "learning_rate": 4.796375e-05, "loss": 1.9635, "step": 32580 }, { "epoch": 0.05, "learning_rate": 4.7963125e-05, "loss": 1.755, "step": 32590 }, { "epoch": 0.05, "learning_rate": 4.796250000000001e-05, "loss": 1.8477, "step": 32600 }, { "epoch": 0.05, "learning_rate": 4.7961875e-05, "loss": 1.7845, "step": 32610 }, { "epoch": 0.05, "learning_rate": 4.796125e-05, "loss": 1.7814, "step": 32620 }, { "epoch": 0.05, "learning_rate": 4.7960625000000005e-05, "loss": 1.9914, "step": 32630 }, { "epoch": 0.05, "learning_rate": 4.796e-05, "loss": 2.1774, "step": 32640 }, { "epoch": 0.05, "learning_rate": 4.7959375e-05, "loss": 1.8057, "step": 32650 }, { "epoch": 0.05, "learning_rate": 4.795875e-05, "loss": 1.9164, "step": 32660 }, { "epoch": 0.05, "learning_rate": 4.7958125000000006e-05, "loss": 2.0505, "step": 32670 }, { "epoch": 0.05, "learning_rate": 4.79575e-05, "loss": 2.2517, "step": 32680 }, { "epoch": 0.05, "learning_rate": 4.7956875e-05, "loss": 2.0312, "step": 32690 }, { "epoch": 0.05, "learning_rate": 4.7956250000000004e-05, "loss": 1.9371, "step": 32700 }, { "epoch": 0.05, "learning_rate": 4.795562500000001e-05, "loss": 1.9309, "step": 32710 }, { "epoch": 0.05, "learning_rate": 4.7955e-05, "loss": 2.0945, "step": 32720 }, { "epoch": 0.05, "learning_rate": 4.7954375e-05, "loss": 2.119, "step": 32730 }, { "epoch": 0.05, "learning_rate": 4.7953750000000005e-05, "loss": 1.9744, "step": 32740 }, { "epoch": 0.05, "learning_rate": 4.7953125e-05, "loss": 1.8734, "step": 32750 }, { "epoch": 0.05, "learning_rate": 4.79525e-05, "loss": 2.0804, "step": 32760 }, { "epoch": 0.05, "learning_rate": 4.7951875e-05, "loss": 2.1273, "step": 32770 }, { "epoch": 0.05, "learning_rate": 4.7951250000000006e-05, "loss": 1.9973, "step": 32780 }, { "epoch": 0.05, "learning_rate": 4.7950625e-05, "loss": 2.0019, "step": 32790 }, { "epoch": 0.05, "learning_rate": 4.795e-05, "loss": 1.8959, "step": 32800 }, { "epoch": 0.05, "learning_rate": 4.7949375000000004e-05, "loss": 1.8536, "step": 32810 }, { "epoch": 0.05, "learning_rate": 4.794875e-05, "loss": 2.057, "step": 32820 }, { "epoch": 0.05, "learning_rate": 4.7948125000000004e-05, "loss": 1.849, "step": 32830 }, { "epoch": 0.05, "learning_rate": 4.79475e-05, "loss": 2.0732, "step": 32840 }, { "epoch": 0.05, "learning_rate": 4.7946875000000005e-05, "loss": 1.7534, "step": 32850 }, { "epoch": 0.05, "learning_rate": 4.794625e-05, "loss": 1.8869, "step": 32860 }, { "epoch": 0.05, "learning_rate": 4.7945625e-05, "loss": 2.0258, "step": 32870 }, { "epoch": 0.05, "learning_rate": 4.7945e-05, "loss": 2.0176, "step": 32880 }, { "epoch": 0.05, "learning_rate": 4.7944375e-05, "loss": 1.8588, "step": 32890 }, { "epoch": 0.05, "learning_rate": 4.794375e-05, "loss": 2.0175, "step": 32900 }, { "epoch": 0.05, "learning_rate": 4.7943125e-05, "loss": 2.263, "step": 32910 }, { "epoch": 0.05, "learning_rate": 4.7942500000000004e-05, "loss": 1.8814, "step": 32920 }, { "epoch": 0.05, "learning_rate": 4.7941875e-05, "loss": 1.9588, "step": 32930 }, { "epoch": 0.05, "learning_rate": 4.7941250000000005e-05, "loss": 2.0548, "step": 32940 }, { "epoch": 0.05, "learning_rate": 4.7940625e-05, "loss": 1.5689, "step": 32950 }, { "epoch": 0.05, "learning_rate": 4.794e-05, "loss": 1.7461, "step": 32960 }, { "epoch": 0.05, "learning_rate": 4.7939375e-05, "loss": 1.8076, "step": 32970 }, { "epoch": 0.05, "learning_rate": 4.7938750000000006e-05, "loss": 1.8384, "step": 32980 }, { "epoch": 0.05, "learning_rate": 4.7938125e-05, "loss": 2.1165, "step": 32990 }, { "epoch": 0.05, "learning_rate": 4.79375e-05, "loss": 1.7169, "step": 33000 }, { "epoch": 0.05, "eval_loss": 1.876387357711792, "eval_runtime": 1291.6914, "eval_samples_per_second": 44.93, "eval_steps_per_second": 5.617, "step": 33000 }, { "epoch": 0.05, "learning_rate": 4.7936875000000003e-05, "loss": 1.953, "step": 33010 }, { "epoch": 0.05, "learning_rate": 4.793625000000001e-05, "loss": 2.1594, "step": 33020 }, { "epoch": 0.05, "learning_rate": 4.7935625000000004e-05, "loss": 1.9948, "step": 33030 }, { "epoch": 0.05, "learning_rate": 4.7935e-05, "loss": 1.9036, "step": 33040 }, { "epoch": 0.05, "learning_rate": 4.7934375000000005e-05, "loss": 1.8007, "step": 33050 }, { "epoch": 0.05, "learning_rate": 4.793375000000001e-05, "loss": 1.7974, "step": 33060 }, { "epoch": 0.05, "learning_rate": 4.7933125e-05, "loss": 1.7934, "step": 33070 }, { "epoch": 0.05, "learning_rate": 4.79325e-05, "loss": 2.0985, "step": 33080 }, { "epoch": 0.05, "learning_rate": 4.7931875000000006e-05, "loss": 2.0378, "step": 33090 }, { "epoch": 0.05, "learning_rate": 4.793125e-05, "loss": 2.0686, "step": 33100 }, { "epoch": 0.05, "learning_rate": 4.7930625e-05, "loss": 1.7591, "step": 33110 }, { "epoch": 0.05, "learning_rate": 4.7930000000000004e-05, "loss": 2.0447, "step": 33120 }, { "epoch": 0.05, "learning_rate": 4.792937500000001e-05, "loss": 1.8924, "step": 33130 }, { "epoch": 0.05, "learning_rate": 4.792875e-05, "loss": 1.9175, "step": 33140 }, { "epoch": 0.05, "learning_rate": 4.7928125e-05, "loss": 2.0445, "step": 33150 }, { "epoch": 0.05, "learning_rate": 4.7927500000000005e-05, "loss": 2.0152, "step": 33160 }, { "epoch": 0.05, "learning_rate": 4.7926875e-05, "loss": 1.9436, "step": 33170 }, { "epoch": 0.05, "learning_rate": 4.792625e-05, "loss": 2.2059, "step": 33180 }, { "epoch": 0.05, "learning_rate": 4.7925625e-05, "loss": 2.0952, "step": 33190 }, { "epoch": 0.05, "learning_rate": 4.7925000000000006e-05, "loss": 2.1516, "step": 33200 }, { "epoch": 0.05, "learning_rate": 4.7924375e-05, "loss": 2.0034, "step": 33210 }, { "epoch": 0.05, "learning_rate": 4.792375e-05, "loss": 1.918, "step": 33220 }, { "epoch": 0.05, "learning_rate": 4.7923125000000004e-05, "loss": 1.9558, "step": 33230 }, { "epoch": 0.05, "learning_rate": 4.79225e-05, "loss": 1.9711, "step": 33240 }, { "epoch": 0.05, "learning_rate": 4.7921875000000004e-05, "loss": 1.9741, "step": 33250 }, { "epoch": 0.05, "learning_rate": 4.792125e-05, "loss": 1.9691, "step": 33260 }, { "epoch": 0.05, "learning_rate": 4.7920625000000005e-05, "loss": 1.9127, "step": 33270 }, { "epoch": 0.05, "learning_rate": 4.792e-05, "loss": 1.9895, "step": 33280 }, { "epoch": 0.05, "learning_rate": 4.7919375e-05, "loss": 2.0704, "step": 33290 }, { "epoch": 0.05, "learning_rate": 4.791875e-05, "loss": 1.8655, "step": 33300 }, { "epoch": 0.05, "learning_rate": 4.7918125e-05, "loss": 1.9302, "step": 33310 }, { "epoch": 0.05, "learning_rate": 4.79175e-05, "loss": 2.0027, "step": 33320 }, { "epoch": 0.05, "learning_rate": 4.7916875e-05, "loss": 2.0899, "step": 33330 }, { "epoch": 0.05, "learning_rate": 4.7916250000000004e-05, "loss": 1.5174, "step": 33340 }, { "epoch": 0.05, "learning_rate": 4.7915625e-05, "loss": 1.8076, "step": 33350 }, { "epoch": 0.05, "learning_rate": 4.7915000000000005e-05, "loss": 1.9232, "step": 33360 }, { "epoch": 0.05, "learning_rate": 4.7914375e-05, "loss": 1.4847, "step": 33370 }, { "epoch": 0.05, "learning_rate": 4.791375e-05, "loss": 2.0086, "step": 33380 }, { "epoch": 0.05, "learning_rate": 4.7913125e-05, "loss": 1.8723, "step": 33390 }, { "epoch": 0.05, "learning_rate": 4.7912500000000006e-05, "loss": 1.8752, "step": 33400 }, { "epoch": 0.05, "learning_rate": 4.7911875e-05, "loss": 2.3164, "step": 33410 }, { "epoch": 0.05, "learning_rate": 4.791125e-05, "loss": 1.7839, "step": 33420 }, { "epoch": 0.05, "learning_rate": 4.7910625000000003e-05, "loss": 1.986, "step": 33430 }, { "epoch": 0.05, "learning_rate": 4.791000000000001e-05, "loss": 1.7754, "step": 33440 }, { "epoch": 0.05, "learning_rate": 4.7909375000000004e-05, "loss": 1.8179, "step": 33450 }, { "epoch": 0.05, "learning_rate": 4.790875e-05, "loss": 1.969, "step": 33460 }, { "epoch": 0.05, "learning_rate": 4.7908125000000005e-05, "loss": 2.1683, "step": 33470 }, { "epoch": 0.05, "learning_rate": 4.79075e-05, "loss": 1.7473, "step": 33480 }, { "epoch": 0.05, "learning_rate": 4.7906875e-05, "loss": 1.8963, "step": 33490 }, { "epoch": 0.05, "learning_rate": 4.790625e-05, "loss": 2.0034, "step": 33500 }, { "epoch": 0.05, "learning_rate": 4.7905625000000006e-05, "loss": 1.8832, "step": 33510 }, { "epoch": 0.05, "learning_rate": 4.7905e-05, "loss": 2.0489, "step": 33520 }, { "epoch": 0.05, "learning_rate": 4.7904375e-05, "loss": 2.299, "step": 33530 }, { "epoch": 0.05, "learning_rate": 4.7903750000000004e-05, "loss": 1.7241, "step": 33540 }, { "epoch": 0.05, "learning_rate": 4.790312500000001e-05, "loss": 2.0578, "step": 33550 }, { "epoch": 0.05, "learning_rate": 4.79025e-05, "loss": 1.8094, "step": 33560 }, { "epoch": 0.05, "learning_rate": 4.7901875e-05, "loss": 2.0455, "step": 33570 }, { "epoch": 0.05, "learning_rate": 4.7901250000000005e-05, "loss": 1.739, "step": 33580 }, { "epoch": 0.05, "learning_rate": 4.7900625e-05, "loss": 1.7518, "step": 33590 }, { "epoch": 0.05, "learning_rate": 4.79e-05, "loss": 2.1102, "step": 33600 }, { "epoch": 0.05, "learning_rate": 4.7899375e-05, "loss": 2.1141, "step": 33610 }, { "epoch": 0.05, "learning_rate": 4.7898750000000006e-05, "loss": 1.6743, "step": 33620 }, { "epoch": 0.05, "learning_rate": 4.7898125e-05, "loss": 1.8269, "step": 33630 }, { "epoch": 0.05, "learning_rate": 4.78975e-05, "loss": 1.7748, "step": 33640 }, { "epoch": 0.05, "learning_rate": 4.7896875000000004e-05, "loss": 2.0108, "step": 33650 }, { "epoch": 0.05, "learning_rate": 4.789625e-05, "loss": 1.8149, "step": 33660 }, { "epoch": 0.05, "learning_rate": 4.7895625e-05, "loss": 1.9661, "step": 33670 }, { "epoch": 0.05, "learning_rate": 4.7895e-05, "loss": 2.0664, "step": 33680 }, { "epoch": 0.05, "learning_rate": 4.7894375000000005e-05, "loss": 1.7713, "step": 33690 }, { "epoch": 0.05, "learning_rate": 4.789375e-05, "loss": 2.017, "step": 33700 }, { "epoch": 0.05, "learning_rate": 4.7893125e-05, "loss": 2.0228, "step": 33710 }, { "epoch": 0.05, "learning_rate": 4.78925e-05, "loss": 1.9631, "step": 33720 }, { "epoch": 0.05, "learning_rate": 4.7891875e-05, "loss": 1.9419, "step": 33730 }, { "epoch": 0.05, "learning_rate": 4.789125e-05, "loss": 2.0859, "step": 33740 }, { "epoch": 0.05, "learning_rate": 4.7890625e-05, "loss": 1.9148, "step": 33750 }, { "epoch": 0.05, "learning_rate": 4.7890000000000004e-05, "loss": 1.8627, "step": 33760 }, { "epoch": 0.05, "learning_rate": 4.7889375e-05, "loss": 1.8502, "step": 33770 }, { "epoch": 0.05, "learning_rate": 4.7888750000000005e-05, "loss": 1.9011, "step": 33780 }, { "epoch": 0.05, "learning_rate": 4.7888125e-05, "loss": 1.9422, "step": 33790 }, { "epoch": 0.05, "learning_rate": 4.78875e-05, "loss": 1.8229, "step": 33800 }, { "epoch": 0.05, "learning_rate": 4.7886875e-05, "loss": 1.92, "step": 33810 }, { "epoch": 0.05, "learning_rate": 4.7886250000000006e-05, "loss": 1.9793, "step": 33820 }, { "epoch": 0.05, "learning_rate": 4.7885625e-05, "loss": 1.7922, "step": 33830 }, { "epoch": 0.05, "learning_rate": 4.7885e-05, "loss": 1.8837, "step": 33840 }, { "epoch": 0.05, "learning_rate": 4.7884375000000003e-05, "loss": 2.2836, "step": 33850 }, { "epoch": 0.05, "learning_rate": 4.788375000000001e-05, "loss": 1.7749, "step": 33860 }, { "epoch": 0.05, "learning_rate": 4.7883125000000004e-05, "loss": 1.8051, "step": 33870 }, { "epoch": 0.05, "learning_rate": 4.78825e-05, "loss": 2.0434, "step": 33880 }, { "epoch": 0.05, "learning_rate": 4.7881875000000005e-05, "loss": 1.9635, "step": 33890 }, { "epoch": 0.05, "learning_rate": 4.788125e-05, "loss": 1.8907, "step": 33900 }, { "epoch": 0.05, "learning_rate": 4.7880625e-05, "loss": 1.7319, "step": 33910 }, { "epoch": 0.05, "learning_rate": 4.788e-05, "loss": 1.815, "step": 33920 }, { "epoch": 0.05, "learning_rate": 4.7879375000000006e-05, "loss": 2.0388, "step": 33930 }, { "epoch": 0.05, "learning_rate": 4.787875e-05, "loss": 1.8116, "step": 33940 }, { "epoch": 0.05, "learning_rate": 4.7878125e-05, "loss": 1.7884, "step": 33950 }, { "epoch": 0.05, "learning_rate": 4.7877500000000004e-05, "loss": 1.9742, "step": 33960 }, { "epoch": 0.05, "learning_rate": 4.787687500000001e-05, "loss": 1.621, "step": 33970 }, { "epoch": 0.05, "learning_rate": 4.787625e-05, "loss": 1.921, "step": 33980 }, { "epoch": 0.05, "learning_rate": 4.7875625e-05, "loss": 1.7999, "step": 33990 }, { "epoch": 0.05, "learning_rate": 4.7875000000000005e-05, "loss": 2.004, "step": 34000 }, { "epoch": 0.05, "eval_loss": 1.8621490001678467, "eval_runtime": 1290.2482, "eval_samples_per_second": 44.98, "eval_steps_per_second": 5.623, "step": 34000 }, { "epoch": 0.05, "learning_rate": 4.7874375e-05, "loss": 1.9137, "step": 34010 }, { "epoch": 0.05, "learning_rate": 4.787375e-05, "loss": 1.7157, "step": 34020 }, { "epoch": 0.05, "learning_rate": 4.7873125e-05, "loss": 1.9056, "step": 34030 }, { "epoch": 0.05, "learning_rate": 4.7872500000000006e-05, "loss": 1.7675, "step": 34040 }, { "epoch": 0.05, "learning_rate": 4.7871875e-05, "loss": 1.8798, "step": 34050 }, { "epoch": 0.05, "learning_rate": 4.787125e-05, "loss": 2.1137, "step": 34060 }, { "epoch": 0.05, "learning_rate": 4.7870625000000004e-05, "loss": 1.9747, "step": 34070 }, { "epoch": 0.05, "learning_rate": 4.787e-05, "loss": 1.7618, "step": 34080 }, { "epoch": 0.05, "learning_rate": 4.7869375e-05, "loss": 2.247, "step": 34090 }, { "epoch": 0.05, "learning_rate": 4.786875e-05, "loss": 1.7526, "step": 34100 }, { "epoch": 0.05, "learning_rate": 4.7868125000000005e-05, "loss": 1.8642, "step": 34110 }, { "epoch": 0.05, "learning_rate": 4.78675e-05, "loss": 2.2651, "step": 34120 }, { "epoch": 0.05, "learning_rate": 4.7866875e-05, "loss": 1.8924, "step": 34130 }, { "epoch": 0.05, "learning_rate": 4.786625e-05, "loss": 1.8775, "step": 34140 }, { "epoch": 0.05, "learning_rate": 4.7865625e-05, "loss": 1.8935, "step": 34150 }, { "epoch": 0.05, "learning_rate": 4.7865e-05, "loss": 1.931, "step": 34160 }, { "epoch": 0.05, "learning_rate": 4.7864375e-05, "loss": 1.9221, "step": 34170 }, { "epoch": 0.05, "learning_rate": 4.7863750000000004e-05, "loss": 1.8526, "step": 34180 }, { "epoch": 0.05, "learning_rate": 4.7863125e-05, "loss": 1.9751, "step": 34190 }, { "epoch": 0.05, "learning_rate": 4.7862500000000005e-05, "loss": 1.9056, "step": 34200 }, { "epoch": 0.05, "learning_rate": 4.7861875e-05, "loss": 1.6843, "step": 34210 }, { "epoch": 0.05, "learning_rate": 4.7861250000000005e-05, "loss": 1.8537, "step": 34220 }, { "epoch": 0.05, "learning_rate": 4.7860625e-05, "loss": 1.817, "step": 34230 }, { "epoch": 0.05, "learning_rate": 4.7860000000000006e-05, "loss": 1.69, "step": 34240 }, { "epoch": 0.05, "learning_rate": 4.7859375e-05, "loss": 1.8352, "step": 34250 }, { "epoch": 0.05, "learning_rate": 4.785875e-05, "loss": 1.7547, "step": 34260 }, { "epoch": 0.05, "learning_rate": 4.7858125000000004e-05, "loss": 2.1589, "step": 34270 }, { "epoch": 0.05, "learning_rate": 4.78575e-05, "loss": 1.9646, "step": 34280 }, { "epoch": 0.05, "learning_rate": 4.7856875000000004e-05, "loss": 1.8536, "step": 34290 }, { "epoch": 0.05, "learning_rate": 4.785625e-05, "loss": 2.021, "step": 34300 }, { "epoch": 0.05, "learning_rate": 4.7855625000000005e-05, "loss": 2.087, "step": 34310 }, { "epoch": 0.05, "learning_rate": 4.7855e-05, "loss": 1.7616, "step": 34320 }, { "epoch": 0.05, "learning_rate": 4.7854375e-05, "loss": 1.777, "step": 34330 }, { "epoch": 0.05, "learning_rate": 4.785375e-05, "loss": 2.0006, "step": 34340 }, { "epoch": 0.05, "learning_rate": 4.7853125000000006e-05, "loss": 2.3521, "step": 34350 }, { "epoch": 0.05, "learning_rate": 4.78525e-05, "loss": 1.7724, "step": 34360 }, { "epoch": 0.05, "learning_rate": 4.7851875e-05, "loss": 1.9834, "step": 34370 }, { "epoch": 0.05, "learning_rate": 4.7851250000000004e-05, "loss": 1.8372, "step": 34380 }, { "epoch": 0.05, "learning_rate": 4.785062500000001e-05, "loss": 2.1185, "step": 34390 }, { "epoch": 0.05, "learning_rate": 4.785e-05, "loss": 2.1841, "step": 34400 }, { "epoch": 0.05, "learning_rate": 4.7849375e-05, "loss": 1.8182, "step": 34410 }, { "epoch": 0.05, "learning_rate": 4.7848750000000005e-05, "loss": 1.9293, "step": 34420 }, { "epoch": 0.05, "learning_rate": 4.7848125e-05, "loss": 2.0547, "step": 34430 }, { "epoch": 0.05, "learning_rate": 4.78475e-05, "loss": 2.1279, "step": 34440 }, { "epoch": 0.05, "learning_rate": 4.7846875e-05, "loss": 1.8028, "step": 34450 }, { "epoch": 0.05, "learning_rate": 4.7846250000000006e-05, "loss": 2.0697, "step": 34460 }, { "epoch": 0.05, "learning_rate": 4.7845625e-05, "loss": 1.7271, "step": 34470 }, { "epoch": 0.05, "learning_rate": 4.7845e-05, "loss": 1.708, "step": 34480 }, { "epoch": 0.05, "learning_rate": 4.7844375000000004e-05, "loss": 1.9845, "step": 34490 }, { "epoch": 0.05, "learning_rate": 4.784375e-05, "loss": 2.0077, "step": 34500 }, { "epoch": 0.05, "learning_rate": 4.7843125e-05, "loss": 2.1005, "step": 34510 }, { "epoch": 0.05, "learning_rate": 4.78425e-05, "loss": 1.8351, "step": 34520 }, { "epoch": 0.05, "learning_rate": 4.7841875000000005e-05, "loss": 1.9116, "step": 34530 }, { "epoch": 0.05, "learning_rate": 4.784125e-05, "loss": 1.9304, "step": 34540 }, { "epoch": 0.05, "learning_rate": 4.7840625e-05, "loss": 1.8715, "step": 34550 }, { "epoch": 0.05, "learning_rate": 4.784e-05, "loss": 1.795, "step": 34560 }, { "epoch": 0.05, "learning_rate": 4.7839375e-05, "loss": 2.0551, "step": 34570 }, { "epoch": 0.05, "learning_rate": 4.7838750000000003e-05, "loss": 1.7124, "step": 34580 }, { "epoch": 0.05, "learning_rate": 4.7838125e-05, "loss": 1.9183, "step": 34590 }, { "epoch": 0.05, "learning_rate": 4.7837500000000004e-05, "loss": 1.899, "step": 34600 }, { "epoch": 0.05, "learning_rate": 4.7836875e-05, "loss": 2.1717, "step": 34610 }, { "epoch": 0.05, "learning_rate": 4.7836250000000005e-05, "loss": 2.0116, "step": 34620 }, { "epoch": 0.05, "learning_rate": 4.7835625e-05, "loss": 1.872, "step": 34630 }, { "epoch": 0.05, "learning_rate": 4.7835000000000005e-05, "loss": 1.6108, "step": 34640 }, { "epoch": 0.05, "learning_rate": 4.7834375e-05, "loss": 1.9554, "step": 34650 }, { "epoch": 0.05, "learning_rate": 4.7833750000000006e-05, "loss": 1.6949, "step": 34660 }, { "epoch": 0.05, "learning_rate": 4.7833125e-05, "loss": 1.9988, "step": 34670 }, { "epoch": 0.05, "learning_rate": 4.78325e-05, "loss": 1.8675, "step": 34680 }, { "epoch": 0.05, "learning_rate": 4.7831875000000004e-05, "loss": 1.6452, "step": 34690 }, { "epoch": 0.05, "learning_rate": 4.783125e-05, "loss": 1.816, "step": 34700 }, { "epoch": 0.05, "learning_rate": 4.7830625000000004e-05, "loss": 2.0339, "step": 34710 }, { "epoch": 0.05, "learning_rate": 4.783e-05, "loss": 1.9986, "step": 34720 }, { "epoch": 0.05, "learning_rate": 4.7829375000000005e-05, "loss": 1.9872, "step": 34730 }, { "epoch": 0.05, "learning_rate": 4.782875e-05, "loss": 1.777, "step": 34740 }, { "epoch": 0.05, "learning_rate": 4.7828125e-05, "loss": 1.8596, "step": 34750 }, { "epoch": 0.05, "learning_rate": 4.78275e-05, "loss": 1.9181, "step": 34760 }, { "epoch": 0.05, "learning_rate": 4.7826875000000006e-05, "loss": 1.8649, "step": 34770 }, { "epoch": 0.05, "learning_rate": 4.782625e-05, "loss": 1.7269, "step": 34780 }, { "epoch": 0.05, "learning_rate": 4.7825625e-05, "loss": 2.148, "step": 34790 }, { "epoch": 0.05, "learning_rate": 4.7825000000000004e-05, "loss": 1.5936, "step": 34800 }, { "epoch": 0.05, "learning_rate": 4.782437500000001e-05, "loss": 1.9922, "step": 34810 }, { "epoch": 0.05, "learning_rate": 4.782375e-05, "loss": 1.9094, "step": 34820 }, { "epoch": 0.05, "learning_rate": 4.7823125e-05, "loss": 1.983, "step": 34830 }, { "epoch": 0.05, "learning_rate": 4.7822500000000005e-05, "loss": 2.1165, "step": 34840 }, { "epoch": 0.05, "learning_rate": 4.7821875e-05, "loss": 1.9838, "step": 34850 }, { "epoch": 0.05, "learning_rate": 4.782125e-05, "loss": 1.6863, "step": 34860 }, { "epoch": 0.05, "learning_rate": 4.7820625e-05, "loss": 1.9946, "step": 34870 }, { "epoch": 0.05, "learning_rate": 4.7820000000000006e-05, "loss": 2.0562, "step": 34880 }, { "epoch": 0.05, "learning_rate": 4.7819374999999996e-05, "loss": 1.8985, "step": 34890 }, { "epoch": 0.05, "learning_rate": 4.781875e-05, "loss": 1.7226, "step": 34900 }, { "epoch": 0.05, "learning_rate": 4.7818125000000004e-05, "loss": 1.7945, "step": 34910 }, { "epoch": 0.05, "learning_rate": 4.78175e-05, "loss": 1.7987, "step": 34920 }, { "epoch": 0.05, "learning_rate": 4.7816875e-05, "loss": 1.8398, "step": 34930 }, { "epoch": 0.05, "learning_rate": 4.781625e-05, "loss": 1.9161, "step": 34940 }, { "epoch": 0.05, "learning_rate": 4.7815625000000005e-05, "loss": 1.8482, "step": 34950 }, { "epoch": 0.05, "learning_rate": 4.7815e-05, "loss": 1.8984, "step": 34960 }, { "epoch": 0.05, "learning_rate": 4.7814375e-05, "loss": 1.8585, "step": 34970 }, { "epoch": 0.05, "learning_rate": 4.781375e-05, "loss": 1.9706, "step": 34980 }, { "epoch": 0.05, "learning_rate": 4.7813125e-05, "loss": 2.0558, "step": 34990 }, { "epoch": 0.05, "learning_rate": 4.7812500000000003e-05, "loss": 1.9291, "step": 35000 }, { "epoch": 0.05, "eval_loss": 1.8351171016693115, "eval_runtime": 1290.2819, "eval_samples_per_second": 44.979, "eval_steps_per_second": 5.623, "step": 35000 }, { "epoch": 0.05, "learning_rate": 4.7811875e-05, "loss": 1.6658, "step": 35010 }, { "epoch": 0.05, "learning_rate": 4.7811250000000004e-05, "loss": 1.9562, "step": 35020 }, { "epoch": 0.05, "learning_rate": 4.7810625e-05, "loss": 1.9485, "step": 35030 }, { "epoch": 0.05, "learning_rate": 4.7810000000000005e-05, "loss": 1.6632, "step": 35040 }, { "epoch": 0.05, "learning_rate": 4.7809375e-05, "loss": 2.0312, "step": 35050 }, { "epoch": 0.05, "learning_rate": 4.7808750000000005e-05, "loss": 2.2212, "step": 35060 }, { "epoch": 0.05, "learning_rate": 4.7808125e-05, "loss": 1.7524, "step": 35070 }, { "epoch": 0.05, "learning_rate": 4.7807500000000006e-05, "loss": 1.8983, "step": 35080 }, { "epoch": 0.05, "learning_rate": 4.7806875e-05, "loss": 2.2251, "step": 35090 }, { "epoch": 0.05, "learning_rate": 4.780625e-05, "loss": 1.8051, "step": 35100 }, { "epoch": 0.05, "learning_rate": 4.7805625000000004e-05, "loss": 2.1604, "step": 35110 }, { "epoch": 0.05, "learning_rate": 4.7805e-05, "loss": 1.7425, "step": 35120 }, { "epoch": 0.05, "learning_rate": 4.7804375000000004e-05, "loss": 1.8786, "step": 35130 }, { "epoch": 0.05, "learning_rate": 4.780375e-05, "loss": 1.7791, "step": 35140 }, { "epoch": 0.05, "learning_rate": 4.7803125000000005e-05, "loss": 1.7856, "step": 35150 }, { "epoch": 0.05, "learning_rate": 4.78025e-05, "loss": 1.9229, "step": 35160 }, { "epoch": 0.05, "learning_rate": 4.7801875e-05, "loss": 1.789, "step": 35170 }, { "epoch": 0.05, "learning_rate": 4.780125e-05, "loss": 2.1723, "step": 35180 }, { "epoch": 0.05, "learning_rate": 4.7800625000000006e-05, "loss": 1.7903, "step": 35190 }, { "epoch": 0.05, "learning_rate": 4.78e-05, "loss": 1.8416, "step": 35200 }, { "epoch": 0.05, "learning_rate": 4.7799375e-05, "loss": 2.0472, "step": 35210 }, { "epoch": 0.05, "learning_rate": 4.7798750000000004e-05, "loss": 1.9557, "step": 35220 }, { "epoch": 0.05, "learning_rate": 4.779812500000001e-05, "loss": 1.7217, "step": 35230 }, { "epoch": 0.05, "learning_rate": 4.77975e-05, "loss": 1.7162, "step": 35240 }, { "epoch": 0.05, "learning_rate": 4.7796875e-05, "loss": 1.859, "step": 35250 }, { "epoch": 0.05, "learning_rate": 4.7796250000000005e-05, "loss": 1.8162, "step": 35260 }, { "epoch": 0.05, "learning_rate": 4.7795625e-05, "loss": 1.9607, "step": 35270 }, { "epoch": 0.05, "learning_rate": 4.7795e-05, "loss": 1.8577, "step": 35280 }, { "epoch": 0.05, "learning_rate": 4.7794375e-05, "loss": 2.0184, "step": 35290 }, { "epoch": 0.05, "learning_rate": 4.7793750000000006e-05, "loss": 1.6874, "step": 35300 }, { "epoch": 0.05, "learning_rate": 4.7793124999999997e-05, "loss": 1.8375, "step": 35310 }, { "epoch": 0.05, "learning_rate": 4.77925e-05, "loss": 1.8951, "step": 35320 }, { "epoch": 0.05, "learning_rate": 4.7791875000000004e-05, "loss": 1.9508, "step": 35330 }, { "epoch": 0.05, "learning_rate": 4.779125e-05, "loss": 1.6976, "step": 35340 }, { "epoch": 0.05, "learning_rate": 4.7790625e-05, "loss": 1.785, "step": 35350 }, { "epoch": 0.05, "learning_rate": 4.779e-05, "loss": 1.996, "step": 35360 }, { "epoch": 0.05, "learning_rate": 4.7789375000000005e-05, "loss": 2.0532, "step": 35370 }, { "epoch": 0.05, "learning_rate": 4.778875e-05, "loss": 1.9884, "step": 35380 }, { "epoch": 0.05, "learning_rate": 4.7788125e-05, "loss": 1.8329, "step": 35390 }, { "epoch": 0.05, "learning_rate": 4.77875e-05, "loss": 1.7806, "step": 35400 }, { "epoch": 0.05, "learning_rate": 4.7786875000000007e-05, "loss": 1.9867, "step": 35410 }, { "epoch": 0.05, "learning_rate": 4.7786250000000003e-05, "loss": 1.8853, "step": 35420 }, { "epoch": 0.05, "learning_rate": 4.7785625e-05, "loss": 1.7674, "step": 35430 }, { "epoch": 0.05, "learning_rate": 4.7785000000000004e-05, "loss": 2.0278, "step": 35440 }, { "epoch": 0.05, "learning_rate": 4.7784375e-05, "loss": 1.7131, "step": 35450 }, { "epoch": 0.05, "learning_rate": 4.7783750000000005e-05, "loss": 1.8397, "step": 35460 }, { "epoch": 0.05, "learning_rate": 4.7783125e-05, "loss": 1.7247, "step": 35470 }, { "epoch": 0.05, "learning_rate": 4.7782500000000005e-05, "loss": 2.1521, "step": 35480 }, { "epoch": 0.05, "learning_rate": 4.7781875e-05, "loss": 2.0474, "step": 35490 }, { "epoch": 0.05, "learning_rate": 4.778125e-05, "loss": 2.1594, "step": 35500 }, { "epoch": 0.05, "learning_rate": 4.7780625e-05, "loss": 1.8714, "step": 35510 }, { "epoch": 0.05, "learning_rate": 4.778e-05, "loss": 1.8563, "step": 35520 }, { "epoch": 0.05, "learning_rate": 4.7779375000000004e-05, "loss": 2.0078, "step": 35530 }, { "epoch": 0.05, "learning_rate": 4.777875e-05, "loss": 1.9348, "step": 35540 }, { "epoch": 0.05, "learning_rate": 4.7778125000000004e-05, "loss": 1.7989, "step": 35550 }, { "epoch": 0.05, "learning_rate": 4.77775e-05, "loss": 1.7751, "step": 35560 }, { "epoch": 0.05, "learning_rate": 4.7776875000000005e-05, "loss": 1.8869, "step": 35570 }, { "epoch": 0.05, "learning_rate": 4.777625e-05, "loss": 1.8341, "step": 35580 }, { "epoch": 0.05, "learning_rate": 4.7775625e-05, "loss": 1.8053, "step": 35590 }, { "epoch": 0.05, "learning_rate": 4.7775e-05, "loss": 1.9299, "step": 35600 }, { "epoch": 0.05, "learning_rate": 4.7774375000000006e-05, "loss": 1.622, "step": 35610 }, { "epoch": 0.05, "learning_rate": 4.777375e-05, "loss": 2.0138, "step": 35620 }, { "epoch": 0.05, "learning_rate": 4.7773125e-05, "loss": 1.9599, "step": 35630 }, { "epoch": 0.05, "learning_rate": 4.7772500000000004e-05, "loss": 2.0098, "step": 35640 }, { "epoch": 0.05, "learning_rate": 4.777187500000001e-05, "loss": 1.7618, "step": 35650 }, { "epoch": 0.05, "learning_rate": 4.777125e-05, "loss": 1.9748, "step": 35660 }, { "epoch": 0.05, "learning_rate": 4.7770625e-05, "loss": 1.9043, "step": 35670 }, { "epoch": 0.05, "learning_rate": 4.7770000000000005e-05, "loss": 1.8744, "step": 35680 }, { "epoch": 0.05, "learning_rate": 4.7769375e-05, "loss": 1.8892, "step": 35690 }, { "epoch": 0.05, "learning_rate": 4.776875e-05, "loss": 1.7591, "step": 35700 }, { "epoch": 0.05, "learning_rate": 4.7768125e-05, "loss": 1.8297, "step": 35710 }, { "epoch": 0.05, "learning_rate": 4.7767500000000006e-05, "loss": 1.9755, "step": 35720 }, { "epoch": 0.05, "learning_rate": 4.7766874999999997e-05, "loss": 1.8396, "step": 35730 }, { "epoch": 0.05, "learning_rate": 4.776625e-05, "loss": 1.7812, "step": 35740 }, { "epoch": 0.05, "learning_rate": 4.7765625000000004e-05, "loss": 1.9026, "step": 35750 }, { "epoch": 0.05, "learning_rate": 4.7765e-05, "loss": 1.9492, "step": 35760 }, { "epoch": 0.05, "learning_rate": 4.7764375e-05, "loss": 1.8244, "step": 35770 }, { "epoch": 0.05, "learning_rate": 4.776375e-05, "loss": 1.906, "step": 35780 }, { "epoch": 0.05, "learning_rate": 4.7763125000000005e-05, "loss": 1.9022, "step": 35790 }, { "epoch": 0.05, "learning_rate": 4.77625e-05, "loss": 2.0257, "step": 35800 }, { "epoch": 0.05, "learning_rate": 4.7761875e-05, "loss": 1.9519, "step": 35810 }, { "epoch": 0.05, "learning_rate": 4.776125e-05, "loss": 1.908, "step": 35820 }, { "epoch": 0.05, "learning_rate": 4.7760625000000007e-05, "loss": 1.7639, "step": 35830 }, { "epoch": 0.05, "learning_rate": 4.7760000000000004e-05, "loss": 1.7981, "step": 35840 }, { "epoch": 0.05, "learning_rate": 4.7759375e-05, "loss": 1.7601, "step": 35850 }, { "epoch": 0.05, "learning_rate": 4.7758750000000004e-05, "loss": 2.0247, "step": 35860 }, { "epoch": 0.05, "learning_rate": 4.7758125e-05, "loss": 1.913, "step": 35870 }, { "epoch": 0.05, "learning_rate": 4.7757500000000005e-05, "loss": 2.0386, "step": 35880 }, { "epoch": 0.05, "learning_rate": 4.7756875e-05, "loss": 1.8681, "step": 35890 }, { "epoch": 0.05, "learning_rate": 4.7756250000000005e-05, "loss": 2.2044, "step": 35900 }, { "epoch": 0.05, "learning_rate": 4.7755625e-05, "loss": 1.965, "step": 35910 }, { "epoch": 0.05, "learning_rate": 4.7755e-05, "loss": 1.8394, "step": 35920 }, { "epoch": 0.05, "learning_rate": 4.7754375e-05, "loss": 1.9106, "step": 35930 }, { "epoch": 0.05, "learning_rate": 4.775375e-05, "loss": 1.7685, "step": 35940 }, { "epoch": 0.05, "learning_rate": 4.7753125000000004e-05, "loss": 1.7439, "step": 35950 }, { "epoch": 0.05, "learning_rate": 4.77525e-05, "loss": 1.9464, "step": 35960 }, { "epoch": 0.05, "learning_rate": 4.7751875000000004e-05, "loss": 2.0623, "step": 35970 }, { "epoch": 0.05, "learning_rate": 4.775125e-05, "loss": 2.0217, "step": 35980 }, { "epoch": 0.05, "learning_rate": 4.7750625000000005e-05, "loss": 1.8045, "step": 35990 }, { "epoch": 0.05, "learning_rate": 4.775e-05, "loss": 1.9284, "step": 36000 }, { "epoch": 0.05, "eval_loss": 1.82748544216156, "eval_runtime": 1290.3179, "eval_samples_per_second": 44.978, "eval_steps_per_second": 5.623, "step": 36000 }, { "epoch": 0.05, "learning_rate": 4.7749375e-05, "loss": 1.9621, "step": 36010 }, { "epoch": 0.05, "learning_rate": 4.774875e-05, "loss": 2.0798, "step": 36020 }, { "epoch": 0.05, "learning_rate": 4.7748125000000006e-05, "loss": 1.9609, "step": 36030 }, { "epoch": 0.05, "learning_rate": 4.77475e-05, "loss": 1.8347, "step": 36040 }, { "epoch": 0.05, "learning_rate": 4.7746875e-05, "loss": 2.0107, "step": 36050 }, { "epoch": 0.05, "learning_rate": 4.7746250000000004e-05, "loss": 1.8513, "step": 36060 }, { "epoch": 0.05, "learning_rate": 4.774562500000001e-05, "loss": 1.7518, "step": 36070 }, { "epoch": 0.05, "learning_rate": 4.7745e-05, "loss": 1.8397, "step": 36080 }, { "epoch": 0.05, "learning_rate": 4.7744375e-05, "loss": 1.8601, "step": 36090 }, { "epoch": 0.05, "learning_rate": 4.7743750000000005e-05, "loss": 2.089, "step": 36100 }, { "epoch": 0.05, "learning_rate": 4.7743125e-05, "loss": 1.8646, "step": 36110 }, { "epoch": 0.05, "learning_rate": 4.77425e-05, "loss": 1.6432, "step": 36120 }, { "epoch": 0.05, "learning_rate": 4.7741875e-05, "loss": 1.8036, "step": 36130 }, { "epoch": 0.05, "learning_rate": 4.7741250000000006e-05, "loss": 1.9601, "step": 36140 }, { "epoch": 0.05, "learning_rate": 4.7740624999999997e-05, "loss": 1.5963, "step": 36150 }, { "epoch": 0.05, "learning_rate": 4.774e-05, "loss": 1.8645, "step": 36160 }, { "epoch": 0.05, "learning_rate": 4.7739375000000004e-05, "loss": 2.0682, "step": 36170 }, { "epoch": 0.05, "learning_rate": 4.773875e-05, "loss": 2.0506, "step": 36180 }, { "epoch": 0.05, "learning_rate": 4.7738125e-05, "loss": 1.7459, "step": 36190 }, { "epoch": 0.05, "learning_rate": 4.77375e-05, "loss": 2.1069, "step": 36200 }, { "epoch": 0.05, "learning_rate": 4.7736875000000005e-05, "loss": 1.7064, "step": 36210 }, { "epoch": 0.05, "learning_rate": 4.773625e-05, "loss": 1.9819, "step": 36220 }, { "epoch": 0.05, "learning_rate": 4.7735625e-05, "loss": 1.7439, "step": 36230 }, { "epoch": 0.05, "learning_rate": 4.7735e-05, "loss": 2.0901, "step": 36240 }, { "epoch": 0.05, "learning_rate": 4.773437500000001e-05, "loss": 1.8379, "step": 36250 }, { "epoch": 0.05, "learning_rate": 4.7733750000000004e-05, "loss": 1.9712, "step": 36260 }, { "epoch": 0.05, "learning_rate": 4.7733125e-05, "loss": 1.8888, "step": 36270 }, { "epoch": 0.05, "learning_rate": 4.7732500000000004e-05, "loss": 1.9248, "step": 36280 }, { "epoch": 0.05, "learning_rate": 4.7731875e-05, "loss": 1.7085, "step": 36290 }, { "epoch": 0.05, "learning_rate": 4.773125e-05, "loss": 1.7587, "step": 36300 }, { "epoch": 0.05, "learning_rate": 4.7730625e-05, "loss": 1.962, "step": 36310 }, { "epoch": 0.05, "learning_rate": 4.7730000000000005e-05, "loss": 2.0222, "step": 36320 }, { "epoch": 0.05, "learning_rate": 4.7729375e-05, "loss": 1.8073, "step": 36330 }, { "epoch": 0.05, "learning_rate": 4.772875e-05, "loss": 1.9189, "step": 36340 }, { "epoch": 0.05, "learning_rate": 4.7728125e-05, "loss": 2.0682, "step": 36350 }, { "epoch": 0.05, "learning_rate": 4.77275e-05, "loss": 1.719, "step": 36360 }, { "epoch": 0.05, "learning_rate": 4.7726875000000004e-05, "loss": 1.8962, "step": 36370 }, { "epoch": 0.05, "learning_rate": 4.772625e-05, "loss": 1.7124, "step": 36380 }, { "epoch": 0.05, "learning_rate": 4.7725625000000004e-05, "loss": 1.8467, "step": 36390 }, { "epoch": 0.05, "learning_rate": 4.7725e-05, "loss": 2.1321, "step": 36400 }, { "epoch": 0.05, "learning_rate": 4.7724375000000005e-05, "loss": 1.8844, "step": 36410 }, { "epoch": 0.05, "learning_rate": 4.772375e-05, "loss": 1.7957, "step": 36420 }, { "epoch": 0.05, "learning_rate": 4.7723125e-05, "loss": 1.902, "step": 36430 }, { "epoch": 0.05, "learning_rate": 4.77225e-05, "loss": 1.7382, "step": 36440 }, { "epoch": 0.05, "learning_rate": 4.7721875000000006e-05, "loss": 1.9399, "step": 36450 }, { "epoch": 0.05, "learning_rate": 4.772125e-05, "loss": 1.8373, "step": 36460 }, { "epoch": 0.05, "learning_rate": 4.7720625e-05, "loss": 1.5948, "step": 36470 }, { "epoch": 0.05, "learning_rate": 4.7720000000000004e-05, "loss": 1.8, "step": 36480 }, { "epoch": 0.05, "learning_rate": 4.771937500000001e-05, "loss": 1.8053, "step": 36490 }, { "epoch": 0.05, "learning_rate": 4.771875e-05, "loss": 1.7642, "step": 36500 }, { "epoch": 0.05, "learning_rate": 4.7718125e-05, "loss": 2.0961, "step": 36510 }, { "epoch": 0.05, "learning_rate": 4.7717500000000005e-05, "loss": 1.8729, "step": 36520 }, { "epoch": 0.05, "learning_rate": 4.7716875e-05, "loss": 1.7822, "step": 36530 }, { "epoch": 0.05, "learning_rate": 4.771625e-05, "loss": 1.8552, "step": 36540 }, { "epoch": 0.05, "learning_rate": 4.7715625e-05, "loss": 1.7221, "step": 36550 }, { "epoch": 0.05, "learning_rate": 4.7715000000000006e-05, "loss": 2.2144, "step": 36560 }, { "epoch": 0.05, "learning_rate": 4.7714375000000003e-05, "loss": 1.7691, "step": 36570 }, { "epoch": 0.05, "learning_rate": 4.771375e-05, "loss": 1.7387, "step": 36580 }, { "epoch": 0.05, "learning_rate": 4.7713125000000004e-05, "loss": 2.1977, "step": 36590 }, { "epoch": 0.05, "learning_rate": 4.771250000000001e-05, "loss": 1.7618, "step": 36600 }, { "epoch": 0.05, "learning_rate": 4.7711875e-05, "loss": 1.78, "step": 36610 }, { "epoch": 0.05, "learning_rate": 4.771125e-05, "loss": 1.8566, "step": 36620 }, { "epoch": 0.05, "learning_rate": 4.7710625000000005e-05, "loss": 1.6561, "step": 36630 }, { "epoch": 0.05, "learning_rate": 4.771e-05, "loss": 2.0168, "step": 36640 }, { "epoch": 0.05, "learning_rate": 4.7709375e-05, "loss": 1.7676, "step": 36650 }, { "epoch": 0.05, "learning_rate": 4.770875e-05, "loss": 2.0478, "step": 36660 }, { "epoch": 0.05, "learning_rate": 4.770812500000001e-05, "loss": 1.8673, "step": 36670 }, { "epoch": 0.05, "learning_rate": 4.7707500000000004e-05, "loss": 2.0325, "step": 36680 }, { "epoch": 0.05, "learning_rate": 4.7706875e-05, "loss": 1.6393, "step": 36690 }, { "epoch": 0.05, "learning_rate": 4.7706250000000004e-05, "loss": 1.9381, "step": 36700 }, { "epoch": 0.05, "learning_rate": 4.7705625e-05, "loss": 1.9652, "step": 36710 }, { "epoch": 0.05, "learning_rate": 4.7705e-05, "loss": 1.7687, "step": 36720 }, { "epoch": 0.05, "learning_rate": 4.7704375e-05, "loss": 2.0176, "step": 36730 }, { "epoch": 0.05, "learning_rate": 4.7703750000000006e-05, "loss": 1.878, "step": 36740 }, { "epoch": 0.05, "learning_rate": 4.7703125e-05, "loss": 2.0247, "step": 36750 }, { "epoch": 0.05, "learning_rate": 4.77025e-05, "loss": 1.8935, "step": 36760 }, { "epoch": 0.05, "learning_rate": 4.7701875e-05, "loss": 1.9047, "step": 36770 }, { "epoch": 0.05, "learning_rate": 4.770125e-05, "loss": 1.8791, "step": 36780 }, { "epoch": 0.05, "learning_rate": 4.7700625000000004e-05, "loss": 1.8682, "step": 36790 }, { "epoch": 0.05, "learning_rate": 4.77e-05, "loss": 2.1092, "step": 36800 }, { "epoch": 0.05, "learning_rate": 4.7699375000000004e-05, "loss": 2.1354, "step": 36810 }, { "epoch": 0.05, "learning_rate": 4.769875e-05, "loss": 1.9009, "step": 36820 }, { "epoch": 0.05, "learning_rate": 4.7698125000000005e-05, "loss": 2.0186, "step": 36830 }, { "epoch": 0.05, "learning_rate": 4.76975e-05, "loss": 1.7371, "step": 36840 }, { "epoch": 0.05, "learning_rate": 4.7696875e-05, "loss": 2.0347, "step": 36850 }, { "epoch": 0.05, "learning_rate": 4.769625e-05, "loss": 1.9052, "step": 36860 }, { "epoch": 0.05, "learning_rate": 4.7695625000000006e-05, "loss": 1.7468, "step": 36870 }, { "epoch": 0.05, "learning_rate": 4.7695e-05, "loss": 1.5869, "step": 36880 }, { "epoch": 0.05, "learning_rate": 4.7694375e-05, "loss": 1.8645, "step": 36890 }, { "epoch": 0.05, "learning_rate": 4.7693750000000004e-05, "loss": 1.8059, "step": 36900 }, { "epoch": 0.05, "learning_rate": 4.7693125e-05, "loss": 2.0106, "step": 36910 }, { "epoch": 0.05, "learning_rate": 4.76925e-05, "loss": 2.0138, "step": 36920 }, { "epoch": 0.05, "learning_rate": 4.7691875e-05, "loss": 2.2234, "step": 36930 }, { "epoch": 0.05, "learning_rate": 4.7691250000000005e-05, "loss": 1.8606, "step": 36940 }, { "epoch": 0.05, "learning_rate": 4.7690625e-05, "loss": 1.7729, "step": 36950 }, { "epoch": 0.05, "learning_rate": 4.769e-05, "loss": 1.8833, "step": 36960 }, { "epoch": 0.05, "learning_rate": 4.7689375e-05, "loss": 1.6008, "step": 36970 }, { "epoch": 0.05, "learning_rate": 4.7688750000000007e-05, "loss": 1.8195, "step": 36980 }, { "epoch": 0.05, "learning_rate": 4.7688125000000003e-05, "loss": 1.71, "step": 36990 }, { "epoch": 0.05, "learning_rate": 4.76875e-05, "loss": 1.7978, "step": 37000 }, { "epoch": 0.05, "eval_loss": 1.804318904876709, "eval_runtime": 1290.3023, "eval_samples_per_second": 44.979, "eval_steps_per_second": 5.623, "step": 37000 }, { "epoch": 0.05, "learning_rate": 4.7686875000000004e-05, "loss": 1.978, "step": 37010 }, { "epoch": 0.05, "learning_rate": 4.768625000000001e-05, "loss": 1.9303, "step": 37020 }, { "epoch": 0.05, "learning_rate": 4.7685625e-05, "loss": 1.811, "step": 37030 }, { "epoch": 0.05, "learning_rate": 4.7685e-05, "loss": 1.7242, "step": 37040 }, { "epoch": 0.05, "learning_rate": 4.7684375000000005e-05, "loss": 1.8592, "step": 37050 }, { "epoch": 0.05, "learning_rate": 4.768375e-05, "loss": 2.2945, "step": 37060 }, { "epoch": 0.05, "learning_rate": 4.7683125e-05, "loss": 1.9471, "step": 37070 }, { "epoch": 0.05, "learning_rate": 4.76825e-05, "loss": 1.8535, "step": 37080 }, { "epoch": 0.05, "learning_rate": 4.768187500000001e-05, "loss": 1.7293, "step": 37090 }, { "epoch": 0.05, "learning_rate": 4.768125e-05, "loss": 1.8882, "step": 37100 }, { "epoch": 0.05, "learning_rate": 4.7680625e-05, "loss": 1.941, "step": 37110 }, { "epoch": 0.05, "learning_rate": 4.7680000000000004e-05, "loss": 1.9084, "step": 37120 }, { "epoch": 0.05, "learning_rate": 4.7679375e-05, "loss": 1.909, "step": 37130 }, { "epoch": 0.05, "learning_rate": 4.767875e-05, "loss": 2.008, "step": 37140 }, { "epoch": 0.05, "learning_rate": 4.7678125e-05, "loss": 1.8433, "step": 37150 }, { "epoch": 0.05, "learning_rate": 4.7677500000000006e-05, "loss": 1.7505, "step": 37160 }, { "epoch": 0.05, "learning_rate": 4.7676875e-05, "loss": 1.7738, "step": 37170 }, { "epoch": 0.05, "learning_rate": 4.767625e-05, "loss": 1.8902, "step": 37180 }, { "epoch": 0.05, "learning_rate": 4.7675625e-05, "loss": 1.8987, "step": 37190 }, { "epoch": 0.05, "learning_rate": 4.7675e-05, "loss": 2.051, "step": 37200 }, { "epoch": 0.05, "learning_rate": 4.7674375000000004e-05, "loss": 2.1636, "step": 37210 }, { "epoch": 0.05, "learning_rate": 4.767375e-05, "loss": 1.9012, "step": 37220 }, { "epoch": 0.05, "learning_rate": 4.7673125000000004e-05, "loss": 1.8635, "step": 37230 }, { "epoch": 0.05, "learning_rate": 4.76725e-05, "loss": 2.0375, "step": 37240 }, { "epoch": 0.05, "learning_rate": 4.7671875000000005e-05, "loss": 1.9729, "step": 37250 }, { "epoch": 0.05, "learning_rate": 4.767125e-05, "loss": 2.0928, "step": 37260 }, { "epoch": 0.05, "learning_rate": 4.7670625e-05, "loss": 1.9347, "step": 37270 }, { "epoch": 0.05, "learning_rate": 4.767e-05, "loss": 1.9824, "step": 37280 }, { "epoch": 0.05, "learning_rate": 4.7669375000000006e-05, "loss": 1.6513, "step": 37290 }, { "epoch": 0.05, "learning_rate": 4.766875e-05, "loss": 2.0493, "step": 37300 }, { "epoch": 0.05, "learning_rate": 4.7668125e-05, "loss": 2.0319, "step": 37310 }, { "epoch": 0.05, "learning_rate": 4.7667500000000004e-05, "loss": 1.8988, "step": 37320 }, { "epoch": 0.05, "learning_rate": 4.7666875e-05, "loss": 2.0423, "step": 37330 }, { "epoch": 0.05, "learning_rate": 4.766625e-05, "loss": 1.9619, "step": 37340 }, { "epoch": 0.05, "learning_rate": 4.7665625e-05, "loss": 1.7455, "step": 37350 }, { "epoch": 0.05, "learning_rate": 4.7665000000000005e-05, "loss": 1.7025, "step": 37360 }, { "epoch": 0.05, "learning_rate": 4.7664375e-05, "loss": 1.9645, "step": 37370 }, { "epoch": 0.05, "learning_rate": 4.766375e-05, "loss": 2.0817, "step": 37380 }, { "epoch": 0.05, "learning_rate": 4.7663125e-05, "loss": 1.7928, "step": 37390 }, { "epoch": 0.05, "learning_rate": 4.7662500000000007e-05, "loss": 2.0841, "step": 37400 }, { "epoch": 0.05, "learning_rate": 4.7661875000000004e-05, "loss": 2.0437, "step": 37410 }, { "epoch": 0.05, "learning_rate": 4.766125e-05, "loss": 1.9482, "step": 37420 }, { "epoch": 0.05, "learning_rate": 4.7660625000000004e-05, "loss": 1.7612, "step": 37430 }, { "epoch": 0.05, "learning_rate": 4.766000000000001e-05, "loss": 1.8704, "step": 37440 }, { "epoch": 0.05, "learning_rate": 4.7659375e-05, "loss": 1.9724, "step": 37450 }, { "epoch": 0.05, "learning_rate": 4.765875e-05, "loss": 1.845, "step": 37460 }, { "epoch": 0.05, "learning_rate": 4.7658125000000005e-05, "loss": 1.9388, "step": 37470 }, { "epoch": 0.05, "learning_rate": 4.76575e-05, "loss": 1.8594, "step": 37480 }, { "epoch": 0.05, "learning_rate": 4.7656875e-05, "loss": 2.0354, "step": 37490 }, { "epoch": 0.05, "learning_rate": 4.765625e-05, "loss": 2.0756, "step": 37500 }, { "epoch": 0.05, "learning_rate": 4.765562500000001e-05, "loss": 1.942, "step": 37510 }, { "epoch": 0.05, "learning_rate": 4.7655e-05, "loss": 1.8913, "step": 37520 }, { "epoch": 0.05, "learning_rate": 4.7654375e-05, "loss": 1.9566, "step": 37530 }, { "epoch": 0.05, "learning_rate": 4.7653750000000004e-05, "loss": 2.0589, "step": 37540 }, { "epoch": 0.05, "learning_rate": 4.7653125e-05, "loss": 2.0882, "step": 37550 }, { "epoch": 0.05, "learning_rate": 4.76525e-05, "loss": 2.0744, "step": 37560 }, { "epoch": 0.05, "learning_rate": 4.7651875e-05, "loss": 1.8786, "step": 37570 }, { "epoch": 0.05, "learning_rate": 4.7651250000000006e-05, "loss": 1.627, "step": 37580 }, { "epoch": 0.05, "learning_rate": 4.7650625e-05, "loss": 1.8693, "step": 37590 }, { "epoch": 0.05, "learning_rate": 4.765e-05, "loss": 2.095, "step": 37600 }, { "epoch": 0.05, "learning_rate": 4.7649375e-05, "loss": 1.8685, "step": 37610 }, { "epoch": 0.05, "learning_rate": 4.764875e-05, "loss": 2.1033, "step": 37620 }, { "epoch": 0.05, "learning_rate": 4.7648125000000004e-05, "loss": 1.6717, "step": 37630 }, { "epoch": 0.05, "learning_rate": 4.76475e-05, "loss": 1.7696, "step": 37640 }, { "epoch": 0.05, "learning_rate": 4.7646875000000004e-05, "loss": 1.7869, "step": 37650 }, { "epoch": 0.05, "learning_rate": 4.764625e-05, "loss": 2.0677, "step": 37660 }, { "epoch": 0.05, "learning_rate": 4.7645625000000005e-05, "loss": 1.9784, "step": 37670 }, { "epoch": 0.05, "learning_rate": 4.7645e-05, "loss": 1.9624, "step": 37680 }, { "epoch": 0.05, "learning_rate": 4.7644375e-05, "loss": 1.8535, "step": 37690 }, { "epoch": 0.05, "learning_rate": 4.764375e-05, "loss": 1.6494, "step": 37700 }, { "epoch": 0.05, "learning_rate": 4.7643125e-05, "loss": 1.9548, "step": 37710 }, { "epoch": 0.05, "learning_rate": 4.76425e-05, "loss": 1.8606, "step": 37720 }, { "epoch": 0.05, "learning_rate": 4.7641875e-05, "loss": 1.8621, "step": 37730 }, { "epoch": 0.05, "learning_rate": 4.7641250000000004e-05, "loss": 1.7156, "step": 37740 }, { "epoch": 0.05, "learning_rate": 4.7640625e-05, "loss": 1.8828, "step": 37750 }, { "epoch": 0.05, "learning_rate": 4.7640000000000005e-05, "loss": 1.7826, "step": 37760 }, { "epoch": 0.05, "learning_rate": 4.7639375e-05, "loss": 1.8366, "step": 37770 }, { "epoch": 0.05, "learning_rate": 4.7638750000000005e-05, "loss": 1.903, "step": 37780 }, { "epoch": 0.05, "learning_rate": 4.7638125e-05, "loss": 1.8379, "step": 37790 }, { "epoch": 0.05, "learning_rate": 4.76375e-05, "loss": 1.7327, "step": 37800 }, { "epoch": 0.05, "learning_rate": 4.7636875e-05, "loss": 1.9951, "step": 37810 }, { "epoch": 0.05, "learning_rate": 4.7636250000000007e-05, "loss": 1.9611, "step": 37820 }, { "epoch": 0.05, "learning_rate": 4.7635625000000004e-05, "loss": 1.7196, "step": 37830 }, { "epoch": 0.05, "learning_rate": 4.7635e-05, "loss": 1.8056, "step": 37840 }, { "epoch": 0.05, "learning_rate": 4.7634375000000004e-05, "loss": 1.7873, "step": 37850 }, { "epoch": 0.05, "learning_rate": 4.763375000000001e-05, "loss": 1.8186, "step": 37860 }, { "epoch": 0.05, "learning_rate": 4.7633125e-05, "loss": 2.0886, "step": 37870 }, { "epoch": 0.05, "learning_rate": 4.76325e-05, "loss": 1.8171, "step": 37880 }, { "epoch": 0.05, "learning_rate": 4.7631875000000005e-05, "loss": 1.8861, "step": 37890 }, { "epoch": 0.05, "learning_rate": 4.763125e-05, "loss": 1.8915, "step": 37900 }, { "epoch": 0.05, "learning_rate": 4.7630625e-05, "loss": 1.9659, "step": 37910 }, { "epoch": 0.05, "learning_rate": 4.763e-05, "loss": 1.7432, "step": 37920 }, { "epoch": 0.05, "learning_rate": 4.762937500000001e-05, "loss": 1.8377, "step": 37930 }, { "epoch": 0.05, "learning_rate": 4.762875e-05, "loss": 1.8472, "step": 37940 }, { "epoch": 0.05, "learning_rate": 4.7628125e-05, "loss": 1.7117, "step": 37950 }, { "epoch": 0.05, "learning_rate": 4.7627500000000004e-05, "loss": 1.9975, "step": 37960 }, { "epoch": 0.05, "learning_rate": 4.7626875e-05, "loss": 1.8648, "step": 37970 }, { "epoch": 0.05, "learning_rate": 4.762625e-05, "loss": 1.9037, "step": 37980 }, { "epoch": 0.05, "learning_rate": 4.7625625e-05, "loss": 1.9362, "step": 37990 }, { "epoch": 0.05, "learning_rate": 4.7625000000000006e-05, "loss": 1.9216, "step": 38000 }, { "epoch": 0.05, "eval_loss": 1.788551688194275, "eval_runtime": 1290.3623, "eval_samples_per_second": 44.977, "eval_steps_per_second": 5.622, "step": 38000 }, { "epoch": 0.05, "learning_rate": 4.7624375e-05, "loss": 2.0043, "step": 38010 }, { "epoch": 0.05, "learning_rate": 4.762375e-05, "loss": 1.7192, "step": 38020 }, { "epoch": 0.05, "learning_rate": 4.7623125e-05, "loss": 1.8607, "step": 38030 }, { "epoch": 0.05, "learning_rate": 4.76225e-05, "loss": 1.9694, "step": 38040 }, { "epoch": 0.05, "learning_rate": 4.7621875000000004e-05, "loss": 1.9105, "step": 38050 }, { "epoch": 0.05, "learning_rate": 4.762125e-05, "loss": 1.8408, "step": 38060 }, { "epoch": 0.05, "learning_rate": 4.7620625000000005e-05, "loss": 2.1094, "step": 38070 }, { "epoch": 0.05, "learning_rate": 4.762e-05, "loss": 1.8223, "step": 38080 }, { "epoch": 0.05, "learning_rate": 4.7619375000000005e-05, "loss": 2.0037, "step": 38090 }, { "epoch": 0.05, "learning_rate": 4.761875e-05, "loss": 1.7194, "step": 38100 }, { "epoch": 0.05, "learning_rate": 4.7618125e-05, "loss": 1.8571, "step": 38110 }, { "epoch": 0.05, "learning_rate": 4.76175e-05, "loss": 1.7923, "step": 38120 }, { "epoch": 0.05, "learning_rate": 4.7616875e-05, "loss": 2.0386, "step": 38130 }, { "epoch": 0.05, "learning_rate": 4.7616250000000003e-05, "loss": 2.0493, "step": 38140 }, { "epoch": 0.05, "learning_rate": 4.7615625e-05, "loss": 1.8419, "step": 38150 }, { "epoch": 0.05, "learning_rate": 4.7615000000000004e-05, "loss": 1.816, "step": 38160 }, { "epoch": 0.05, "learning_rate": 4.7614375e-05, "loss": 1.7983, "step": 38170 }, { "epoch": 0.05, "learning_rate": 4.7613750000000005e-05, "loss": 1.9202, "step": 38180 }, { "epoch": 0.05, "learning_rate": 4.7613125e-05, "loss": 1.9441, "step": 38190 }, { "epoch": 0.05, "learning_rate": 4.7612500000000005e-05, "loss": 1.8249, "step": 38200 }, { "epoch": 0.05, "learning_rate": 4.7611875e-05, "loss": 1.7957, "step": 38210 }, { "epoch": 0.05, "learning_rate": 4.761125e-05, "loss": 1.8627, "step": 38220 }, { "epoch": 0.05, "learning_rate": 4.7610625e-05, "loss": 1.9506, "step": 38230 }, { "epoch": 0.05, "learning_rate": 4.761000000000001e-05, "loss": 2.0524, "step": 38240 }, { "epoch": 0.05, "learning_rate": 4.7609375000000004e-05, "loss": 1.9718, "step": 38250 }, { "epoch": 0.05, "learning_rate": 4.760875e-05, "loss": 1.8649, "step": 38260 }, { "epoch": 0.05, "learning_rate": 4.7608125000000004e-05, "loss": 1.8648, "step": 38270 }, { "epoch": 0.05, "learning_rate": 4.760750000000001e-05, "loss": 2.0251, "step": 38280 }, { "epoch": 0.05, "learning_rate": 4.7606875e-05, "loss": 1.873, "step": 38290 }, { "epoch": 0.05, "learning_rate": 4.760625e-05, "loss": 1.7119, "step": 38300 }, { "epoch": 0.05, "learning_rate": 4.7605625000000006e-05, "loss": 2.0184, "step": 38310 }, { "epoch": 0.05, "learning_rate": 4.7605e-05, "loss": 2.0551, "step": 38320 }, { "epoch": 0.05, "learning_rate": 4.7604375e-05, "loss": 2.0649, "step": 38330 }, { "epoch": 0.05, "learning_rate": 4.760375e-05, "loss": 1.7424, "step": 38340 }, { "epoch": 0.05, "learning_rate": 4.760312500000001e-05, "loss": 1.8256, "step": 38350 }, { "epoch": 0.05, "learning_rate": 4.76025e-05, "loss": 1.7901, "step": 38360 }, { "epoch": 0.05, "learning_rate": 4.7601875e-05, "loss": 2.0526, "step": 38370 }, { "epoch": 0.05, "learning_rate": 4.7601250000000004e-05, "loss": 1.801, "step": 38380 }, { "epoch": 0.05, "learning_rate": 4.7600625e-05, "loss": 2.0096, "step": 38390 }, { "epoch": 0.05, "learning_rate": 4.76e-05, "loss": 1.8109, "step": 38400 }, { "epoch": 0.05, "learning_rate": 4.7599375e-05, "loss": 1.9376, "step": 38410 }, { "epoch": 0.05, "learning_rate": 4.7598750000000006e-05, "loss": 1.7576, "step": 38420 }, { "epoch": 0.05, "learning_rate": 4.7598125e-05, "loss": 1.6923, "step": 38430 }, { "epoch": 0.05, "learning_rate": 4.75975e-05, "loss": 1.833, "step": 38440 }, { "epoch": 0.05, "learning_rate": 4.7596875e-05, "loss": 2.0206, "step": 38450 }, { "epoch": 0.05, "learning_rate": 4.759625e-05, "loss": 1.7478, "step": 38460 }, { "epoch": 0.05, "learning_rate": 4.7595625000000004e-05, "loss": 1.9773, "step": 38470 }, { "epoch": 0.05, "learning_rate": 4.7595e-05, "loss": 1.8628, "step": 38480 }, { "epoch": 0.05, "learning_rate": 4.7594375000000005e-05, "loss": 1.9451, "step": 38490 }, { "epoch": 0.05, "learning_rate": 4.759375e-05, "loss": 1.8679, "step": 38500 }, { "epoch": 0.05, "learning_rate": 4.7593125e-05, "loss": 1.8024, "step": 38510 }, { "epoch": 0.05, "learning_rate": 4.75925e-05, "loss": 1.9672, "step": 38520 }, { "epoch": 0.05, "learning_rate": 4.7591875e-05, "loss": 1.8556, "step": 38530 }, { "epoch": 0.05, "learning_rate": 4.759125e-05, "loss": 1.8069, "step": 38540 }, { "epoch": 0.05, "learning_rate": 4.7590625e-05, "loss": 1.8674, "step": 38550 }, { "epoch": 0.05, "learning_rate": 4.7590000000000003e-05, "loss": 1.8699, "step": 38560 }, { "epoch": 0.05, "learning_rate": 4.7589375e-05, "loss": 1.9463, "step": 38570 }, { "epoch": 0.05, "learning_rate": 4.7588750000000004e-05, "loss": 1.7963, "step": 38580 }, { "epoch": 0.05, "learning_rate": 4.7588125e-05, "loss": 1.7481, "step": 38590 }, { "epoch": 0.05, "learning_rate": 4.7587500000000005e-05, "loss": 1.8682, "step": 38600 }, { "epoch": 0.05, "learning_rate": 4.7586875e-05, "loss": 1.9868, "step": 38610 }, { "epoch": 0.05, "learning_rate": 4.7586250000000005e-05, "loss": 2.2456, "step": 38620 }, { "epoch": 0.05, "learning_rate": 4.7585625e-05, "loss": 1.8259, "step": 38630 }, { "epoch": 0.05, "learning_rate": 4.7585e-05, "loss": 1.9398, "step": 38640 }, { "epoch": 0.05, "learning_rate": 4.7584375e-05, "loss": 2.1259, "step": 38650 }, { "epoch": 0.05, "learning_rate": 4.758375000000001e-05, "loss": 1.9324, "step": 38660 }, { "epoch": 0.05, "learning_rate": 4.7583125000000004e-05, "loss": 1.9431, "step": 38670 }, { "epoch": 0.05, "learning_rate": 4.75825e-05, "loss": 1.7239, "step": 38680 }, { "epoch": 0.05, "learning_rate": 4.7581875000000004e-05, "loss": 2.1235, "step": 38690 }, { "epoch": 0.05, "learning_rate": 4.758125000000001e-05, "loss": 2.0347, "step": 38700 }, { "epoch": 0.05, "learning_rate": 4.7580625e-05, "loss": 1.7219, "step": 38710 }, { "epoch": 0.05, "learning_rate": 4.758e-05, "loss": 1.9692, "step": 38720 }, { "epoch": 0.05, "learning_rate": 4.7579375000000006e-05, "loss": 1.9706, "step": 38730 }, { "epoch": 0.05, "learning_rate": 4.757875e-05, "loss": 1.8129, "step": 38740 }, { "epoch": 0.05, "learning_rate": 4.7578125e-05, "loss": 1.8428, "step": 38750 }, { "epoch": 0.05, "learning_rate": 4.75775e-05, "loss": 1.925, "step": 38760 }, { "epoch": 0.05, "learning_rate": 4.757687500000001e-05, "loss": 1.7075, "step": 38770 }, { "epoch": 0.05, "learning_rate": 4.757625e-05, "loss": 2.0858, "step": 38780 }, { "epoch": 0.05, "learning_rate": 4.7575625e-05, "loss": 1.7549, "step": 38790 }, { "epoch": 0.05, "learning_rate": 4.7575000000000004e-05, "loss": 2.0991, "step": 38800 }, { "epoch": 0.05, "learning_rate": 4.7574375e-05, "loss": 1.7445, "step": 38810 }, { "epoch": 0.05, "learning_rate": 4.757375e-05, "loss": 1.7088, "step": 38820 }, { "epoch": 0.05, "learning_rate": 4.7573125e-05, "loss": 1.6395, "step": 38830 }, { "epoch": 0.05, "learning_rate": 4.7572500000000006e-05, "loss": 1.9175, "step": 38840 }, { "epoch": 0.05, "learning_rate": 4.7571875e-05, "loss": 1.7331, "step": 38850 }, { "epoch": 0.05, "learning_rate": 4.757125e-05, "loss": 1.7758, "step": 38860 }, { "epoch": 0.05, "learning_rate": 4.7570625e-05, "loss": 2.0721, "step": 38870 }, { "epoch": 0.05, "learning_rate": 4.757e-05, "loss": 1.7196, "step": 38880 }, { "epoch": 0.05, "learning_rate": 4.7569375000000004e-05, "loss": 2.1702, "step": 38890 }, { "epoch": 0.05, "learning_rate": 4.756875e-05, "loss": 1.9188, "step": 38900 }, { "epoch": 0.05, "learning_rate": 4.7568125000000005e-05, "loss": 1.8599, "step": 38910 }, { "epoch": 0.05, "learning_rate": 4.75675e-05, "loss": 2.1392, "step": 38920 }, { "epoch": 0.05, "learning_rate": 4.7566875e-05, "loss": 1.7976, "step": 38930 }, { "epoch": 0.05, "learning_rate": 4.756625e-05, "loss": 1.774, "step": 38940 }, { "epoch": 0.05, "learning_rate": 4.7565625000000006e-05, "loss": 1.8073, "step": 38950 }, { "epoch": 0.05, "learning_rate": 4.7565e-05, "loss": 1.6487, "step": 38960 }, { "epoch": 0.05, "learning_rate": 4.7564375e-05, "loss": 1.6775, "step": 38970 }, { "epoch": 0.05, "learning_rate": 4.7563750000000003e-05, "loss": 1.904, "step": 38980 }, { "epoch": 0.05, "learning_rate": 4.7563125e-05, "loss": 1.7283, "step": 38990 }, { "epoch": 0.05, "learning_rate": 4.7562500000000004e-05, "loss": 1.7947, "step": 39000 }, { "epoch": 0.05, "eval_loss": 1.7819639444351196, "eval_runtime": 1290.5313, "eval_samples_per_second": 44.971, "eval_steps_per_second": 5.622, "step": 39000 }, { "epoch": 0.05, "learning_rate": 4.7561875e-05, "loss": 1.966, "step": 39010 }, { "epoch": 0.05, "learning_rate": 4.7561250000000005e-05, "loss": 1.7337, "step": 39020 }, { "epoch": 0.05, "learning_rate": 4.7560625e-05, "loss": 1.7603, "step": 39030 }, { "epoch": 0.05, "learning_rate": 4.7560000000000005e-05, "loss": 1.6526, "step": 39040 }, { "epoch": 0.05, "learning_rate": 4.7559375e-05, "loss": 1.9758, "step": 39050 }, { "epoch": 0.05, "learning_rate": 4.755875e-05, "loss": 1.9289, "step": 39060 }, { "epoch": 0.05, "learning_rate": 4.7558125e-05, "loss": 1.7916, "step": 39070 }, { "epoch": 0.05, "learning_rate": 4.755750000000001e-05, "loss": 1.9425, "step": 39080 }, { "epoch": 0.05, "learning_rate": 4.7556875000000004e-05, "loss": 1.9771, "step": 39090 }, { "epoch": 0.05, "learning_rate": 4.755625e-05, "loss": 1.9054, "step": 39100 }, { "epoch": 0.05, "learning_rate": 4.7555625000000004e-05, "loss": 1.9352, "step": 39110 }, { "epoch": 0.05, "learning_rate": 4.7555e-05, "loss": 1.8426, "step": 39120 }, { "epoch": 0.05, "learning_rate": 4.7554375e-05, "loss": 1.6436, "step": 39130 }, { "epoch": 0.05, "learning_rate": 4.755375e-05, "loss": 1.7216, "step": 39140 }, { "epoch": 0.05, "learning_rate": 4.7553125000000006e-05, "loss": 1.8475, "step": 39150 }, { "epoch": 0.05, "learning_rate": 4.75525e-05, "loss": 2.0502, "step": 39160 }, { "epoch": 0.05, "learning_rate": 4.7551875e-05, "loss": 1.9811, "step": 39170 }, { "epoch": 0.05, "learning_rate": 4.755125e-05, "loss": 1.8339, "step": 39180 }, { "epoch": 0.05, "learning_rate": 4.755062500000001e-05, "loss": 2.02, "step": 39190 }, { "epoch": 0.05, "learning_rate": 4.755e-05, "loss": 2.0822, "step": 39200 }, { "epoch": 0.05, "learning_rate": 4.7549375e-05, "loss": 1.6523, "step": 39210 }, { "epoch": 0.05, "learning_rate": 4.7548750000000004e-05, "loss": 1.8884, "step": 39220 }, { "epoch": 0.05, "learning_rate": 4.7548125e-05, "loss": 1.9505, "step": 39230 }, { "epoch": 0.05, "learning_rate": 4.75475e-05, "loss": 2.0082, "step": 39240 }, { "epoch": 0.05, "learning_rate": 4.7546875e-05, "loss": 1.7718, "step": 39250 }, { "epoch": 0.05, "learning_rate": 4.7546250000000006e-05, "loss": 1.9047, "step": 39260 }, { "epoch": 0.05, "learning_rate": 4.7545625e-05, "loss": 1.6863, "step": 39270 }, { "epoch": 0.05, "learning_rate": 4.7545e-05, "loss": 1.9439, "step": 39280 }, { "epoch": 0.05, "learning_rate": 4.7544375e-05, "loss": 1.8532, "step": 39290 }, { "epoch": 0.05, "learning_rate": 4.754375e-05, "loss": 1.9837, "step": 39300 }, { "epoch": 0.05, "learning_rate": 4.7543125000000004e-05, "loss": 2.1564, "step": 39310 }, { "epoch": 0.05, "learning_rate": 4.75425e-05, "loss": 1.9712, "step": 39320 }, { "epoch": 0.05, "learning_rate": 4.7541875000000005e-05, "loss": 1.8383, "step": 39330 }, { "epoch": 0.05, "learning_rate": 4.754125e-05, "loss": 1.8254, "step": 39340 }, { "epoch": 0.05, "learning_rate": 4.7540625e-05, "loss": 2.0118, "step": 39350 }, { "epoch": 0.05, "learning_rate": 4.754e-05, "loss": 1.6709, "step": 39360 }, { "epoch": 0.05, "learning_rate": 4.7539375000000006e-05, "loss": 2.0321, "step": 39370 }, { "epoch": 0.05, "learning_rate": 4.753875e-05, "loss": 1.8733, "step": 39380 }, { "epoch": 0.05, "learning_rate": 4.7538125e-05, "loss": 1.6651, "step": 39390 }, { "epoch": 0.05, "learning_rate": 4.7537500000000004e-05, "loss": 1.8941, "step": 39400 }, { "epoch": 0.05, "learning_rate": 4.7536875e-05, "loss": 1.6987, "step": 39410 }, { "epoch": 0.05, "learning_rate": 4.7536250000000004e-05, "loss": 1.9529, "step": 39420 }, { "epoch": 0.05, "learning_rate": 4.7535625e-05, "loss": 1.7105, "step": 39430 }, { "epoch": 0.05, "learning_rate": 4.7535000000000005e-05, "loss": 1.6233, "step": 39440 }, { "epoch": 0.05, "learning_rate": 4.7534375e-05, "loss": 1.7487, "step": 39450 }, { "epoch": 0.05, "learning_rate": 4.7533750000000005e-05, "loss": 1.8293, "step": 39460 }, { "epoch": 0.05, "learning_rate": 4.7533125e-05, "loss": 1.7073, "step": 39470 }, { "epoch": 0.05, "learning_rate": 4.75325e-05, "loss": 1.8988, "step": 39480 }, { "epoch": 0.05, "learning_rate": 4.7531875e-05, "loss": 1.6529, "step": 39490 }, { "epoch": 0.05, "learning_rate": 4.753125000000001e-05, "loss": 1.874, "step": 39500 }, { "epoch": 0.06, "learning_rate": 4.7530625000000004e-05, "loss": 2.0305, "step": 39510 }, { "epoch": 0.06, "learning_rate": 4.753e-05, "loss": 2.0514, "step": 39520 }, { "epoch": 0.06, "learning_rate": 4.7529375000000004e-05, "loss": 1.7246, "step": 39530 }, { "epoch": 0.06, "learning_rate": 4.752875e-05, "loss": 1.9092, "step": 39540 }, { "epoch": 0.06, "learning_rate": 4.7528125e-05, "loss": 2.0552, "step": 39550 }, { "epoch": 0.06, "learning_rate": 4.75275e-05, "loss": 1.9631, "step": 39560 }, { "epoch": 0.06, "learning_rate": 4.7526875000000006e-05, "loss": 1.7127, "step": 39570 }, { "epoch": 0.06, "learning_rate": 4.752625e-05, "loss": 1.9575, "step": 39580 }, { "epoch": 0.06, "learning_rate": 4.7525625e-05, "loss": 1.8563, "step": 39590 }, { "epoch": 0.06, "learning_rate": 4.7525e-05, "loss": 1.9701, "step": 39600 }, { "epoch": 0.06, "learning_rate": 4.752437500000001e-05, "loss": 2.0038, "step": 39610 }, { "epoch": 0.06, "learning_rate": 4.752375e-05, "loss": 1.8298, "step": 39620 }, { "epoch": 0.06, "learning_rate": 4.7523125e-05, "loss": 1.7138, "step": 39630 }, { "epoch": 0.06, "learning_rate": 4.7522500000000005e-05, "loss": 1.8898, "step": 39640 }, { "epoch": 0.06, "learning_rate": 4.7521875e-05, "loss": 1.9916, "step": 39650 }, { "epoch": 0.06, "learning_rate": 4.752125e-05, "loss": 1.8462, "step": 39660 }, { "epoch": 0.06, "learning_rate": 4.7520625e-05, "loss": 1.8433, "step": 39670 }, { "epoch": 0.06, "learning_rate": 4.7520000000000006e-05, "loss": 1.7469, "step": 39680 }, { "epoch": 0.06, "learning_rate": 4.7519375e-05, "loss": 1.8853, "step": 39690 }, { "epoch": 0.06, "learning_rate": 4.751875e-05, "loss": 1.9428, "step": 39700 }, { "epoch": 0.06, "learning_rate": 4.7518125000000003e-05, "loss": 1.8027, "step": 39710 }, { "epoch": 0.06, "learning_rate": 4.75175e-05, "loss": 2.0026, "step": 39720 }, { "epoch": 0.06, "learning_rate": 4.7516875e-05, "loss": 1.7866, "step": 39730 }, { "epoch": 0.06, "learning_rate": 4.751625e-05, "loss": 1.8319, "step": 39740 }, { "epoch": 0.06, "learning_rate": 4.7515625000000005e-05, "loss": 1.7326, "step": 39750 }, { "epoch": 0.06, "learning_rate": 4.7515e-05, "loss": 1.7863, "step": 39760 }, { "epoch": 0.06, "learning_rate": 4.7514375e-05, "loss": 1.8585, "step": 39770 }, { "epoch": 0.06, "learning_rate": 4.751375e-05, "loss": 1.737, "step": 39780 }, { "epoch": 0.06, "learning_rate": 4.7513125000000006e-05, "loss": 1.9222, "step": 39790 }, { "epoch": 0.06, "learning_rate": 4.75125e-05, "loss": 1.8813, "step": 39800 }, { "epoch": 0.06, "learning_rate": 4.7511875e-05, "loss": 1.8389, "step": 39810 }, { "epoch": 0.06, "learning_rate": 4.7511250000000004e-05, "loss": 1.9041, "step": 39820 }, { "epoch": 0.06, "learning_rate": 4.7510625e-05, "loss": 1.8281, "step": 39830 }, { "epoch": 0.06, "learning_rate": 4.7510000000000004e-05, "loss": 1.8486, "step": 39840 }, { "epoch": 0.06, "learning_rate": 4.7509375e-05, "loss": 1.7455, "step": 39850 }, { "epoch": 0.06, "learning_rate": 4.7508750000000005e-05, "loss": 1.7348, "step": 39860 }, { "epoch": 0.06, "learning_rate": 4.7508125e-05, "loss": 1.7541, "step": 39870 }, { "epoch": 0.06, "learning_rate": 4.7507500000000006e-05, "loss": 1.8452, "step": 39880 }, { "epoch": 0.06, "learning_rate": 4.7506875e-05, "loss": 1.7422, "step": 39890 }, { "epoch": 0.06, "learning_rate": 4.750625e-05, "loss": 1.8202, "step": 39900 }, { "epoch": 0.06, "learning_rate": 4.7505625e-05, "loss": 1.7824, "step": 39910 }, { "epoch": 0.06, "learning_rate": 4.7505e-05, "loss": 1.6884, "step": 39920 }, { "epoch": 0.06, "learning_rate": 4.7504375000000004e-05, "loss": 1.7898, "step": 39930 }, { "epoch": 0.06, "learning_rate": 4.750375e-05, "loss": 1.9212, "step": 39940 }, { "epoch": 0.06, "learning_rate": 4.7503125000000004e-05, "loss": 1.7576, "step": 39950 }, { "epoch": 0.06, "learning_rate": 4.75025e-05, "loss": 1.949, "step": 39960 }, { "epoch": 0.06, "learning_rate": 4.7501875e-05, "loss": 1.9134, "step": 39970 }, { "epoch": 0.06, "learning_rate": 4.750125e-05, "loss": 2.0797, "step": 39980 }, { "epoch": 0.06, "learning_rate": 4.7500625000000006e-05, "loss": 1.7467, "step": 39990 }, { "epoch": 0.06, "learning_rate": 4.75e-05, "loss": 1.6799, "step": 40000 }, { "epoch": 0.06, "eval_loss": 1.7663819789886475, "eval_runtime": 1290.4549, "eval_samples_per_second": 44.973, "eval_steps_per_second": 5.622, "step": 40000 }, { "epoch": 0.06, "learning_rate": 4.7499375e-05, "loss": 1.8614, "step": 40010 }, { "epoch": 0.06, "learning_rate": 4.749875e-05, "loss": 1.6513, "step": 40020 }, { "epoch": 0.06, "learning_rate": 4.749812500000001e-05, "loss": 1.9872, "step": 40030 }, { "epoch": 0.06, "learning_rate": 4.74975e-05, "loss": 1.8434, "step": 40040 }, { "epoch": 0.06, "learning_rate": 4.7496875e-05, "loss": 1.7036, "step": 40050 }, { "epoch": 0.06, "learning_rate": 4.7496250000000005e-05, "loss": 2.0236, "step": 40060 }, { "epoch": 0.06, "learning_rate": 4.7495625e-05, "loss": 1.718, "step": 40070 }, { "epoch": 0.06, "learning_rate": 4.7495e-05, "loss": 1.7367, "step": 40080 }, { "epoch": 0.06, "learning_rate": 4.7494375e-05, "loss": 2.1927, "step": 40090 }, { "epoch": 0.06, "learning_rate": 4.7493750000000006e-05, "loss": 1.7274, "step": 40100 }, { "epoch": 0.06, "learning_rate": 4.7493125e-05, "loss": 1.9842, "step": 40110 }, { "epoch": 0.06, "learning_rate": 4.74925e-05, "loss": 1.633, "step": 40120 }, { "epoch": 0.06, "learning_rate": 4.7491875000000003e-05, "loss": 1.8612, "step": 40130 }, { "epoch": 0.06, "learning_rate": 4.749125000000001e-05, "loss": 1.9322, "step": 40140 }, { "epoch": 0.06, "learning_rate": 4.7490625e-05, "loss": 1.9672, "step": 40150 }, { "epoch": 0.06, "learning_rate": 4.749e-05, "loss": 1.8579, "step": 40160 }, { "epoch": 0.06, "learning_rate": 4.7489375000000005e-05, "loss": 1.8101, "step": 40170 }, { "epoch": 0.06, "learning_rate": 4.748875e-05, "loss": 1.7613, "step": 40180 }, { "epoch": 0.06, "learning_rate": 4.7488125e-05, "loss": 1.7347, "step": 40190 }, { "epoch": 0.06, "learning_rate": 4.74875e-05, "loss": 1.8793, "step": 40200 }, { "epoch": 0.06, "learning_rate": 4.7486875000000006e-05, "loss": 1.695, "step": 40210 }, { "epoch": 0.06, "learning_rate": 4.748625e-05, "loss": 1.895, "step": 40220 }, { "epoch": 0.06, "learning_rate": 4.7485625e-05, "loss": 1.715, "step": 40230 }, { "epoch": 0.06, "learning_rate": 4.7485000000000004e-05, "loss": 1.7217, "step": 40240 }, { "epoch": 0.06, "learning_rate": 4.7484375e-05, "loss": 2.0232, "step": 40250 }, { "epoch": 0.06, "learning_rate": 4.7483750000000004e-05, "loss": 1.7936, "step": 40260 }, { "epoch": 0.06, "learning_rate": 4.7483125e-05, "loss": 1.7299, "step": 40270 }, { "epoch": 0.06, "learning_rate": 4.7482500000000005e-05, "loss": 1.8462, "step": 40280 }, { "epoch": 0.06, "learning_rate": 4.7481875e-05, "loss": 1.9007, "step": 40290 }, { "epoch": 0.06, "learning_rate": 4.7481250000000006e-05, "loss": 1.9367, "step": 40300 }, { "epoch": 0.06, "learning_rate": 4.7480625e-05, "loss": 1.8836, "step": 40310 }, { "epoch": 0.06, "learning_rate": 4.748e-05, "loss": 1.7913, "step": 40320 }, { "epoch": 0.06, "learning_rate": 4.7479375e-05, "loss": 2.0071, "step": 40330 }, { "epoch": 0.06, "learning_rate": 4.747875e-05, "loss": 1.8969, "step": 40340 }, { "epoch": 0.06, "learning_rate": 4.7478125000000004e-05, "loss": 1.9244, "step": 40350 }, { "epoch": 0.06, "learning_rate": 4.74775e-05, "loss": 1.8021, "step": 40360 }, { "epoch": 0.06, "learning_rate": 4.7476875000000004e-05, "loss": 1.7381, "step": 40370 }, { "epoch": 0.06, "learning_rate": 4.747625e-05, "loss": 2.015, "step": 40380 }, { "epoch": 0.06, "learning_rate": 4.7475625e-05, "loss": 1.5689, "step": 40390 }, { "epoch": 0.06, "learning_rate": 4.7475e-05, "loss": 1.6576, "step": 40400 }, { "epoch": 0.06, "learning_rate": 4.7474375000000006e-05, "loss": 1.9053, "step": 40410 }, { "epoch": 0.06, "learning_rate": 4.747375e-05, "loss": 1.7028, "step": 40420 }, { "epoch": 0.06, "learning_rate": 4.7473125e-05, "loss": 1.87, "step": 40430 }, { "epoch": 0.06, "learning_rate": 4.74725e-05, "loss": 1.986, "step": 40440 }, { "epoch": 0.06, "learning_rate": 4.747187500000001e-05, "loss": 1.7129, "step": 40450 }, { "epoch": 0.06, "learning_rate": 4.747125e-05, "loss": 1.6178, "step": 40460 }, { "epoch": 0.06, "learning_rate": 4.7470625e-05, "loss": 1.8038, "step": 40470 }, { "epoch": 0.06, "learning_rate": 4.7470000000000005e-05, "loss": 1.8384, "step": 40480 }, { "epoch": 0.06, "learning_rate": 4.7469375e-05, "loss": 1.8382, "step": 40490 }, { "epoch": 0.06, "learning_rate": 4.746875e-05, "loss": 1.9659, "step": 40500 }, { "epoch": 0.06, "learning_rate": 4.7468125e-05, "loss": 1.8406, "step": 40510 }, { "epoch": 0.06, "learning_rate": 4.7467500000000006e-05, "loss": 1.8499, "step": 40520 }, { "epoch": 0.06, "learning_rate": 4.7466875e-05, "loss": 1.8015, "step": 40530 }, { "epoch": 0.06, "learning_rate": 4.746625e-05, "loss": 1.8779, "step": 40540 }, { "epoch": 0.06, "learning_rate": 4.7465625000000003e-05, "loss": 1.6974, "step": 40550 }, { "epoch": 0.06, "learning_rate": 4.746500000000001e-05, "loss": 1.8121, "step": 40560 }, { "epoch": 0.06, "learning_rate": 4.7464375e-05, "loss": 1.8442, "step": 40570 }, { "epoch": 0.06, "learning_rate": 4.746375e-05, "loss": 1.8524, "step": 40580 }, { "epoch": 0.06, "learning_rate": 4.7463125000000005e-05, "loss": 2.1738, "step": 40590 }, { "epoch": 0.06, "learning_rate": 4.74625e-05, "loss": 1.8909, "step": 40600 }, { "epoch": 0.06, "learning_rate": 4.7461875e-05, "loss": 1.8698, "step": 40610 }, { "epoch": 0.06, "learning_rate": 4.746125e-05, "loss": 1.707, "step": 40620 }, { "epoch": 0.06, "learning_rate": 4.7460625000000006e-05, "loss": 1.9757, "step": 40630 }, { "epoch": 0.06, "learning_rate": 4.746e-05, "loss": 2.0009, "step": 40640 }, { "epoch": 0.06, "learning_rate": 4.7459375e-05, "loss": 1.9788, "step": 40650 }, { "epoch": 0.06, "learning_rate": 4.7458750000000004e-05, "loss": 1.7586, "step": 40660 }, { "epoch": 0.06, "learning_rate": 4.7458125e-05, "loss": 1.843, "step": 40670 }, { "epoch": 0.06, "learning_rate": 4.7457500000000004e-05, "loss": 2.0364, "step": 40680 }, { "epoch": 0.06, "learning_rate": 4.7456875e-05, "loss": 2.1099, "step": 40690 }, { "epoch": 0.06, "learning_rate": 4.7456250000000005e-05, "loss": 1.727, "step": 40700 }, { "epoch": 0.06, "learning_rate": 4.7455625e-05, "loss": 1.7767, "step": 40710 }, { "epoch": 0.06, "learning_rate": 4.7455000000000006e-05, "loss": 2.005, "step": 40720 }, { "epoch": 0.06, "learning_rate": 4.7454375e-05, "loss": 1.8663, "step": 40730 }, { "epoch": 0.06, "learning_rate": 4.745375e-05, "loss": 1.979, "step": 40740 }, { "epoch": 0.06, "learning_rate": 4.7453125e-05, "loss": 2.028, "step": 40750 }, { "epoch": 0.06, "learning_rate": 4.74525e-05, "loss": 1.9894, "step": 40760 }, { "epoch": 0.06, "learning_rate": 4.7451875000000004e-05, "loss": 1.88, "step": 40770 }, { "epoch": 0.06, "learning_rate": 4.745125e-05, "loss": 1.6251, "step": 40780 }, { "epoch": 0.06, "learning_rate": 4.7450625000000004e-05, "loss": 1.8684, "step": 40790 }, { "epoch": 0.06, "learning_rate": 4.745e-05, "loss": 1.9218, "step": 40800 }, { "epoch": 0.06, "learning_rate": 4.7449375e-05, "loss": 1.8795, "step": 40810 }, { "epoch": 0.06, "learning_rate": 4.744875e-05, "loss": 1.8399, "step": 40820 }, { "epoch": 0.06, "learning_rate": 4.7448125000000006e-05, "loss": 1.6334, "step": 40830 }, { "epoch": 0.06, "learning_rate": 4.74475e-05, "loss": 1.8289, "step": 40840 }, { "epoch": 0.06, "learning_rate": 4.7446875e-05, "loss": 1.9678, "step": 40850 }, { "epoch": 0.06, "learning_rate": 4.744625e-05, "loss": 1.6384, "step": 40860 }, { "epoch": 0.06, "learning_rate": 4.744562500000001e-05, "loss": 1.9367, "step": 40870 }, { "epoch": 0.06, "learning_rate": 4.7445e-05, "loss": 1.6028, "step": 40880 }, { "epoch": 0.06, "learning_rate": 4.7444375e-05, "loss": 1.9656, "step": 40890 }, { "epoch": 0.06, "learning_rate": 4.7443750000000005e-05, "loss": 1.8465, "step": 40900 }, { "epoch": 0.06, "learning_rate": 4.7443125e-05, "loss": 1.9843, "step": 40910 }, { "epoch": 0.06, "learning_rate": 4.74425e-05, "loss": 1.9459, "step": 40920 }, { "epoch": 0.06, "learning_rate": 4.7441875e-05, "loss": 1.8457, "step": 40930 }, { "epoch": 0.06, "learning_rate": 4.7441250000000006e-05, "loss": 1.7285, "step": 40940 }, { "epoch": 0.06, "learning_rate": 4.7440625e-05, "loss": 1.8194, "step": 40950 }, { "epoch": 0.06, "learning_rate": 4.744e-05, "loss": 1.7946, "step": 40960 }, { "epoch": 0.06, "learning_rate": 4.7439375000000004e-05, "loss": 1.868, "step": 40970 }, { "epoch": 0.06, "learning_rate": 4.743875000000001e-05, "loss": 1.7323, "step": 40980 }, { "epoch": 0.06, "learning_rate": 4.7438125e-05, "loss": 1.9237, "step": 40990 }, { "epoch": 0.06, "learning_rate": 4.74375e-05, "loss": 2.0093, "step": 41000 }, { "epoch": 0.06, "eval_loss": 1.750366449356079, "eval_runtime": 1291.221, "eval_samples_per_second": 44.947, "eval_steps_per_second": 5.619, "step": 41000 }, { "epoch": 0.06, "learning_rate": 4.7436875000000005e-05, "loss": 1.892, "step": 41010 }, { "epoch": 0.06, "learning_rate": 4.743625e-05, "loss": 1.8785, "step": 41020 }, { "epoch": 0.06, "learning_rate": 4.7435625e-05, "loss": 1.5227, "step": 41030 }, { "epoch": 0.06, "learning_rate": 4.7435e-05, "loss": 1.9773, "step": 41040 }, { "epoch": 0.06, "learning_rate": 4.7434375000000006e-05, "loss": 1.7073, "step": 41050 }, { "epoch": 0.06, "learning_rate": 4.743375e-05, "loss": 1.7092, "step": 41060 }, { "epoch": 0.06, "learning_rate": 4.7433125e-05, "loss": 1.8297, "step": 41070 }, { "epoch": 0.06, "learning_rate": 4.7432500000000004e-05, "loss": 1.7396, "step": 41080 }, { "epoch": 0.06, "learning_rate": 4.7431875e-05, "loss": 1.8093, "step": 41090 }, { "epoch": 0.06, "learning_rate": 4.7431250000000004e-05, "loss": 1.7229, "step": 41100 }, { "epoch": 0.06, "learning_rate": 4.7430625e-05, "loss": 1.922, "step": 41110 }, { "epoch": 0.06, "learning_rate": 4.7430000000000005e-05, "loss": 1.7895, "step": 41120 }, { "epoch": 0.06, "learning_rate": 4.7429375e-05, "loss": 1.8217, "step": 41130 }, { "epoch": 0.06, "learning_rate": 4.742875e-05, "loss": 1.9071, "step": 41140 }, { "epoch": 0.06, "learning_rate": 4.7428125e-05, "loss": 1.6832, "step": 41150 }, { "epoch": 0.06, "learning_rate": 4.74275e-05, "loss": 1.6396, "step": 41160 }, { "epoch": 0.06, "learning_rate": 4.7426875e-05, "loss": 1.8101, "step": 41170 }, { "epoch": 0.06, "learning_rate": 4.742625e-05, "loss": 2.1735, "step": 41180 }, { "epoch": 0.06, "learning_rate": 4.7425625000000004e-05, "loss": 2.1129, "step": 41190 }, { "epoch": 0.06, "learning_rate": 4.7425e-05, "loss": 1.9592, "step": 41200 }, { "epoch": 0.06, "learning_rate": 4.7424375000000005e-05, "loss": 1.9685, "step": 41210 }, { "epoch": 0.06, "learning_rate": 4.742375e-05, "loss": 1.7483, "step": 41220 }, { "epoch": 0.06, "learning_rate": 4.7423125e-05, "loss": 1.8423, "step": 41230 }, { "epoch": 0.06, "learning_rate": 4.74225e-05, "loss": 1.715, "step": 41240 }, { "epoch": 0.06, "learning_rate": 4.7421875000000006e-05, "loss": 1.8776, "step": 41250 }, { "epoch": 0.06, "learning_rate": 4.742125e-05, "loss": 1.7386, "step": 41260 }, { "epoch": 0.06, "learning_rate": 4.7420625e-05, "loss": 1.8495, "step": 41270 }, { "epoch": 0.06, "learning_rate": 4.742e-05, "loss": 1.8096, "step": 41280 }, { "epoch": 0.06, "learning_rate": 4.741937500000001e-05, "loss": 1.6582, "step": 41290 }, { "epoch": 0.06, "learning_rate": 4.7418750000000004e-05, "loss": 1.9139, "step": 41300 }, { "epoch": 0.06, "learning_rate": 4.7418125e-05, "loss": 1.9468, "step": 41310 }, { "epoch": 0.06, "learning_rate": 4.7417500000000005e-05, "loss": 1.7785, "step": 41320 }, { "epoch": 0.06, "learning_rate": 4.741687500000001e-05, "loss": 1.5241, "step": 41330 }, { "epoch": 0.06, "learning_rate": 4.741625e-05, "loss": 1.8533, "step": 41340 }, { "epoch": 0.06, "learning_rate": 4.7415625e-05, "loss": 1.901, "step": 41350 }, { "epoch": 0.06, "learning_rate": 4.7415000000000006e-05, "loss": 1.9809, "step": 41360 }, { "epoch": 0.06, "learning_rate": 4.7414375e-05, "loss": 1.6525, "step": 41370 }, { "epoch": 0.06, "learning_rate": 4.741375e-05, "loss": 1.7763, "step": 41380 }, { "epoch": 0.06, "learning_rate": 4.7413125000000004e-05, "loss": 1.9575, "step": 41390 }, { "epoch": 0.06, "learning_rate": 4.741250000000001e-05, "loss": 1.7569, "step": 41400 }, { "epoch": 0.06, "learning_rate": 4.7411875e-05, "loss": 1.6209, "step": 41410 }, { "epoch": 0.06, "learning_rate": 4.741125e-05, "loss": 1.8091, "step": 41420 }, { "epoch": 0.06, "learning_rate": 4.7410625000000005e-05, "loss": 1.9167, "step": 41430 }, { "epoch": 0.06, "learning_rate": 4.741e-05, "loss": 1.8726, "step": 41440 }, { "epoch": 0.06, "learning_rate": 4.7409375e-05, "loss": 1.8503, "step": 41450 }, { "epoch": 0.06, "learning_rate": 4.740875e-05, "loss": 1.8196, "step": 41460 }, { "epoch": 0.06, "learning_rate": 4.7408125000000006e-05, "loss": 1.9426, "step": 41470 }, { "epoch": 0.06, "learning_rate": 4.74075e-05, "loss": 1.7305, "step": 41480 }, { "epoch": 0.06, "learning_rate": 4.7406875e-05, "loss": 1.5896, "step": 41490 }, { "epoch": 0.06, "learning_rate": 4.7406250000000004e-05, "loss": 1.7932, "step": 41500 }, { "epoch": 0.06, "learning_rate": 4.7405625e-05, "loss": 1.8214, "step": 41510 }, { "epoch": 0.06, "learning_rate": 4.7405000000000004e-05, "loss": 1.8779, "step": 41520 }, { "epoch": 0.06, "learning_rate": 4.7404375e-05, "loss": 1.8739, "step": 41530 }, { "epoch": 0.06, "learning_rate": 4.7403750000000005e-05, "loss": 1.8813, "step": 41540 }, { "epoch": 0.06, "learning_rate": 4.7403125e-05, "loss": 1.6647, "step": 41550 }, { "epoch": 0.06, "learning_rate": 4.74025e-05, "loss": 2.1551, "step": 41560 }, { "epoch": 0.06, "learning_rate": 4.7401875e-05, "loss": 1.7142, "step": 41570 }, { "epoch": 0.06, "learning_rate": 4.740125e-05, "loss": 1.7725, "step": 41580 }, { "epoch": 0.06, "learning_rate": 4.7400625e-05, "loss": 1.5932, "step": 41590 }, { "epoch": 0.06, "learning_rate": 4.74e-05, "loss": 1.8546, "step": 41600 }, { "epoch": 0.06, "learning_rate": 4.7399375000000004e-05, "loss": 1.6365, "step": 41610 }, { "epoch": 0.06, "learning_rate": 4.739875e-05, "loss": 1.8703, "step": 41620 }, { "epoch": 0.06, "learning_rate": 4.7398125000000005e-05, "loss": 1.9789, "step": 41630 }, { "epoch": 0.06, "learning_rate": 4.73975e-05, "loss": 1.5946, "step": 41640 }, { "epoch": 0.06, "learning_rate": 4.7396875e-05, "loss": 2.1466, "step": 41650 }, { "epoch": 0.06, "learning_rate": 4.739625e-05, "loss": 1.8644, "step": 41660 }, { "epoch": 0.06, "learning_rate": 4.7395625000000006e-05, "loss": 1.8227, "step": 41670 }, { "epoch": 0.06, "learning_rate": 4.7395e-05, "loss": 1.8206, "step": 41680 }, { "epoch": 0.06, "learning_rate": 4.7394375e-05, "loss": 1.6507, "step": 41690 }, { "epoch": 0.06, "learning_rate": 4.7393750000000003e-05, "loss": 1.8826, "step": 41700 }, { "epoch": 0.06, "learning_rate": 4.739312500000001e-05, "loss": 1.7979, "step": 41710 }, { "epoch": 0.06, "learning_rate": 4.7392500000000004e-05, "loss": 1.7188, "step": 41720 }, { "epoch": 0.06, "learning_rate": 4.7391875e-05, "loss": 1.8157, "step": 41730 }, { "epoch": 0.06, "learning_rate": 4.7391250000000005e-05, "loss": 2.0555, "step": 41740 }, { "epoch": 0.06, "learning_rate": 4.7390625e-05, "loss": 1.877, "step": 41750 }, { "epoch": 0.06, "learning_rate": 4.739e-05, "loss": 1.9434, "step": 41760 }, { "epoch": 0.06, "learning_rate": 4.7389375e-05, "loss": 1.7599, "step": 41770 }, { "epoch": 0.06, "learning_rate": 4.7388750000000006e-05, "loss": 1.7289, "step": 41780 }, { "epoch": 0.06, "learning_rate": 4.7388125e-05, "loss": 1.998, "step": 41790 }, { "epoch": 0.06, "learning_rate": 4.73875e-05, "loss": 1.7548, "step": 41800 }, { "epoch": 0.06, "learning_rate": 4.7386875000000004e-05, "loss": 1.7636, "step": 41810 }, { "epoch": 0.06, "learning_rate": 4.738625000000001e-05, "loss": 1.7804, "step": 41820 }, { "epoch": 0.06, "learning_rate": 4.7385625e-05, "loss": 1.7897, "step": 41830 }, { "epoch": 0.06, "learning_rate": 4.7385e-05, "loss": 1.9118, "step": 41840 }, { "epoch": 0.06, "learning_rate": 4.7384375000000005e-05, "loss": 2.045, "step": 41850 }, { "epoch": 0.06, "learning_rate": 4.738375e-05, "loss": 1.8007, "step": 41860 }, { "epoch": 0.06, "learning_rate": 4.7383125e-05, "loss": 1.8342, "step": 41870 }, { "epoch": 0.06, "learning_rate": 4.73825e-05, "loss": 1.7599, "step": 41880 }, { "epoch": 0.06, "learning_rate": 4.7381875000000006e-05, "loss": 1.703, "step": 41890 }, { "epoch": 0.06, "learning_rate": 4.738125e-05, "loss": 1.8652, "step": 41900 }, { "epoch": 0.06, "learning_rate": 4.7380625e-05, "loss": 1.7088, "step": 41910 }, { "epoch": 0.06, "learning_rate": 4.7380000000000004e-05, "loss": 1.6325, "step": 41920 }, { "epoch": 0.06, "learning_rate": 4.7379375e-05, "loss": 1.7416, "step": 41930 }, { "epoch": 0.06, "learning_rate": 4.737875e-05, "loss": 1.8406, "step": 41940 }, { "epoch": 0.06, "learning_rate": 4.7378125e-05, "loss": 1.7935, "step": 41950 }, { "epoch": 0.06, "learning_rate": 4.7377500000000005e-05, "loss": 1.8347, "step": 41960 }, { "epoch": 0.06, "learning_rate": 4.7376875e-05, "loss": 1.9084, "step": 41970 }, { "epoch": 0.06, "learning_rate": 4.737625e-05, "loss": 1.8318, "step": 41980 }, { "epoch": 0.06, "learning_rate": 4.7375625e-05, "loss": 1.692, "step": 41990 }, { "epoch": 0.06, "learning_rate": 4.7375e-05, "loss": 1.736, "step": 42000 }, { "epoch": 0.06, "eval_loss": 1.7409533262252808, "eval_runtime": 1292.0194, "eval_samples_per_second": 44.919, "eval_steps_per_second": 5.615, "step": 42000 }, { "epoch": 0.06, "learning_rate": 4.7374375e-05, "loss": 1.8084, "step": 42010 }, { "epoch": 0.06, "learning_rate": 4.737375e-05, "loss": 1.6139, "step": 42020 }, { "epoch": 0.06, "learning_rate": 4.7373125000000004e-05, "loss": 1.896, "step": 42030 }, { "epoch": 0.06, "learning_rate": 4.73725e-05, "loss": 1.7725, "step": 42040 }, { "epoch": 0.06, "learning_rate": 4.7371875000000005e-05, "loss": 1.8969, "step": 42050 }, { "epoch": 0.06, "learning_rate": 4.737125e-05, "loss": 1.7035, "step": 42060 }, { "epoch": 0.06, "learning_rate": 4.7370625e-05, "loss": 1.9471, "step": 42070 }, { "epoch": 0.06, "learning_rate": 4.737e-05, "loss": 1.7803, "step": 42080 }, { "epoch": 0.06, "learning_rate": 4.7369375000000006e-05, "loss": 1.9218, "step": 42090 }, { "epoch": 0.06, "learning_rate": 4.736875e-05, "loss": 1.7082, "step": 42100 }, { "epoch": 0.06, "learning_rate": 4.7368125e-05, "loss": 1.7725, "step": 42110 }, { "epoch": 0.06, "learning_rate": 4.7367500000000003e-05, "loss": 1.8865, "step": 42120 }, { "epoch": 0.06, "learning_rate": 4.736687500000001e-05, "loss": 1.7743, "step": 42130 }, { "epoch": 0.06, "learning_rate": 4.7366250000000004e-05, "loss": 1.8025, "step": 42140 }, { "epoch": 0.06, "learning_rate": 4.7365625e-05, "loss": 1.9809, "step": 42150 }, { "epoch": 0.06, "learning_rate": 4.7365000000000005e-05, "loss": 1.9816, "step": 42160 }, { "epoch": 0.06, "learning_rate": 4.7364375e-05, "loss": 1.6758, "step": 42170 }, { "epoch": 0.06, "learning_rate": 4.736375e-05, "loss": 1.9266, "step": 42180 }, { "epoch": 0.06, "learning_rate": 4.7363125e-05, "loss": 1.6891, "step": 42190 }, { "epoch": 0.06, "learning_rate": 4.7362500000000006e-05, "loss": 1.7985, "step": 42200 }, { "epoch": 0.06, "learning_rate": 4.7361875e-05, "loss": 1.4937, "step": 42210 }, { "epoch": 0.06, "learning_rate": 4.736125e-05, "loss": 1.6772, "step": 42220 }, { "epoch": 0.06, "learning_rate": 4.7360625000000004e-05, "loss": 1.804, "step": 42230 }, { "epoch": 0.06, "learning_rate": 4.736000000000001e-05, "loss": 1.6445, "step": 42240 }, { "epoch": 0.06, "learning_rate": 4.7359375e-05, "loss": 1.6791, "step": 42250 }, { "epoch": 0.06, "learning_rate": 4.735875e-05, "loss": 1.5834, "step": 42260 }, { "epoch": 0.06, "learning_rate": 4.7358125000000005e-05, "loss": 1.7952, "step": 42270 }, { "epoch": 0.06, "learning_rate": 4.73575e-05, "loss": 1.7682, "step": 42280 }, { "epoch": 0.06, "learning_rate": 4.7356875e-05, "loss": 1.9422, "step": 42290 }, { "epoch": 0.06, "learning_rate": 4.735625e-05, "loss": 1.9386, "step": 42300 }, { "epoch": 0.06, "learning_rate": 4.7355625000000006e-05, "loss": 1.9103, "step": 42310 }, { "epoch": 0.06, "learning_rate": 4.7355e-05, "loss": 1.8514, "step": 42320 }, { "epoch": 0.06, "learning_rate": 4.7354375e-05, "loss": 1.7891, "step": 42330 }, { "epoch": 0.06, "learning_rate": 4.7353750000000004e-05, "loss": 1.8582, "step": 42340 }, { "epoch": 0.06, "learning_rate": 4.7353125e-05, "loss": 1.93, "step": 42350 }, { "epoch": 0.06, "learning_rate": 4.73525e-05, "loss": 1.9614, "step": 42360 }, { "epoch": 0.06, "learning_rate": 4.7351875e-05, "loss": 1.7835, "step": 42370 }, { "epoch": 0.06, "learning_rate": 4.7351250000000005e-05, "loss": 1.8575, "step": 42380 }, { "epoch": 0.06, "learning_rate": 4.7350625e-05, "loss": 1.8859, "step": 42390 }, { "epoch": 0.06, "learning_rate": 4.735e-05, "loss": 1.8596, "step": 42400 }, { "epoch": 0.06, "learning_rate": 4.7349375e-05, "loss": 1.9759, "step": 42410 }, { "epoch": 0.06, "learning_rate": 4.734875e-05, "loss": 1.6899, "step": 42420 }, { "epoch": 0.06, "learning_rate": 4.7348125e-05, "loss": 1.992, "step": 42430 }, { "epoch": 0.06, "learning_rate": 4.73475e-05, "loss": 1.7573, "step": 42440 }, { "epoch": 0.06, "learning_rate": 4.7346875000000004e-05, "loss": 1.8078, "step": 42450 }, { "epoch": 0.06, "learning_rate": 4.734625e-05, "loss": 1.6626, "step": 42460 }, { "epoch": 0.06, "learning_rate": 4.7345625000000005e-05, "loss": 1.7624, "step": 42470 }, { "epoch": 0.06, "learning_rate": 4.7345e-05, "loss": 2.0004, "step": 42480 }, { "epoch": 0.06, "learning_rate": 4.7344375000000005e-05, "loss": 1.7809, "step": 42490 }, { "epoch": 0.06, "learning_rate": 4.734375e-05, "loss": 1.7256, "step": 42500 }, { "epoch": 0.06, "learning_rate": 4.7343125000000006e-05, "loss": 2.034, "step": 42510 }, { "epoch": 0.06, "learning_rate": 4.73425e-05, "loss": 1.8678, "step": 42520 }, { "epoch": 0.06, "learning_rate": 4.7341875e-05, "loss": 1.8055, "step": 42530 }, { "epoch": 0.06, "learning_rate": 4.7341250000000003e-05, "loss": 1.7104, "step": 42540 }, { "epoch": 0.06, "learning_rate": 4.7340625e-05, "loss": 1.7407, "step": 42550 }, { "epoch": 0.06, "learning_rate": 4.7340000000000004e-05, "loss": 1.7692, "step": 42560 }, { "epoch": 0.06, "learning_rate": 4.7339375e-05, "loss": 1.8927, "step": 42570 }, { "epoch": 0.06, "learning_rate": 4.7338750000000005e-05, "loss": 1.7943, "step": 42580 }, { "epoch": 0.06, "learning_rate": 4.7338125e-05, "loss": 1.8255, "step": 42590 }, { "epoch": 0.06, "learning_rate": 4.73375e-05, "loss": 1.9352, "step": 42600 }, { "epoch": 0.06, "learning_rate": 4.7336875e-05, "loss": 1.8487, "step": 42610 }, { "epoch": 0.06, "learning_rate": 4.7336250000000006e-05, "loss": 1.9709, "step": 42620 }, { "epoch": 0.06, "learning_rate": 4.7335625e-05, "loss": 1.535, "step": 42630 }, { "epoch": 0.06, "learning_rate": 4.7335e-05, "loss": 1.8657, "step": 42640 }, { "epoch": 0.06, "learning_rate": 4.7334375000000004e-05, "loss": 1.853, "step": 42650 }, { "epoch": 0.06, "learning_rate": 4.733375000000001e-05, "loss": 1.7979, "step": 42660 }, { "epoch": 0.06, "learning_rate": 4.7333125e-05, "loss": 1.8783, "step": 42670 }, { "epoch": 0.06, "learning_rate": 4.73325e-05, "loss": 1.7149, "step": 42680 }, { "epoch": 0.06, "learning_rate": 4.7331875000000005e-05, "loss": 2.0726, "step": 42690 }, { "epoch": 0.06, "learning_rate": 4.733125e-05, "loss": 1.9401, "step": 42700 }, { "epoch": 0.06, "learning_rate": 4.7330625e-05, "loss": 1.7612, "step": 42710 }, { "epoch": 0.06, "learning_rate": 4.733e-05, "loss": 1.9319, "step": 42720 }, { "epoch": 0.06, "learning_rate": 4.7329375000000006e-05, "loss": 1.7144, "step": 42730 }, { "epoch": 0.06, "learning_rate": 4.732875e-05, "loss": 1.7888, "step": 42740 }, { "epoch": 0.06, "learning_rate": 4.7328125e-05, "loss": 1.9734, "step": 42750 }, { "epoch": 0.06, "learning_rate": 4.7327500000000004e-05, "loss": 1.8592, "step": 42760 }, { "epoch": 0.06, "learning_rate": 4.7326875e-05, "loss": 1.6255, "step": 42770 }, { "epoch": 0.06, "learning_rate": 4.732625e-05, "loss": 1.9404, "step": 42780 }, { "epoch": 0.06, "learning_rate": 4.7325625e-05, "loss": 1.8102, "step": 42790 }, { "epoch": 0.06, "learning_rate": 4.7325000000000005e-05, "loss": 1.8315, "step": 42800 }, { "epoch": 0.06, "learning_rate": 4.7324375e-05, "loss": 1.8428, "step": 42810 }, { "epoch": 0.06, "learning_rate": 4.732375e-05, "loss": 1.7412, "step": 42820 }, { "epoch": 0.06, "learning_rate": 4.7323125e-05, "loss": 1.7948, "step": 42830 }, { "epoch": 0.06, "learning_rate": 4.73225e-05, "loss": 1.932, "step": 42840 }, { "epoch": 0.06, "learning_rate": 4.7321875e-05, "loss": 1.827, "step": 42850 }, { "epoch": 0.06, "learning_rate": 4.732125e-05, "loss": 1.7592, "step": 42860 }, { "epoch": 0.06, "learning_rate": 4.7320625000000004e-05, "loss": 1.7021, "step": 42870 }, { "epoch": 0.06, "learning_rate": 4.732e-05, "loss": 1.7778, "step": 42880 }, { "epoch": 0.06, "learning_rate": 4.7319375000000005e-05, "loss": 1.6949, "step": 42890 }, { "epoch": 0.06, "learning_rate": 4.731875e-05, "loss": 2.0371, "step": 42900 }, { "epoch": 0.06, "learning_rate": 4.7318125000000005e-05, "loss": 1.7595, "step": 42910 }, { "epoch": 0.06, "learning_rate": 4.73175e-05, "loss": 1.898, "step": 42920 }, { "epoch": 0.06, "learning_rate": 4.7316875000000006e-05, "loss": 1.8569, "step": 42930 }, { "epoch": 0.06, "learning_rate": 4.731625e-05, "loss": 1.6592, "step": 42940 }, { "epoch": 0.06, "learning_rate": 4.7315625e-05, "loss": 1.83, "step": 42950 }, { "epoch": 0.06, "learning_rate": 4.7315000000000004e-05, "loss": 1.8758, "step": 42960 }, { "epoch": 0.06, "learning_rate": 4.7314375e-05, "loss": 1.752, "step": 42970 }, { "epoch": 0.06, "learning_rate": 4.7313750000000004e-05, "loss": 1.8817, "step": 42980 }, { "epoch": 0.06, "learning_rate": 4.7313125e-05, "loss": 1.797, "step": 42990 }, { "epoch": 0.06, "learning_rate": 4.7312500000000005e-05, "loss": 1.8302, "step": 43000 }, { "epoch": 0.06, "eval_loss": 1.7298684120178223, "eval_runtime": 1291.8163, "eval_samples_per_second": 44.926, "eval_steps_per_second": 5.616, "step": 43000 }, { "epoch": 0.06, "learning_rate": 4.7311875e-05, "loss": 1.9517, "step": 43010 }, { "epoch": 0.06, "learning_rate": 4.731125e-05, "loss": 1.7316, "step": 43020 }, { "epoch": 0.06, "learning_rate": 4.7310625e-05, "loss": 1.6622, "step": 43030 }, { "epoch": 0.06, "learning_rate": 4.7310000000000006e-05, "loss": 1.8151, "step": 43040 }, { "epoch": 0.06, "learning_rate": 4.7309375e-05, "loss": 1.7946, "step": 43050 }, { "epoch": 0.06, "learning_rate": 4.730875e-05, "loss": 1.9389, "step": 43060 }, { "epoch": 0.06, "learning_rate": 4.7308125000000004e-05, "loss": 1.8067, "step": 43070 }, { "epoch": 0.06, "learning_rate": 4.730750000000001e-05, "loss": 1.9435, "step": 43080 }, { "epoch": 0.06, "learning_rate": 4.7306875e-05, "loss": 1.6887, "step": 43090 }, { "epoch": 0.06, "learning_rate": 4.730625e-05, "loss": 1.8396, "step": 43100 }, { "epoch": 0.06, "learning_rate": 4.7305625000000005e-05, "loss": 1.8342, "step": 43110 }, { "epoch": 0.06, "learning_rate": 4.7305e-05, "loss": 1.6045, "step": 43120 }, { "epoch": 0.06, "learning_rate": 4.7304375e-05, "loss": 2.0939, "step": 43130 }, { "epoch": 0.06, "learning_rate": 4.730375e-05, "loss": 1.8553, "step": 43140 }, { "epoch": 0.06, "learning_rate": 4.7303125000000006e-05, "loss": 1.6576, "step": 43150 }, { "epoch": 0.06, "learning_rate": 4.7302499999999996e-05, "loss": 1.8036, "step": 43160 }, { "epoch": 0.06, "learning_rate": 4.7301875e-05, "loss": 2.1891, "step": 43170 }, { "epoch": 0.06, "learning_rate": 4.7301250000000004e-05, "loss": 1.7399, "step": 43180 }, { "epoch": 0.06, "learning_rate": 4.7300625e-05, "loss": 1.6814, "step": 43190 }, { "epoch": 0.06, "learning_rate": 4.73e-05, "loss": 1.8087, "step": 43200 }, { "epoch": 0.06, "learning_rate": 4.7299375e-05, "loss": 1.8076, "step": 43210 }, { "epoch": 0.06, "learning_rate": 4.7298750000000005e-05, "loss": 2.0111, "step": 43220 }, { "epoch": 0.06, "learning_rate": 4.7298125e-05, "loss": 1.8063, "step": 43230 }, { "epoch": 0.06, "learning_rate": 4.72975e-05, "loss": 1.9937, "step": 43240 }, { "epoch": 0.06, "learning_rate": 4.7296875e-05, "loss": 1.8349, "step": 43250 }, { "epoch": 0.06, "learning_rate": 4.729625e-05, "loss": 1.6207, "step": 43260 }, { "epoch": 0.06, "learning_rate": 4.7295625000000003e-05, "loss": 1.6997, "step": 43270 }, { "epoch": 0.06, "learning_rate": 4.7295e-05, "loss": 1.8353, "step": 43280 }, { "epoch": 0.06, "learning_rate": 4.7294375000000004e-05, "loss": 1.7584, "step": 43290 }, { "epoch": 0.06, "learning_rate": 4.729375e-05, "loss": 1.8998, "step": 43300 }, { "epoch": 0.06, "learning_rate": 4.7293125000000005e-05, "loss": 1.8236, "step": 43310 }, { "epoch": 0.06, "learning_rate": 4.72925e-05, "loss": 1.6643, "step": 43320 }, { "epoch": 0.06, "learning_rate": 4.7291875000000005e-05, "loss": 1.6009, "step": 43330 }, { "epoch": 0.06, "learning_rate": 4.729125e-05, "loss": 1.881, "step": 43340 }, { "epoch": 0.06, "learning_rate": 4.7290625e-05, "loss": 1.8804, "step": 43350 }, { "epoch": 0.06, "learning_rate": 4.729e-05, "loss": 1.9072, "step": 43360 }, { "epoch": 0.06, "learning_rate": 4.7289375e-05, "loss": 1.8836, "step": 43370 }, { "epoch": 0.06, "learning_rate": 4.7288750000000004e-05, "loss": 1.4735, "step": 43380 }, { "epoch": 0.06, "learning_rate": 4.7288125e-05, "loss": 1.905, "step": 43390 }, { "epoch": 0.06, "learning_rate": 4.7287500000000004e-05, "loss": 1.7099, "step": 43400 }, { "epoch": 0.06, "learning_rate": 4.7286875e-05, "loss": 1.7133, "step": 43410 }, { "epoch": 0.06, "learning_rate": 4.7286250000000005e-05, "loss": 1.71, "step": 43420 }, { "epoch": 0.06, "learning_rate": 4.7285625e-05, "loss": 1.7913, "step": 43430 }, { "epoch": 0.06, "learning_rate": 4.7285e-05, "loss": 1.9222, "step": 43440 }, { "epoch": 0.06, "learning_rate": 4.7284375e-05, "loss": 1.5931, "step": 43450 }, { "epoch": 0.06, "learning_rate": 4.7283750000000006e-05, "loss": 1.4661, "step": 43460 }, { "epoch": 0.06, "learning_rate": 4.7283125e-05, "loss": 1.8088, "step": 43470 }, { "epoch": 0.06, "learning_rate": 4.72825e-05, "loss": 1.8222, "step": 43480 }, { "epoch": 0.06, "learning_rate": 4.7281875000000004e-05, "loss": 1.8122, "step": 43490 }, { "epoch": 0.06, "learning_rate": 4.728125000000001e-05, "loss": 1.9352, "step": 43500 }, { "epoch": 0.06, "learning_rate": 4.7280625e-05, "loss": 1.9828, "step": 43510 }, { "epoch": 0.06, "learning_rate": 4.728e-05, "loss": 1.8909, "step": 43520 }, { "epoch": 0.06, "learning_rate": 4.7279375000000005e-05, "loss": 1.9158, "step": 43530 }, { "epoch": 0.06, "learning_rate": 4.727875e-05, "loss": 1.895, "step": 43540 }, { "epoch": 0.06, "learning_rate": 4.7278125e-05, "loss": 1.6892, "step": 43550 }, { "epoch": 0.06, "learning_rate": 4.72775e-05, "loss": 1.5279, "step": 43560 }, { "epoch": 0.06, "learning_rate": 4.7276875000000006e-05, "loss": 1.6588, "step": 43570 }, { "epoch": 0.06, "learning_rate": 4.7276249999999996e-05, "loss": 1.9166, "step": 43580 }, { "epoch": 0.06, "learning_rate": 4.7275625e-05, "loss": 1.8177, "step": 43590 }, { "epoch": 0.06, "learning_rate": 4.7275000000000004e-05, "loss": 1.556, "step": 43600 }, { "epoch": 0.06, "learning_rate": 4.7274375e-05, "loss": 1.6942, "step": 43610 }, { "epoch": 0.06, "learning_rate": 4.727375e-05, "loss": 1.8864, "step": 43620 }, { "epoch": 0.06, "learning_rate": 4.7273125e-05, "loss": 1.9201, "step": 43630 }, { "epoch": 0.06, "learning_rate": 4.7272500000000005e-05, "loss": 1.7666, "step": 43640 }, { "epoch": 0.06, "learning_rate": 4.7271875e-05, "loss": 1.968, "step": 43650 }, { "epoch": 0.06, "learning_rate": 4.727125e-05, "loss": 1.7986, "step": 43660 }, { "epoch": 0.06, "learning_rate": 4.7270625e-05, "loss": 1.8219, "step": 43670 }, { "epoch": 0.06, "learning_rate": 4.7270000000000007e-05, "loss": 1.9003, "step": 43680 }, { "epoch": 0.06, "learning_rate": 4.7269375000000003e-05, "loss": 2.0106, "step": 43690 }, { "epoch": 0.06, "learning_rate": 4.726875e-05, "loss": 1.8224, "step": 43700 }, { "epoch": 0.06, "learning_rate": 4.7268125000000004e-05, "loss": 1.7112, "step": 43710 }, { "epoch": 0.06, "learning_rate": 4.72675e-05, "loss": 1.9649, "step": 43720 }, { "epoch": 0.06, "learning_rate": 4.7266875000000005e-05, "loss": 1.6427, "step": 43730 }, { "epoch": 0.06, "learning_rate": 4.726625e-05, "loss": 1.6065, "step": 43740 }, { "epoch": 0.06, "learning_rate": 4.7265625000000005e-05, "loss": 1.7581, "step": 43750 }, { "epoch": 0.06, "learning_rate": 4.7265e-05, "loss": 1.6802, "step": 43760 }, { "epoch": 0.06, "learning_rate": 4.7264375e-05, "loss": 1.9348, "step": 43770 }, { "epoch": 0.06, "learning_rate": 4.726375e-05, "loss": 1.6604, "step": 43780 }, { "epoch": 0.06, "learning_rate": 4.7263125e-05, "loss": 1.9367, "step": 43790 }, { "epoch": 0.06, "learning_rate": 4.7262500000000004e-05, "loss": 1.9242, "step": 43800 }, { "epoch": 0.06, "learning_rate": 4.7261875e-05, "loss": 1.7078, "step": 43810 }, { "epoch": 0.06, "learning_rate": 4.7261250000000004e-05, "loss": 1.7648, "step": 43820 }, { "epoch": 0.06, "learning_rate": 4.7260625e-05, "loss": 1.7629, "step": 43830 }, { "epoch": 0.06, "learning_rate": 4.7260000000000005e-05, "loss": 1.6884, "step": 43840 }, { "epoch": 0.06, "learning_rate": 4.7259375e-05, "loss": 1.7772, "step": 43850 }, { "epoch": 0.06, "learning_rate": 4.725875e-05, "loss": 1.9439, "step": 43860 }, { "epoch": 0.06, "learning_rate": 4.7258125e-05, "loss": 1.7857, "step": 43870 }, { "epoch": 0.06, "learning_rate": 4.7257500000000006e-05, "loss": 1.62, "step": 43880 }, { "epoch": 0.06, "learning_rate": 4.7256875e-05, "loss": 1.8845, "step": 43890 }, { "epoch": 0.06, "learning_rate": 4.725625e-05, "loss": 1.789, "step": 43900 }, { "epoch": 0.06, "learning_rate": 4.7255625000000004e-05, "loss": 1.7423, "step": 43910 }, { "epoch": 0.06, "learning_rate": 4.725500000000001e-05, "loss": 1.961, "step": 43920 }, { "epoch": 0.06, "learning_rate": 4.7254375e-05, "loss": 1.5733, "step": 43930 }, { "epoch": 0.06, "learning_rate": 4.725375e-05, "loss": 1.8342, "step": 43940 }, { "epoch": 0.06, "learning_rate": 4.7253125000000005e-05, "loss": 2.0233, "step": 43950 }, { "epoch": 0.06, "learning_rate": 4.72525e-05, "loss": 1.9809, "step": 43960 }, { "epoch": 0.06, "learning_rate": 4.7251875e-05, "loss": 1.8476, "step": 43970 }, { "epoch": 0.06, "learning_rate": 4.725125e-05, "loss": 1.8672, "step": 43980 }, { "epoch": 0.06, "learning_rate": 4.7250625000000006e-05, "loss": 1.8433, "step": 43990 }, { "epoch": 0.06, "learning_rate": 4.7249999999999997e-05, "loss": 1.8581, "step": 44000 }, { "epoch": 0.06, "eval_loss": 1.7216545343399048, "eval_runtime": 1291.7709, "eval_samples_per_second": 44.927, "eval_steps_per_second": 5.616, "step": 44000 }, { "epoch": 0.06, "learning_rate": 4.7249375e-05, "loss": 1.8798, "step": 44010 }, { "epoch": 0.06, "learning_rate": 4.7248750000000004e-05, "loss": 1.7796, "step": 44020 }, { "epoch": 0.06, "learning_rate": 4.7248125e-05, "loss": 1.8128, "step": 44030 }, { "epoch": 0.06, "learning_rate": 4.72475e-05, "loss": 1.9106, "step": 44040 }, { "epoch": 0.06, "learning_rate": 4.7246875e-05, "loss": 1.5999, "step": 44050 }, { "epoch": 0.06, "learning_rate": 4.7246250000000005e-05, "loss": 1.7884, "step": 44060 }, { "epoch": 0.06, "learning_rate": 4.7245625e-05, "loss": 1.5667, "step": 44070 }, { "epoch": 0.06, "learning_rate": 4.7245e-05, "loss": 1.8351, "step": 44080 }, { "epoch": 0.06, "learning_rate": 4.7244375e-05, "loss": 1.8528, "step": 44090 }, { "epoch": 0.06, "learning_rate": 4.7243750000000007e-05, "loss": 1.859, "step": 44100 }, { "epoch": 0.06, "learning_rate": 4.7243125000000003e-05, "loss": 1.9549, "step": 44110 }, { "epoch": 0.06, "learning_rate": 4.72425e-05, "loss": 2.128, "step": 44120 }, { "epoch": 0.06, "learning_rate": 4.7241875000000004e-05, "loss": 1.9208, "step": 44130 }, { "epoch": 0.06, "learning_rate": 4.724125e-05, "loss": 1.8389, "step": 44140 }, { "epoch": 0.06, "learning_rate": 4.7240625000000005e-05, "loss": 1.7259, "step": 44150 }, { "epoch": 0.06, "learning_rate": 4.724e-05, "loss": 2.0701, "step": 44160 }, { "epoch": 0.06, "learning_rate": 4.7239375000000005e-05, "loss": 1.4354, "step": 44170 }, { "epoch": 0.06, "learning_rate": 4.723875e-05, "loss": 1.7075, "step": 44180 }, { "epoch": 0.06, "learning_rate": 4.7238125e-05, "loss": 1.7326, "step": 44190 }, { "epoch": 0.06, "learning_rate": 4.72375e-05, "loss": 1.9741, "step": 44200 }, { "epoch": 0.06, "learning_rate": 4.7236875e-05, "loss": 1.8586, "step": 44210 }, { "epoch": 0.06, "learning_rate": 4.7236250000000004e-05, "loss": 1.7806, "step": 44220 }, { "epoch": 0.06, "learning_rate": 4.7235625e-05, "loss": 1.953, "step": 44230 }, { "epoch": 0.06, "learning_rate": 4.7235000000000004e-05, "loss": 1.7884, "step": 44240 }, { "epoch": 0.06, "learning_rate": 4.7234375e-05, "loss": 1.939, "step": 44250 }, { "epoch": 0.06, "learning_rate": 4.7233750000000005e-05, "loss": 1.7052, "step": 44260 }, { "epoch": 0.06, "learning_rate": 4.7233125e-05, "loss": 1.6968, "step": 44270 }, { "epoch": 0.06, "learning_rate": 4.72325e-05, "loss": 1.9433, "step": 44280 }, { "epoch": 0.06, "learning_rate": 4.7231875e-05, "loss": 1.6662, "step": 44290 }, { "epoch": 0.06, "learning_rate": 4.7231250000000006e-05, "loss": 1.7596, "step": 44300 }, { "epoch": 0.06, "learning_rate": 4.7230625e-05, "loss": 1.7765, "step": 44310 }, { "epoch": 0.06, "learning_rate": 4.723e-05, "loss": 1.9014, "step": 44320 }, { "epoch": 0.06, "learning_rate": 4.7229375000000004e-05, "loss": 1.922, "step": 44330 }, { "epoch": 0.06, "learning_rate": 4.722875000000001e-05, "loss": 1.835, "step": 44340 }, { "epoch": 0.06, "learning_rate": 4.7228125e-05, "loss": 1.7834, "step": 44350 }, { "epoch": 0.06, "learning_rate": 4.72275e-05, "loss": 1.693, "step": 44360 }, { "epoch": 0.06, "learning_rate": 4.7226875000000005e-05, "loss": 1.937, "step": 44370 }, { "epoch": 0.06, "learning_rate": 4.722625e-05, "loss": 1.7218, "step": 44380 }, { "epoch": 0.06, "learning_rate": 4.7225625e-05, "loss": 1.6935, "step": 44390 }, { "epoch": 0.06, "learning_rate": 4.7225e-05, "loss": 1.8082, "step": 44400 }, { "epoch": 0.06, "learning_rate": 4.7224375000000006e-05, "loss": 1.864, "step": 44410 }, { "epoch": 0.06, "learning_rate": 4.7223749999999997e-05, "loss": 1.7171, "step": 44420 }, { "epoch": 0.06, "learning_rate": 4.7223125e-05, "loss": 1.9102, "step": 44430 }, { "epoch": 0.06, "learning_rate": 4.7222500000000004e-05, "loss": 1.7662, "step": 44440 }, { "epoch": 0.06, "learning_rate": 4.7221875e-05, "loss": 1.8459, "step": 44450 }, { "epoch": 0.06, "learning_rate": 4.722125e-05, "loss": 2.0537, "step": 44460 }, { "epoch": 0.06, "learning_rate": 4.7220625e-05, "loss": 1.8709, "step": 44470 }, { "epoch": 0.06, "learning_rate": 4.7220000000000005e-05, "loss": 1.8053, "step": 44480 }, { "epoch": 0.06, "learning_rate": 4.7219375e-05, "loss": 1.7616, "step": 44490 }, { "epoch": 0.06, "learning_rate": 4.721875e-05, "loss": 1.7863, "step": 44500 }, { "epoch": 0.06, "learning_rate": 4.7218125e-05, "loss": 1.8371, "step": 44510 }, { "epoch": 0.06, "learning_rate": 4.7217500000000007e-05, "loss": 1.6726, "step": 44520 }, { "epoch": 0.06, "learning_rate": 4.7216875000000004e-05, "loss": 1.877, "step": 44530 }, { "epoch": 0.06, "learning_rate": 4.721625e-05, "loss": 1.8842, "step": 44540 }, { "epoch": 0.06, "learning_rate": 4.7215625000000004e-05, "loss": 1.7038, "step": 44550 }, { "epoch": 0.06, "learning_rate": 4.7215e-05, "loss": 1.828, "step": 44560 }, { "epoch": 0.06, "learning_rate": 4.7214375e-05, "loss": 1.8854, "step": 44570 }, { "epoch": 0.06, "learning_rate": 4.721375e-05, "loss": 1.6815, "step": 44580 }, { "epoch": 0.06, "learning_rate": 4.7213125000000005e-05, "loss": 1.8213, "step": 44590 }, { "epoch": 0.06, "learning_rate": 4.72125e-05, "loss": 1.8636, "step": 44600 }, { "epoch": 0.06, "learning_rate": 4.7211875e-05, "loss": 1.6916, "step": 44610 }, { "epoch": 0.06, "learning_rate": 4.721125e-05, "loss": 1.7947, "step": 44620 }, { "epoch": 0.06, "learning_rate": 4.7210625e-05, "loss": 1.4856, "step": 44630 }, { "epoch": 0.06, "learning_rate": 4.7210000000000004e-05, "loss": 1.9632, "step": 44640 }, { "epoch": 0.06, "learning_rate": 4.7209375e-05, "loss": 1.8773, "step": 44650 }, { "epoch": 0.06, "learning_rate": 4.7208750000000004e-05, "loss": 1.9642, "step": 44660 }, { "epoch": 0.06, "learning_rate": 4.7208125e-05, "loss": 1.7513, "step": 44670 }, { "epoch": 0.06, "learning_rate": 4.7207500000000005e-05, "loss": 1.8971, "step": 44680 }, { "epoch": 0.06, "learning_rate": 4.7206875e-05, "loss": 1.6724, "step": 44690 }, { "epoch": 0.06, "learning_rate": 4.720625e-05, "loss": 1.8366, "step": 44700 }, { "epoch": 0.06, "learning_rate": 4.7205625e-05, "loss": 1.9381, "step": 44710 }, { "epoch": 0.06, "learning_rate": 4.7205000000000006e-05, "loss": 1.8222, "step": 44720 }, { "epoch": 0.06, "learning_rate": 4.7204375e-05, "loss": 1.8921, "step": 44730 }, { "epoch": 0.06, "learning_rate": 4.720375e-05, "loss": 1.8071, "step": 44740 }, { "epoch": 0.06, "learning_rate": 4.7203125000000004e-05, "loss": 1.8612, "step": 44750 }, { "epoch": 0.06, "learning_rate": 4.72025e-05, "loss": 1.546, "step": 44760 }, { "epoch": 0.06, "learning_rate": 4.7201875e-05, "loss": 1.7917, "step": 44770 }, { "epoch": 0.06, "learning_rate": 4.720125e-05, "loss": 1.8986, "step": 44780 }, { "epoch": 0.06, "learning_rate": 4.7200625000000005e-05, "loss": 1.6076, "step": 44790 }, { "epoch": 0.06, "learning_rate": 4.72e-05, "loss": 1.8012, "step": 44800 }, { "epoch": 0.06, "learning_rate": 4.7199375e-05, "loss": 1.5808, "step": 44810 }, { "epoch": 0.06, "learning_rate": 4.719875e-05, "loss": 1.8845, "step": 44820 }, { "epoch": 0.06, "learning_rate": 4.7198125000000006e-05, "loss": 1.6921, "step": 44830 }, { "epoch": 0.06, "learning_rate": 4.7197500000000003e-05, "loss": 1.598, "step": 44840 }, { "epoch": 0.06, "learning_rate": 4.7196875e-05, "loss": 1.7634, "step": 44850 }, { "epoch": 0.06, "learning_rate": 4.7196250000000004e-05, "loss": 1.6995, "step": 44860 }, { "epoch": 0.06, "learning_rate": 4.719562500000001e-05, "loss": 1.8602, "step": 44870 }, { "epoch": 0.06, "learning_rate": 4.7195e-05, "loss": 1.7653, "step": 44880 }, { "epoch": 0.06, "learning_rate": 4.7194375e-05, "loss": 1.7563, "step": 44890 }, { "epoch": 0.06, "learning_rate": 4.7193750000000005e-05, "loss": 1.822, "step": 44900 }, { "epoch": 0.06, "learning_rate": 4.7193125e-05, "loss": 1.981, "step": 44910 }, { "epoch": 0.06, "learning_rate": 4.71925e-05, "loss": 1.7421, "step": 44920 }, { "epoch": 0.06, "learning_rate": 4.7191875e-05, "loss": 1.7292, "step": 44930 }, { "epoch": 0.06, "learning_rate": 4.719125000000001e-05, "loss": 1.7228, "step": 44940 }, { "epoch": 0.06, "learning_rate": 4.7190625000000004e-05, "loss": 1.7848, "step": 44950 }, { "epoch": 0.06, "learning_rate": 4.719e-05, "loss": 1.7758, "step": 44960 }, { "epoch": 0.06, "learning_rate": 4.7189375000000004e-05, "loss": 1.9499, "step": 44970 }, { "epoch": 0.06, "learning_rate": 4.718875e-05, "loss": 1.7048, "step": 44980 }, { "epoch": 0.06, "learning_rate": 4.7188125e-05, "loss": 1.9674, "step": 44990 }, { "epoch": 0.06, "learning_rate": 4.71875e-05, "loss": 1.805, "step": 45000 }, { "epoch": 0.06, "eval_loss": 1.7015135288238525, "eval_runtime": 1297.0354, "eval_samples_per_second": 44.745, "eval_steps_per_second": 5.594, "step": 45000 }, { "epoch": 0.06, "learning_rate": 4.7186875000000006e-05, "loss": 1.763, "step": 45010 }, { "epoch": 0.06, "learning_rate": 4.718625e-05, "loss": 1.9919, "step": 45020 }, { "epoch": 0.06, "learning_rate": 4.7185625e-05, "loss": 1.8759, "step": 45030 }, { "epoch": 0.06, "learning_rate": 4.7185e-05, "loss": 1.8578, "step": 45040 }, { "epoch": 0.06, "learning_rate": 4.7184375e-05, "loss": 1.9324, "step": 45050 }, { "epoch": 0.06, "learning_rate": 4.7183750000000004e-05, "loss": 1.7332, "step": 45060 }, { "epoch": 0.06, "learning_rate": 4.7183125e-05, "loss": 1.7259, "step": 45070 }, { "epoch": 0.06, "learning_rate": 4.7182500000000004e-05, "loss": 1.7969, "step": 45080 }, { "epoch": 0.06, "learning_rate": 4.7181875e-05, "loss": 1.6927, "step": 45090 }, { "epoch": 0.06, "learning_rate": 4.7181250000000005e-05, "loss": 1.5938, "step": 45100 }, { "epoch": 0.06, "learning_rate": 4.7180625e-05, "loss": 1.7781, "step": 45110 }, { "epoch": 0.06, "learning_rate": 4.718e-05, "loss": 2.0091, "step": 45120 }, { "epoch": 0.06, "learning_rate": 4.7179375e-05, "loss": 1.7, "step": 45130 }, { "epoch": 0.06, "learning_rate": 4.7178750000000006e-05, "loss": 1.8682, "step": 45140 }, { "epoch": 0.06, "learning_rate": 4.7178125e-05, "loss": 2.0323, "step": 45150 }, { "epoch": 0.06, "learning_rate": 4.71775e-05, "loss": 1.7032, "step": 45160 }, { "epoch": 0.06, "learning_rate": 4.7176875000000004e-05, "loss": 1.4568, "step": 45170 }, { "epoch": 0.06, "learning_rate": 4.717625e-05, "loss": 1.6979, "step": 45180 }, { "epoch": 0.06, "learning_rate": 4.7175625e-05, "loss": 1.7648, "step": 45190 }, { "epoch": 0.06, "learning_rate": 4.7175e-05, "loss": 1.9138, "step": 45200 }, { "epoch": 0.06, "learning_rate": 4.7174375000000005e-05, "loss": 1.7125, "step": 45210 }, { "epoch": 0.06, "learning_rate": 4.717375e-05, "loss": 1.6842, "step": 45220 }, { "epoch": 0.06, "learning_rate": 4.7173125e-05, "loss": 1.7782, "step": 45230 }, { "epoch": 0.06, "learning_rate": 4.71725e-05, "loss": 1.7899, "step": 45240 }, { "epoch": 0.06, "learning_rate": 4.7171875000000006e-05, "loss": 1.8359, "step": 45250 }, { "epoch": 0.06, "learning_rate": 4.7171250000000003e-05, "loss": 1.8095, "step": 45260 }, { "epoch": 0.06, "learning_rate": 4.7170625e-05, "loss": 1.7324, "step": 45270 }, { "epoch": 0.06, "learning_rate": 4.7170000000000004e-05, "loss": 1.6982, "step": 45280 }, { "epoch": 0.06, "learning_rate": 4.716937500000001e-05, "loss": 1.8083, "step": 45290 }, { "epoch": 0.06, "learning_rate": 4.716875e-05, "loss": 1.8258, "step": 45300 }, { "epoch": 0.06, "learning_rate": 4.7168125e-05, "loss": 1.6617, "step": 45310 }, { "epoch": 0.06, "learning_rate": 4.7167500000000005e-05, "loss": 1.8154, "step": 45320 }, { "epoch": 0.06, "learning_rate": 4.7166875e-05, "loss": 1.8319, "step": 45330 }, { "epoch": 0.06, "learning_rate": 4.716625e-05, "loss": 1.8681, "step": 45340 }, { "epoch": 0.06, "learning_rate": 4.7165625e-05, "loss": 1.6457, "step": 45350 }, { "epoch": 0.06, "learning_rate": 4.716500000000001e-05, "loss": 1.7752, "step": 45360 }, { "epoch": 0.06, "learning_rate": 4.7164375e-05, "loss": 1.8327, "step": 45370 }, { "epoch": 0.06, "learning_rate": 4.716375e-05, "loss": 1.7542, "step": 45380 }, { "epoch": 0.06, "learning_rate": 4.7163125000000004e-05, "loss": 1.7213, "step": 45390 }, { "epoch": 0.06, "learning_rate": 4.71625e-05, "loss": 1.7522, "step": 45400 }, { "epoch": 0.06, "learning_rate": 4.7161875e-05, "loss": 1.6275, "step": 45410 }, { "epoch": 0.06, "learning_rate": 4.716125e-05, "loss": 1.7074, "step": 45420 }, { "epoch": 0.06, "learning_rate": 4.7160625000000006e-05, "loss": 1.7488, "step": 45430 }, { "epoch": 0.06, "learning_rate": 4.716e-05, "loss": 1.7566, "step": 45440 }, { "epoch": 0.06, "learning_rate": 4.7159375e-05, "loss": 1.7236, "step": 45450 }, { "epoch": 0.06, "learning_rate": 4.715875e-05, "loss": 1.8958, "step": 45460 }, { "epoch": 0.06, "learning_rate": 4.7158125e-05, "loss": 1.8227, "step": 45470 }, { "epoch": 0.06, "learning_rate": 4.7157500000000004e-05, "loss": 1.7644, "step": 45480 }, { "epoch": 0.06, "learning_rate": 4.7156875e-05, "loss": 1.7846, "step": 45490 }, { "epoch": 0.06, "learning_rate": 4.7156250000000004e-05, "loss": 1.6317, "step": 45500 }, { "epoch": 0.06, "learning_rate": 4.7155625e-05, "loss": 2.0053, "step": 45510 }, { "epoch": 0.06, "learning_rate": 4.7155000000000005e-05, "loss": 1.6729, "step": 45520 }, { "epoch": 0.06, "learning_rate": 4.7154375e-05, "loss": 1.8501, "step": 45530 }, { "epoch": 0.06, "learning_rate": 4.715375e-05, "loss": 1.3343, "step": 45540 }, { "epoch": 0.06, "learning_rate": 4.7153125e-05, "loss": 1.9074, "step": 45550 }, { "epoch": 0.06, "learning_rate": 4.7152500000000006e-05, "loss": 1.7482, "step": 45560 }, { "epoch": 0.06, "learning_rate": 4.7151875e-05, "loss": 1.7097, "step": 45570 }, { "epoch": 0.06, "learning_rate": 4.715125e-05, "loss": 1.6367, "step": 45580 }, { "epoch": 0.06, "learning_rate": 4.7150625000000004e-05, "loss": 1.4809, "step": 45590 }, { "epoch": 0.06, "learning_rate": 4.715e-05, "loss": 1.7594, "step": 45600 }, { "epoch": 0.06, "learning_rate": 4.7149375e-05, "loss": 1.8156, "step": 45610 }, { "epoch": 0.06, "learning_rate": 4.714875e-05, "loss": 1.834, "step": 45620 }, { "epoch": 0.06, "learning_rate": 4.7148125000000005e-05, "loss": 1.7413, "step": 45630 }, { "epoch": 0.06, "learning_rate": 4.71475e-05, "loss": 1.9541, "step": 45640 }, { "epoch": 0.06, "learning_rate": 4.7146875e-05, "loss": 1.8431, "step": 45650 }, { "epoch": 0.06, "learning_rate": 4.714625e-05, "loss": 1.7105, "step": 45660 }, { "epoch": 0.06, "learning_rate": 4.7145625000000007e-05, "loss": 1.7726, "step": 45670 }, { "epoch": 0.06, "learning_rate": 4.7145000000000003e-05, "loss": 1.729, "step": 45680 }, { "epoch": 0.06, "learning_rate": 4.7144375e-05, "loss": 1.7431, "step": 45690 }, { "epoch": 0.06, "learning_rate": 4.7143750000000004e-05, "loss": 1.6678, "step": 45700 }, { "epoch": 0.06, "learning_rate": 4.714312500000001e-05, "loss": 1.7223, "step": 45710 }, { "epoch": 0.06, "learning_rate": 4.71425e-05, "loss": 1.8814, "step": 45720 }, { "epoch": 0.06, "learning_rate": 4.7141875e-05, "loss": 1.5896, "step": 45730 }, { "epoch": 0.06, "learning_rate": 4.7141250000000005e-05, "loss": 1.8072, "step": 45740 }, { "epoch": 0.06, "learning_rate": 4.7140625e-05, "loss": 1.8689, "step": 45750 }, { "epoch": 0.06, "learning_rate": 4.714e-05, "loss": 1.4492, "step": 45760 }, { "epoch": 0.06, "learning_rate": 4.7139375e-05, "loss": 1.4516, "step": 45770 }, { "epoch": 0.06, "learning_rate": 4.713875000000001e-05, "loss": 1.7013, "step": 45780 }, { "epoch": 0.06, "learning_rate": 4.7138125e-05, "loss": 1.6682, "step": 45790 }, { "epoch": 0.06, "learning_rate": 4.71375e-05, "loss": 1.6022, "step": 45800 }, { "epoch": 0.06, "learning_rate": 4.7136875000000004e-05, "loss": 1.8164, "step": 45810 }, { "epoch": 0.06, "learning_rate": 4.713625e-05, "loss": 1.8416, "step": 45820 }, { "epoch": 0.06, "learning_rate": 4.7135625e-05, "loss": 1.7352, "step": 45830 }, { "epoch": 0.06, "learning_rate": 4.7135e-05, "loss": 1.7221, "step": 45840 }, { "epoch": 0.06, "learning_rate": 4.7134375000000006e-05, "loss": 1.7225, "step": 45850 }, { "epoch": 0.06, "learning_rate": 4.713375e-05, "loss": 1.7311, "step": 45860 }, { "epoch": 0.06, "learning_rate": 4.7133125e-05, "loss": 1.7921, "step": 45870 }, { "epoch": 0.06, "learning_rate": 4.71325e-05, "loss": 1.6265, "step": 45880 }, { "epoch": 0.06, "learning_rate": 4.7131875e-05, "loss": 1.8644, "step": 45890 }, { "epoch": 0.06, "learning_rate": 4.7131250000000004e-05, "loss": 1.6974, "step": 45900 }, { "epoch": 0.06, "learning_rate": 4.7130625e-05, "loss": 1.822, "step": 45910 }, { "epoch": 0.06, "learning_rate": 4.7130000000000004e-05, "loss": 1.6458, "step": 45920 }, { "epoch": 0.06, "learning_rate": 4.7129375e-05, "loss": 1.5856, "step": 45930 }, { "epoch": 0.06, "learning_rate": 4.7128750000000005e-05, "loss": 1.5907, "step": 45940 }, { "epoch": 0.06, "learning_rate": 4.7128125e-05, "loss": 1.7892, "step": 45950 }, { "epoch": 0.06, "learning_rate": 4.71275e-05, "loss": 1.8874, "step": 45960 }, { "epoch": 0.06, "learning_rate": 4.7126875e-05, "loss": 1.7746, "step": 45970 }, { "epoch": 0.06, "learning_rate": 4.712625e-05, "loss": 1.5604, "step": 45980 }, { "epoch": 0.06, "learning_rate": 4.7125625e-05, "loss": 1.7271, "step": 45990 }, { "epoch": 0.06, "learning_rate": 4.7125e-05, "loss": 1.9193, "step": 46000 }, { "epoch": 0.06, "eval_loss": 1.691484808921814, "eval_runtime": 1298.4459, "eval_samples_per_second": 44.697, "eval_steps_per_second": 5.587, "step": 46000 }, { "epoch": 0.06, "learning_rate": 4.7124375000000004e-05, "loss": 1.8959, "step": 46010 }, { "epoch": 0.06, "learning_rate": 4.712375e-05, "loss": 1.8281, "step": 46020 }, { "epoch": 0.06, "learning_rate": 4.7123125000000005e-05, "loss": 1.6912, "step": 46030 }, { "epoch": 0.06, "learning_rate": 4.71225e-05, "loss": 1.7982, "step": 46040 }, { "epoch": 0.06, "learning_rate": 4.7121875000000005e-05, "loss": 1.6593, "step": 46050 }, { "epoch": 0.06, "learning_rate": 4.712125e-05, "loss": 1.7255, "step": 46060 }, { "epoch": 0.06, "learning_rate": 4.7120625e-05, "loss": 1.8109, "step": 46070 }, { "epoch": 0.06, "learning_rate": 4.712e-05, "loss": 1.7013, "step": 46080 }, { "epoch": 0.06, "learning_rate": 4.7119375000000007e-05, "loss": 1.5886, "step": 46090 }, { "epoch": 0.06, "learning_rate": 4.7118750000000004e-05, "loss": 1.6897, "step": 46100 }, { "epoch": 0.06, "learning_rate": 4.7118125e-05, "loss": 1.8029, "step": 46110 }, { "epoch": 0.06, "learning_rate": 4.7117500000000004e-05, "loss": 1.7934, "step": 46120 }, { "epoch": 0.06, "learning_rate": 4.711687500000001e-05, "loss": 1.7299, "step": 46130 }, { "epoch": 0.06, "learning_rate": 4.711625e-05, "loss": 1.7967, "step": 46140 }, { "epoch": 0.06, "learning_rate": 4.7115625e-05, "loss": 1.847, "step": 46150 }, { "epoch": 0.06, "learning_rate": 4.7115000000000005e-05, "loss": 1.8315, "step": 46160 }, { "epoch": 0.06, "learning_rate": 4.7114375e-05, "loss": 1.8153, "step": 46170 }, { "epoch": 0.06, "learning_rate": 4.711375e-05, "loss": 2.0441, "step": 46180 }, { "epoch": 0.06, "learning_rate": 4.7113125e-05, "loss": 1.8645, "step": 46190 }, { "epoch": 0.06, "learning_rate": 4.711250000000001e-05, "loss": 1.5601, "step": 46200 }, { "epoch": 0.06, "learning_rate": 4.7111875e-05, "loss": 1.7114, "step": 46210 }, { "epoch": 0.06, "learning_rate": 4.711125e-05, "loss": 1.9689, "step": 46220 }, { "epoch": 0.06, "learning_rate": 4.7110625000000004e-05, "loss": 1.618, "step": 46230 }, { "epoch": 0.06, "learning_rate": 4.711e-05, "loss": 1.6445, "step": 46240 }, { "epoch": 0.06, "learning_rate": 4.7109375e-05, "loss": 1.658, "step": 46250 }, { "epoch": 0.06, "learning_rate": 4.710875e-05, "loss": 1.9047, "step": 46260 }, { "epoch": 0.06, "learning_rate": 4.7108125000000006e-05, "loss": 1.609, "step": 46270 }, { "epoch": 0.06, "learning_rate": 4.71075e-05, "loss": 1.9605, "step": 46280 }, { "epoch": 0.06, "learning_rate": 4.7106875e-05, "loss": 1.6838, "step": 46290 }, { "epoch": 0.06, "learning_rate": 4.710625e-05, "loss": 1.9252, "step": 46300 }, { "epoch": 0.06, "learning_rate": 4.7105625e-05, "loss": 1.6935, "step": 46310 }, { "epoch": 0.06, "learning_rate": 4.7105000000000004e-05, "loss": 1.8132, "step": 46320 }, { "epoch": 0.06, "learning_rate": 4.7104375e-05, "loss": 1.8969, "step": 46330 }, { "epoch": 0.06, "learning_rate": 4.7103750000000005e-05, "loss": 1.7965, "step": 46340 }, { "epoch": 0.06, "learning_rate": 4.7103125e-05, "loss": 1.8416, "step": 46350 }, { "epoch": 0.06, "learning_rate": 4.7102500000000005e-05, "loss": 1.9083, "step": 46360 }, { "epoch": 0.06, "learning_rate": 4.7101875e-05, "loss": 1.8758, "step": 46370 }, { "epoch": 0.06, "learning_rate": 4.710125e-05, "loss": 1.9706, "step": 46380 }, { "epoch": 0.06, "learning_rate": 4.7100625e-05, "loss": 1.7405, "step": 46390 }, { "epoch": 0.06, "learning_rate": 4.71e-05, "loss": 1.4552, "step": 46400 }, { "epoch": 0.06, "learning_rate": 4.7099375e-05, "loss": 1.6857, "step": 46410 }, { "epoch": 0.06, "learning_rate": 4.709875e-05, "loss": 1.5975, "step": 46420 }, { "epoch": 0.06, "learning_rate": 4.7098125000000004e-05, "loss": 1.9716, "step": 46430 }, { "epoch": 0.06, "learning_rate": 4.70975e-05, "loss": 1.8754, "step": 46440 }, { "epoch": 0.06, "learning_rate": 4.7096875000000005e-05, "loss": 2.0014, "step": 46450 }, { "epoch": 0.06, "learning_rate": 4.709625e-05, "loss": 1.8097, "step": 46460 }, { "epoch": 0.06, "learning_rate": 4.7095625000000005e-05, "loss": 1.6657, "step": 46470 }, { "epoch": 0.06, "learning_rate": 4.7095e-05, "loss": 1.7314, "step": 46480 }, { "epoch": 0.06, "learning_rate": 4.7094375e-05, "loss": 1.838, "step": 46490 }, { "epoch": 0.06, "learning_rate": 4.709375e-05, "loss": 1.7141, "step": 46500 }, { "epoch": 0.06, "learning_rate": 4.709312500000001e-05, "loss": 1.7565, "step": 46510 }, { "epoch": 0.06, "learning_rate": 4.7092500000000004e-05, "loss": 1.7611, "step": 46520 }, { "epoch": 0.06, "learning_rate": 4.7091875e-05, "loss": 1.6871, "step": 46530 }, { "epoch": 0.06, "learning_rate": 4.7091250000000004e-05, "loss": 2.0683, "step": 46540 }, { "epoch": 0.06, "learning_rate": 4.709062500000001e-05, "loss": 1.9455, "step": 46550 }, { "epoch": 0.06, "learning_rate": 4.709e-05, "loss": 1.7699, "step": 46560 }, { "epoch": 0.06, "learning_rate": 4.7089375e-05, "loss": 1.7574, "step": 46570 }, { "epoch": 0.06, "learning_rate": 4.7088750000000005e-05, "loss": 1.5528, "step": 46580 }, { "epoch": 0.06, "learning_rate": 4.7088125e-05, "loss": 1.8084, "step": 46590 }, { "epoch": 0.06, "learning_rate": 4.70875e-05, "loss": 1.6627, "step": 46600 }, { "epoch": 0.06, "learning_rate": 4.7086875e-05, "loss": 1.8545, "step": 46610 }, { "epoch": 0.06, "learning_rate": 4.708625000000001e-05, "loss": 1.8095, "step": 46620 }, { "epoch": 0.06, "learning_rate": 4.7085625e-05, "loss": 1.5896, "step": 46630 }, { "epoch": 0.06, "learning_rate": 4.7085e-05, "loss": 1.8501, "step": 46640 }, { "epoch": 0.06, "learning_rate": 4.7084375000000004e-05, "loss": 1.7758, "step": 46650 }, { "epoch": 0.06, "learning_rate": 4.708375e-05, "loss": 1.5861, "step": 46660 }, { "epoch": 0.06, "learning_rate": 4.7083125e-05, "loss": 1.8786, "step": 46670 }, { "epoch": 0.06, "learning_rate": 4.70825e-05, "loss": 1.8421, "step": 46680 }, { "epoch": 0.07, "learning_rate": 4.7081875000000006e-05, "loss": 1.6198, "step": 46690 }, { "epoch": 0.07, "learning_rate": 4.708125e-05, "loss": 1.7773, "step": 46700 }, { "epoch": 0.07, "learning_rate": 4.7080625e-05, "loss": 1.4585, "step": 46710 }, { "epoch": 0.07, "learning_rate": 4.708e-05, "loss": 1.8123, "step": 46720 }, { "epoch": 0.07, "learning_rate": 4.7079375e-05, "loss": 1.6608, "step": 46730 }, { "epoch": 0.07, "learning_rate": 4.7078750000000004e-05, "loss": 1.4882, "step": 46740 }, { "epoch": 0.07, "learning_rate": 4.7078125e-05, "loss": 1.9458, "step": 46750 }, { "epoch": 0.07, "learning_rate": 4.7077500000000005e-05, "loss": 2.002, "step": 46760 }, { "epoch": 0.07, "learning_rate": 4.7076875e-05, "loss": 1.7838, "step": 46770 }, { "epoch": 0.07, "learning_rate": 4.707625e-05, "loss": 1.8363, "step": 46780 }, { "epoch": 0.07, "learning_rate": 4.7075625e-05, "loss": 1.7806, "step": 46790 }, { "epoch": 0.07, "learning_rate": 4.7075e-05, "loss": 1.5177, "step": 46800 }, { "epoch": 0.07, "learning_rate": 4.7074375e-05, "loss": 1.7043, "step": 46810 }, { "epoch": 0.07, "learning_rate": 4.707375e-05, "loss": 1.8277, "step": 46820 }, { "epoch": 0.07, "learning_rate": 4.7073125000000003e-05, "loss": 1.8911, "step": 46830 }, { "epoch": 0.07, "learning_rate": 4.70725e-05, "loss": 1.8976, "step": 46840 }, { "epoch": 0.07, "learning_rate": 4.7071875000000004e-05, "loss": 1.6003, "step": 46850 }, { "epoch": 0.07, "learning_rate": 4.707125e-05, "loss": 1.7409, "step": 46860 }, { "epoch": 0.07, "learning_rate": 4.7070625000000005e-05, "loss": 1.9016, "step": 46870 }, { "epoch": 0.07, "learning_rate": 4.707e-05, "loss": 1.5142, "step": 46880 }, { "epoch": 0.07, "learning_rate": 4.7069375000000005e-05, "loss": 1.8242, "step": 46890 }, { "epoch": 0.07, "learning_rate": 4.706875e-05, "loss": 1.6512, "step": 46900 }, { "epoch": 0.07, "learning_rate": 4.7068125e-05, "loss": 1.656, "step": 46910 }, { "epoch": 0.07, "learning_rate": 4.70675e-05, "loss": 1.786, "step": 46920 }, { "epoch": 0.07, "learning_rate": 4.706687500000001e-05, "loss": 1.651, "step": 46930 }, { "epoch": 0.07, "learning_rate": 4.7066250000000004e-05, "loss": 1.7643, "step": 46940 }, { "epoch": 0.07, "learning_rate": 4.7065625e-05, "loss": 1.8514, "step": 46950 }, { "epoch": 0.07, "learning_rate": 4.7065000000000004e-05, "loss": 1.8097, "step": 46960 }, { "epoch": 0.07, "learning_rate": 4.706437500000001e-05, "loss": 1.6851, "step": 46970 }, { "epoch": 0.07, "learning_rate": 4.706375e-05, "loss": 1.8732, "step": 46980 }, { "epoch": 0.07, "learning_rate": 4.7063125e-05, "loss": 1.8865, "step": 46990 }, { "epoch": 0.07, "learning_rate": 4.7062500000000006e-05, "loss": 1.6821, "step": 47000 }, { "epoch": 0.07, "eval_loss": 1.682392954826355, "eval_runtime": 1328.9479, "eval_samples_per_second": 43.671, "eval_steps_per_second": 5.459, "step": 47000 }, { "epoch": 0.07, "learning_rate": 4.7061875e-05, "loss": 1.6824, "step": 47010 }, { "epoch": 0.07, "learning_rate": 4.706125e-05, "loss": 1.7649, "step": 47020 }, { "epoch": 0.07, "learning_rate": 4.7060625e-05, "loss": 1.7233, "step": 47030 }, { "epoch": 0.07, "learning_rate": 4.706000000000001e-05, "loss": 1.5288, "step": 47040 }, { "epoch": 0.07, "learning_rate": 4.7059375e-05, "loss": 1.8885, "step": 47050 }, { "epoch": 0.07, "learning_rate": 4.705875e-05, "loss": 1.7922, "step": 47060 }, { "epoch": 0.07, "learning_rate": 4.7058125000000004e-05, "loss": 1.7274, "step": 47070 }, { "epoch": 0.07, "learning_rate": 4.70575e-05, "loss": 1.7168, "step": 47080 }, { "epoch": 0.07, "learning_rate": 4.7056875e-05, "loss": 1.5844, "step": 47090 }, { "epoch": 0.07, "learning_rate": 4.705625e-05, "loss": 1.5497, "step": 47100 }, { "epoch": 0.07, "learning_rate": 4.7055625000000006e-05, "loss": 1.7436, "step": 47110 }, { "epoch": 0.07, "learning_rate": 4.7055e-05, "loss": 1.751, "step": 47120 }, { "epoch": 0.07, "learning_rate": 4.7054375e-05, "loss": 1.8959, "step": 47130 }, { "epoch": 0.07, "learning_rate": 4.705375e-05, "loss": 1.8778, "step": 47140 }, { "epoch": 0.07, "learning_rate": 4.7053125e-05, "loss": 1.8243, "step": 47150 }, { "epoch": 0.07, "learning_rate": 4.7052500000000004e-05, "loss": 1.7414, "step": 47160 }, { "epoch": 0.07, "learning_rate": 4.7051875e-05, "loss": 1.7727, "step": 47170 }, { "epoch": 0.07, "learning_rate": 4.7051250000000005e-05, "loss": 1.5553, "step": 47180 }, { "epoch": 0.07, "learning_rate": 4.7050625e-05, "loss": 1.5671, "step": 47190 }, { "epoch": 0.07, "learning_rate": 4.705e-05, "loss": 1.7702, "step": 47200 }, { "epoch": 0.07, "learning_rate": 4.7049375e-05, "loss": 1.8159, "step": 47210 }, { "epoch": 0.07, "learning_rate": 4.7048750000000006e-05, "loss": 1.8589, "step": 47220 }, { "epoch": 0.07, "learning_rate": 4.7048125e-05, "loss": 1.8105, "step": 47230 }, { "epoch": 0.07, "learning_rate": 4.70475e-05, "loss": 1.7047, "step": 47240 }, { "epoch": 0.07, "learning_rate": 4.7046875000000003e-05, "loss": 1.878, "step": 47250 }, { "epoch": 0.07, "learning_rate": 4.704625e-05, "loss": 1.6006, "step": 47260 }, { "epoch": 0.07, "learning_rate": 4.7045625000000004e-05, "loss": 1.7871, "step": 47270 }, { "epoch": 0.07, "learning_rate": 4.7045e-05, "loss": 1.8282, "step": 47280 }, { "epoch": 0.07, "learning_rate": 4.7044375000000005e-05, "loss": 1.7589, "step": 47290 }, { "epoch": 0.07, "learning_rate": 4.704375e-05, "loss": 1.5551, "step": 47300 }, { "epoch": 0.07, "learning_rate": 4.7043125000000005e-05, "loss": 1.7461, "step": 47310 }, { "epoch": 0.07, "learning_rate": 4.70425e-05, "loss": 1.7396, "step": 47320 }, { "epoch": 0.07, "learning_rate": 4.7041875e-05, "loss": 1.8538, "step": 47330 }, { "epoch": 0.07, "learning_rate": 4.704125e-05, "loss": 1.8016, "step": 47340 }, { "epoch": 0.07, "learning_rate": 4.704062500000001e-05, "loss": 1.7603, "step": 47350 }, { "epoch": 0.07, "learning_rate": 4.7040000000000004e-05, "loss": 1.768, "step": 47360 }, { "epoch": 0.07, "learning_rate": 4.7039375e-05, "loss": 1.6452, "step": 47370 }, { "epoch": 0.07, "learning_rate": 4.7038750000000004e-05, "loss": 1.6384, "step": 47380 }, { "epoch": 0.07, "learning_rate": 4.7038125e-05, "loss": 1.9357, "step": 47390 }, { "epoch": 0.07, "learning_rate": 4.70375e-05, "loss": 1.5978, "step": 47400 }, { "epoch": 0.07, "learning_rate": 4.7036875e-05, "loss": 1.6547, "step": 47410 }, { "epoch": 0.07, "learning_rate": 4.7036250000000006e-05, "loss": 1.8335, "step": 47420 }, { "epoch": 0.07, "learning_rate": 4.7035625e-05, "loss": 1.628, "step": 47430 }, { "epoch": 0.07, "learning_rate": 4.7035e-05, "loss": 1.8228, "step": 47440 }, { "epoch": 0.07, "learning_rate": 4.7034375e-05, "loss": 1.7545, "step": 47450 }, { "epoch": 0.07, "learning_rate": 4.703375000000001e-05, "loss": 1.6737, "step": 47460 }, { "epoch": 0.07, "learning_rate": 4.7033125e-05, "loss": 1.6821, "step": 47470 }, { "epoch": 0.07, "learning_rate": 4.70325e-05, "loss": 1.7149, "step": 47480 }, { "epoch": 0.07, "learning_rate": 4.7031875000000004e-05, "loss": 1.5556, "step": 47490 }, { "epoch": 0.07, "learning_rate": 4.703125e-05, "loss": 1.5792, "step": 47500 }, { "epoch": 0.07, "learning_rate": 4.7030625e-05, "loss": 1.9228, "step": 47510 }, { "epoch": 0.07, "learning_rate": 4.703e-05, "loss": 2.0023, "step": 47520 }, { "epoch": 0.07, "learning_rate": 4.7029375000000006e-05, "loss": 1.817, "step": 47530 }, { "epoch": 0.07, "learning_rate": 4.702875e-05, "loss": 1.7949, "step": 47540 }, { "epoch": 0.07, "learning_rate": 4.7028125e-05, "loss": 1.8855, "step": 47550 }, { "epoch": 0.07, "learning_rate": 4.70275e-05, "loss": 1.6467, "step": 47560 }, { "epoch": 0.07, "learning_rate": 4.7026875e-05, "loss": 1.7353, "step": 47570 }, { "epoch": 0.07, "learning_rate": 4.7026250000000004e-05, "loss": 1.8955, "step": 47580 }, { "epoch": 0.07, "learning_rate": 4.7025625e-05, "loss": 1.5979, "step": 47590 }, { "epoch": 0.07, "learning_rate": 4.7025000000000005e-05, "loss": 1.6725, "step": 47600 }, { "epoch": 0.07, "learning_rate": 4.7024375e-05, "loss": 1.7496, "step": 47610 }, { "epoch": 0.07, "learning_rate": 4.702375e-05, "loss": 1.6653, "step": 47620 }, { "epoch": 0.07, "learning_rate": 4.7023125e-05, "loss": 1.4771, "step": 47630 }, { "epoch": 0.07, "learning_rate": 4.7022500000000006e-05, "loss": 2.0641, "step": 47640 }, { "epoch": 0.07, "learning_rate": 4.7021875e-05, "loss": 1.6933, "step": 47650 }, { "epoch": 0.07, "learning_rate": 4.702125e-05, "loss": 1.4306, "step": 47660 }, { "epoch": 0.07, "learning_rate": 4.7020625000000003e-05, "loss": 1.5739, "step": 47670 }, { "epoch": 0.07, "learning_rate": 4.702e-05, "loss": 1.6415, "step": 47680 }, { "epoch": 0.07, "learning_rate": 4.7019375000000004e-05, "loss": 1.8337, "step": 47690 }, { "epoch": 0.07, "learning_rate": 4.701875e-05, "loss": 1.7281, "step": 47700 }, { "epoch": 0.07, "learning_rate": 4.7018125000000005e-05, "loss": 1.6337, "step": 47710 }, { "epoch": 0.07, "learning_rate": 4.70175e-05, "loss": 1.8221, "step": 47720 }, { "epoch": 0.07, "learning_rate": 4.7016875000000005e-05, "loss": 1.5906, "step": 47730 }, { "epoch": 0.07, "learning_rate": 4.701625e-05, "loss": 1.6432, "step": 47740 }, { "epoch": 0.07, "learning_rate": 4.7015625e-05, "loss": 1.8354, "step": 47750 }, { "epoch": 0.07, "learning_rate": 4.7015e-05, "loss": 1.4755, "step": 47760 }, { "epoch": 0.07, "learning_rate": 4.701437500000001e-05, "loss": 1.7458, "step": 47770 }, { "epoch": 0.07, "learning_rate": 4.7013750000000004e-05, "loss": 1.6076, "step": 47780 }, { "epoch": 0.07, "learning_rate": 4.7013125e-05, "loss": 1.5722, "step": 47790 }, { "epoch": 0.07, "learning_rate": 4.7012500000000004e-05, "loss": 1.6865, "step": 47800 }, { "epoch": 0.07, "learning_rate": 4.7011875e-05, "loss": 1.6873, "step": 47810 }, { "epoch": 0.07, "learning_rate": 4.701125e-05, "loss": 1.7008, "step": 47820 }, { "epoch": 0.07, "learning_rate": 4.7010625e-05, "loss": 1.7259, "step": 47830 }, { "epoch": 0.07, "learning_rate": 4.7010000000000006e-05, "loss": 1.8682, "step": 47840 }, { "epoch": 0.07, "learning_rate": 4.7009375e-05, "loss": 1.633, "step": 47850 }, { "epoch": 0.07, "learning_rate": 4.700875e-05, "loss": 1.7121, "step": 47860 }, { "epoch": 0.07, "learning_rate": 4.7008125e-05, "loss": 1.6795, "step": 47870 }, { "epoch": 0.07, "learning_rate": 4.700750000000001e-05, "loss": 1.6594, "step": 47880 }, { "epoch": 0.07, "learning_rate": 4.7006875e-05, "loss": 1.6286, "step": 47890 }, { "epoch": 0.07, "learning_rate": 4.700625e-05, "loss": 1.7596, "step": 47900 }, { "epoch": 0.07, "learning_rate": 4.7005625000000004e-05, "loss": 1.6736, "step": 47910 }, { "epoch": 0.07, "learning_rate": 4.7005e-05, "loss": 1.5747, "step": 47920 }, { "epoch": 0.07, "learning_rate": 4.7004375e-05, "loss": 1.6828, "step": 47930 }, { "epoch": 0.07, "learning_rate": 4.700375e-05, "loss": 1.7275, "step": 47940 }, { "epoch": 0.07, "learning_rate": 4.7003125000000006e-05, "loss": 1.6942, "step": 47950 }, { "epoch": 0.07, "learning_rate": 4.70025e-05, "loss": 1.4673, "step": 47960 }, { "epoch": 0.07, "learning_rate": 4.7001875e-05, "loss": 1.801, "step": 47970 }, { "epoch": 0.07, "learning_rate": 4.700125e-05, "loss": 1.8147, "step": 47980 }, { "epoch": 0.07, "learning_rate": 4.7000625e-05, "loss": 2.1111, "step": 47990 }, { "epoch": 0.07, "learning_rate": 4.7e-05, "loss": 1.7247, "step": 48000 }, { "epoch": 0.07, "eval_loss": 1.6827584505081177, "eval_runtime": 1327.8355, "eval_samples_per_second": 43.707, "eval_steps_per_second": 5.464, "step": 48000 }, { "epoch": 0.07, "learning_rate": 4.6999375e-05, "loss": 1.6552, "step": 48010 }, { "epoch": 0.07, "learning_rate": 4.6998750000000005e-05, "loss": 1.6061, "step": 48020 }, { "epoch": 0.07, "learning_rate": 4.6998125e-05, "loss": 1.7625, "step": 48030 }, { "epoch": 0.07, "learning_rate": 4.69975e-05, "loss": 1.6149, "step": 48040 }, { "epoch": 0.07, "learning_rate": 4.6996875e-05, "loss": 1.8497, "step": 48050 }, { "epoch": 0.07, "learning_rate": 4.6996250000000006e-05, "loss": 1.9117, "step": 48060 }, { "epoch": 0.07, "learning_rate": 4.6995625e-05, "loss": 1.9146, "step": 48070 }, { "epoch": 0.07, "learning_rate": 4.6995e-05, "loss": 1.7347, "step": 48080 }, { "epoch": 0.07, "learning_rate": 4.6994375000000004e-05, "loss": 1.7672, "step": 48090 }, { "epoch": 0.07, "learning_rate": 4.699375e-05, "loss": 1.557, "step": 48100 }, { "epoch": 0.07, "learning_rate": 4.6993125000000004e-05, "loss": 1.5431, "step": 48110 }, { "epoch": 0.07, "learning_rate": 4.69925e-05, "loss": 1.4898, "step": 48120 }, { "epoch": 0.07, "learning_rate": 4.6991875000000005e-05, "loss": 1.642, "step": 48130 }, { "epoch": 0.07, "learning_rate": 4.699125e-05, "loss": 1.5947, "step": 48140 }, { "epoch": 0.07, "learning_rate": 4.6990625000000005e-05, "loss": 1.5966, "step": 48150 }, { "epoch": 0.07, "learning_rate": 4.699e-05, "loss": 1.8295, "step": 48160 }, { "epoch": 0.07, "learning_rate": 4.6989375e-05, "loss": 1.9641, "step": 48170 }, { "epoch": 0.07, "learning_rate": 4.698875e-05, "loss": 1.5316, "step": 48180 }, { "epoch": 0.07, "learning_rate": 4.6988125e-05, "loss": 1.9691, "step": 48190 }, { "epoch": 0.07, "learning_rate": 4.6987500000000004e-05, "loss": 1.5519, "step": 48200 }, { "epoch": 0.07, "learning_rate": 4.6986875e-05, "loss": 1.7338, "step": 48210 }, { "epoch": 0.07, "learning_rate": 4.6986250000000004e-05, "loss": 1.5956, "step": 48220 }, { "epoch": 0.07, "learning_rate": 4.6985625e-05, "loss": 1.7107, "step": 48230 }, { "epoch": 0.07, "learning_rate": 4.6985e-05, "loss": 1.7571, "step": 48240 }, { "epoch": 0.07, "learning_rate": 4.6984375e-05, "loss": 1.4452, "step": 48250 }, { "epoch": 0.07, "learning_rate": 4.6983750000000006e-05, "loss": 1.7728, "step": 48260 }, { "epoch": 0.07, "learning_rate": 4.6983125e-05, "loss": 1.5964, "step": 48270 }, { "epoch": 0.07, "learning_rate": 4.69825e-05, "loss": 1.5917, "step": 48280 }, { "epoch": 0.07, "learning_rate": 4.6981875e-05, "loss": 1.7987, "step": 48290 }, { "epoch": 0.07, "learning_rate": 4.698125000000001e-05, "loss": 1.7431, "step": 48300 }, { "epoch": 0.07, "learning_rate": 4.6980625e-05, "loss": 1.6221, "step": 48310 }, { "epoch": 0.07, "learning_rate": 4.698e-05, "loss": 1.6824, "step": 48320 }, { "epoch": 0.07, "learning_rate": 4.6979375000000005e-05, "loss": 1.7554, "step": 48330 }, { "epoch": 0.07, "learning_rate": 4.697875e-05, "loss": 1.7718, "step": 48340 }, { "epoch": 0.07, "learning_rate": 4.6978125e-05, "loss": 1.8378, "step": 48350 }, { "epoch": 0.07, "learning_rate": 4.69775e-05, "loss": 1.7128, "step": 48360 }, { "epoch": 0.07, "learning_rate": 4.6976875000000006e-05, "loss": 1.7037, "step": 48370 }, { "epoch": 0.07, "learning_rate": 4.697625e-05, "loss": 1.8306, "step": 48380 }, { "epoch": 0.07, "learning_rate": 4.6975625e-05, "loss": 1.5489, "step": 48390 }, { "epoch": 0.07, "learning_rate": 4.6975000000000003e-05, "loss": 1.7996, "step": 48400 }, { "epoch": 0.07, "learning_rate": 4.697437500000001e-05, "loss": 1.7621, "step": 48410 }, { "epoch": 0.07, "learning_rate": 4.697375e-05, "loss": 1.6789, "step": 48420 }, { "epoch": 0.07, "learning_rate": 4.6973125e-05, "loss": 1.8648, "step": 48430 }, { "epoch": 0.07, "learning_rate": 4.6972500000000005e-05, "loss": 1.7058, "step": 48440 }, { "epoch": 0.07, "learning_rate": 4.6971875e-05, "loss": 1.5696, "step": 48450 }, { "epoch": 0.07, "learning_rate": 4.697125e-05, "loss": 1.723, "step": 48460 }, { "epoch": 0.07, "learning_rate": 4.6970625e-05, "loss": 1.8439, "step": 48470 }, { "epoch": 0.07, "learning_rate": 4.6970000000000006e-05, "loss": 1.7877, "step": 48480 }, { "epoch": 0.07, "learning_rate": 4.6969375e-05, "loss": 2.0074, "step": 48490 }, { "epoch": 0.07, "learning_rate": 4.696875e-05, "loss": 1.8098, "step": 48500 }, { "epoch": 0.07, "learning_rate": 4.6968125000000004e-05, "loss": 1.9502, "step": 48510 }, { "epoch": 0.07, "learning_rate": 4.69675e-05, "loss": 1.6452, "step": 48520 }, { "epoch": 0.07, "learning_rate": 4.6966875000000004e-05, "loss": 1.7543, "step": 48530 }, { "epoch": 0.07, "learning_rate": 4.696625e-05, "loss": 1.9316, "step": 48540 }, { "epoch": 0.07, "learning_rate": 4.6965625000000005e-05, "loss": 1.5814, "step": 48550 }, { "epoch": 0.07, "learning_rate": 4.6965e-05, "loss": 1.6987, "step": 48560 }, { "epoch": 0.07, "learning_rate": 4.6964375000000006e-05, "loss": 1.7932, "step": 48570 }, { "epoch": 0.07, "learning_rate": 4.696375e-05, "loss": 1.868, "step": 48580 }, { "epoch": 0.07, "learning_rate": 4.6963125e-05, "loss": 1.4799, "step": 48590 }, { "epoch": 0.07, "learning_rate": 4.69625e-05, "loss": 1.8285, "step": 48600 }, { "epoch": 0.07, "learning_rate": 4.6961875e-05, "loss": 1.787, "step": 48610 }, { "epoch": 0.07, "learning_rate": 4.6961250000000004e-05, "loss": 1.6871, "step": 48620 }, { "epoch": 0.07, "learning_rate": 4.6960625e-05, "loss": 1.508, "step": 48630 }, { "epoch": 0.07, "learning_rate": 4.6960000000000004e-05, "loss": 1.7582, "step": 48640 }, { "epoch": 0.07, "learning_rate": 4.6959375e-05, "loss": 1.8005, "step": 48650 }, { "epoch": 0.07, "learning_rate": 4.695875e-05, "loss": 1.9249, "step": 48660 }, { "epoch": 0.07, "learning_rate": 4.6958125e-05, "loss": 1.7856, "step": 48670 }, { "epoch": 0.07, "learning_rate": 4.6957500000000006e-05, "loss": 1.7103, "step": 48680 }, { "epoch": 0.07, "learning_rate": 4.6956875e-05, "loss": 1.9194, "step": 48690 }, { "epoch": 0.07, "learning_rate": 4.695625e-05, "loss": 1.4847, "step": 48700 }, { "epoch": 0.07, "learning_rate": 4.6955625e-05, "loss": 1.6888, "step": 48710 }, { "epoch": 0.07, "learning_rate": 4.695500000000001e-05, "loss": 1.7777, "step": 48720 }, { "epoch": 0.07, "learning_rate": 4.6954375e-05, "loss": 1.8284, "step": 48730 }, { "epoch": 0.07, "learning_rate": 4.695375e-05, "loss": 1.6662, "step": 48740 }, { "epoch": 0.07, "learning_rate": 4.6953125000000005e-05, "loss": 1.5582, "step": 48750 }, { "epoch": 0.07, "learning_rate": 4.69525e-05, "loss": 1.9744, "step": 48760 }, { "epoch": 0.07, "learning_rate": 4.6951875e-05, "loss": 1.5618, "step": 48770 }, { "epoch": 0.07, "learning_rate": 4.695125e-05, "loss": 1.9778, "step": 48780 }, { "epoch": 0.07, "learning_rate": 4.6950625000000006e-05, "loss": 1.6662, "step": 48790 }, { "epoch": 0.07, "learning_rate": 4.695e-05, "loss": 1.8352, "step": 48800 }, { "epoch": 0.07, "learning_rate": 4.6949375e-05, "loss": 1.5396, "step": 48810 }, { "epoch": 0.07, "learning_rate": 4.6948750000000003e-05, "loss": 1.7946, "step": 48820 }, { "epoch": 0.07, "learning_rate": 4.694812500000001e-05, "loss": 1.6593, "step": 48830 }, { "epoch": 0.07, "learning_rate": 4.69475e-05, "loss": 1.634, "step": 48840 }, { "epoch": 0.07, "learning_rate": 4.6946875e-05, "loss": 1.9781, "step": 48850 }, { "epoch": 0.07, "learning_rate": 4.6946250000000005e-05, "loss": 1.9453, "step": 48860 }, { "epoch": 0.07, "learning_rate": 4.6945625e-05, "loss": 1.6003, "step": 48870 }, { "epoch": 0.07, "learning_rate": 4.6945e-05, "loss": 1.8916, "step": 48880 }, { "epoch": 0.07, "learning_rate": 4.6944375e-05, "loss": 1.6673, "step": 48890 }, { "epoch": 0.07, "learning_rate": 4.6943750000000006e-05, "loss": 1.7402, "step": 48900 }, { "epoch": 0.07, "learning_rate": 4.6943125e-05, "loss": 1.6876, "step": 48910 }, { "epoch": 0.07, "learning_rate": 4.69425e-05, "loss": 1.538, "step": 48920 }, { "epoch": 0.07, "learning_rate": 4.6941875000000004e-05, "loss": 1.6167, "step": 48930 }, { "epoch": 0.07, "learning_rate": 4.694125e-05, "loss": 1.9429, "step": 48940 }, { "epoch": 0.07, "learning_rate": 4.6940625000000004e-05, "loss": 1.6842, "step": 48950 }, { "epoch": 0.07, "learning_rate": 4.694e-05, "loss": 1.6071, "step": 48960 }, { "epoch": 0.07, "learning_rate": 4.6939375000000005e-05, "loss": 1.6419, "step": 48970 }, { "epoch": 0.07, "learning_rate": 4.693875e-05, "loss": 1.7951, "step": 48980 }, { "epoch": 0.07, "learning_rate": 4.6938125000000006e-05, "loss": 1.9549, "step": 48990 }, { "epoch": 0.07, "learning_rate": 4.69375e-05, "loss": 1.7082, "step": 49000 }, { "epoch": 0.07, "eval_loss": 1.6669126749038696, "eval_runtime": 1293.9192, "eval_samples_per_second": 44.853, "eval_steps_per_second": 5.607, "step": 49000 }, { "epoch": 0.07, "learning_rate": 4.6936875e-05, "loss": 1.7864, "step": 49010 }, { "epoch": 0.07, "learning_rate": 4.693625e-05, "loss": 1.6495, "step": 49020 }, { "epoch": 0.07, "learning_rate": 4.6935625e-05, "loss": 1.718, "step": 49030 }, { "epoch": 0.07, "learning_rate": 4.6935000000000004e-05, "loss": 1.7556, "step": 49040 }, { "epoch": 0.07, "learning_rate": 4.6934375e-05, "loss": 1.7374, "step": 49050 }, { "epoch": 0.07, "learning_rate": 4.6933750000000004e-05, "loss": 1.8438, "step": 49060 }, { "epoch": 0.07, "learning_rate": 4.6933125e-05, "loss": 1.9412, "step": 49070 }, { "epoch": 0.07, "learning_rate": 4.69325e-05, "loss": 1.7973, "step": 49080 }, { "epoch": 0.07, "learning_rate": 4.6931875e-05, "loss": 1.7877, "step": 49090 }, { "epoch": 0.07, "learning_rate": 4.6931250000000006e-05, "loss": 1.9075, "step": 49100 }, { "epoch": 0.07, "learning_rate": 4.6930625e-05, "loss": 1.6436, "step": 49110 }, { "epoch": 0.07, "learning_rate": 4.693e-05, "loss": 1.8801, "step": 49120 }, { "epoch": 0.07, "learning_rate": 4.6929375e-05, "loss": 1.7587, "step": 49130 }, { "epoch": 0.07, "learning_rate": 4.692875000000001e-05, "loss": 1.6872, "step": 49140 }, { "epoch": 0.07, "learning_rate": 4.6928125e-05, "loss": 1.6874, "step": 49150 }, { "epoch": 0.07, "learning_rate": 4.69275e-05, "loss": 1.7026, "step": 49160 }, { "epoch": 0.07, "learning_rate": 4.6926875000000005e-05, "loss": 1.6897, "step": 49170 }, { "epoch": 0.07, "learning_rate": 4.692625e-05, "loss": 1.931, "step": 49180 }, { "epoch": 0.07, "learning_rate": 4.6925625e-05, "loss": 1.9775, "step": 49190 }, { "epoch": 0.07, "learning_rate": 4.6925e-05, "loss": 1.6127, "step": 49200 }, { "epoch": 0.07, "learning_rate": 4.6924375000000006e-05, "loss": 1.5568, "step": 49210 }, { "epoch": 0.07, "learning_rate": 4.692375e-05, "loss": 1.5898, "step": 49220 }, { "epoch": 0.07, "learning_rate": 4.6923125e-05, "loss": 1.6359, "step": 49230 }, { "epoch": 0.07, "learning_rate": 4.6922500000000003e-05, "loss": 1.7508, "step": 49240 }, { "epoch": 0.07, "learning_rate": 4.692187500000001e-05, "loss": 1.6958, "step": 49250 }, { "epoch": 0.07, "learning_rate": 4.692125e-05, "loss": 1.8264, "step": 49260 }, { "epoch": 0.07, "learning_rate": 4.6920625e-05, "loss": 1.7009, "step": 49270 }, { "epoch": 0.07, "learning_rate": 4.6920000000000005e-05, "loss": 1.6218, "step": 49280 }, { "epoch": 0.07, "learning_rate": 4.6919375e-05, "loss": 1.5956, "step": 49290 }, { "epoch": 0.07, "learning_rate": 4.691875e-05, "loss": 1.753, "step": 49300 }, { "epoch": 0.07, "learning_rate": 4.6918125e-05, "loss": 1.7618, "step": 49310 }, { "epoch": 0.07, "learning_rate": 4.6917500000000006e-05, "loss": 1.6442, "step": 49320 }, { "epoch": 0.07, "learning_rate": 4.6916875e-05, "loss": 1.6361, "step": 49330 }, { "epoch": 0.07, "learning_rate": 4.691625e-05, "loss": 1.7239, "step": 49340 }, { "epoch": 0.07, "learning_rate": 4.6915625000000004e-05, "loss": 1.7736, "step": 49350 }, { "epoch": 0.07, "learning_rate": 4.6915e-05, "loss": 1.7147, "step": 49360 }, { "epoch": 0.07, "learning_rate": 4.6914375000000004e-05, "loss": 1.7432, "step": 49370 }, { "epoch": 0.07, "learning_rate": 4.691375e-05, "loss": 1.935, "step": 49380 }, { "epoch": 0.07, "learning_rate": 4.6913125000000005e-05, "loss": 1.6111, "step": 49390 }, { "epoch": 0.07, "learning_rate": 4.69125e-05, "loss": 1.791, "step": 49400 }, { "epoch": 0.07, "learning_rate": 4.6911875e-05, "loss": 1.8555, "step": 49410 }, { "epoch": 0.07, "learning_rate": 4.691125e-05, "loss": 1.6804, "step": 49420 }, { "epoch": 0.07, "learning_rate": 4.6910625e-05, "loss": 1.8422, "step": 49430 }, { "epoch": 0.07, "learning_rate": 4.691e-05, "loss": 1.7493, "step": 49440 }, { "epoch": 0.07, "learning_rate": 4.6909375e-05, "loss": 1.5206, "step": 49450 }, { "epoch": 0.07, "learning_rate": 4.6908750000000004e-05, "loss": 1.6937, "step": 49460 }, { "epoch": 0.07, "learning_rate": 4.6908125e-05, "loss": 1.4895, "step": 49470 }, { "epoch": 0.07, "learning_rate": 4.6907500000000004e-05, "loss": 1.6194, "step": 49480 }, { "epoch": 0.07, "learning_rate": 4.6906875e-05, "loss": 1.7321, "step": 49490 }, { "epoch": 0.07, "learning_rate": 4.690625e-05, "loss": 1.6497, "step": 49500 }, { "epoch": 0.07, "learning_rate": 4.6905625e-05, "loss": 1.7364, "step": 49510 }, { "epoch": 0.07, "learning_rate": 4.6905000000000006e-05, "loss": 1.634, "step": 49520 }, { "epoch": 0.07, "learning_rate": 4.6904375e-05, "loss": 1.5539, "step": 49530 }, { "epoch": 0.07, "learning_rate": 4.690375e-05, "loss": 1.7851, "step": 49540 }, { "epoch": 0.07, "learning_rate": 4.6903125e-05, "loss": 1.6193, "step": 49550 }, { "epoch": 0.07, "learning_rate": 4.690250000000001e-05, "loss": 1.8082, "step": 49560 }, { "epoch": 0.07, "learning_rate": 4.6901875000000004e-05, "loss": 1.8603, "step": 49570 }, { "epoch": 0.07, "learning_rate": 4.690125e-05, "loss": 1.9281, "step": 49580 }, { "epoch": 0.07, "learning_rate": 4.6900625000000005e-05, "loss": 1.8961, "step": 49590 }, { "epoch": 0.07, "learning_rate": 4.69e-05, "loss": 1.8256, "step": 49600 }, { "epoch": 0.07, "learning_rate": 4.6899375e-05, "loss": 1.7576, "step": 49610 }, { "epoch": 0.07, "learning_rate": 4.689875e-05, "loss": 1.7173, "step": 49620 }, { "epoch": 0.07, "learning_rate": 4.6898125000000006e-05, "loss": 1.7852, "step": 49630 }, { "epoch": 0.07, "learning_rate": 4.68975e-05, "loss": 1.6088, "step": 49640 }, { "epoch": 0.07, "learning_rate": 4.6896875e-05, "loss": 1.6805, "step": 49650 }, { "epoch": 0.07, "learning_rate": 4.6896250000000004e-05, "loss": 1.6099, "step": 49660 }, { "epoch": 0.07, "learning_rate": 4.689562500000001e-05, "loss": 1.6782, "step": 49670 }, { "epoch": 0.07, "learning_rate": 4.6895e-05, "loss": 1.7812, "step": 49680 }, { "epoch": 0.07, "learning_rate": 4.6894375e-05, "loss": 1.6638, "step": 49690 }, { "epoch": 0.07, "learning_rate": 4.6893750000000005e-05, "loss": 1.5904, "step": 49700 }, { "epoch": 0.07, "learning_rate": 4.6893125e-05, "loss": 1.6991, "step": 49710 }, { "epoch": 0.07, "learning_rate": 4.68925e-05, "loss": 1.8575, "step": 49720 }, { "epoch": 0.07, "learning_rate": 4.6891875e-05, "loss": 1.6415, "step": 49730 }, { "epoch": 0.07, "learning_rate": 4.6891250000000006e-05, "loss": 1.6404, "step": 49740 }, { "epoch": 0.07, "learning_rate": 4.6890625e-05, "loss": 1.9822, "step": 49750 }, { "epoch": 0.07, "learning_rate": 4.689e-05, "loss": 1.6369, "step": 49760 }, { "epoch": 0.07, "learning_rate": 4.6889375000000004e-05, "loss": 1.5872, "step": 49770 }, { "epoch": 0.07, "learning_rate": 4.688875e-05, "loss": 1.5277, "step": 49780 }, { "epoch": 0.07, "learning_rate": 4.6888125000000004e-05, "loss": 1.8342, "step": 49790 }, { "epoch": 0.07, "learning_rate": 4.68875e-05, "loss": 1.7174, "step": 49800 }, { "epoch": 0.07, "learning_rate": 4.6886875000000005e-05, "loss": 1.8072, "step": 49810 }, { "epoch": 0.07, "learning_rate": 4.688625e-05, "loss": 1.7179, "step": 49820 }, { "epoch": 0.07, "learning_rate": 4.6885625e-05, "loss": 1.6672, "step": 49830 }, { "epoch": 0.07, "learning_rate": 4.6885e-05, "loss": 1.9375, "step": 49840 }, { "epoch": 0.07, "learning_rate": 4.6884375e-05, "loss": 2.0935, "step": 49850 }, { "epoch": 0.07, "learning_rate": 4.688375e-05, "loss": 1.7959, "step": 49860 }, { "epoch": 0.07, "learning_rate": 4.6883125e-05, "loss": 1.6921, "step": 49870 }, { "epoch": 0.07, "learning_rate": 4.6882500000000004e-05, "loss": 1.8885, "step": 49880 }, { "epoch": 0.07, "learning_rate": 4.6881875e-05, "loss": 1.7175, "step": 49890 }, { "epoch": 0.07, "learning_rate": 4.6881250000000005e-05, "loss": 1.6437, "step": 49900 }, { "epoch": 0.07, "learning_rate": 4.6880625e-05, "loss": 1.8227, "step": 49910 }, { "epoch": 0.07, "learning_rate": 4.688e-05, "loss": 1.6353, "step": 49920 }, { "epoch": 0.07, "learning_rate": 4.6879375e-05, "loss": 1.8072, "step": 49930 }, { "epoch": 0.07, "learning_rate": 4.6878750000000006e-05, "loss": 1.8961, "step": 49940 }, { "epoch": 0.07, "learning_rate": 4.6878125e-05, "loss": 1.6595, "step": 49950 }, { "epoch": 0.07, "learning_rate": 4.68775e-05, "loss": 1.823, "step": 49960 }, { "epoch": 0.07, "learning_rate": 4.6876875000000003e-05, "loss": 1.8803, "step": 49970 }, { "epoch": 0.07, "learning_rate": 4.687625000000001e-05, "loss": 1.6092, "step": 49980 }, { "epoch": 0.07, "learning_rate": 4.6875625000000004e-05, "loss": 1.8178, "step": 49990 }, { "epoch": 0.07, "learning_rate": 4.6875e-05, "loss": 1.6066, "step": 50000 }, { "epoch": 0.07, "eval_loss": 1.658612608909607, "eval_runtime": 1297.6788, "eval_samples_per_second": 44.723, "eval_steps_per_second": 5.591, "step": 50000 }, { "epoch": 0.07, "learning_rate": 4.6874375000000005e-05, "loss": 1.722, "step": 50010 }, { "epoch": 0.07, "learning_rate": 4.687375e-05, "loss": 1.744, "step": 50020 }, { "epoch": 0.07, "learning_rate": 4.6873125e-05, "loss": 1.66, "step": 50030 }, { "epoch": 0.07, "learning_rate": 4.68725e-05, "loss": 1.7563, "step": 50040 }, { "epoch": 0.07, "learning_rate": 4.6871875000000006e-05, "loss": 1.7619, "step": 50050 }, { "epoch": 0.07, "learning_rate": 4.687125e-05, "loss": 1.7177, "step": 50060 }, { "epoch": 0.07, "learning_rate": 4.6870625e-05, "loss": 1.6865, "step": 50070 }, { "epoch": 0.07, "learning_rate": 4.6870000000000004e-05, "loss": 1.6033, "step": 50080 }, { "epoch": 0.07, "learning_rate": 4.686937500000001e-05, "loss": 1.7307, "step": 50090 }, { "epoch": 0.07, "learning_rate": 4.686875e-05, "loss": 1.672, "step": 50100 }, { "epoch": 0.07, "learning_rate": 4.6868125e-05, "loss": 1.8587, "step": 50110 }, { "epoch": 0.07, "learning_rate": 4.6867500000000005e-05, "loss": 1.9373, "step": 50120 }, { "epoch": 0.07, "learning_rate": 4.6866875e-05, "loss": 1.8593, "step": 50130 }, { "epoch": 0.07, "learning_rate": 4.686625e-05, "loss": 1.8185, "step": 50140 }, { "epoch": 0.07, "learning_rate": 4.6865625e-05, "loss": 1.6982, "step": 50150 }, { "epoch": 0.07, "learning_rate": 4.6865000000000006e-05, "loss": 1.7812, "step": 50160 }, { "epoch": 0.07, "learning_rate": 4.6864375e-05, "loss": 1.7878, "step": 50170 }, { "epoch": 0.07, "learning_rate": 4.686375e-05, "loss": 1.666, "step": 50180 }, { "epoch": 0.07, "learning_rate": 4.6863125000000004e-05, "loss": 1.6908, "step": 50190 }, { "epoch": 0.07, "learning_rate": 4.68625e-05, "loss": 1.706, "step": 50200 }, { "epoch": 0.07, "learning_rate": 4.6861875e-05, "loss": 1.7812, "step": 50210 }, { "epoch": 0.07, "learning_rate": 4.686125e-05, "loss": 1.8501, "step": 50220 }, { "epoch": 0.07, "learning_rate": 4.6860625000000005e-05, "loss": 1.7039, "step": 50230 }, { "epoch": 0.07, "learning_rate": 4.686e-05, "loss": 1.8289, "step": 50240 }, { "epoch": 0.07, "learning_rate": 4.6859375e-05, "loss": 1.5037, "step": 50250 }, { "epoch": 0.07, "learning_rate": 4.685875e-05, "loss": 1.9023, "step": 50260 }, { "epoch": 0.07, "learning_rate": 4.6858125e-05, "loss": 1.8388, "step": 50270 }, { "epoch": 0.07, "learning_rate": 4.68575e-05, "loss": 1.7466, "step": 50280 }, { "epoch": 0.07, "learning_rate": 4.6856875e-05, "loss": 1.8722, "step": 50290 }, { "epoch": 0.07, "learning_rate": 4.6856250000000004e-05, "loss": 1.6342, "step": 50300 }, { "epoch": 0.07, "learning_rate": 4.6855625e-05, "loss": 1.6475, "step": 50310 }, { "epoch": 0.07, "learning_rate": 4.6855000000000005e-05, "loss": 2.116, "step": 50320 }, { "epoch": 0.07, "learning_rate": 4.6854375e-05, "loss": 1.585, "step": 50330 }, { "epoch": 0.07, "learning_rate": 4.685375e-05, "loss": 1.6364, "step": 50340 }, { "epoch": 0.07, "learning_rate": 4.6853125e-05, "loss": 1.6543, "step": 50350 }, { "epoch": 0.07, "learning_rate": 4.6852500000000006e-05, "loss": 1.7592, "step": 50360 }, { "epoch": 0.07, "learning_rate": 4.6851875e-05, "loss": 1.5888, "step": 50370 }, { "epoch": 0.07, "learning_rate": 4.685125e-05, "loss": 1.7258, "step": 50380 }, { "epoch": 0.07, "learning_rate": 4.6850625000000003e-05, "loss": 1.5153, "step": 50390 }, { "epoch": 0.07, "learning_rate": 4.685000000000001e-05, "loss": 1.658, "step": 50400 }, { "epoch": 0.07, "learning_rate": 4.6849375000000004e-05, "loss": 1.8125, "step": 50410 }, { "epoch": 0.07, "learning_rate": 4.684875e-05, "loss": 1.9476, "step": 50420 }, { "epoch": 0.07, "learning_rate": 4.6848125000000005e-05, "loss": 1.6821, "step": 50430 }, { "epoch": 0.07, "learning_rate": 4.68475e-05, "loss": 1.7993, "step": 50440 }, { "epoch": 0.07, "learning_rate": 4.6846875e-05, "loss": 1.8898, "step": 50450 }, { "epoch": 0.07, "learning_rate": 4.684625e-05, "loss": 1.6968, "step": 50460 }, { "epoch": 0.07, "learning_rate": 4.6845625000000006e-05, "loss": 1.7394, "step": 50470 }, { "epoch": 0.07, "learning_rate": 4.6845e-05, "loss": 1.7421, "step": 50480 }, { "epoch": 0.07, "learning_rate": 4.6844375e-05, "loss": 1.67, "step": 50490 }, { "epoch": 0.07, "learning_rate": 4.6843750000000004e-05, "loss": 1.5264, "step": 50500 }, { "epoch": 0.07, "learning_rate": 4.684312500000001e-05, "loss": 1.7611, "step": 50510 }, { "epoch": 0.07, "learning_rate": 4.68425e-05, "loss": 1.7691, "step": 50520 }, { "epoch": 0.07, "learning_rate": 4.6841875e-05, "loss": 1.8658, "step": 50530 }, { "epoch": 0.07, "learning_rate": 4.6841250000000005e-05, "loss": 1.8618, "step": 50540 }, { "epoch": 0.07, "learning_rate": 4.6840625e-05, "loss": 1.9388, "step": 50550 }, { "epoch": 0.07, "learning_rate": 4.684e-05, "loss": 1.7986, "step": 50560 }, { "epoch": 0.07, "learning_rate": 4.6839375e-05, "loss": 1.8591, "step": 50570 }, { "epoch": 0.07, "learning_rate": 4.6838750000000006e-05, "loss": 1.8515, "step": 50580 }, { "epoch": 0.07, "learning_rate": 4.6838125e-05, "loss": 1.5888, "step": 50590 }, { "epoch": 0.07, "learning_rate": 4.68375e-05, "loss": 1.6863, "step": 50600 }, { "epoch": 0.07, "learning_rate": 4.6836875000000004e-05, "loss": 1.561, "step": 50610 }, { "epoch": 0.07, "learning_rate": 4.683625e-05, "loss": 1.777, "step": 50620 }, { "epoch": 0.07, "learning_rate": 4.6835625e-05, "loss": 1.8876, "step": 50630 }, { "epoch": 0.07, "learning_rate": 4.6835e-05, "loss": 1.753, "step": 50640 }, { "epoch": 0.07, "learning_rate": 4.6834375000000005e-05, "loss": 1.6216, "step": 50650 }, { "epoch": 0.07, "learning_rate": 4.683375e-05, "loss": 1.5277, "step": 50660 }, { "epoch": 0.07, "learning_rate": 4.6833125e-05, "loss": 1.717, "step": 50670 }, { "epoch": 0.07, "learning_rate": 4.68325e-05, "loss": 1.7986, "step": 50680 }, { "epoch": 0.07, "learning_rate": 4.6831875e-05, "loss": 1.7685, "step": 50690 }, { "epoch": 0.07, "learning_rate": 4.683125e-05, "loss": 1.9036, "step": 50700 }, { "epoch": 0.07, "learning_rate": 4.6830625e-05, "loss": 1.8071, "step": 50710 }, { "epoch": 0.07, "learning_rate": 4.6830000000000004e-05, "loss": 1.6758, "step": 50720 }, { "epoch": 0.07, "learning_rate": 4.6829375e-05, "loss": 1.78, "step": 50730 }, { "epoch": 0.07, "learning_rate": 4.6828750000000005e-05, "loss": 1.7867, "step": 50740 }, { "epoch": 0.07, "learning_rate": 4.6828125e-05, "loss": 1.6651, "step": 50750 }, { "epoch": 0.07, "learning_rate": 4.6827500000000005e-05, "loss": 1.6369, "step": 50760 }, { "epoch": 0.07, "learning_rate": 4.6826875e-05, "loss": 1.6343, "step": 50770 }, { "epoch": 0.07, "learning_rate": 4.6826250000000006e-05, "loss": 1.4813, "step": 50780 }, { "epoch": 0.07, "learning_rate": 4.6825625e-05, "loss": 1.9196, "step": 50790 }, { "epoch": 0.07, "learning_rate": 4.6825e-05, "loss": 1.7277, "step": 50800 }, { "epoch": 0.07, "learning_rate": 4.6824375000000003e-05, "loss": 1.4326, "step": 50810 }, { "epoch": 0.07, "learning_rate": 4.682375e-05, "loss": 1.7486, "step": 50820 }, { "epoch": 0.07, "learning_rate": 4.6823125000000004e-05, "loss": 1.6783, "step": 50830 }, { "epoch": 0.07, "learning_rate": 4.68225e-05, "loss": 1.8815, "step": 50840 }, { "epoch": 0.07, "learning_rate": 4.6821875000000005e-05, "loss": 1.7911, "step": 50850 }, { "epoch": 0.07, "learning_rate": 4.682125e-05, "loss": 1.7195, "step": 50860 }, { "epoch": 0.07, "learning_rate": 4.6820625e-05, "loss": 1.5909, "step": 50870 }, { "epoch": 0.07, "learning_rate": 4.682e-05, "loss": 1.8311, "step": 50880 }, { "epoch": 0.07, "learning_rate": 4.6819375000000006e-05, "loss": 1.6191, "step": 50890 }, { "epoch": 0.07, "learning_rate": 4.681875e-05, "loss": 1.6494, "step": 50900 }, { "epoch": 0.07, "learning_rate": 4.6818125e-05, "loss": 1.9209, "step": 50910 }, { "epoch": 0.07, "learning_rate": 4.6817500000000004e-05, "loss": 1.5889, "step": 50920 }, { "epoch": 0.07, "learning_rate": 4.681687500000001e-05, "loss": 1.583, "step": 50930 }, { "epoch": 0.07, "learning_rate": 4.681625e-05, "loss": 1.8616, "step": 50940 }, { "epoch": 0.07, "learning_rate": 4.6815625e-05, "loss": 1.7652, "step": 50950 }, { "epoch": 0.07, "learning_rate": 4.6815000000000005e-05, "loss": 1.7704, "step": 50960 }, { "epoch": 0.07, "learning_rate": 4.6814375e-05, "loss": 1.6383, "step": 50970 }, { "epoch": 0.07, "learning_rate": 4.681375e-05, "loss": 1.8624, "step": 50980 }, { "epoch": 0.07, "learning_rate": 4.6813125e-05, "loss": 1.535, "step": 50990 }, { "epoch": 0.07, "learning_rate": 4.6812500000000006e-05, "loss": 1.6241, "step": 51000 }, { "epoch": 0.07, "eval_loss": 1.6507177352905273, "eval_runtime": 1303.2019, "eval_samples_per_second": 44.533, "eval_steps_per_second": 5.567, "step": 51000 }, { "epoch": 0.07, "learning_rate": 4.6811874999999996e-05, "loss": 1.6333, "step": 51010 }, { "epoch": 0.07, "learning_rate": 4.681125e-05, "loss": 1.6048, "step": 51020 }, { "epoch": 0.07, "learning_rate": 4.6810625000000004e-05, "loss": 1.828, "step": 51030 }, { "epoch": 0.07, "learning_rate": 4.681e-05, "loss": 1.6446, "step": 51040 }, { "epoch": 0.07, "learning_rate": 4.6809375e-05, "loss": 1.6051, "step": 51050 }, { "epoch": 0.07, "learning_rate": 4.680875e-05, "loss": 1.5643, "step": 51060 }, { "epoch": 0.07, "learning_rate": 4.6808125000000005e-05, "loss": 1.7384, "step": 51070 }, { "epoch": 0.07, "learning_rate": 4.68075e-05, "loss": 1.6978, "step": 51080 }, { "epoch": 0.07, "learning_rate": 4.6806875e-05, "loss": 1.807, "step": 51090 }, { "epoch": 0.07, "learning_rate": 4.680625e-05, "loss": 1.8683, "step": 51100 }, { "epoch": 0.07, "learning_rate": 4.6805625e-05, "loss": 1.6666, "step": 51110 }, { "epoch": 0.07, "learning_rate": 4.6805e-05, "loss": 1.8414, "step": 51120 }, { "epoch": 0.07, "learning_rate": 4.6804375e-05, "loss": 1.5178, "step": 51130 }, { "epoch": 0.07, "learning_rate": 4.6803750000000004e-05, "loss": 1.4666, "step": 51140 }, { "epoch": 0.07, "learning_rate": 4.6803125e-05, "loss": 1.7737, "step": 51150 }, { "epoch": 0.07, "learning_rate": 4.6802500000000005e-05, "loss": 1.8206, "step": 51160 }, { "epoch": 0.07, "learning_rate": 4.6801875e-05, "loss": 1.8413, "step": 51170 }, { "epoch": 0.07, "learning_rate": 4.6801250000000005e-05, "loss": 1.8434, "step": 51180 }, { "epoch": 0.07, "learning_rate": 4.6800625e-05, "loss": 1.7064, "step": 51190 }, { "epoch": 0.07, "learning_rate": 4.6800000000000006e-05, "loss": 1.8489, "step": 51200 }, { "epoch": 0.07, "learning_rate": 4.6799375e-05, "loss": 1.7956, "step": 51210 }, { "epoch": 0.07, "learning_rate": 4.679875e-05, "loss": 1.7132, "step": 51220 }, { "epoch": 0.07, "learning_rate": 4.6798125000000004e-05, "loss": 1.8306, "step": 51230 }, { "epoch": 0.07, "learning_rate": 4.67975e-05, "loss": 1.7726, "step": 51240 }, { "epoch": 0.07, "learning_rate": 4.6796875000000004e-05, "loss": 1.7175, "step": 51250 }, { "epoch": 0.07, "learning_rate": 4.679625e-05, "loss": 1.9108, "step": 51260 }, { "epoch": 0.07, "learning_rate": 4.6795625000000005e-05, "loss": 1.8897, "step": 51270 }, { "epoch": 0.07, "learning_rate": 4.6795e-05, "loss": 1.8002, "step": 51280 }, { "epoch": 0.07, "learning_rate": 4.6794375e-05, "loss": 1.7067, "step": 51290 }, { "epoch": 0.07, "learning_rate": 4.679375e-05, "loss": 1.6431, "step": 51300 }, { "epoch": 0.07, "learning_rate": 4.6793125000000006e-05, "loss": 1.9066, "step": 51310 }, { "epoch": 0.07, "learning_rate": 4.67925e-05, "loss": 1.734, "step": 51320 }, { "epoch": 0.07, "learning_rate": 4.6791875e-05, "loss": 1.8239, "step": 51330 }, { "epoch": 0.07, "learning_rate": 4.6791250000000004e-05, "loss": 1.6917, "step": 51340 }, { "epoch": 0.07, "learning_rate": 4.679062500000001e-05, "loss": 1.6987, "step": 51350 }, { "epoch": 0.07, "learning_rate": 4.679e-05, "loss": 1.7196, "step": 51360 }, { "epoch": 0.07, "learning_rate": 4.6789375e-05, "loss": 1.7576, "step": 51370 }, { "epoch": 0.07, "learning_rate": 4.6788750000000005e-05, "loss": 1.6645, "step": 51380 }, { "epoch": 0.07, "learning_rate": 4.6788125e-05, "loss": 1.6059, "step": 51390 }, { "epoch": 0.07, "learning_rate": 4.67875e-05, "loss": 1.7867, "step": 51400 }, { "epoch": 0.07, "learning_rate": 4.6786875e-05, "loss": 1.8981, "step": 51410 }, { "epoch": 0.07, "learning_rate": 4.6786250000000006e-05, "loss": 1.7025, "step": 51420 }, { "epoch": 0.07, "learning_rate": 4.6785624999999996e-05, "loss": 1.6667, "step": 51430 }, { "epoch": 0.07, "learning_rate": 4.6785e-05, "loss": 1.7149, "step": 51440 }, { "epoch": 0.07, "learning_rate": 4.6784375000000004e-05, "loss": 1.7046, "step": 51450 }, { "epoch": 0.07, "learning_rate": 4.678375e-05, "loss": 1.9565, "step": 51460 }, { "epoch": 0.07, "learning_rate": 4.6783125e-05, "loss": 1.7115, "step": 51470 }, { "epoch": 0.07, "learning_rate": 4.67825e-05, "loss": 1.7786, "step": 51480 }, { "epoch": 0.07, "learning_rate": 4.6781875000000005e-05, "loss": 1.6451, "step": 51490 }, { "epoch": 0.07, "learning_rate": 4.678125e-05, "loss": 1.7866, "step": 51500 }, { "epoch": 0.07, "learning_rate": 4.6780625e-05, "loss": 1.7635, "step": 51510 }, { "epoch": 0.07, "learning_rate": 4.678e-05, "loss": 1.7342, "step": 51520 }, { "epoch": 0.07, "learning_rate": 4.6779375000000006e-05, "loss": 1.5117, "step": 51530 }, { "epoch": 0.07, "learning_rate": 4.677875e-05, "loss": 1.6896, "step": 51540 }, { "epoch": 0.07, "learning_rate": 4.6778125e-05, "loss": 1.7512, "step": 51550 }, { "epoch": 0.07, "learning_rate": 4.6777500000000004e-05, "loss": 1.6944, "step": 51560 }, { "epoch": 0.07, "learning_rate": 4.6776875e-05, "loss": 1.866, "step": 51570 }, { "epoch": 0.07, "learning_rate": 4.6776250000000005e-05, "loss": 1.5319, "step": 51580 }, { "epoch": 0.07, "learning_rate": 4.6775625e-05, "loss": 1.7187, "step": 51590 }, { "epoch": 0.07, "learning_rate": 4.6775000000000005e-05, "loss": 1.7569, "step": 51600 }, { "epoch": 0.07, "learning_rate": 4.6774375e-05, "loss": 1.618, "step": 51610 }, { "epoch": 0.07, "learning_rate": 4.677375e-05, "loss": 1.7393, "step": 51620 }, { "epoch": 0.07, "learning_rate": 4.6773125e-05, "loss": 1.7812, "step": 51630 }, { "epoch": 0.07, "learning_rate": 4.67725e-05, "loss": 1.7057, "step": 51640 }, { "epoch": 0.07, "learning_rate": 4.6771875000000004e-05, "loss": 1.734, "step": 51650 }, { "epoch": 0.07, "learning_rate": 4.677125e-05, "loss": 1.8554, "step": 51660 }, { "epoch": 0.07, "learning_rate": 4.6770625000000004e-05, "loss": 1.7728, "step": 51670 }, { "epoch": 0.07, "learning_rate": 4.677e-05, "loss": 1.6162, "step": 51680 }, { "epoch": 0.07, "learning_rate": 4.6769375000000005e-05, "loss": 1.684, "step": 51690 }, { "epoch": 0.07, "learning_rate": 4.676875e-05, "loss": 1.8217, "step": 51700 }, { "epoch": 0.07, "learning_rate": 4.6768125e-05, "loss": 1.7072, "step": 51710 }, { "epoch": 0.07, "learning_rate": 4.67675e-05, "loss": 1.4842, "step": 51720 }, { "epoch": 0.07, "learning_rate": 4.6766875000000006e-05, "loss": 1.8945, "step": 51730 }, { "epoch": 0.07, "learning_rate": 4.676625e-05, "loss": 1.8066, "step": 51740 }, { "epoch": 0.07, "learning_rate": 4.6765625e-05, "loss": 1.7784, "step": 51750 }, { "epoch": 0.07, "learning_rate": 4.6765000000000004e-05, "loss": 1.5156, "step": 51760 }, { "epoch": 0.07, "learning_rate": 4.676437500000001e-05, "loss": 1.9619, "step": 51770 }, { "epoch": 0.07, "learning_rate": 4.676375e-05, "loss": 2.0464, "step": 51780 }, { "epoch": 0.07, "learning_rate": 4.6763125e-05, "loss": 1.7863, "step": 51790 }, { "epoch": 0.07, "learning_rate": 4.6762500000000005e-05, "loss": 1.606, "step": 51800 }, { "epoch": 0.07, "learning_rate": 4.6761875e-05, "loss": 1.6556, "step": 51810 }, { "epoch": 0.07, "learning_rate": 4.676125e-05, "loss": 1.7749, "step": 51820 }, { "epoch": 0.07, "learning_rate": 4.6760625e-05, "loss": 1.4962, "step": 51830 }, { "epoch": 0.07, "learning_rate": 4.6760000000000006e-05, "loss": 1.5784, "step": 51840 }, { "epoch": 0.07, "learning_rate": 4.6759374999999996e-05, "loss": 1.7027, "step": 51850 }, { "epoch": 0.07, "learning_rate": 4.675875e-05, "loss": 1.6718, "step": 51860 }, { "epoch": 0.07, "learning_rate": 4.6758125000000004e-05, "loss": 1.5759, "step": 51870 }, { "epoch": 0.07, "learning_rate": 4.67575e-05, "loss": 1.7681, "step": 51880 }, { "epoch": 0.07, "learning_rate": 4.6756875e-05, "loss": 1.7036, "step": 51890 }, { "epoch": 0.07, "learning_rate": 4.675625e-05, "loss": 1.6287, "step": 51900 }, { "epoch": 0.07, "learning_rate": 4.6755625000000005e-05, "loss": 1.5796, "step": 51910 }, { "epoch": 0.07, "learning_rate": 4.6755e-05, "loss": 1.4784, "step": 51920 }, { "epoch": 0.07, "learning_rate": 4.6754375e-05, "loss": 1.8728, "step": 51930 }, { "epoch": 0.07, "learning_rate": 4.675375e-05, "loss": 1.679, "step": 51940 }, { "epoch": 0.07, "learning_rate": 4.6753125000000006e-05, "loss": 1.585, "step": 51950 }, { "epoch": 0.07, "learning_rate": 4.6752500000000003e-05, "loss": 1.818, "step": 51960 }, { "epoch": 0.07, "learning_rate": 4.6751875e-05, "loss": 1.5936, "step": 51970 }, { "epoch": 0.07, "learning_rate": 4.6751250000000004e-05, "loss": 1.6312, "step": 51980 }, { "epoch": 0.07, "learning_rate": 4.6750625e-05, "loss": 1.6819, "step": 51990 }, { "epoch": 0.07, "learning_rate": 4.6750000000000005e-05, "loss": 1.5339, "step": 52000 }, { "epoch": 0.07, "eval_loss": 1.634942889213562, "eval_runtime": 1615.546, "eval_samples_per_second": 35.923, "eval_steps_per_second": 4.491, "step": 52000 }, { "epoch": 0.07, "learning_rate": 4.6749375e-05, "loss": 1.7347, "step": 52010 }, { "epoch": 0.07, "learning_rate": 4.6748750000000005e-05, "loss": 1.9073, "step": 52020 }, { "epoch": 0.07, "learning_rate": 4.6748125e-05, "loss": 1.7182, "step": 52030 }, { "epoch": 0.07, "learning_rate": 4.67475e-05, "loss": 1.7732, "step": 52040 }, { "epoch": 0.07, "learning_rate": 4.6746875e-05, "loss": 1.9719, "step": 52050 }, { "epoch": 0.07, "learning_rate": 4.674625e-05, "loss": 2.0019, "step": 52060 }, { "epoch": 0.07, "learning_rate": 4.6745625000000004e-05, "loss": 1.782, "step": 52070 }, { "epoch": 0.07, "learning_rate": 4.6745e-05, "loss": 1.6748, "step": 52080 }, { "epoch": 0.07, "learning_rate": 4.6744375000000004e-05, "loss": 1.4765, "step": 52090 }, { "epoch": 0.07, "learning_rate": 4.674375e-05, "loss": 1.4723, "step": 52100 }, { "epoch": 0.07, "learning_rate": 4.6743125000000005e-05, "loss": 1.7067, "step": 52110 }, { "epoch": 0.07, "learning_rate": 4.67425e-05, "loss": 1.5654, "step": 52120 }, { "epoch": 0.07, "learning_rate": 4.6741875e-05, "loss": 1.7474, "step": 52130 }, { "epoch": 0.07, "learning_rate": 4.674125e-05, "loss": 1.7748, "step": 52140 }, { "epoch": 0.07, "learning_rate": 4.6740625000000006e-05, "loss": 1.6811, "step": 52150 }, { "epoch": 0.07, "learning_rate": 4.674e-05, "loss": 1.6392, "step": 52160 }, { "epoch": 0.07, "learning_rate": 4.6739375e-05, "loss": 1.8113, "step": 52170 }, { "epoch": 0.07, "learning_rate": 4.6738750000000004e-05, "loss": 1.9515, "step": 52180 }, { "epoch": 0.07, "learning_rate": 4.673812500000001e-05, "loss": 1.5027, "step": 52190 }, { "epoch": 0.07, "learning_rate": 4.67375e-05, "loss": 1.749, "step": 52200 }, { "epoch": 0.07, "learning_rate": 4.6736875e-05, "loss": 1.655, "step": 52210 }, { "epoch": 0.07, "learning_rate": 4.6736250000000005e-05, "loss": 1.7387, "step": 52220 }, { "epoch": 0.07, "learning_rate": 4.6735625e-05, "loss": 1.3379, "step": 52230 }, { "epoch": 0.07, "learning_rate": 4.6735e-05, "loss": 1.8746, "step": 52240 }, { "epoch": 0.07, "learning_rate": 4.6734375e-05, "loss": 1.6607, "step": 52250 }, { "epoch": 0.07, "learning_rate": 4.6733750000000006e-05, "loss": 1.7507, "step": 52260 }, { "epoch": 0.07, "learning_rate": 4.6733124999999997e-05, "loss": 1.6635, "step": 52270 }, { "epoch": 0.07, "learning_rate": 4.67325e-05, "loss": 1.6515, "step": 52280 }, { "epoch": 0.07, "learning_rate": 4.6731875000000004e-05, "loss": 1.9204, "step": 52290 }, { "epoch": 0.07, "learning_rate": 4.673125e-05, "loss": 1.6677, "step": 52300 }, { "epoch": 0.07, "learning_rate": 4.6730625e-05, "loss": 1.5703, "step": 52310 }, { "epoch": 0.07, "learning_rate": 4.673e-05, "loss": 1.7086, "step": 52320 }, { "epoch": 0.07, "learning_rate": 4.6729375000000005e-05, "loss": 1.7725, "step": 52330 }, { "epoch": 0.07, "learning_rate": 4.672875e-05, "loss": 1.7024, "step": 52340 }, { "epoch": 0.07, "learning_rate": 4.6728125e-05, "loss": 1.6484, "step": 52350 }, { "epoch": 0.07, "learning_rate": 4.67275e-05, "loss": 1.7945, "step": 52360 }, { "epoch": 0.07, "learning_rate": 4.6726875000000007e-05, "loss": 1.6567, "step": 52370 }, { "epoch": 0.07, "learning_rate": 4.6726250000000003e-05, "loss": 1.7105, "step": 52380 }, { "epoch": 0.07, "learning_rate": 4.6725625e-05, "loss": 1.4856, "step": 52390 }, { "epoch": 0.07, "learning_rate": 4.6725000000000004e-05, "loss": 1.5057, "step": 52400 }, { "epoch": 0.07, "learning_rate": 4.6724375e-05, "loss": 1.6771, "step": 52410 }, { "epoch": 0.07, "learning_rate": 4.672375e-05, "loss": 1.6193, "step": 52420 }, { "epoch": 0.07, "learning_rate": 4.6723125e-05, "loss": 1.8544, "step": 52430 }, { "epoch": 0.07, "learning_rate": 4.6722500000000005e-05, "loss": 1.6249, "step": 52440 }, { "epoch": 0.07, "learning_rate": 4.6721875e-05, "loss": 1.7543, "step": 52450 }, { "epoch": 0.07, "learning_rate": 4.672125e-05, "loss": 1.4759, "step": 52460 }, { "epoch": 0.07, "learning_rate": 4.6720625e-05, "loss": 1.7931, "step": 52470 }, { "epoch": 0.07, "learning_rate": 4.672e-05, "loss": 1.694, "step": 52480 }, { "epoch": 0.07, "learning_rate": 4.6719375000000004e-05, "loss": 1.4691, "step": 52490 }, { "epoch": 0.07, "learning_rate": 4.671875e-05, "loss": 1.7594, "step": 52500 }, { "epoch": 0.07, "learning_rate": 4.6718125000000004e-05, "loss": 1.43, "step": 52510 }, { "epoch": 0.07, "learning_rate": 4.67175e-05, "loss": 1.8434, "step": 52520 }, { "epoch": 0.07, "learning_rate": 4.6716875000000005e-05, "loss": 1.6528, "step": 52530 }, { "epoch": 0.07, "learning_rate": 4.671625e-05, "loss": 1.6947, "step": 52540 }, { "epoch": 0.07, "learning_rate": 4.6715625e-05, "loss": 1.7568, "step": 52550 }, { "epoch": 0.07, "learning_rate": 4.6715e-05, "loss": 1.6516, "step": 52560 }, { "epoch": 0.07, "learning_rate": 4.6714375000000006e-05, "loss": 1.4133, "step": 52570 }, { "epoch": 0.07, "learning_rate": 4.671375e-05, "loss": 1.5438, "step": 52580 }, { "epoch": 0.07, "learning_rate": 4.6713125e-05, "loss": 1.6065, "step": 52590 }, { "epoch": 0.07, "learning_rate": 4.6712500000000004e-05, "loss": 1.6711, "step": 52600 }, { "epoch": 0.07, "learning_rate": 4.671187500000001e-05, "loss": 1.5398, "step": 52610 }, { "epoch": 0.07, "learning_rate": 4.671125e-05, "loss": 1.7131, "step": 52620 }, { "epoch": 0.07, "learning_rate": 4.6710625e-05, "loss": 1.6139, "step": 52630 }, { "epoch": 0.07, "learning_rate": 4.6710000000000005e-05, "loss": 1.7604, "step": 52640 }, { "epoch": 0.07, "learning_rate": 4.6709375e-05, "loss": 1.731, "step": 52650 }, { "epoch": 0.07, "learning_rate": 4.670875e-05, "loss": 1.5123, "step": 52660 }, { "epoch": 0.07, "learning_rate": 4.6708125e-05, "loss": 1.9538, "step": 52670 }, { "epoch": 0.07, "learning_rate": 4.6707500000000006e-05, "loss": 1.6214, "step": 52680 }, { "epoch": 0.07, "learning_rate": 4.6706874999999997e-05, "loss": 1.7117, "step": 52690 }, { "epoch": 0.07, "learning_rate": 4.670625e-05, "loss": 1.5816, "step": 52700 }, { "epoch": 0.07, "learning_rate": 4.6705625000000004e-05, "loss": 1.498, "step": 52710 }, { "epoch": 0.07, "learning_rate": 4.670500000000001e-05, "loss": 1.6573, "step": 52720 }, { "epoch": 0.07, "learning_rate": 4.6704375e-05, "loss": 1.894, "step": 52730 }, { "epoch": 0.07, "learning_rate": 4.670375e-05, "loss": 1.6858, "step": 52740 }, { "epoch": 0.07, "learning_rate": 4.6703125000000005e-05, "loss": 1.7126, "step": 52750 }, { "epoch": 0.07, "learning_rate": 4.67025e-05, "loss": 1.782, "step": 52760 }, { "epoch": 0.07, "learning_rate": 4.6701875e-05, "loss": 1.4994, "step": 52770 }, { "epoch": 0.07, "learning_rate": 4.670125e-05, "loss": 1.7076, "step": 52780 }, { "epoch": 0.07, "learning_rate": 4.6700625000000007e-05, "loss": 1.8002, "step": 52790 }, { "epoch": 0.07, "learning_rate": 4.6700000000000003e-05, "loss": 1.7831, "step": 52800 }, { "epoch": 0.07, "learning_rate": 4.6699375e-05, "loss": 1.7057, "step": 52810 }, { "epoch": 0.07, "learning_rate": 4.6698750000000004e-05, "loss": 1.686, "step": 52820 }, { "epoch": 0.07, "learning_rate": 4.6698125e-05, "loss": 1.6386, "step": 52830 }, { "epoch": 0.07, "learning_rate": 4.66975e-05, "loss": 1.663, "step": 52840 }, { "epoch": 0.07, "learning_rate": 4.6696875e-05, "loss": 1.6739, "step": 52850 }, { "epoch": 0.07, "learning_rate": 4.6696250000000005e-05, "loss": 1.727, "step": 52860 }, { "epoch": 0.07, "learning_rate": 4.6695625e-05, "loss": 1.6101, "step": 52870 }, { "epoch": 0.07, "learning_rate": 4.6695e-05, "loss": 1.6009, "step": 52880 }, { "epoch": 0.07, "learning_rate": 4.6694375e-05, "loss": 1.5431, "step": 52890 }, { "epoch": 0.07, "learning_rate": 4.669375e-05, "loss": 1.8463, "step": 52900 }, { "epoch": 0.07, "learning_rate": 4.6693125000000004e-05, "loss": 1.6704, "step": 52910 }, { "epoch": 0.07, "learning_rate": 4.66925e-05, "loss": 1.6787, "step": 52920 }, { "epoch": 0.07, "learning_rate": 4.6691875000000004e-05, "loss": 1.7641, "step": 52930 }, { "epoch": 0.07, "learning_rate": 4.669125e-05, "loss": 1.4238, "step": 52940 }, { "epoch": 0.07, "learning_rate": 4.6690625000000005e-05, "loss": 1.6755, "step": 52950 }, { "epoch": 0.07, "learning_rate": 4.669e-05, "loss": 1.8033, "step": 52960 }, { "epoch": 0.07, "learning_rate": 4.6689375e-05, "loss": 1.8124, "step": 52970 }, { "epoch": 0.07, "learning_rate": 4.668875e-05, "loss": 1.9401, "step": 52980 }, { "epoch": 0.07, "learning_rate": 4.6688125000000006e-05, "loss": 1.7022, "step": 52990 }, { "epoch": 0.07, "learning_rate": 4.66875e-05, "loss": 1.5145, "step": 53000 }, { "epoch": 0.07, "eval_loss": 1.6293290853500366, "eval_runtime": 1302.4811, "eval_samples_per_second": 44.558, "eval_steps_per_second": 5.57, "step": 53000 }, { "epoch": 0.07, "learning_rate": 4.6686875e-05, "loss": 1.8011, "step": 53010 }, { "epoch": 0.07, "learning_rate": 4.6686250000000004e-05, "loss": 1.5098, "step": 53020 }, { "epoch": 0.07, "learning_rate": 4.6685625e-05, "loss": 1.7923, "step": 53030 }, { "epoch": 0.07, "learning_rate": 4.6685e-05, "loss": 1.6686, "step": 53040 }, { "epoch": 0.07, "learning_rate": 4.6684375e-05, "loss": 1.5578, "step": 53050 }, { "epoch": 0.07, "learning_rate": 4.6683750000000005e-05, "loss": 1.9234, "step": 53060 }, { "epoch": 0.07, "learning_rate": 4.6683125e-05, "loss": 1.4329, "step": 53070 }, { "epoch": 0.07, "learning_rate": 4.66825e-05, "loss": 1.7339, "step": 53080 }, { "epoch": 0.07, "learning_rate": 4.6681875e-05, "loss": 1.6959, "step": 53090 }, { "epoch": 0.07, "learning_rate": 4.6681250000000006e-05, "loss": 1.7505, "step": 53100 }, { "epoch": 0.07, "learning_rate": 4.6680625e-05, "loss": 1.7128, "step": 53110 }, { "epoch": 0.07, "learning_rate": 4.668e-05, "loss": 1.7826, "step": 53120 }, { "epoch": 0.07, "learning_rate": 4.6679375000000004e-05, "loss": 1.5619, "step": 53130 }, { "epoch": 0.07, "learning_rate": 4.667875000000001e-05, "loss": 1.5591, "step": 53140 }, { "epoch": 0.07, "learning_rate": 4.6678125e-05, "loss": 1.9825, "step": 53150 }, { "epoch": 0.07, "learning_rate": 4.66775e-05, "loss": 1.7273, "step": 53160 }, { "epoch": 0.07, "learning_rate": 4.6676875000000005e-05, "loss": 1.8471, "step": 53170 }, { "epoch": 0.07, "learning_rate": 4.667625e-05, "loss": 1.7333, "step": 53180 }, { "epoch": 0.07, "learning_rate": 4.6675625e-05, "loss": 1.6622, "step": 53190 }, { "epoch": 0.07, "learning_rate": 4.6675e-05, "loss": 1.6068, "step": 53200 }, { "epoch": 0.07, "learning_rate": 4.6674375000000007e-05, "loss": 1.8316, "step": 53210 }, { "epoch": 0.07, "learning_rate": 4.6673750000000004e-05, "loss": 1.6864, "step": 53220 }, { "epoch": 0.07, "learning_rate": 4.6673125e-05, "loss": 1.9585, "step": 53230 }, { "epoch": 0.07, "learning_rate": 4.6672500000000004e-05, "loss": 1.5371, "step": 53240 }, { "epoch": 0.07, "learning_rate": 4.6671875e-05, "loss": 1.8233, "step": 53250 }, { "epoch": 0.07, "learning_rate": 4.667125e-05, "loss": 1.643, "step": 53260 }, { "epoch": 0.07, "learning_rate": 4.6670625e-05, "loss": 1.7144, "step": 53270 }, { "epoch": 0.07, "learning_rate": 4.6670000000000005e-05, "loss": 1.7963, "step": 53280 }, { "epoch": 0.07, "learning_rate": 4.6669375e-05, "loss": 1.8884, "step": 53290 }, { "epoch": 0.07, "learning_rate": 4.666875e-05, "loss": 1.5729, "step": 53300 }, { "epoch": 0.07, "learning_rate": 4.6668125e-05, "loss": 1.8046, "step": 53310 }, { "epoch": 0.07, "learning_rate": 4.66675e-05, "loss": 1.6322, "step": 53320 }, { "epoch": 0.07, "learning_rate": 4.6666875000000004e-05, "loss": 1.4607, "step": 53330 }, { "epoch": 0.07, "learning_rate": 4.666625e-05, "loss": 1.5511, "step": 53340 }, { "epoch": 0.07, "learning_rate": 4.6665625000000004e-05, "loss": 1.66, "step": 53350 }, { "epoch": 0.07, "learning_rate": 4.6665e-05, "loss": 1.7905, "step": 53360 }, { "epoch": 0.07, "learning_rate": 4.6664375000000005e-05, "loss": 1.6406, "step": 53370 }, { "epoch": 0.07, "learning_rate": 4.666375e-05, "loss": 1.8506, "step": 53380 }, { "epoch": 0.07, "learning_rate": 4.6663125e-05, "loss": 1.7791, "step": 53390 }, { "epoch": 0.07, "learning_rate": 4.66625e-05, "loss": 1.5792, "step": 53400 }, { "epoch": 0.07, "learning_rate": 4.6661875000000006e-05, "loss": 1.5355, "step": 53410 }, { "epoch": 0.07, "learning_rate": 4.666125e-05, "loss": 1.628, "step": 53420 }, { "epoch": 0.07, "learning_rate": 4.6660625e-05, "loss": 1.6507, "step": 53430 }, { "epoch": 0.07, "learning_rate": 4.6660000000000004e-05, "loss": 1.3912, "step": 53440 }, { "epoch": 0.07, "learning_rate": 4.6659375e-05, "loss": 1.455, "step": 53450 }, { "epoch": 0.07, "learning_rate": 4.665875e-05, "loss": 1.4983, "step": 53460 }, { "epoch": 0.07, "learning_rate": 4.6658125e-05, "loss": 1.686, "step": 53470 }, { "epoch": 0.07, "learning_rate": 4.6657500000000005e-05, "loss": 1.8411, "step": 53480 }, { "epoch": 0.07, "learning_rate": 4.6656875e-05, "loss": 1.5894, "step": 53490 }, { "epoch": 0.07, "learning_rate": 4.665625e-05, "loss": 1.8492, "step": 53500 }, { "epoch": 0.07, "learning_rate": 4.6655625e-05, "loss": 1.8134, "step": 53510 }, { "epoch": 0.07, "learning_rate": 4.6655000000000006e-05, "loss": 1.8323, "step": 53520 }, { "epoch": 0.07, "learning_rate": 4.6654375000000003e-05, "loss": 1.8186, "step": 53530 }, { "epoch": 0.07, "learning_rate": 4.665375e-05, "loss": 1.6008, "step": 53540 }, { "epoch": 0.07, "learning_rate": 4.6653125000000004e-05, "loss": 1.5588, "step": 53550 }, { "epoch": 0.07, "learning_rate": 4.665250000000001e-05, "loss": 1.697, "step": 53560 }, { "epoch": 0.07, "learning_rate": 4.6651875e-05, "loss": 1.5972, "step": 53570 }, { "epoch": 0.07, "learning_rate": 4.665125e-05, "loss": 1.5196, "step": 53580 }, { "epoch": 0.07, "learning_rate": 4.6650625000000005e-05, "loss": 1.6161, "step": 53590 }, { "epoch": 0.07, "learning_rate": 4.665e-05, "loss": 1.6872, "step": 53600 }, { "epoch": 0.07, "learning_rate": 4.6649375e-05, "loss": 1.7467, "step": 53610 }, { "epoch": 0.07, "learning_rate": 4.664875e-05, "loss": 1.6045, "step": 53620 }, { "epoch": 0.07, "learning_rate": 4.664812500000001e-05, "loss": 1.8252, "step": 53630 }, { "epoch": 0.07, "learning_rate": 4.66475e-05, "loss": 1.6844, "step": 53640 }, { "epoch": 0.07, "learning_rate": 4.6646875e-05, "loss": 1.5866, "step": 53650 }, { "epoch": 0.07, "learning_rate": 4.6646250000000004e-05, "loss": 1.7875, "step": 53660 }, { "epoch": 0.07, "learning_rate": 4.6645625e-05, "loss": 1.6806, "step": 53670 }, { "epoch": 0.07, "learning_rate": 4.6645e-05, "loss": 1.9668, "step": 53680 }, { "epoch": 0.07, "learning_rate": 4.6644375e-05, "loss": 1.6755, "step": 53690 }, { "epoch": 0.07, "learning_rate": 4.6643750000000006e-05, "loss": 1.6746, "step": 53700 }, { "epoch": 0.07, "learning_rate": 4.6643125e-05, "loss": 1.6573, "step": 53710 }, { "epoch": 0.07, "learning_rate": 4.66425e-05, "loss": 1.8109, "step": 53720 }, { "epoch": 0.07, "learning_rate": 4.6641875e-05, "loss": 1.5692, "step": 53730 }, { "epoch": 0.07, "learning_rate": 4.664125e-05, "loss": 1.6178, "step": 53740 }, { "epoch": 0.07, "learning_rate": 4.6640625000000004e-05, "loss": 1.6659, "step": 53750 }, { "epoch": 0.07, "learning_rate": 4.664e-05, "loss": 1.8817, "step": 53760 }, { "epoch": 0.07, "learning_rate": 4.6639375000000004e-05, "loss": 1.7911, "step": 53770 }, { "epoch": 0.07, "learning_rate": 4.663875e-05, "loss": 1.5175, "step": 53780 }, { "epoch": 0.07, "learning_rate": 4.6638125000000005e-05, "loss": 1.6875, "step": 53790 }, { "epoch": 0.07, "learning_rate": 4.66375e-05, "loss": 1.7644, "step": 53800 }, { "epoch": 0.07, "learning_rate": 4.6636875e-05, "loss": 1.6254, "step": 53810 }, { "epoch": 0.07, "learning_rate": 4.663625e-05, "loss": 1.8498, "step": 53820 }, { "epoch": 0.07, "learning_rate": 4.6635625000000006e-05, "loss": 1.7081, "step": 53830 }, { "epoch": 0.07, "learning_rate": 4.6635e-05, "loss": 1.6127, "step": 53840 }, { "epoch": 0.07, "learning_rate": 4.6634375e-05, "loss": 1.8763, "step": 53850 }, { "epoch": 0.07, "learning_rate": 4.6633750000000004e-05, "loss": 1.5747, "step": 53860 }, { "epoch": 0.08, "learning_rate": 4.6633125e-05, "loss": 1.63, "step": 53870 }, { "epoch": 0.08, "learning_rate": 4.6632500000000005e-05, "loss": 1.7227, "step": 53880 }, { "epoch": 0.08, "learning_rate": 4.6631875e-05, "loss": 1.6947, "step": 53890 }, { "epoch": 0.08, "learning_rate": 4.6631250000000005e-05, "loss": 1.4937, "step": 53900 }, { "epoch": 0.08, "learning_rate": 4.6630625e-05, "loss": 1.7372, "step": 53910 }, { "epoch": 0.08, "learning_rate": 4.663e-05, "loss": 1.4943, "step": 53920 }, { "epoch": 0.08, "learning_rate": 4.6629375e-05, "loss": 1.6798, "step": 53930 }, { "epoch": 0.08, "learning_rate": 4.6628750000000007e-05, "loss": 1.7687, "step": 53940 }, { "epoch": 0.08, "learning_rate": 4.6628125000000003e-05, "loss": 1.9173, "step": 53950 }, { "epoch": 0.08, "learning_rate": 4.66275e-05, "loss": 1.556, "step": 53960 }, { "epoch": 0.08, "learning_rate": 4.6626875000000004e-05, "loss": 1.5685, "step": 53970 }, { "epoch": 0.08, "learning_rate": 4.662625000000001e-05, "loss": 1.6573, "step": 53980 }, { "epoch": 0.08, "learning_rate": 4.6625625e-05, "loss": 1.6648, "step": 53990 }, { "epoch": 0.08, "learning_rate": 4.6625e-05, "loss": 1.4827, "step": 54000 }, { "epoch": 0.08, "eval_loss": 1.6160304546356201, "eval_runtime": 1298.5116, "eval_samples_per_second": 44.694, "eval_steps_per_second": 5.587, "step": 54000 }, { "epoch": 0.08, "learning_rate": 4.6624375000000005e-05, "loss": 1.5033, "step": 54010 }, { "epoch": 0.08, "learning_rate": 4.662375e-05, "loss": 1.595, "step": 54020 }, { "epoch": 0.08, "learning_rate": 4.6623125e-05, "loss": 1.6211, "step": 54030 }, { "epoch": 0.08, "learning_rate": 4.66225e-05, "loss": 1.8656, "step": 54040 }, { "epoch": 0.08, "learning_rate": 4.662187500000001e-05, "loss": 1.574, "step": 54050 }, { "epoch": 0.08, "learning_rate": 4.662125e-05, "loss": 1.629, "step": 54060 }, { "epoch": 0.08, "learning_rate": 4.6620625e-05, "loss": 1.8519, "step": 54070 }, { "epoch": 0.08, "learning_rate": 4.6620000000000004e-05, "loss": 1.5904, "step": 54080 }, { "epoch": 0.08, "learning_rate": 4.6619375e-05, "loss": 1.6985, "step": 54090 }, { "epoch": 0.08, "learning_rate": 4.661875e-05, "loss": 1.5382, "step": 54100 }, { "epoch": 0.08, "learning_rate": 4.6618125e-05, "loss": 1.5362, "step": 54110 }, { "epoch": 0.08, "learning_rate": 4.6617500000000006e-05, "loss": 1.6746, "step": 54120 }, { "epoch": 0.08, "learning_rate": 4.6616875e-05, "loss": 1.759, "step": 54130 }, { "epoch": 0.08, "learning_rate": 4.661625e-05, "loss": 1.5995, "step": 54140 }, { "epoch": 0.08, "learning_rate": 4.6615625e-05, "loss": 1.61, "step": 54150 }, { "epoch": 0.08, "learning_rate": 4.6615e-05, "loss": 1.412, "step": 54160 }, { "epoch": 0.08, "learning_rate": 4.6614375000000004e-05, "loss": 1.6876, "step": 54170 }, { "epoch": 0.08, "learning_rate": 4.661375e-05, "loss": 1.6216, "step": 54180 }, { "epoch": 0.08, "learning_rate": 4.6613125000000004e-05, "loss": 1.7109, "step": 54190 }, { "epoch": 0.08, "learning_rate": 4.66125e-05, "loss": 1.706, "step": 54200 }, { "epoch": 0.08, "learning_rate": 4.6611875000000005e-05, "loss": 1.7698, "step": 54210 }, { "epoch": 0.08, "learning_rate": 4.661125e-05, "loss": 1.6639, "step": 54220 }, { "epoch": 0.08, "learning_rate": 4.6610625e-05, "loss": 1.5807, "step": 54230 }, { "epoch": 0.08, "learning_rate": 4.661e-05, "loss": 1.6401, "step": 54240 }, { "epoch": 0.08, "learning_rate": 4.6609375e-05, "loss": 1.7555, "step": 54250 }, { "epoch": 0.08, "learning_rate": 4.660875e-05, "loss": 1.5865, "step": 54260 }, { "epoch": 0.08, "learning_rate": 4.6608125e-05, "loss": 1.5233, "step": 54270 }, { "epoch": 0.08, "learning_rate": 4.6607500000000004e-05, "loss": 1.72, "step": 54280 }, { "epoch": 0.08, "learning_rate": 4.6606875e-05, "loss": 1.5445, "step": 54290 }, { "epoch": 0.08, "learning_rate": 4.6606250000000005e-05, "loss": 1.8215, "step": 54300 }, { "epoch": 0.08, "learning_rate": 4.6605625e-05, "loss": 1.9139, "step": 54310 }, { "epoch": 0.08, "learning_rate": 4.6605000000000005e-05, "loss": 1.9816, "step": 54320 }, { "epoch": 0.08, "learning_rate": 4.6604375e-05, "loss": 1.7246, "step": 54330 }, { "epoch": 0.08, "learning_rate": 4.660375e-05, "loss": 1.7539, "step": 54340 }, { "epoch": 0.08, "learning_rate": 4.6603125e-05, "loss": 1.5447, "step": 54350 }, { "epoch": 0.08, "learning_rate": 4.6602500000000007e-05, "loss": 1.6378, "step": 54360 }, { "epoch": 0.08, "learning_rate": 4.6601875000000003e-05, "loss": 1.8461, "step": 54370 }, { "epoch": 0.08, "learning_rate": 4.660125e-05, "loss": 1.5154, "step": 54380 }, { "epoch": 0.08, "learning_rate": 4.6600625000000004e-05, "loss": 1.76, "step": 54390 }, { "epoch": 0.08, "learning_rate": 4.660000000000001e-05, "loss": 1.527, "step": 54400 }, { "epoch": 0.08, "learning_rate": 4.6599375e-05, "loss": 1.7648, "step": 54410 }, { "epoch": 0.08, "learning_rate": 4.659875e-05, "loss": 1.6433, "step": 54420 }, { "epoch": 0.08, "learning_rate": 4.6598125000000005e-05, "loss": 1.556, "step": 54430 }, { "epoch": 0.08, "learning_rate": 4.65975e-05, "loss": 1.5563, "step": 54440 }, { "epoch": 0.08, "learning_rate": 4.6596875e-05, "loss": 1.8231, "step": 54450 }, { "epoch": 0.08, "learning_rate": 4.659625e-05, "loss": 1.8122, "step": 54460 }, { "epoch": 0.08, "learning_rate": 4.659562500000001e-05, "loss": 1.8833, "step": 54470 }, { "epoch": 0.08, "learning_rate": 4.6595e-05, "loss": 1.5243, "step": 54480 }, { "epoch": 0.08, "learning_rate": 4.6594375e-05, "loss": 1.7002, "step": 54490 }, { "epoch": 0.08, "learning_rate": 4.6593750000000004e-05, "loss": 1.6476, "step": 54500 }, { "epoch": 0.08, "learning_rate": 4.6593125e-05, "loss": 1.6566, "step": 54510 }, { "epoch": 0.08, "learning_rate": 4.65925e-05, "loss": 1.7714, "step": 54520 }, { "epoch": 0.08, "learning_rate": 4.6591875e-05, "loss": 1.7089, "step": 54530 }, { "epoch": 0.08, "learning_rate": 4.6591250000000006e-05, "loss": 1.5556, "step": 54540 }, { "epoch": 0.08, "learning_rate": 4.6590625e-05, "loss": 1.7306, "step": 54550 }, { "epoch": 0.08, "learning_rate": 4.659e-05, "loss": 1.3302, "step": 54560 }, { "epoch": 0.08, "learning_rate": 4.6589375e-05, "loss": 1.7926, "step": 54570 }, { "epoch": 0.08, "learning_rate": 4.658875e-05, "loss": 1.7469, "step": 54580 }, { "epoch": 0.08, "learning_rate": 4.6588125000000004e-05, "loss": 1.5194, "step": 54590 }, { "epoch": 0.08, "learning_rate": 4.65875e-05, "loss": 1.6426, "step": 54600 }, { "epoch": 0.08, "learning_rate": 4.6586875000000004e-05, "loss": 1.5154, "step": 54610 }, { "epoch": 0.08, "learning_rate": 4.658625e-05, "loss": 1.7646, "step": 54620 }, { "epoch": 0.08, "learning_rate": 4.6585625000000005e-05, "loss": 1.7282, "step": 54630 }, { "epoch": 0.08, "learning_rate": 4.6585e-05, "loss": 1.6044, "step": 54640 }, { "epoch": 0.08, "learning_rate": 4.6584375e-05, "loss": 1.8084, "step": 54650 }, { "epoch": 0.08, "learning_rate": 4.658375e-05, "loss": 1.5572, "step": 54660 }, { "epoch": 0.08, "learning_rate": 4.6583125e-05, "loss": 1.8036, "step": 54670 }, { "epoch": 0.08, "learning_rate": 4.65825e-05, "loss": 1.5242, "step": 54680 }, { "epoch": 0.08, "learning_rate": 4.6581875e-05, "loss": 1.4967, "step": 54690 }, { "epoch": 0.08, "learning_rate": 4.6581250000000004e-05, "loss": 1.6094, "step": 54700 }, { "epoch": 0.08, "learning_rate": 4.6580625e-05, "loss": 1.7732, "step": 54710 }, { "epoch": 0.08, "learning_rate": 4.6580000000000005e-05, "loss": 1.6703, "step": 54720 }, { "epoch": 0.08, "learning_rate": 4.6579375e-05, "loss": 1.6137, "step": 54730 }, { "epoch": 0.08, "learning_rate": 4.6578750000000005e-05, "loss": 1.6013, "step": 54740 }, { "epoch": 0.08, "learning_rate": 4.6578125e-05, "loss": 1.5115, "step": 54750 }, { "epoch": 0.08, "learning_rate": 4.65775e-05, "loss": 1.816, "step": 54760 }, { "epoch": 0.08, "learning_rate": 4.6576875e-05, "loss": 1.6636, "step": 54770 }, { "epoch": 0.08, "learning_rate": 4.6576250000000007e-05, "loss": 1.6161, "step": 54780 }, { "epoch": 0.08, "learning_rate": 4.6575625000000004e-05, "loss": 1.6181, "step": 54790 }, { "epoch": 0.08, "learning_rate": 4.6575e-05, "loss": 1.578, "step": 54800 }, { "epoch": 0.08, "learning_rate": 4.6574375000000004e-05, "loss": 1.5344, "step": 54810 }, { "epoch": 0.08, "learning_rate": 4.657375000000001e-05, "loss": 1.5546, "step": 54820 }, { "epoch": 0.08, "learning_rate": 4.6573125e-05, "loss": 1.4665, "step": 54830 }, { "epoch": 0.08, "learning_rate": 4.65725e-05, "loss": 1.3409, "step": 54840 }, { "epoch": 0.08, "learning_rate": 4.6571875000000005e-05, "loss": 1.6005, "step": 54850 }, { "epoch": 0.08, "learning_rate": 4.657125e-05, "loss": 1.6587, "step": 54860 }, { "epoch": 0.08, "learning_rate": 4.6570625e-05, "loss": 1.6535, "step": 54870 }, { "epoch": 0.08, "learning_rate": 4.657e-05, "loss": 1.819, "step": 54880 }, { "epoch": 0.08, "learning_rate": 4.656937500000001e-05, "loss": 1.7112, "step": 54890 }, { "epoch": 0.08, "learning_rate": 4.656875e-05, "loss": 1.4443, "step": 54900 }, { "epoch": 0.08, "learning_rate": 4.6568125e-05, "loss": 1.5825, "step": 54910 }, { "epoch": 0.08, "learning_rate": 4.6567500000000004e-05, "loss": 1.879, "step": 54920 }, { "epoch": 0.08, "learning_rate": 4.6566875e-05, "loss": 1.7632, "step": 54930 }, { "epoch": 0.08, "learning_rate": 4.656625e-05, "loss": 1.4482, "step": 54940 }, { "epoch": 0.08, "learning_rate": 4.6565625e-05, "loss": 1.6933, "step": 54950 }, { "epoch": 0.08, "learning_rate": 4.6565000000000006e-05, "loss": 1.6237, "step": 54960 }, { "epoch": 0.08, "learning_rate": 4.6564375e-05, "loss": 1.6373, "step": 54970 }, { "epoch": 0.08, "learning_rate": 4.656375e-05, "loss": 1.72, "step": 54980 }, { "epoch": 0.08, "learning_rate": 4.6563125e-05, "loss": 1.7994, "step": 54990 }, { "epoch": 0.08, "learning_rate": 4.65625e-05, "loss": 1.6221, "step": 55000 }, { "epoch": 0.08, "eval_loss": 1.6151599884033203, "eval_runtime": 1392.8037, "eval_samples_per_second": 41.668, "eval_steps_per_second": 5.209, "step": 55000 }, { "epoch": 0.08, "learning_rate": 4.6561875000000004e-05, "loss": 1.8737, "step": 55010 }, { "epoch": 0.08, "learning_rate": 4.656125e-05, "loss": 1.59, "step": 55020 }, { "epoch": 0.08, "learning_rate": 4.6560625000000005e-05, "loss": 1.7522, "step": 55030 }, { "epoch": 0.08, "learning_rate": 4.656e-05, "loss": 1.6603, "step": 55040 }, { "epoch": 0.08, "learning_rate": 4.6559375e-05, "loss": 1.74, "step": 55050 }, { "epoch": 0.08, "learning_rate": 4.655875e-05, "loss": 1.5713, "step": 55060 }, { "epoch": 0.08, "learning_rate": 4.6558125000000006e-05, "loss": 1.6959, "step": 55070 }, { "epoch": 0.08, "learning_rate": 4.65575e-05, "loss": 1.952, "step": 55080 }, { "epoch": 0.08, "learning_rate": 4.6556875e-05, "loss": 1.5731, "step": 55090 }, { "epoch": 0.08, "learning_rate": 4.6556250000000003e-05, "loss": 1.9377, "step": 55100 }, { "epoch": 0.08, "learning_rate": 4.6555625e-05, "loss": 1.8469, "step": 55110 }, { "epoch": 0.08, "learning_rate": 4.6555000000000004e-05, "loss": 1.5358, "step": 55120 }, { "epoch": 0.08, "learning_rate": 4.6554375e-05, "loss": 1.7001, "step": 55130 }, { "epoch": 0.08, "learning_rate": 4.6553750000000005e-05, "loss": 1.5995, "step": 55140 }, { "epoch": 0.08, "learning_rate": 4.6553125e-05, "loss": 1.682, "step": 55150 }, { "epoch": 0.08, "learning_rate": 4.6552500000000005e-05, "loss": 1.5456, "step": 55160 }, { "epoch": 0.08, "learning_rate": 4.6551875e-05, "loss": 1.5437, "step": 55170 }, { "epoch": 0.08, "learning_rate": 4.655125e-05, "loss": 1.7268, "step": 55180 }, { "epoch": 0.08, "learning_rate": 4.6550625e-05, "loss": 1.6096, "step": 55190 }, { "epoch": 0.08, "learning_rate": 4.655000000000001e-05, "loss": 1.5518, "step": 55200 }, { "epoch": 0.08, "learning_rate": 4.6549375000000004e-05, "loss": 1.7129, "step": 55210 }, { "epoch": 0.08, "learning_rate": 4.654875e-05, "loss": 1.6555, "step": 55220 }, { "epoch": 0.08, "learning_rate": 4.6548125000000004e-05, "loss": 1.8355, "step": 55230 }, { "epoch": 0.08, "learning_rate": 4.654750000000001e-05, "loss": 1.7297, "step": 55240 }, { "epoch": 0.08, "learning_rate": 4.6546875e-05, "loss": 1.724, "step": 55250 }, { "epoch": 0.08, "learning_rate": 4.654625e-05, "loss": 1.7345, "step": 55260 }, { "epoch": 0.08, "learning_rate": 4.6545625000000006e-05, "loss": 1.5383, "step": 55270 }, { "epoch": 0.08, "learning_rate": 4.6545e-05, "loss": 1.9229, "step": 55280 }, { "epoch": 0.08, "learning_rate": 4.6544375e-05, "loss": 1.8324, "step": 55290 }, { "epoch": 0.08, "learning_rate": 4.654375e-05, "loss": 1.5833, "step": 55300 }, { "epoch": 0.08, "learning_rate": 4.654312500000001e-05, "loss": 1.7998, "step": 55310 }, { "epoch": 0.08, "learning_rate": 4.65425e-05, "loss": 1.7434, "step": 55320 }, { "epoch": 0.08, "learning_rate": 4.6541875e-05, "loss": 1.9544, "step": 55330 }, { "epoch": 0.08, "learning_rate": 4.6541250000000004e-05, "loss": 1.764, "step": 55340 }, { "epoch": 0.08, "learning_rate": 4.6540625e-05, "loss": 1.5111, "step": 55350 }, { "epoch": 0.08, "learning_rate": 4.654e-05, "loss": 1.7395, "step": 55360 }, { "epoch": 0.08, "learning_rate": 4.6539375e-05, "loss": 1.7367, "step": 55370 }, { "epoch": 0.08, "learning_rate": 4.6538750000000006e-05, "loss": 1.5874, "step": 55380 }, { "epoch": 0.08, "learning_rate": 4.6538125e-05, "loss": 1.376, "step": 55390 }, { "epoch": 0.08, "learning_rate": 4.65375e-05, "loss": 1.7791, "step": 55400 }, { "epoch": 0.08, "learning_rate": 4.6536875e-05, "loss": 1.682, "step": 55410 }, { "epoch": 0.08, "learning_rate": 4.653625e-05, "loss": 1.5618, "step": 55420 }, { "epoch": 0.08, "learning_rate": 4.6535625000000004e-05, "loss": 1.7616, "step": 55430 }, { "epoch": 0.08, "learning_rate": 4.6535e-05, "loss": 1.6987, "step": 55440 }, { "epoch": 0.08, "learning_rate": 4.6534375000000005e-05, "loss": 1.7672, "step": 55450 }, { "epoch": 0.08, "learning_rate": 4.653375e-05, "loss": 1.6337, "step": 55460 }, { "epoch": 0.08, "learning_rate": 4.6533125e-05, "loss": 1.7526, "step": 55470 }, { "epoch": 0.08, "learning_rate": 4.65325e-05, "loss": 1.7056, "step": 55480 }, { "epoch": 0.08, "learning_rate": 4.6531875000000006e-05, "loss": 1.6762, "step": 55490 }, { "epoch": 0.08, "learning_rate": 4.653125e-05, "loss": 1.6309, "step": 55500 }, { "epoch": 0.08, "learning_rate": 4.6530625e-05, "loss": 1.5244, "step": 55510 }, { "epoch": 0.08, "learning_rate": 4.6530000000000003e-05, "loss": 1.7906, "step": 55520 }, { "epoch": 0.08, "learning_rate": 4.6529375e-05, "loss": 1.8878, "step": 55530 }, { "epoch": 0.08, "learning_rate": 4.6528750000000004e-05, "loss": 1.7191, "step": 55540 }, { "epoch": 0.08, "learning_rate": 4.6528125e-05, "loss": 1.713, "step": 55550 }, { "epoch": 0.08, "learning_rate": 4.6527500000000005e-05, "loss": 1.5791, "step": 55560 }, { "epoch": 0.08, "learning_rate": 4.6526875e-05, "loss": 1.6179, "step": 55570 }, { "epoch": 0.08, "learning_rate": 4.6526250000000005e-05, "loss": 1.4373, "step": 55580 }, { "epoch": 0.08, "learning_rate": 4.6525625e-05, "loss": 1.8017, "step": 55590 }, { "epoch": 0.08, "learning_rate": 4.6525e-05, "loss": 1.6828, "step": 55600 }, { "epoch": 0.08, "learning_rate": 4.6524375e-05, "loss": 1.5602, "step": 55610 }, { "epoch": 0.08, "learning_rate": 4.652375000000001e-05, "loss": 1.53, "step": 55620 }, { "epoch": 0.08, "learning_rate": 4.6523125000000004e-05, "loss": 1.7417, "step": 55630 }, { "epoch": 0.08, "learning_rate": 4.65225e-05, "loss": 1.8854, "step": 55640 }, { "epoch": 0.08, "learning_rate": 4.6521875000000004e-05, "loss": 1.7818, "step": 55650 }, { "epoch": 0.08, "learning_rate": 4.652125e-05, "loss": 1.7373, "step": 55660 }, { "epoch": 0.08, "learning_rate": 4.6520625e-05, "loss": 1.5846, "step": 55670 }, { "epoch": 0.08, "learning_rate": 4.652e-05, "loss": 1.7261, "step": 55680 }, { "epoch": 0.08, "learning_rate": 4.6519375000000006e-05, "loss": 1.6143, "step": 55690 }, { "epoch": 0.08, "learning_rate": 4.651875e-05, "loss": 1.8989, "step": 55700 }, { "epoch": 0.08, "learning_rate": 4.6518125e-05, "loss": 1.7225, "step": 55710 }, { "epoch": 0.08, "learning_rate": 4.65175e-05, "loss": 1.7025, "step": 55720 }, { "epoch": 0.08, "learning_rate": 4.651687500000001e-05, "loss": 1.5824, "step": 55730 }, { "epoch": 0.08, "learning_rate": 4.651625e-05, "loss": 1.4748, "step": 55740 }, { "epoch": 0.08, "learning_rate": 4.6515625e-05, "loss": 1.5442, "step": 55750 }, { "epoch": 0.08, "learning_rate": 4.6515000000000004e-05, "loss": 1.621, "step": 55760 }, { "epoch": 0.08, "learning_rate": 4.6514375e-05, "loss": 1.3683, "step": 55770 }, { "epoch": 0.08, "learning_rate": 4.651375e-05, "loss": 1.6425, "step": 55780 }, { "epoch": 0.08, "learning_rate": 4.6513125e-05, "loss": 1.8599, "step": 55790 }, { "epoch": 0.08, "learning_rate": 4.6512500000000006e-05, "loss": 1.6363, "step": 55800 }, { "epoch": 0.08, "learning_rate": 4.6511875e-05, "loss": 1.6414, "step": 55810 }, { "epoch": 0.08, "learning_rate": 4.651125e-05, "loss": 1.69, "step": 55820 }, { "epoch": 0.08, "learning_rate": 4.6510625e-05, "loss": 1.8015, "step": 55830 }, { "epoch": 0.08, "learning_rate": 4.651e-05, "loss": 1.6646, "step": 55840 }, { "epoch": 0.08, "learning_rate": 4.6509375e-05, "loss": 1.9129, "step": 55850 }, { "epoch": 0.08, "learning_rate": 4.650875e-05, "loss": 1.5059, "step": 55860 }, { "epoch": 0.08, "learning_rate": 4.6508125000000005e-05, "loss": 1.7856, "step": 55870 }, { "epoch": 0.08, "learning_rate": 4.65075e-05, "loss": 1.8651, "step": 55880 }, { "epoch": 0.08, "learning_rate": 4.6506875e-05, "loss": 1.7844, "step": 55890 }, { "epoch": 0.08, "learning_rate": 4.650625e-05, "loss": 1.8804, "step": 55900 }, { "epoch": 0.08, "learning_rate": 4.6505625000000006e-05, "loss": 1.6321, "step": 55910 }, { "epoch": 0.08, "learning_rate": 4.6505e-05, "loss": 1.483, "step": 55920 }, { "epoch": 0.08, "learning_rate": 4.6504375e-05, "loss": 1.7352, "step": 55930 }, { "epoch": 0.08, "learning_rate": 4.6503750000000003e-05, "loss": 1.7291, "step": 55940 }, { "epoch": 0.08, "learning_rate": 4.6503125e-05, "loss": 1.6951, "step": 55950 }, { "epoch": 0.08, "learning_rate": 4.6502500000000004e-05, "loss": 1.3214, "step": 55960 }, { "epoch": 0.08, "learning_rate": 4.6501875e-05, "loss": 1.8627, "step": 55970 }, { "epoch": 0.08, "learning_rate": 4.6501250000000005e-05, "loss": 1.5398, "step": 55980 }, { "epoch": 0.08, "learning_rate": 4.6500625e-05, "loss": 1.8298, "step": 55990 }, { "epoch": 0.08, "learning_rate": 4.6500000000000005e-05, "loss": 1.5918, "step": 56000 }, { "epoch": 0.08, "eval_loss": 1.600969910621643, "eval_runtime": 1313.89, "eval_samples_per_second": 44.171, "eval_steps_per_second": 5.522, "step": 56000 }, { "epoch": 0.08, "learning_rate": 4.6499375e-05, "loss": 1.6223, "step": 56010 }, { "epoch": 0.08, "learning_rate": 4.649875e-05, "loss": 1.6704, "step": 56020 }, { "epoch": 0.08, "learning_rate": 4.6498125e-05, "loss": 1.6367, "step": 56030 }, { "epoch": 0.08, "learning_rate": 4.649750000000001e-05, "loss": 1.7945, "step": 56040 }, { "epoch": 0.08, "learning_rate": 4.6496875000000004e-05, "loss": 1.7127, "step": 56050 }, { "epoch": 0.08, "learning_rate": 4.649625e-05, "loss": 1.7794, "step": 56060 }, { "epoch": 0.08, "learning_rate": 4.6495625000000004e-05, "loss": 1.7182, "step": 56070 }, { "epoch": 0.08, "learning_rate": 4.6495e-05, "loss": 1.6012, "step": 56080 }, { "epoch": 0.08, "learning_rate": 4.6494375e-05, "loss": 1.7075, "step": 56090 }, { "epoch": 0.08, "learning_rate": 4.649375e-05, "loss": 1.6442, "step": 56100 }, { "epoch": 0.08, "learning_rate": 4.6493125000000006e-05, "loss": 1.7448, "step": 56110 }, { "epoch": 0.08, "learning_rate": 4.64925e-05, "loss": 1.4941, "step": 56120 }, { "epoch": 0.08, "learning_rate": 4.6491875e-05, "loss": 1.5836, "step": 56130 }, { "epoch": 0.08, "learning_rate": 4.649125e-05, "loss": 1.7771, "step": 56140 }, { "epoch": 0.08, "learning_rate": 4.649062500000001e-05, "loss": 1.5346, "step": 56150 }, { "epoch": 0.08, "learning_rate": 4.649e-05, "loss": 1.8583, "step": 56160 }, { "epoch": 0.08, "learning_rate": 4.6489375e-05, "loss": 1.667, "step": 56170 }, { "epoch": 0.08, "learning_rate": 4.6488750000000004e-05, "loss": 1.551, "step": 56180 }, { "epoch": 0.08, "learning_rate": 4.6488125e-05, "loss": 1.737, "step": 56190 }, { "epoch": 0.08, "learning_rate": 4.64875e-05, "loss": 1.657, "step": 56200 }, { "epoch": 0.08, "learning_rate": 4.6486875e-05, "loss": 1.6131, "step": 56210 }, { "epoch": 0.08, "learning_rate": 4.6486250000000006e-05, "loss": 1.6013, "step": 56220 }, { "epoch": 0.08, "learning_rate": 4.6485625e-05, "loss": 1.6498, "step": 56230 }, { "epoch": 0.08, "learning_rate": 4.6485e-05, "loss": 1.8717, "step": 56240 }, { "epoch": 0.08, "learning_rate": 4.6484375e-05, "loss": 1.657, "step": 56250 }, { "epoch": 0.08, "learning_rate": 4.648375000000001e-05, "loss": 1.584, "step": 56260 }, { "epoch": 0.08, "learning_rate": 4.6483125e-05, "loss": 1.5821, "step": 56270 }, { "epoch": 0.08, "learning_rate": 4.64825e-05, "loss": 1.9301, "step": 56280 }, { "epoch": 0.08, "learning_rate": 4.6481875000000005e-05, "loss": 1.6964, "step": 56290 }, { "epoch": 0.08, "learning_rate": 4.648125e-05, "loss": 1.8161, "step": 56300 }, { "epoch": 0.08, "learning_rate": 4.6480625e-05, "loss": 1.7049, "step": 56310 }, { "epoch": 0.08, "learning_rate": 4.648e-05, "loss": 1.7121, "step": 56320 }, { "epoch": 0.08, "learning_rate": 4.6479375000000006e-05, "loss": 1.6971, "step": 56330 }, { "epoch": 0.08, "learning_rate": 4.647875e-05, "loss": 1.6524, "step": 56340 }, { "epoch": 0.08, "learning_rate": 4.6478125e-05, "loss": 1.552, "step": 56350 }, { "epoch": 0.08, "learning_rate": 4.6477500000000004e-05, "loss": 1.6977, "step": 56360 }, { "epoch": 0.08, "learning_rate": 4.6476875e-05, "loss": 1.6128, "step": 56370 }, { "epoch": 0.08, "learning_rate": 4.6476250000000004e-05, "loss": 1.8397, "step": 56380 }, { "epoch": 0.08, "learning_rate": 4.6475625e-05, "loss": 1.6998, "step": 56390 }, { "epoch": 0.08, "learning_rate": 4.6475000000000005e-05, "loss": 1.605, "step": 56400 }, { "epoch": 0.08, "learning_rate": 4.6474375e-05, "loss": 1.7246, "step": 56410 }, { "epoch": 0.08, "learning_rate": 4.6473750000000005e-05, "loss": 1.8735, "step": 56420 }, { "epoch": 0.08, "learning_rate": 4.6473125e-05, "loss": 1.6023, "step": 56430 }, { "epoch": 0.08, "learning_rate": 4.64725e-05, "loss": 1.7006, "step": 56440 }, { "epoch": 0.08, "learning_rate": 4.6471875e-05, "loss": 1.6973, "step": 56450 }, { "epoch": 0.08, "learning_rate": 4.647125e-05, "loss": 1.5718, "step": 56460 }, { "epoch": 0.08, "learning_rate": 4.6470625000000004e-05, "loss": 1.4929, "step": 56470 }, { "epoch": 0.08, "learning_rate": 4.647e-05, "loss": 1.6566, "step": 56480 }, { "epoch": 0.08, "learning_rate": 4.6469375000000004e-05, "loss": 1.743, "step": 56490 }, { "epoch": 0.08, "learning_rate": 4.646875e-05, "loss": 1.7537, "step": 56500 }, { "epoch": 0.08, "learning_rate": 4.6468125e-05, "loss": 1.4871, "step": 56510 }, { "epoch": 0.08, "learning_rate": 4.64675e-05, "loss": 1.6381, "step": 56520 }, { "epoch": 0.08, "learning_rate": 4.6466875000000006e-05, "loss": 1.5538, "step": 56530 }, { "epoch": 0.08, "learning_rate": 4.646625e-05, "loss": 1.431, "step": 56540 }, { "epoch": 0.08, "learning_rate": 4.6465625e-05, "loss": 1.7136, "step": 56550 }, { "epoch": 0.08, "learning_rate": 4.6465e-05, "loss": 1.5339, "step": 56560 }, { "epoch": 0.08, "learning_rate": 4.646437500000001e-05, "loss": 1.4659, "step": 56570 }, { "epoch": 0.08, "learning_rate": 4.646375e-05, "loss": 1.7406, "step": 56580 }, { "epoch": 0.08, "learning_rate": 4.6463125e-05, "loss": 1.3433, "step": 56590 }, { "epoch": 0.08, "learning_rate": 4.6462500000000005e-05, "loss": 1.6896, "step": 56600 }, { "epoch": 0.08, "learning_rate": 4.6461875e-05, "loss": 1.6008, "step": 56610 }, { "epoch": 0.08, "learning_rate": 4.646125e-05, "loss": 1.8296, "step": 56620 }, { "epoch": 0.08, "learning_rate": 4.6460625e-05, "loss": 1.8681, "step": 56630 }, { "epoch": 0.08, "learning_rate": 4.6460000000000006e-05, "loss": 1.6625, "step": 56640 }, { "epoch": 0.08, "learning_rate": 4.6459375e-05, "loss": 1.6196, "step": 56650 }, { "epoch": 0.08, "learning_rate": 4.645875e-05, "loss": 1.8019, "step": 56660 }, { "epoch": 0.08, "learning_rate": 4.6458125e-05, "loss": 1.4438, "step": 56670 }, { "epoch": 0.08, "learning_rate": 4.645750000000001e-05, "loss": 1.5813, "step": 56680 }, { "epoch": 0.08, "learning_rate": 4.6456875e-05, "loss": 1.5918, "step": 56690 }, { "epoch": 0.08, "learning_rate": 4.645625e-05, "loss": 1.6665, "step": 56700 }, { "epoch": 0.08, "learning_rate": 4.6455625000000005e-05, "loss": 1.6335, "step": 56710 }, { "epoch": 0.08, "learning_rate": 4.6455e-05, "loss": 1.8936, "step": 56720 }, { "epoch": 0.08, "learning_rate": 4.6454375e-05, "loss": 1.5382, "step": 56730 }, { "epoch": 0.08, "learning_rate": 4.645375e-05, "loss": 1.8718, "step": 56740 }, { "epoch": 0.08, "learning_rate": 4.6453125000000006e-05, "loss": 1.6944, "step": 56750 }, { "epoch": 0.08, "learning_rate": 4.64525e-05, "loss": 1.6766, "step": 56760 }, { "epoch": 0.08, "learning_rate": 4.6451875e-05, "loss": 1.6269, "step": 56770 }, { "epoch": 0.08, "learning_rate": 4.6451250000000004e-05, "loss": 1.6182, "step": 56780 }, { "epoch": 0.08, "learning_rate": 4.6450625e-05, "loss": 1.7546, "step": 56790 }, { "epoch": 0.08, "learning_rate": 4.6450000000000004e-05, "loss": 1.5757, "step": 56800 }, { "epoch": 0.08, "learning_rate": 4.6449375e-05, "loss": 1.4759, "step": 56810 }, { "epoch": 0.08, "learning_rate": 4.6448750000000005e-05, "loss": 1.6675, "step": 56820 }, { "epoch": 0.08, "learning_rate": 4.6448125e-05, "loss": 1.6235, "step": 56830 }, { "epoch": 0.08, "learning_rate": 4.6447500000000005e-05, "loss": 1.8928, "step": 56840 }, { "epoch": 0.08, "learning_rate": 4.6446875e-05, "loss": 1.7254, "step": 56850 }, { "epoch": 0.08, "learning_rate": 4.644625e-05, "loss": 1.5205, "step": 56860 }, { "epoch": 0.08, "learning_rate": 4.6445625e-05, "loss": 1.692, "step": 56870 }, { "epoch": 0.08, "learning_rate": 4.6445e-05, "loss": 1.688, "step": 56880 }, { "epoch": 0.08, "learning_rate": 4.6444375000000004e-05, "loss": 1.683, "step": 56890 }, { "epoch": 0.08, "learning_rate": 4.644375e-05, "loss": 1.802, "step": 56900 }, { "epoch": 0.08, "learning_rate": 4.6443125000000004e-05, "loss": 1.8281, "step": 56910 }, { "epoch": 0.08, "learning_rate": 4.64425e-05, "loss": 1.5358, "step": 56920 }, { "epoch": 0.08, "learning_rate": 4.6441875e-05, "loss": 1.587, "step": 56930 }, { "epoch": 0.08, "learning_rate": 4.644125e-05, "loss": 1.6151, "step": 56940 }, { "epoch": 0.08, "learning_rate": 4.6440625000000006e-05, "loss": 1.6717, "step": 56950 }, { "epoch": 0.08, "learning_rate": 4.644e-05, "loss": 1.5656, "step": 56960 }, { "epoch": 0.08, "learning_rate": 4.6439375e-05, "loss": 1.7376, "step": 56970 }, { "epoch": 0.08, "learning_rate": 4.643875e-05, "loss": 1.7101, "step": 56980 }, { "epoch": 0.08, "learning_rate": 4.643812500000001e-05, "loss": 1.544, "step": 56990 }, { "epoch": 0.08, "learning_rate": 4.64375e-05, "loss": 1.6419, "step": 57000 }, { "epoch": 0.08, "eval_loss": 1.6000648736953735, "eval_runtime": 1320.6233, "eval_samples_per_second": 43.946, "eval_steps_per_second": 5.494, "step": 57000 }, { "epoch": 0.08, "learning_rate": 4.6436875e-05, "loss": 1.6581, "step": 57010 }, { "epoch": 0.08, "learning_rate": 4.6436250000000005e-05, "loss": 1.628, "step": 57020 }, { "epoch": 0.08, "learning_rate": 4.6435625e-05, "loss": 1.6841, "step": 57030 }, { "epoch": 0.08, "learning_rate": 4.6435e-05, "loss": 1.5182, "step": 57040 }, { "epoch": 0.08, "learning_rate": 4.6434375e-05, "loss": 1.7142, "step": 57050 }, { "epoch": 0.08, "learning_rate": 4.6433750000000006e-05, "loss": 1.7386, "step": 57060 }, { "epoch": 0.08, "learning_rate": 4.6433125e-05, "loss": 1.6874, "step": 57070 }, { "epoch": 0.08, "learning_rate": 4.64325e-05, "loss": 1.5234, "step": 57080 }, { "epoch": 0.08, "learning_rate": 4.6431875000000003e-05, "loss": 1.4447, "step": 57090 }, { "epoch": 0.08, "learning_rate": 4.643125000000001e-05, "loss": 1.6999, "step": 57100 }, { "epoch": 0.08, "learning_rate": 4.6430625e-05, "loss": 1.8383, "step": 57110 }, { "epoch": 0.08, "learning_rate": 4.643e-05, "loss": 1.5208, "step": 57120 }, { "epoch": 0.08, "learning_rate": 4.6429375000000005e-05, "loss": 1.6412, "step": 57130 }, { "epoch": 0.08, "learning_rate": 4.642875e-05, "loss": 1.7595, "step": 57140 }, { "epoch": 0.08, "learning_rate": 4.6428125e-05, "loss": 1.6314, "step": 57150 }, { "epoch": 0.08, "learning_rate": 4.64275e-05, "loss": 1.8498, "step": 57160 }, { "epoch": 0.08, "learning_rate": 4.6426875000000006e-05, "loss": 1.6613, "step": 57170 }, { "epoch": 0.08, "learning_rate": 4.642625e-05, "loss": 1.7716, "step": 57180 }, { "epoch": 0.08, "learning_rate": 4.6425625e-05, "loss": 1.7088, "step": 57190 }, { "epoch": 0.08, "learning_rate": 4.6425000000000004e-05, "loss": 1.8855, "step": 57200 }, { "epoch": 0.08, "learning_rate": 4.6424375e-05, "loss": 1.4825, "step": 57210 }, { "epoch": 0.08, "learning_rate": 4.6423750000000004e-05, "loss": 1.8204, "step": 57220 }, { "epoch": 0.08, "learning_rate": 4.6423125e-05, "loss": 1.7568, "step": 57230 }, { "epoch": 0.08, "learning_rate": 4.6422500000000005e-05, "loss": 1.6563, "step": 57240 }, { "epoch": 0.08, "learning_rate": 4.6421875e-05, "loss": 1.5936, "step": 57250 }, { "epoch": 0.08, "learning_rate": 4.642125e-05, "loss": 1.4401, "step": 57260 }, { "epoch": 0.08, "learning_rate": 4.6420625e-05, "loss": 1.87, "step": 57270 }, { "epoch": 0.08, "learning_rate": 4.642e-05, "loss": 1.7379, "step": 57280 }, { "epoch": 0.08, "learning_rate": 4.6419375e-05, "loss": 1.6225, "step": 57290 }, { "epoch": 0.08, "learning_rate": 4.641875e-05, "loss": 1.7092, "step": 57300 }, { "epoch": 0.08, "learning_rate": 4.6418125000000004e-05, "loss": 1.8752, "step": 57310 }, { "epoch": 0.08, "learning_rate": 4.64175e-05, "loss": 1.6119, "step": 57320 }, { "epoch": 0.08, "learning_rate": 4.6416875000000004e-05, "loss": 1.5311, "step": 57330 }, { "epoch": 0.08, "learning_rate": 4.641625e-05, "loss": 1.6514, "step": 57340 }, { "epoch": 0.08, "learning_rate": 4.6415625e-05, "loss": 1.521, "step": 57350 }, { "epoch": 0.08, "learning_rate": 4.6415e-05, "loss": 1.7888, "step": 57360 }, { "epoch": 0.08, "learning_rate": 4.6414375000000006e-05, "loss": 1.7105, "step": 57370 }, { "epoch": 0.08, "learning_rate": 4.641375e-05, "loss": 1.5745, "step": 57380 }, { "epoch": 0.08, "learning_rate": 4.6413125e-05, "loss": 1.7222, "step": 57390 }, { "epoch": 0.08, "learning_rate": 4.64125e-05, "loss": 1.6464, "step": 57400 }, { "epoch": 0.08, "learning_rate": 4.641187500000001e-05, "loss": 1.4821, "step": 57410 }, { "epoch": 0.08, "learning_rate": 4.6411250000000004e-05, "loss": 1.7374, "step": 57420 }, { "epoch": 0.08, "learning_rate": 4.6410625e-05, "loss": 1.6299, "step": 57430 }, { "epoch": 0.08, "learning_rate": 4.6410000000000005e-05, "loss": 1.8286, "step": 57440 }, { "epoch": 0.08, "learning_rate": 4.640937500000001e-05, "loss": 1.6751, "step": 57450 }, { "epoch": 0.08, "learning_rate": 4.640875e-05, "loss": 1.6891, "step": 57460 }, { "epoch": 0.08, "learning_rate": 4.6408125e-05, "loss": 1.9798, "step": 57470 }, { "epoch": 0.08, "learning_rate": 4.6407500000000006e-05, "loss": 1.7499, "step": 57480 }, { "epoch": 0.08, "learning_rate": 4.6406875e-05, "loss": 1.5399, "step": 57490 }, { "epoch": 0.08, "learning_rate": 4.640625e-05, "loss": 1.7479, "step": 57500 }, { "epoch": 0.08, "learning_rate": 4.6405625000000003e-05, "loss": 1.6206, "step": 57510 }, { "epoch": 0.08, "learning_rate": 4.640500000000001e-05, "loss": 1.6498, "step": 57520 }, { "epoch": 0.08, "learning_rate": 4.6404375e-05, "loss": 1.5607, "step": 57530 }, { "epoch": 0.08, "learning_rate": 4.640375e-05, "loss": 1.7368, "step": 57540 }, { "epoch": 0.08, "learning_rate": 4.6403125000000005e-05, "loss": 1.6685, "step": 57550 }, { "epoch": 0.08, "learning_rate": 4.64025e-05, "loss": 1.5855, "step": 57560 }, { "epoch": 0.08, "learning_rate": 4.6401875e-05, "loss": 1.7931, "step": 57570 }, { "epoch": 0.08, "learning_rate": 4.640125e-05, "loss": 1.8092, "step": 57580 }, { "epoch": 0.08, "learning_rate": 4.6400625000000006e-05, "loss": 1.7496, "step": 57590 }, { "epoch": 0.08, "learning_rate": 4.64e-05, "loss": 1.5407, "step": 57600 }, { "epoch": 0.08, "learning_rate": 4.6399375e-05, "loss": 1.811, "step": 57610 }, { "epoch": 0.08, "learning_rate": 4.6398750000000004e-05, "loss": 1.7931, "step": 57620 }, { "epoch": 0.08, "learning_rate": 4.6398125e-05, "loss": 1.5446, "step": 57630 }, { "epoch": 0.08, "learning_rate": 4.6397500000000004e-05, "loss": 1.6158, "step": 57640 }, { "epoch": 0.08, "learning_rate": 4.6396875e-05, "loss": 1.4546, "step": 57650 }, { "epoch": 0.08, "learning_rate": 4.6396250000000005e-05, "loss": 1.8069, "step": 57660 }, { "epoch": 0.08, "learning_rate": 4.6395625e-05, "loss": 1.5074, "step": 57670 }, { "epoch": 0.08, "learning_rate": 4.6395e-05, "loss": 1.6116, "step": 57680 }, { "epoch": 0.08, "learning_rate": 4.6394375e-05, "loss": 1.9137, "step": 57690 }, { "epoch": 0.08, "learning_rate": 4.639375e-05, "loss": 1.5356, "step": 57700 }, { "epoch": 0.08, "learning_rate": 4.6393125e-05, "loss": 1.6157, "step": 57710 }, { "epoch": 0.08, "learning_rate": 4.63925e-05, "loss": 1.5015, "step": 57720 }, { "epoch": 0.08, "learning_rate": 4.6391875000000004e-05, "loss": 1.5613, "step": 57730 }, { "epoch": 0.08, "learning_rate": 4.639125e-05, "loss": 1.6942, "step": 57740 }, { "epoch": 0.08, "learning_rate": 4.6390625000000004e-05, "loss": 1.5727, "step": 57750 }, { "epoch": 0.08, "learning_rate": 4.639e-05, "loss": 1.6184, "step": 57760 }, { "epoch": 0.08, "learning_rate": 4.6389375e-05, "loss": 1.9358, "step": 57770 }, { "epoch": 0.08, "learning_rate": 4.638875e-05, "loss": 1.6133, "step": 57780 }, { "epoch": 0.08, "learning_rate": 4.6388125000000006e-05, "loss": 1.6682, "step": 57790 }, { "epoch": 0.08, "learning_rate": 4.63875e-05, "loss": 1.7915, "step": 57800 }, { "epoch": 0.08, "learning_rate": 4.6386875e-05, "loss": 1.5229, "step": 57810 }, { "epoch": 0.08, "learning_rate": 4.638625e-05, "loss": 1.5316, "step": 57820 }, { "epoch": 0.08, "learning_rate": 4.638562500000001e-05, "loss": 1.7727, "step": 57830 }, { "epoch": 0.08, "learning_rate": 4.6385000000000004e-05, "loss": 1.9108, "step": 57840 }, { "epoch": 0.08, "learning_rate": 4.6384375e-05, "loss": 1.8294, "step": 57850 }, { "epoch": 0.08, "learning_rate": 4.6383750000000005e-05, "loss": 1.8966, "step": 57860 }, { "epoch": 0.08, "learning_rate": 4.6383125e-05, "loss": 1.441, "step": 57870 }, { "epoch": 0.08, "learning_rate": 4.63825e-05, "loss": 1.7027, "step": 57880 }, { "epoch": 0.08, "learning_rate": 4.6381875e-05, "loss": 1.5524, "step": 57890 }, { "epoch": 0.08, "learning_rate": 4.6381250000000006e-05, "loss": 1.6584, "step": 57900 }, { "epoch": 0.08, "learning_rate": 4.6380625e-05, "loss": 1.863, "step": 57910 }, { "epoch": 0.08, "learning_rate": 4.638e-05, "loss": 1.547, "step": 57920 }, { "epoch": 0.08, "learning_rate": 4.6379375000000003e-05, "loss": 1.6267, "step": 57930 }, { "epoch": 0.08, "learning_rate": 4.637875000000001e-05, "loss": 1.614, "step": 57940 }, { "epoch": 0.08, "learning_rate": 4.6378125e-05, "loss": 1.6027, "step": 57950 }, { "epoch": 0.08, "learning_rate": 4.63775e-05, "loss": 1.6055, "step": 57960 }, { "epoch": 0.08, "learning_rate": 4.6376875000000005e-05, "loss": 1.5986, "step": 57970 }, { "epoch": 0.08, "learning_rate": 4.637625e-05, "loss": 1.5885, "step": 57980 }, { "epoch": 0.08, "learning_rate": 4.6375625e-05, "loss": 1.5415, "step": 57990 }, { "epoch": 0.08, "learning_rate": 4.6375e-05, "loss": 1.8135, "step": 58000 }, { "epoch": 0.08, "eval_loss": 1.5844295024871826, "eval_runtime": 1297.1857, "eval_samples_per_second": 44.74, "eval_steps_per_second": 5.593, "step": 58000 }, { "epoch": 0.08, "learning_rate": 4.6374375000000006e-05, "loss": 1.7768, "step": 58010 }, { "epoch": 0.08, "learning_rate": 4.637375e-05, "loss": 1.6917, "step": 58020 }, { "epoch": 0.08, "learning_rate": 4.6373125e-05, "loss": 1.4816, "step": 58030 }, { "epoch": 0.08, "learning_rate": 4.6372500000000004e-05, "loss": 1.5226, "step": 58040 }, { "epoch": 0.08, "learning_rate": 4.6371875e-05, "loss": 1.65, "step": 58050 }, { "epoch": 0.08, "learning_rate": 4.6371250000000004e-05, "loss": 1.6544, "step": 58060 }, { "epoch": 0.08, "learning_rate": 4.6370625e-05, "loss": 1.6248, "step": 58070 }, { "epoch": 0.08, "learning_rate": 4.6370000000000005e-05, "loss": 1.5201, "step": 58080 }, { "epoch": 0.08, "learning_rate": 4.6369375e-05, "loss": 1.8594, "step": 58090 }, { "epoch": 0.08, "learning_rate": 4.636875e-05, "loss": 1.6947, "step": 58100 }, { "epoch": 0.08, "learning_rate": 4.6368125e-05, "loss": 1.528, "step": 58110 }, { "epoch": 0.08, "learning_rate": 4.63675e-05, "loss": 1.7539, "step": 58120 }, { "epoch": 0.08, "learning_rate": 4.6366875e-05, "loss": 1.914, "step": 58130 }, { "epoch": 0.08, "learning_rate": 4.636625e-05, "loss": 1.6054, "step": 58140 }, { "epoch": 0.08, "learning_rate": 4.6365625000000004e-05, "loss": 1.4889, "step": 58150 }, { "epoch": 0.08, "learning_rate": 4.6365e-05, "loss": 1.6818, "step": 58160 }, { "epoch": 0.08, "learning_rate": 4.6364375000000004e-05, "loss": 1.5758, "step": 58170 }, { "epoch": 0.08, "learning_rate": 4.636375e-05, "loss": 1.6046, "step": 58180 }, { "epoch": 0.08, "learning_rate": 4.6363125e-05, "loss": 1.5451, "step": 58190 }, { "epoch": 0.08, "learning_rate": 4.63625e-05, "loss": 1.5652, "step": 58200 }, { "epoch": 0.08, "learning_rate": 4.6361875000000006e-05, "loss": 1.5586, "step": 58210 }, { "epoch": 0.08, "learning_rate": 4.636125e-05, "loss": 1.5532, "step": 58220 }, { "epoch": 0.08, "learning_rate": 4.6360625e-05, "loss": 1.6318, "step": 58230 }, { "epoch": 0.08, "learning_rate": 4.636e-05, "loss": 1.8131, "step": 58240 }, { "epoch": 0.08, "learning_rate": 4.635937500000001e-05, "loss": 1.569, "step": 58250 }, { "epoch": 0.08, "learning_rate": 4.6358750000000004e-05, "loss": 1.808, "step": 58260 }, { "epoch": 0.08, "learning_rate": 4.6358125e-05, "loss": 1.7606, "step": 58270 }, { "epoch": 0.08, "learning_rate": 4.6357500000000005e-05, "loss": 1.6959, "step": 58280 }, { "epoch": 0.08, "learning_rate": 4.6356875e-05, "loss": 1.7111, "step": 58290 }, { "epoch": 0.08, "learning_rate": 4.635625e-05, "loss": 1.5941, "step": 58300 }, { "epoch": 0.08, "learning_rate": 4.6355625e-05, "loss": 1.449, "step": 58310 }, { "epoch": 0.08, "learning_rate": 4.6355000000000006e-05, "loss": 1.8471, "step": 58320 }, { "epoch": 0.08, "learning_rate": 4.6354375e-05, "loss": 2.1279, "step": 58330 }, { "epoch": 0.08, "learning_rate": 4.635375e-05, "loss": 1.6026, "step": 58340 }, { "epoch": 0.08, "learning_rate": 4.6353125000000004e-05, "loss": 1.7212, "step": 58350 }, { "epoch": 0.08, "learning_rate": 4.635250000000001e-05, "loss": 1.5485, "step": 58360 }, { "epoch": 0.08, "learning_rate": 4.6351875e-05, "loss": 1.5212, "step": 58370 }, { "epoch": 0.08, "learning_rate": 4.635125e-05, "loss": 1.6585, "step": 58380 }, { "epoch": 0.08, "learning_rate": 4.6350625000000005e-05, "loss": 1.5166, "step": 58390 }, { "epoch": 0.08, "learning_rate": 4.635e-05, "loss": 1.5669, "step": 58400 }, { "epoch": 0.08, "learning_rate": 4.6349375e-05, "loss": 1.5149, "step": 58410 }, { "epoch": 0.08, "learning_rate": 4.634875e-05, "loss": 1.6991, "step": 58420 }, { "epoch": 0.08, "learning_rate": 4.6348125000000006e-05, "loss": 1.6292, "step": 58430 }, { "epoch": 0.08, "learning_rate": 4.63475e-05, "loss": 1.6457, "step": 58440 }, { "epoch": 0.08, "learning_rate": 4.6346875e-05, "loss": 1.6623, "step": 58450 }, { "epoch": 0.08, "learning_rate": 4.6346250000000004e-05, "loss": 1.599, "step": 58460 }, { "epoch": 0.08, "learning_rate": 4.6345625e-05, "loss": 1.7037, "step": 58470 }, { "epoch": 0.08, "learning_rate": 4.6345e-05, "loss": 1.6618, "step": 58480 }, { "epoch": 0.08, "learning_rate": 4.6344375e-05, "loss": 1.4524, "step": 58490 }, { "epoch": 0.08, "learning_rate": 4.6343750000000005e-05, "loss": 1.6084, "step": 58500 }, { "epoch": 0.08, "learning_rate": 4.6343125e-05, "loss": 1.6849, "step": 58510 }, { "epoch": 0.08, "learning_rate": 4.63425e-05, "loss": 1.4787, "step": 58520 }, { "epoch": 0.08, "learning_rate": 4.6341875e-05, "loss": 1.7096, "step": 58530 }, { "epoch": 0.08, "learning_rate": 4.634125e-05, "loss": 1.6999, "step": 58540 }, { "epoch": 0.08, "learning_rate": 4.6340625e-05, "loss": 1.5092, "step": 58550 }, { "epoch": 0.08, "learning_rate": 4.634e-05, "loss": 1.5194, "step": 58560 }, { "epoch": 0.08, "learning_rate": 4.6339375000000004e-05, "loss": 1.7265, "step": 58570 }, { "epoch": 0.08, "learning_rate": 4.633875e-05, "loss": 1.5986, "step": 58580 }, { "epoch": 0.08, "learning_rate": 4.6338125000000005e-05, "loss": 1.5828, "step": 58590 }, { "epoch": 0.08, "learning_rate": 4.63375e-05, "loss": 1.5242, "step": 58600 }, { "epoch": 0.08, "learning_rate": 4.6336875000000005e-05, "loss": 1.7057, "step": 58610 }, { "epoch": 0.08, "learning_rate": 4.633625e-05, "loss": 1.6875, "step": 58620 }, { "epoch": 0.08, "learning_rate": 4.6335625000000006e-05, "loss": 2.0259, "step": 58630 }, { "epoch": 0.08, "learning_rate": 4.6335e-05, "loss": 1.5472, "step": 58640 }, { "epoch": 0.08, "learning_rate": 4.6334375e-05, "loss": 1.6527, "step": 58650 }, { "epoch": 0.08, "learning_rate": 4.6333750000000003e-05, "loss": 1.6659, "step": 58660 }, { "epoch": 0.08, "learning_rate": 4.6333125e-05, "loss": 1.8456, "step": 58670 }, { "epoch": 0.08, "learning_rate": 4.6332500000000004e-05, "loss": 1.7079, "step": 58680 }, { "epoch": 0.08, "learning_rate": 4.6331875e-05, "loss": 1.679, "step": 58690 }, { "epoch": 0.08, "learning_rate": 4.6331250000000005e-05, "loss": 1.6481, "step": 58700 }, { "epoch": 0.08, "learning_rate": 4.6330625e-05, "loss": 1.7071, "step": 58710 }, { "epoch": 0.08, "learning_rate": 4.633e-05, "loss": 1.6217, "step": 58720 }, { "epoch": 0.08, "learning_rate": 4.6329375e-05, "loss": 1.6888, "step": 58730 }, { "epoch": 0.08, "learning_rate": 4.6328750000000006e-05, "loss": 1.6592, "step": 58740 }, { "epoch": 0.08, "learning_rate": 4.6328125e-05, "loss": 1.7118, "step": 58750 }, { "epoch": 0.08, "learning_rate": 4.63275e-05, "loss": 1.7544, "step": 58760 }, { "epoch": 0.08, "learning_rate": 4.6326875000000004e-05, "loss": 1.6047, "step": 58770 }, { "epoch": 0.08, "learning_rate": 4.632625000000001e-05, "loss": 1.6937, "step": 58780 }, { "epoch": 0.08, "learning_rate": 4.6325625e-05, "loss": 1.7463, "step": 58790 }, { "epoch": 0.08, "learning_rate": 4.6325e-05, "loss": 1.4275, "step": 58800 }, { "epoch": 0.08, "learning_rate": 4.6324375000000005e-05, "loss": 1.6878, "step": 58810 }, { "epoch": 0.08, "learning_rate": 4.632375e-05, "loss": 1.5983, "step": 58820 }, { "epoch": 0.08, "learning_rate": 4.6323125e-05, "loss": 1.5888, "step": 58830 }, { "epoch": 0.08, "learning_rate": 4.63225e-05, "loss": 1.6708, "step": 58840 }, { "epoch": 0.08, "learning_rate": 4.6321875000000006e-05, "loss": 1.6753, "step": 58850 }, { "epoch": 0.08, "learning_rate": 4.632125e-05, "loss": 1.4419, "step": 58860 }, { "epoch": 0.08, "learning_rate": 4.6320625e-05, "loss": 1.6108, "step": 58870 }, { "epoch": 0.08, "learning_rate": 4.6320000000000004e-05, "loss": 1.5777, "step": 58880 }, { "epoch": 0.08, "learning_rate": 4.6319375e-05, "loss": 1.7454, "step": 58890 }, { "epoch": 0.08, "learning_rate": 4.631875e-05, "loss": 1.8848, "step": 58900 }, { "epoch": 0.08, "learning_rate": 4.6318125e-05, "loss": 1.6321, "step": 58910 }, { "epoch": 0.08, "learning_rate": 4.6317500000000005e-05, "loss": 1.7482, "step": 58920 }, { "epoch": 0.08, "learning_rate": 4.6316875e-05, "loss": 1.4997, "step": 58930 }, { "epoch": 0.08, "learning_rate": 4.631625e-05, "loss": 1.4372, "step": 58940 }, { "epoch": 0.08, "learning_rate": 4.6315625e-05, "loss": 1.6087, "step": 58950 }, { "epoch": 0.08, "learning_rate": 4.6315e-05, "loss": 1.4892, "step": 58960 }, { "epoch": 0.08, "learning_rate": 4.6314375e-05, "loss": 1.5587, "step": 58970 }, { "epoch": 0.08, "learning_rate": 4.631375e-05, "loss": 1.8591, "step": 58980 }, { "epoch": 0.08, "learning_rate": 4.6313125000000004e-05, "loss": 1.6448, "step": 58990 }, { "epoch": 0.08, "learning_rate": 4.63125e-05, "loss": 1.6829, "step": 59000 }, { "epoch": 0.08, "eval_loss": 1.5832444429397583, "eval_runtime": 1294.7045, "eval_samples_per_second": 44.826, "eval_steps_per_second": 5.604, "step": 59000 }, { "epoch": 0.08, "learning_rate": 4.6311875000000005e-05, "loss": 1.4151, "step": 59010 }, { "epoch": 0.08, "learning_rate": 4.631125e-05, "loss": 1.5869, "step": 59020 }, { "epoch": 0.08, "learning_rate": 4.6310625000000005e-05, "loss": 1.4614, "step": 59030 }, { "epoch": 0.08, "learning_rate": 4.631e-05, "loss": 1.4122, "step": 59040 }, { "epoch": 0.08, "learning_rate": 4.6309375000000006e-05, "loss": 1.5928, "step": 59050 }, { "epoch": 0.08, "learning_rate": 4.630875e-05, "loss": 1.7006, "step": 59060 }, { "epoch": 0.08, "learning_rate": 4.6308125e-05, "loss": 1.6766, "step": 59070 }, { "epoch": 0.08, "learning_rate": 4.6307500000000003e-05, "loss": 1.618, "step": 59080 }, { "epoch": 0.08, "learning_rate": 4.6306875e-05, "loss": 1.7396, "step": 59090 }, { "epoch": 0.08, "learning_rate": 4.6306250000000004e-05, "loss": 1.543, "step": 59100 }, { "epoch": 0.08, "learning_rate": 4.6305625e-05, "loss": 1.6358, "step": 59110 }, { "epoch": 0.08, "learning_rate": 4.6305000000000005e-05, "loss": 1.6894, "step": 59120 }, { "epoch": 0.08, "learning_rate": 4.6304375e-05, "loss": 1.57, "step": 59130 }, { "epoch": 0.08, "learning_rate": 4.630375e-05, "loss": 1.7874, "step": 59140 }, { "epoch": 0.08, "learning_rate": 4.6303125e-05, "loss": 1.7682, "step": 59150 }, { "epoch": 0.08, "learning_rate": 4.6302500000000006e-05, "loss": 1.5962, "step": 59160 }, { "epoch": 0.08, "learning_rate": 4.6301875e-05, "loss": 1.6256, "step": 59170 }, { "epoch": 0.08, "learning_rate": 4.630125e-05, "loss": 1.6473, "step": 59180 }, { "epoch": 0.08, "learning_rate": 4.6300625000000004e-05, "loss": 1.7194, "step": 59190 }, { "epoch": 0.08, "learning_rate": 4.630000000000001e-05, "loss": 1.6263, "step": 59200 }, { "epoch": 0.08, "learning_rate": 4.6299375e-05, "loss": 1.8601, "step": 59210 }, { "epoch": 0.08, "learning_rate": 4.629875e-05, "loss": 1.6696, "step": 59220 }, { "epoch": 0.08, "learning_rate": 4.6298125000000005e-05, "loss": 1.7204, "step": 59230 }, { "epoch": 0.08, "learning_rate": 4.62975e-05, "loss": 1.5827, "step": 59240 }, { "epoch": 0.08, "learning_rate": 4.6296875e-05, "loss": 1.575, "step": 59250 }, { "epoch": 0.08, "learning_rate": 4.629625e-05, "loss": 1.8175, "step": 59260 }, { "epoch": 0.08, "learning_rate": 4.6295625000000006e-05, "loss": 1.7766, "step": 59270 }, { "epoch": 0.08, "learning_rate": 4.6294999999999996e-05, "loss": 1.4543, "step": 59280 }, { "epoch": 0.08, "learning_rate": 4.6294375e-05, "loss": 1.5077, "step": 59290 }, { "epoch": 0.08, "learning_rate": 4.6293750000000004e-05, "loss": 1.6871, "step": 59300 }, { "epoch": 0.08, "learning_rate": 4.6293125e-05, "loss": 1.792, "step": 59310 }, { "epoch": 0.08, "learning_rate": 4.62925e-05, "loss": 1.6179, "step": 59320 }, { "epoch": 0.08, "learning_rate": 4.6291875e-05, "loss": 1.7958, "step": 59330 }, { "epoch": 0.08, "learning_rate": 4.6291250000000005e-05, "loss": 1.6873, "step": 59340 }, { "epoch": 0.08, "learning_rate": 4.6290625e-05, "loss": 1.6884, "step": 59350 }, { "epoch": 0.08, "learning_rate": 4.629e-05, "loss": 1.5627, "step": 59360 }, { "epoch": 0.08, "learning_rate": 4.6289375e-05, "loss": 1.4832, "step": 59370 }, { "epoch": 0.08, "learning_rate": 4.628875e-05, "loss": 1.5604, "step": 59380 }, { "epoch": 0.08, "learning_rate": 4.6288125e-05, "loss": 1.6153, "step": 59390 }, { "epoch": 0.08, "learning_rate": 4.62875e-05, "loss": 1.608, "step": 59400 }, { "epoch": 0.08, "learning_rate": 4.6286875000000004e-05, "loss": 1.593, "step": 59410 }, { "epoch": 0.08, "learning_rate": 4.628625e-05, "loss": 1.4002, "step": 59420 }, { "epoch": 0.08, "learning_rate": 4.6285625000000005e-05, "loss": 1.6601, "step": 59430 }, { "epoch": 0.08, "learning_rate": 4.6285e-05, "loss": 1.5137, "step": 59440 }, { "epoch": 0.08, "learning_rate": 4.6284375000000005e-05, "loss": 1.7638, "step": 59450 }, { "epoch": 0.08, "learning_rate": 4.628375e-05, "loss": 1.6326, "step": 59460 }, { "epoch": 0.08, "learning_rate": 4.6283125000000006e-05, "loss": 1.6052, "step": 59470 }, { "epoch": 0.08, "learning_rate": 4.62825e-05, "loss": 1.7012, "step": 59480 }, { "epoch": 0.08, "learning_rate": 4.6281875e-05, "loss": 1.6156, "step": 59490 }, { "epoch": 0.08, "learning_rate": 4.6281250000000003e-05, "loss": 1.8815, "step": 59500 }, { "epoch": 0.08, "learning_rate": 4.6280625e-05, "loss": 1.5338, "step": 59510 }, { "epoch": 0.08, "learning_rate": 4.6280000000000004e-05, "loss": 1.495, "step": 59520 }, { "epoch": 0.08, "learning_rate": 4.6279375e-05, "loss": 1.6864, "step": 59530 }, { "epoch": 0.08, "learning_rate": 4.6278750000000005e-05, "loss": 1.6927, "step": 59540 }, { "epoch": 0.08, "learning_rate": 4.6278125e-05, "loss": 1.8281, "step": 59550 }, { "epoch": 0.08, "learning_rate": 4.62775e-05, "loss": 1.4548, "step": 59560 }, { "epoch": 0.08, "learning_rate": 4.6276875e-05, "loss": 1.5317, "step": 59570 }, { "epoch": 0.08, "learning_rate": 4.6276250000000006e-05, "loss": 1.3498, "step": 59580 }, { "epoch": 0.08, "learning_rate": 4.6275625e-05, "loss": 1.6638, "step": 59590 }, { "epoch": 0.08, "learning_rate": 4.6275e-05, "loss": 1.6593, "step": 59600 }, { "epoch": 0.08, "learning_rate": 4.6274375000000004e-05, "loss": 1.8363, "step": 59610 }, { "epoch": 0.08, "learning_rate": 4.627375000000001e-05, "loss": 1.3855, "step": 59620 }, { "epoch": 0.08, "learning_rate": 4.6273125e-05, "loss": 1.6407, "step": 59630 }, { "epoch": 0.08, "learning_rate": 4.62725e-05, "loss": 1.7622, "step": 59640 }, { "epoch": 0.08, "learning_rate": 4.6271875000000005e-05, "loss": 1.4451, "step": 59650 }, { "epoch": 0.08, "learning_rate": 4.627125e-05, "loss": 1.639, "step": 59660 }, { "epoch": 0.08, "learning_rate": 4.6270625e-05, "loss": 1.9764, "step": 59670 }, { "epoch": 0.08, "learning_rate": 4.627e-05, "loss": 1.7054, "step": 59680 }, { "epoch": 0.08, "learning_rate": 4.6269375000000006e-05, "loss": 1.5551, "step": 59690 }, { "epoch": 0.08, "learning_rate": 4.6268749999999996e-05, "loss": 1.6161, "step": 59700 }, { "epoch": 0.08, "learning_rate": 4.6268125e-05, "loss": 1.7916, "step": 59710 }, { "epoch": 0.08, "learning_rate": 4.6267500000000004e-05, "loss": 1.6373, "step": 59720 }, { "epoch": 0.08, "learning_rate": 4.6266875e-05, "loss": 1.4854, "step": 59730 }, { "epoch": 0.08, "learning_rate": 4.626625e-05, "loss": 1.8443, "step": 59740 }, { "epoch": 0.08, "learning_rate": 4.6265625e-05, "loss": 1.7494, "step": 59750 }, { "epoch": 0.08, "learning_rate": 4.6265000000000005e-05, "loss": 1.6094, "step": 59760 }, { "epoch": 0.08, "learning_rate": 4.6264375e-05, "loss": 1.456, "step": 59770 }, { "epoch": 0.08, "learning_rate": 4.626375e-05, "loss": 1.82, "step": 59780 }, { "epoch": 0.08, "learning_rate": 4.6263125e-05, "loss": 1.4992, "step": 59790 }, { "epoch": 0.08, "learning_rate": 4.6262500000000006e-05, "loss": 1.5196, "step": 59800 }, { "epoch": 0.08, "learning_rate": 4.6261875e-05, "loss": 1.6277, "step": 59810 }, { "epoch": 0.08, "learning_rate": 4.626125e-05, "loss": 1.4678, "step": 59820 }, { "epoch": 0.08, "learning_rate": 4.6260625000000004e-05, "loss": 1.6296, "step": 59830 }, { "epoch": 0.08, "learning_rate": 4.626e-05, "loss": 1.526, "step": 59840 }, { "epoch": 0.08, "learning_rate": 4.6259375000000005e-05, "loss": 1.8086, "step": 59850 }, { "epoch": 0.08, "learning_rate": 4.625875e-05, "loss": 1.508, "step": 59860 }, { "epoch": 0.08, "learning_rate": 4.6258125000000005e-05, "loss": 1.8053, "step": 59870 }, { "epoch": 0.08, "learning_rate": 4.62575e-05, "loss": 1.7744, "step": 59880 }, { "epoch": 0.08, "learning_rate": 4.6256875e-05, "loss": 1.686, "step": 59890 }, { "epoch": 0.08, "learning_rate": 4.625625e-05, "loss": 1.4996, "step": 59900 }, { "epoch": 0.08, "learning_rate": 4.6255625e-05, "loss": 1.8709, "step": 59910 }, { "epoch": 0.08, "learning_rate": 4.6255000000000004e-05, "loss": 1.6486, "step": 59920 }, { "epoch": 0.08, "learning_rate": 4.6254375e-05, "loss": 1.6523, "step": 59930 }, { "epoch": 0.08, "learning_rate": 4.6253750000000004e-05, "loss": 1.6109, "step": 59940 }, { "epoch": 0.08, "learning_rate": 4.6253125e-05, "loss": 1.7032, "step": 59950 }, { "epoch": 0.08, "learning_rate": 4.6252500000000005e-05, "loss": 1.7277, "step": 59960 }, { "epoch": 0.08, "learning_rate": 4.6251875e-05, "loss": 1.5751, "step": 59970 }, { "epoch": 0.08, "learning_rate": 4.625125e-05, "loss": 1.7408, "step": 59980 }, { "epoch": 0.08, "learning_rate": 4.6250625e-05, "loss": 1.6432, "step": 59990 }, { "epoch": 0.08, "learning_rate": 4.6250000000000006e-05, "loss": 1.9513, "step": 60000 }, { "epoch": 0.08, "eval_loss": 1.5791839361190796, "eval_runtime": 1296.4796, "eval_samples_per_second": 44.764, "eval_steps_per_second": 5.596, "step": 60000 }, { "epoch": 0.08, "learning_rate": 4.6249375e-05, "loss": 1.7064, "step": 60010 }, { "epoch": 0.08, "learning_rate": 4.624875e-05, "loss": 1.6901, "step": 60020 }, { "epoch": 0.08, "learning_rate": 4.6248125000000004e-05, "loss": 1.6833, "step": 60030 }, { "epoch": 0.08, "learning_rate": 4.624750000000001e-05, "loss": 1.6157, "step": 60040 }, { "epoch": 0.08, "learning_rate": 4.6246875e-05, "loss": 1.7337, "step": 60050 }, { "epoch": 0.08, "learning_rate": 4.624625e-05, "loss": 1.5198, "step": 60060 }, { "epoch": 0.08, "learning_rate": 4.6245625000000005e-05, "loss": 1.5685, "step": 60070 }, { "epoch": 0.08, "learning_rate": 4.6245e-05, "loss": 1.9099, "step": 60080 }, { "epoch": 0.08, "learning_rate": 4.6244375e-05, "loss": 1.7891, "step": 60090 }, { "epoch": 0.08, "learning_rate": 4.624375e-05, "loss": 1.426, "step": 60100 }, { "epoch": 0.08, "learning_rate": 4.6243125000000006e-05, "loss": 1.5863, "step": 60110 }, { "epoch": 0.08, "learning_rate": 4.6242499999999996e-05, "loss": 1.8013, "step": 60120 }, { "epoch": 0.08, "learning_rate": 4.6241875e-05, "loss": 1.5835, "step": 60130 }, { "epoch": 0.08, "learning_rate": 4.6241250000000004e-05, "loss": 1.7763, "step": 60140 }, { "epoch": 0.08, "learning_rate": 4.6240625e-05, "loss": 1.6276, "step": 60150 }, { "epoch": 0.08, "learning_rate": 4.624e-05, "loss": 1.5876, "step": 60160 }, { "epoch": 0.08, "learning_rate": 4.6239375e-05, "loss": 1.3543, "step": 60170 }, { "epoch": 0.08, "learning_rate": 4.6238750000000005e-05, "loss": 1.7986, "step": 60180 }, { "epoch": 0.08, "learning_rate": 4.6238125e-05, "loss": 1.6415, "step": 60190 }, { "epoch": 0.08, "learning_rate": 4.62375e-05, "loss": 1.9335, "step": 60200 }, { "epoch": 0.08, "learning_rate": 4.6236875e-05, "loss": 1.7689, "step": 60210 }, { "epoch": 0.08, "learning_rate": 4.6236250000000006e-05, "loss": 1.7437, "step": 60220 }, { "epoch": 0.08, "learning_rate": 4.6235625000000003e-05, "loss": 1.7554, "step": 60230 }, { "epoch": 0.08, "learning_rate": 4.6235e-05, "loss": 1.7162, "step": 60240 }, { "epoch": 0.08, "learning_rate": 4.6234375000000004e-05, "loss": 1.7796, "step": 60250 }, { "epoch": 0.08, "learning_rate": 4.623375e-05, "loss": 1.9238, "step": 60260 }, { "epoch": 0.08, "learning_rate": 4.6233125000000005e-05, "loss": 1.6391, "step": 60270 }, { "epoch": 0.08, "learning_rate": 4.62325e-05, "loss": 1.6331, "step": 60280 }, { "epoch": 0.08, "learning_rate": 4.6231875000000005e-05, "loss": 1.9749, "step": 60290 }, { "epoch": 0.08, "learning_rate": 4.623125e-05, "loss": 1.6407, "step": 60300 }, { "epoch": 0.08, "learning_rate": 4.6230625e-05, "loss": 1.7504, "step": 60310 }, { "epoch": 0.08, "learning_rate": 4.623e-05, "loss": 1.573, "step": 60320 }, { "epoch": 0.08, "learning_rate": 4.6229375e-05, "loss": 1.4738, "step": 60330 }, { "epoch": 0.08, "learning_rate": 4.6228750000000004e-05, "loss": 1.8452, "step": 60340 }, { "epoch": 0.08, "learning_rate": 4.6228125e-05, "loss": 1.5962, "step": 60350 }, { "epoch": 0.08, "learning_rate": 4.6227500000000004e-05, "loss": 1.4444, "step": 60360 }, { "epoch": 0.08, "learning_rate": 4.6226875e-05, "loss": 1.6842, "step": 60370 }, { "epoch": 0.08, "learning_rate": 4.6226250000000005e-05, "loss": 1.8168, "step": 60380 }, { "epoch": 0.08, "learning_rate": 4.6225625e-05, "loss": 1.463, "step": 60390 }, { "epoch": 0.08, "learning_rate": 4.6225e-05, "loss": 1.7293, "step": 60400 }, { "epoch": 0.08, "learning_rate": 4.6224375e-05, "loss": 1.36, "step": 60410 }, { "epoch": 0.08, "learning_rate": 4.6223750000000006e-05, "loss": 1.8073, "step": 60420 }, { "epoch": 0.08, "learning_rate": 4.6223125e-05, "loss": 1.4968, "step": 60430 }, { "epoch": 0.08, "learning_rate": 4.62225e-05, "loss": 1.5852, "step": 60440 }, { "epoch": 0.08, "learning_rate": 4.6221875000000004e-05, "loss": 1.6104, "step": 60450 }, { "epoch": 0.08, "learning_rate": 4.622125000000001e-05, "loss": 1.784, "step": 60460 }, { "epoch": 0.08, "learning_rate": 4.6220625e-05, "loss": 1.6885, "step": 60470 }, { "epoch": 0.08, "learning_rate": 4.622e-05, "loss": 1.5367, "step": 60480 }, { "epoch": 0.08, "learning_rate": 4.6219375000000005e-05, "loss": 1.7017, "step": 60490 }, { "epoch": 0.08, "learning_rate": 4.621875e-05, "loss": 1.4845, "step": 60500 }, { "epoch": 0.08, "learning_rate": 4.6218125e-05, "loss": 1.4614, "step": 60510 }, { "epoch": 0.08, "learning_rate": 4.62175e-05, "loss": 1.5301, "step": 60520 }, { "epoch": 0.08, "learning_rate": 4.6216875000000006e-05, "loss": 1.6848, "step": 60530 }, { "epoch": 0.08, "learning_rate": 4.6216249999999996e-05, "loss": 1.9307, "step": 60540 }, { "epoch": 0.08, "learning_rate": 4.6215625e-05, "loss": 1.5805, "step": 60550 }, { "epoch": 0.08, "learning_rate": 4.6215000000000004e-05, "loss": 1.7333, "step": 60560 }, { "epoch": 0.08, "learning_rate": 4.6214375e-05, "loss": 1.6611, "step": 60570 }, { "epoch": 0.08, "learning_rate": 4.621375e-05, "loss": 1.6588, "step": 60580 }, { "epoch": 0.08, "learning_rate": 4.6213125e-05, "loss": 1.8113, "step": 60590 }, { "epoch": 0.08, "learning_rate": 4.6212500000000005e-05, "loss": 1.5858, "step": 60600 }, { "epoch": 0.08, "learning_rate": 4.6211875e-05, "loss": 1.546, "step": 60610 }, { "epoch": 0.08, "learning_rate": 4.621125e-05, "loss": 1.7197, "step": 60620 }, { "epoch": 0.08, "learning_rate": 4.6210625e-05, "loss": 1.5547, "step": 60630 }, { "epoch": 0.08, "learning_rate": 4.6210000000000006e-05, "loss": 1.5258, "step": 60640 }, { "epoch": 0.08, "learning_rate": 4.6209375000000003e-05, "loss": 1.825, "step": 60650 }, { "epoch": 0.08, "learning_rate": 4.620875e-05, "loss": 1.6941, "step": 60660 }, { "epoch": 0.08, "learning_rate": 4.6208125000000004e-05, "loss": 1.8333, "step": 60670 }, { "epoch": 0.08, "learning_rate": 4.62075e-05, "loss": 1.8705, "step": 60680 }, { "epoch": 0.08, "learning_rate": 4.6206875e-05, "loss": 1.864, "step": 60690 }, { "epoch": 0.08, "learning_rate": 4.620625e-05, "loss": 1.8476, "step": 60700 }, { "epoch": 0.08, "learning_rate": 4.6205625000000005e-05, "loss": 1.6048, "step": 60710 }, { "epoch": 0.08, "learning_rate": 4.6205e-05, "loss": 1.8247, "step": 60720 }, { "epoch": 0.08, "learning_rate": 4.6204375e-05, "loss": 1.6009, "step": 60730 }, { "epoch": 0.08, "learning_rate": 4.620375e-05, "loss": 1.4397, "step": 60740 }, { "epoch": 0.08, "learning_rate": 4.6203125e-05, "loss": 1.559, "step": 60750 }, { "epoch": 0.08, "learning_rate": 4.6202500000000004e-05, "loss": 1.6363, "step": 60760 }, { "epoch": 0.08, "learning_rate": 4.6201875e-05, "loss": 1.7361, "step": 60770 }, { "epoch": 0.08, "learning_rate": 4.6201250000000004e-05, "loss": 1.3863, "step": 60780 }, { "epoch": 0.08, "learning_rate": 4.6200625e-05, "loss": 1.4503, "step": 60790 }, { "epoch": 0.08, "learning_rate": 4.6200000000000005e-05, "loss": 1.6176, "step": 60800 }, { "epoch": 0.08, "learning_rate": 4.6199375e-05, "loss": 1.7508, "step": 60810 }, { "epoch": 0.08, "learning_rate": 4.619875e-05, "loss": 1.691, "step": 60820 }, { "epoch": 0.08, "learning_rate": 4.6198125e-05, "loss": 1.5347, "step": 60830 }, { "epoch": 0.08, "learning_rate": 4.6197500000000006e-05, "loss": 1.6337, "step": 60840 }, { "epoch": 0.08, "learning_rate": 4.6196875e-05, "loss": 1.6305, "step": 60850 }, { "epoch": 0.08, "learning_rate": 4.619625e-05, "loss": 1.6793, "step": 60860 }, { "epoch": 0.08, "learning_rate": 4.6195625000000004e-05, "loss": 1.5712, "step": 60870 }, { "epoch": 0.08, "learning_rate": 4.619500000000001e-05, "loss": 1.787, "step": 60880 }, { "epoch": 0.08, "learning_rate": 4.6194375e-05, "loss": 1.5536, "step": 60890 }, { "epoch": 0.08, "learning_rate": 4.619375e-05, "loss": 1.5218, "step": 60900 }, { "epoch": 0.08, "learning_rate": 4.6193125000000005e-05, "loss": 1.7931, "step": 60910 }, { "epoch": 0.08, "learning_rate": 4.61925e-05, "loss": 1.5634, "step": 60920 }, { "epoch": 0.08, "learning_rate": 4.6191875e-05, "loss": 1.556, "step": 60930 }, { "epoch": 0.08, "learning_rate": 4.619125e-05, "loss": 1.6997, "step": 60940 }, { "epoch": 0.08, "learning_rate": 4.6190625000000006e-05, "loss": 1.6741, "step": 60950 }, { "epoch": 0.08, "learning_rate": 4.619e-05, "loss": 1.539, "step": 60960 }, { "epoch": 0.08, "learning_rate": 4.6189375e-05, "loss": 1.691, "step": 60970 }, { "epoch": 0.08, "learning_rate": 4.6188750000000004e-05, "loss": 1.7231, "step": 60980 }, { "epoch": 0.08, "learning_rate": 4.618812500000001e-05, "loss": 1.5306, "step": 60990 }, { "epoch": 0.08, "learning_rate": 4.61875e-05, "loss": 1.5642, "step": 61000 }, { "epoch": 0.08, "eval_loss": 1.564333438873291, "eval_runtime": 1301.0298, "eval_samples_per_second": 44.608, "eval_steps_per_second": 5.576, "step": 61000 }, { "epoch": 0.08, "learning_rate": 4.6186875e-05, "loss": 1.4256, "step": 61010 }, { "epoch": 0.08, "learning_rate": 4.6186250000000005e-05, "loss": 1.6084, "step": 61020 }, { "epoch": 0.08, "learning_rate": 4.6185625e-05, "loss": 1.5139, "step": 61030 }, { "epoch": 0.08, "learning_rate": 4.6185e-05, "loss": 1.463, "step": 61040 }, { "epoch": 0.09, "learning_rate": 4.6184375e-05, "loss": 1.7538, "step": 61050 }, { "epoch": 0.09, "learning_rate": 4.6183750000000007e-05, "loss": 1.7553, "step": 61060 }, { "epoch": 0.09, "learning_rate": 4.6183125000000003e-05, "loss": 1.6622, "step": 61070 }, { "epoch": 0.09, "learning_rate": 4.61825e-05, "loss": 1.5804, "step": 61080 }, { "epoch": 0.09, "learning_rate": 4.6181875000000004e-05, "loss": 1.6853, "step": 61090 }, { "epoch": 0.09, "learning_rate": 4.618125e-05, "loss": 1.6835, "step": 61100 }, { "epoch": 0.09, "learning_rate": 4.6180625e-05, "loss": 1.5292, "step": 61110 }, { "epoch": 0.09, "learning_rate": 4.618e-05, "loss": 1.4963, "step": 61120 }, { "epoch": 0.09, "learning_rate": 4.6179375000000005e-05, "loss": 1.6677, "step": 61130 }, { "epoch": 0.09, "learning_rate": 4.617875e-05, "loss": 1.9038, "step": 61140 }, { "epoch": 0.09, "learning_rate": 4.6178125e-05, "loss": 1.6011, "step": 61150 }, { "epoch": 0.09, "learning_rate": 4.61775e-05, "loss": 1.7244, "step": 61160 }, { "epoch": 0.09, "learning_rate": 4.6176875e-05, "loss": 1.6954, "step": 61170 }, { "epoch": 0.09, "learning_rate": 4.6176250000000004e-05, "loss": 1.6639, "step": 61180 }, { "epoch": 0.09, "learning_rate": 4.6175625e-05, "loss": 1.664, "step": 61190 }, { "epoch": 0.09, "learning_rate": 4.6175000000000004e-05, "loss": 1.8662, "step": 61200 }, { "epoch": 0.09, "learning_rate": 4.6174375e-05, "loss": 1.6901, "step": 61210 }, { "epoch": 0.09, "learning_rate": 4.6173750000000005e-05, "loss": 1.7625, "step": 61220 }, { "epoch": 0.09, "learning_rate": 4.6173125e-05, "loss": 1.5141, "step": 61230 }, { "epoch": 0.09, "learning_rate": 4.61725e-05, "loss": 1.6611, "step": 61240 }, { "epoch": 0.09, "learning_rate": 4.6171875e-05, "loss": 1.4603, "step": 61250 }, { "epoch": 0.09, "learning_rate": 4.6171250000000006e-05, "loss": 1.4824, "step": 61260 }, { "epoch": 0.09, "learning_rate": 4.6170625e-05, "loss": 1.7011, "step": 61270 }, { "epoch": 0.09, "learning_rate": 4.617e-05, "loss": 1.5796, "step": 61280 }, { "epoch": 0.09, "learning_rate": 4.6169375000000004e-05, "loss": 1.5739, "step": 61290 }, { "epoch": 0.09, "learning_rate": 4.616875e-05, "loss": 1.6345, "step": 61300 }, { "epoch": 0.09, "learning_rate": 4.6168125e-05, "loss": 1.593, "step": 61310 }, { "epoch": 0.09, "learning_rate": 4.61675e-05, "loss": 1.5123, "step": 61320 }, { "epoch": 0.09, "learning_rate": 4.6166875000000005e-05, "loss": 1.5769, "step": 61330 }, { "epoch": 0.09, "learning_rate": 4.616625e-05, "loss": 1.6503, "step": 61340 }, { "epoch": 0.09, "learning_rate": 4.6165625e-05, "loss": 1.4439, "step": 61350 }, { "epoch": 0.09, "learning_rate": 4.6165e-05, "loss": 1.8579, "step": 61360 }, { "epoch": 0.09, "learning_rate": 4.6164375000000006e-05, "loss": 1.3364, "step": 61370 }, { "epoch": 0.09, "learning_rate": 4.616375e-05, "loss": 1.6936, "step": 61380 }, { "epoch": 0.09, "learning_rate": 4.6163125e-05, "loss": 1.5126, "step": 61390 }, { "epoch": 0.09, "learning_rate": 4.6162500000000004e-05, "loss": 1.5833, "step": 61400 }, { "epoch": 0.09, "learning_rate": 4.616187500000001e-05, "loss": 1.4252, "step": 61410 }, { "epoch": 0.09, "learning_rate": 4.616125e-05, "loss": 1.5315, "step": 61420 }, { "epoch": 0.09, "learning_rate": 4.6160625e-05, "loss": 1.5806, "step": 61430 }, { "epoch": 0.09, "learning_rate": 4.6160000000000005e-05, "loss": 1.7139, "step": 61440 }, { "epoch": 0.09, "learning_rate": 4.6159375e-05, "loss": 1.598, "step": 61450 }, { "epoch": 0.09, "learning_rate": 4.615875e-05, "loss": 1.5185, "step": 61460 }, { "epoch": 0.09, "learning_rate": 4.6158125e-05, "loss": 1.6315, "step": 61470 }, { "epoch": 0.09, "learning_rate": 4.6157500000000007e-05, "loss": 1.6148, "step": 61480 }, { "epoch": 0.09, "learning_rate": 4.6156875000000004e-05, "loss": 1.6835, "step": 61490 }, { "epoch": 0.09, "learning_rate": 4.615625e-05, "loss": 1.5004, "step": 61500 }, { "epoch": 0.09, "learning_rate": 4.6155625000000004e-05, "loss": 1.8743, "step": 61510 }, { "epoch": 0.09, "learning_rate": 4.6155e-05, "loss": 1.6131, "step": 61520 }, { "epoch": 0.09, "learning_rate": 4.6154375e-05, "loss": 1.8497, "step": 61530 }, { "epoch": 0.09, "learning_rate": 4.615375e-05, "loss": 1.5664, "step": 61540 }, { "epoch": 0.09, "learning_rate": 4.6153125000000005e-05, "loss": 1.4919, "step": 61550 }, { "epoch": 0.09, "learning_rate": 4.61525e-05, "loss": 1.4118, "step": 61560 }, { "epoch": 0.09, "learning_rate": 4.6151875e-05, "loss": 1.6163, "step": 61570 }, { "epoch": 0.09, "learning_rate": 4.615125e-05, "loss": 1.5115, "step": 61580 }, { "epoch": 0.09, "learning_rate": 4.6150625e-05, "loss": 1.6089, "step": 61590 }, { "epoch": 0.09, "learning_rate": 4.6150000000000004e-05, "loss": 1.644, "step": 61600 }, { "epoch": 0.09, "learning_rate": 4.6149375e-05, "loss": 1.6628, "step": 61610 }, { "epoch": 0.09, "learning_rate": 4.6148750000000004e-05, "loss": 1.5965, "step": 61620 }, { "epoch": 0.09, "learning_rate": 4.6148125e-05, "loss": 1.7326, "step": 61630 }, { "epoch": 0.09, "learning_rate": 4.6147500000000005e-05, "loss": 1.5239, "step": 61640 }, { "epoch": 0.09, "learning_rate": 4.6146875e-05, "loss": 1.7797, "step": 61650 }, { "epoch": 0.09, "learning_rate": 4.614625e-05, "loss": 1.5013, "step": 61660 }, { "epoch": 0.09, "learning_rate": 4.6145625e-05, "loss": 1.6451, "step": 61670 }, { "epoch": 0.09, "learning_rate": 4.6145000000000006e-05, "loss": 1.4768, "step": 61680 }, { "epoch": 0.09, "learning_rate": 4.6144375e-05, "loss": 1.3357, "step": 61690 }, { "epoch": 0.09, "learning_rate": 4.614375e-05, "loss": 1.714, "step": 61700 }, { "epoch": 0.09, "learning_rate": 4.6143125000000004e-05, "loss": 1.6372, "step": 61710 }, { "epoch": 0.09, "learning_rate": 4.61425e-05, "loss": 1.4219, "step": 61720 }, { "epoch": 0.09, "learning_rate": 4.6141875e-05, "loss": 1.6662, "step": 61730 }, { "epoch": 0.09, "learning_rate": 4.614125e-05, "loss": 1.8215, "step": 61740 }, { "epoch": 0.09, "learning_rate": 4.6140625000000005e-05, "loss": 1.5357, "step": 61750 }, { "epoch": 0.09, "learning_rate": 4.614e-05, "loss": 1.6663, "step": 61760 }, { "epoch": 0.09, "learning_rate": 4.6139375e-05, "loss": 1.6441, "step": 61770 }, { "epoch": 0.09, "learning_rate": 4.613875e-05, "loss": 1.7621, "step": 61780 }, { "epoch": 0.09, "learning_rate": 4.6138125000000006e-05, "loss": 1.8232, "step": 61790 }, { "epoch": 0.09, "learning_rate": 4.61375e-05, "loss": 1.756, "step": 61800 }, { "epoch": 0.09, "learning_rate": 4.6136875e-05, "loss": 1.7183, "step": 61810 }, { "epoch": 0.09, "learning_rate": 4.6136250000000004e-05, "loss": 1.5697, "step": 61820 }, { "epoch": 0.09, "learning_rate": 4.613562500000001e-05, "loss": 1.7544, "step": 61830 }, { "epoch": 0.09, "learning_rate": 4.6135e-05, "loss": 1.5294, "step": 61840 }, { "epoch": 0.09, "learning_rate": 4.6134375e-05, "loss": 1.3724, "step": 61850 }, { "epoch": 0.09, "learning_rate": 4.6133750000000005e-05, "loss": 1.4902, "step": 61860 }, { "epoch": 0.09, "learning_rate": 4.6133125e-05, "loss": 1.5041, "step": 61870 }, { "epoch": 0.09, "learning_rate": 4.61325e-05, "loss": 1.7235, "step": 61880 }, { "epoch": 0.09, "learning_rate": 4.6131875e-05, "loss": 1.7758, "step": 61890 }, { "epoch": 0.09, "learning_rate": 4.613125000000001e-05, "loss": 1.5973, "step": 61900 }, { "epoch": 0.09, "learning_rate": 4.6130625e-05, "loss": 1.806, "step": 61910 }, { "epoch": 0.09, "learning_rate": 4.613e-05, "loss": 1.3633, "step": 61920 }, { "epoch": 0.09, "learning_rate": 4.6129375000000004e-05, "loss": 1.6776, "step": 61930 }, { "epoch": 0.09, "learning_rate": 4.612875e-05, "loss": 1.8659, "step": 61940 }, { "epoch": 0.09, "learning_rate": 4.6128125e-05, "loss": 1.7112, "step": 61950 }, { "epoch": 0.09, "learning_rate": 4.61275e-05, "loss": 1.4732, "step": 61960 }, { "epoch": 0.09, "learning_rate": 4.6126875000000005e-05, "loss": 1.5226, "step": 61970 }, { "epoch": 0.09, "learning_rate": 4.612625e-05, "loss": 1.4511, "step": 61980 }, { "epoch": 0.09, "learning_rate": 4.6125625e-05, "loss": 1.8338, "step": 61990 }, { "epoch": 0.09, "learning_rate": 4.6125e-05, "loss": 1.714, "step": 62000 }, { "epoch": 0.09, "eval_loss": 1.56070876121521, "eval_runtime": 1321.3433, "eval_samples_per_second": 43.922, "eval_steps_per_second": 5.491, "step": 62000 }, { "epoch": 0.09, "learning_rate": 4.6124375e-05, "loss": 1.549, "step": 62010 }, { "epoch": 0.09, "learning_rate": 4.6123750000000004e-05, "loss": 1.8428, "step": 62020 }, { "epoch": 0.09, "learning_rate": 4.6123125e-05, "loss": 1.7891, "step": 62030 }, { "epoch": 0.09, "learning_rate": 4.6122500000000004e-05, "loss": 1.6318, "step": 62040 }, { "epoch": 0.09, "learning_rate": 4.6121875e-05, "loss": 1.5175, "step": 62050 }, { "epoch": 0.09, "learning_rate": 4.6121250000000005e-05, "loss": 1.7782, "step": 62060 }, { "epoch": 0.09, "learning_rate": 4.6120625e-05, "loss": 1.4967, "step": 62070 }, { "epoch": 0.09, "learning_rate": 4.612e-05, "loss": 1.5982, "step": 62080 }, { "epoch": 0.09, "learning_rate": 4.6119375e-05, "loss": 1.601, "step": 62090 }, { "epoch": 0.09, "learning_rate": 4.611875e-05, "loss": 1.606, "step": 62100 }, { "epoch": 0.09, "learning_rate": 4.6118125e-05, "loss": 1.2303, "step": 62110 }, { "epoch": 0.09, "learning_rate": 4.61175e-05, "loss": 1.469, "step": 62120 }, { "epoch": 0.09, "learning_rate": 4.6116875000000004e-05, "loss": 1.749, "step": 62130 }, { "epoch": 0.09, "learning_rate": 4.611625e-05, "loss": 1.7061, "step": 62140 }, { "epoch": 0.09, "learning_rate": 4.6115625000000005e-05, "loss": 1.715, "step": 62150 }, { "epoch": 0.09, "learning_rate": 4.6115e-05, "loss": 1.668, "step": 62160 }, { "epoch": 0.09, "learning_rate": 4.6114375000000005e-05, "loss": 1.7941, "step": 62170 }, { "epoch": 0.09, "learning_rate": 4.611375e-05, "loss": 1.4981, "step": 62180 }, { "epoch": 0.09, "learning_rate": 4.6113125e-05, "loss": 1.5401, "step": 62190 }, { "epoch": 0.09, "learning_rate": 4.61125e-05, "loss": 1.4978, "step": 62200 }, { "epoch": 0.09, "learning_rate": 4.6111875000000006e-05, "loss": 1.5816, "step": 62210 }, { "epoch": 0.09, "learning_rate": 4.6111250000000003e-05, "loss": 1.7157, "step": 62220 }, { "epoch": 0.09, "learning_rate": 4.6110625e-05, "loss": 1.6297, "step": 62230 }, { "epoch": 0.09, "learning_rate": 4.6110000000000004e-05, "loss": 1.5231, "step": 62240 }, { "epoch": 0.09, "learning_rate": 4.610937500000001e-05, "loss": 1.4405, "step": 62250 }, { "epoch": 0.09, "learning_rate": 4.610875e-05, "loss": 1.7336, "step": 62260 }, { "epoch": 0.09, "learning_rate": 4.6108125e-05, "loss": 1.601, "step": 62270 }, { "epoch": 0.09, "learning_rate": 4.6107500000000005e-05, "loss": 1.4849, "step": 62280 }, { "epoch": 0.09, "learning_rate": 4.6106875e-05, "loss": 1.6944, "step": 62290 }, { "epoch": 0.09, "learning_rate": 4.610625e-05, "loss": 1.535, "step": 62300 }, { "epoch": 0.09, "learning_rate": 4.6105625e-05, "loss": 1.6011, "step": 62310 }, { "epoch": 0.09, "learning_rate": 4.610500000000001e-05, "loss": 1.5797, "step": 62320 }, { "epoch": 0.09, "learning_rate": 4.6104375e-05, "loss": 1.7939, "step": 62330 }, { "epoch": 0.09, "learning_rate": 4.610375e-05, "loss": 1.7782, "step": 62340 }, { "epoch": 0.09, "learning_rate": 4.6103125000000004e-05, "loss": 1.6367, "step": 62350 }, { "epoch": 0.09, "learning_rate": 4.61025e-05, "loss": 1.7211, "step": 62360 }, { "epoch": 0.09, "learning_rate": 4.6101875e-05, "loss": 1.6301, "step": 62370 }, { "epoch": 0.09, "learning_rate": 4.610125e-05, "loss": 1.704, "step": 62380 }, { "epoch": 0.09, "learning_rate": 4.6100625000000006e-05, "loss": 1.4567, "step": 62390 }, { "epoch": 0.09, "learning_rate": 4.61e-05, "loss": 1.6498, "step": 62400 }, { "epoch": 0.09, "learning_rate": 4.6099375e-05, "loss": 1.5007, "step": 62410 }, { "epoch": 0.09, "learning_rate": 4.609875e-05, "loss": 1.7007, "step": 62420 }, { "epoch": 0.09, "learning_rate": 4.6098125e-05, "loss": 1.7843, "step": 62430 }, { "epoch": 0.09, "learning_rate": 4.6097500000000004e-05, "loss": 1.6483, "step": 62440 }, { "epoch": 0.09, "learning_rate": 4.6096875e-05, "loss": 1.7071, "step": 62450 }, { "epoch": 0.09, "learning_rate": 4.6096250000000004e-05, "loss": 1.4748, "step": 62460 }, { "epoch": 0.09, "learning_rate": 4.6095625e-05, "loss": 1.7109, "step": 62470 }, { "epoch": 0.09, "learning_rate": 4.6095000000000005e-05, "loss": 1.6187, "step": 62480 }, { "epoch": 0.09, "learning_rate": 4.6094375e-05, "loss": 1.6866, "step": 62490 }, { "epoch": 0.09, "learning_rate": 4.609375e-05, "loss": 1.659, "step": 62500 }, { "epoch": 0.09, "learning_rate": 4.6093125e-05, "loss": 1.6315, "step": 62510 }, { "epoch": 0.09, "learning_rate": 4.60925e-05, "loss": 1.7413, "step": 62520 }, { "epoch": 0.09, "learning_rate": 4.6091875e-05, "loss": 1.5781, "step": 62530 }, { "epoch": 0.09, "learning_rate": 4.609125e-05, "loss": 1.6466, "step": 62540 }, { "epoch": 0.09, "learning_rate": 4.6090625000000004e-05, "loss": 1.6167, "step": 62550 }, { "epoch": 0.09, "learning_rate": 4.609e-05, "loss": 1.7752, "step": 62560 }, { "epoch": 0.09, "learning_rate": 4.6089375000000005e-05, "loss": 1.6131, "step": 62570 }, { "epoch": 0.09, "learning_rate": 4.608875e-05, "loss": 1.6339, "step": 62580 }, { "epoch": 0.09, "learning_rate": 4.6088125000000005e-05, "loss": 1.5231, "step": 62590 }, { "epoch": 0.09, "learning_rate": 4.60875e-05, "loss": 1.5771, "step": 62600 }, { "epoch": 0.09, "learning_rate": 4.6086875e-05, "loss": 1.6691, "step": 62610 }, { "epoch": 0.09, "learning_rate": 4.608625e-05, "loss": 1.8729, "step": 62620 }, { "epoch": 0.09, "learning_rate": 4.6085625000000007e-05, "loss": 1.6125, "step": 62630 }, { "epoch": 0.09, "learning_rate": 4.6085000000000003e-05, "loss": 1.6129, "step": 62640 }, { "epoch": 0.09, "learning_rate": 4.6084375e-05, "loss": 1.4669, "step": 62650 }, { "epoch": 0.09, "learning_rate": 4.6083750000000004e-05, "loss": 1.626, "step": 62660 }, { "epoch": 0.09, "learning_rate": 4.608312500000001e-05, "loss": 1.502, "step": 62670 }, { "epoch": 0.09, "learning_rate": 4.60825e-05, "loss": 1.5985, "step": 62680 }, { "epoch": 0.09, "learning_rate": 4.6081875e-05, "loss": 1.4688, "step": 62690 }, { "epoch": 0.09, "learning_rate": 4.6081250000000005e-05, "loss": 1.5525, "step": 62700 }, { "epoch": 0.09, "learning_rate": 4.6080625e-05, "loss": 1.7319, "step": 62710 }, { "epoch": 0.09, "learning_rate": 4.608e-05, "loss": 1.6243, "step": 62720 }, { "epoch": 0.09, "learning_rate": 4.6079375e-05, "loss": 1.6956, "step": 62730 }, { "epoch": 0.09, "learning_rate": 4.607875000000001e-05, "loss": 1.775, "step": 62740 }, { "epoch": 0.09, "learning_rate": 4.6078125e-05, "loss": 1.8437, "step": 62750 }, { "epoch": 0.09, "learning_rate": 4.60775e-05, "loss": 1.4883, "step": 62760 }, { "epoch": 0.09, "learning_rate": 4.6076875000000004e-05, "loss": 1.7679, "step": 62770 }, { "epoch": 0.09, "learning_rate": 4.607625e-05, "loss": 1.4974, "step": 62780 }, { "epoch": 0.09, "learning_rate": 4.6075625e-05, "loss": 1.5163, "step": 62790 }, { "epoch": 0.09, "learning_rate": 4.6075e-05, "loss": 1.6648, "step": 62800 }, { "epoch": 0.09, "learning_rate": 4.6074375000000006e-05, "loss": 1.7579, "step": 62810 }, { "epoch": 0.09, "learning_rate": 4.607375e-05, "loss": 1.5123, "step": 62820 }, { "epoch": 0.09, "learning_rate": 4.6073125e-05, "loss": 1.6614, "step": 62830 }, { "epoch": 0.09, "learning_rate": 4.60725e-05, "loss": 1.6125, "step": 62840 }, { "epoch": 0.09, "learning_rate": 4.6071875e-05, "loss": 1.5894, "step": 62850 }, { "epoch": 0.09, "learning_rate": 4.6071250000000004e-05, "loss": 1.8093, "step": 62860 }, { "epoch": 0.09, "learning_rate": 4.6070625e-05, "loss": 1.6105, "step": 62870 }, { "epoch": 0.09, "learning_rate": 4.6070000000000004e-05, "loss": 1.7657, "step": 62880 }, { "epoch": 0.09, "learning_rate": 4.6069375e-05, "loss": 1.3357, "step": 62890 }, { "epoch": 0.09, "learning_rate": 4.6068750000000005e-05, "loss": 1.4398, "step": 62900 }, { "epoch": 0.09, "learning_rate": 4.6068125e-05, "loss": 1.5411, "step": 62910 }, { "epoch": 0.09, "learning_rate": 4.60675e-05, "loss": 1.5998, "step": 62920 }, { "epoch": 0.09, "learning_rate": 4.6066875e-05, "loss": 1.7979, "step": 62930 }, { "epoch": 0.09, "learning_rate": 4.606625e-05, "loss": 1.5682, "step": 62940 }, { "epoch": 0.09, "learning_rate": 4.6065625e-05, "loss": 1.6442, "step": 62950 }, { "epoch": 0.09, "learning_rate": 4.6065e-05, "loss": 1.558, "step": 62960 }, { "epoch": 0.09, "learning_rate": 4.6064375000000004e-05, "loss": 1.5853, "step": 62970 }, { "epoch": 0.09, "learning_rate": 4.606375e-05, "loss": 1.7556, "step": 62980 }, { "epoch": 0.09, "learning_rate": 4.6063125000000005e-05, "loss": 1.6842, "step": 62990 }, { "epoch": 0.09, "learning_rate": 4.60625e-05, "loss": 1.7619, "step": 63000 }, { "epoch": 0.09, "eval_loss": 1.5541919469833374, "eval_runtime": 1309.5512, "eval_samples_per_second": 44.317, "eval_steps_per_second": 5.54, "step": 63000 }, { "epoch": 0.09, "learning_rate": 4.6061875000000005e-05, "loss": 1.6013, "step": 63010 }, { "epoch": 0.09, "learning_rate": 4.606125e-05, "loss": 1.6575, "step": 63020 }, { "epoch": 0.09, "learning_rate": 4.6060625e-05, "loss": 1.7033, "step": 63030 }, { "epoch": 0.09, "learning_rate": 4.606e-05, "loss": 1.7672, "step": 63040 }, { "epoch": 0.09, "learning_rate": 4.6059375000000007e-05, "loss": 1.3948, "step": 63050 }, { "epoch": 0.09, "learning_rate": 4.6058750000000003e-05, "loss": 1.5589, "step": 63060 }, { "epoch": 0.09, "learning_rate": 4.6058125e-05, "loss": 1.5779, "step": 63070 }, { "epoch": 0.09, "learning_rate": 4.6057500000000004e-05, "loss": 1.6497, "step": 63080 }, { "epoch": 0.09, "learning_rate": 4.605687500000001e-05, "loss": 1.5733, "step": 63090 }, { "epoch": 0.09, "learning_rate": 4.605625e-05, "loss": 1.5852, "step": 63100 }, { "epoch": 0.09, "learning_rate": 4.6055625e-05, "loss": 1.7632, "step": 63110 }, { "epoch": 0.09, "learning_rate": 4.6055000000000005e-05, "loss": 1.6494, "step": 63120 }, { "epoch": 0.09, "learning_rate": 4.6054375e-05, "loss": 1.4412, "step": 63130 }, { "epoch": 0.09, "learning_rate": 4.605375e-05, "loss": 1.6202, "step": 63140 }, { "epoch": 0.09, "learning_rate": 4.6053125e-05, "loss": 1.5076, "step": 63150 }, { "epoch": 0.09, "learning_rate": 4.605250000000001e-05, "loss": 1.4412, "step": 63160 }, { "epoch": 0.09, "learning_rate": 4.6051875e-05, "loss": 1.5077, "step": 63170 }, { "epoch": 0.09, "learning_rate": 4.605125e-05, "loss": 1.5584, "step": 63180 }, { "epoch": 0.09, "learning_rate": 4.6050625000000004e-05, "loss": 1.6218, "step": 63190 }, { "epoch": 0.09, "learning_rate": 4.605e-05, "loss": 1.4626, "step": 63200 }, { "epoch": 0.09, "learning_rate": 4.6049375e-05, "loss": 1.6109, "step": 63210 }, { "epoch": 0.09, "learning_rate": 4.604875e-05, "loss": 1.6805, "step": 63220 }, { "epoch": 0.09, "learning_rate": 4.6048125000000006e-05, "loss": 1.7736, "step": 63230 }, { "epoch": 0.09, "learning_rate": 4.60475e-05, "loss": 1.6952, "step": 63240 }, { "epoch": 0.09, "learning_rate": 4.6046875e-05, "loss": 1.6329, "step": 63250 }, { "epoch": 0.09, "learning_rate": 4.604625e-05, "loss": 1.4135, "step": 63260 }, { "epoch": 0.09, "learning_rate": 4.6045625e-05, "loss": 1.5816, "step": 63270 }, { "epoch": 0.09, "learning_rate": 4.6045000000000004e-05, "loss": 1.6425, "step": 63280 }, { "epoch": 0.09, "learning_rate": 4.6044375e-05, "loss": 1.7586, "step": 63290 }, { "epoch": 0.09, "learning_rate": 4.6043750000000004e-05, "loss": 1.5466, "step": 63300 }, { "epoch": 0.09, "learning_rate": 4.6043125e-05, "loss": 1.6705, "step": 63310 }, { "epoch": 0.09, "learning_rate": 4.60425e-05, "loss": 1.7017, "step": 63320 }, { "epoch": 0.09, "learning_rate": 4.6041875e-05, "loss": 1.8158, "step": 63330 }, { "epoch": 0.09, "learning_rate": 4.6041250000000006e-05, "loss": 1.4333, "step": 63340 }, { "epoch": 0.09, "learning_rate": 4.6040625e-05, "loss": 1.7426, "step": 63350 }, { "epoch": 0.09, "learning_rate": 4.604e-05, "loss": 1.6263, "step": 63360 }, { "epoch": 0.09, "learning_rate": 4.6039375e-05, "loss": 1.633, "step": 63370 }, { "epoch": 0.09, "learning_rate": 4.603875e-05, "loss": 1.6941, "step": 63380 }, { "epoch": 0.09, "learning_rate": 4.6038125000000004e-05, "loss": 1.574, "step": 63390 }, { "epoch": 0.09, "learning_rate": 4.60375e-05, "loss": 1.6801, "step": 63400 }, { "epoch": 0.09, "learning_rate": 4.6036875000000005e-05, "loss": 1.7135, "step": 63410 }, { "epoch": 0.09, "learning_rate": 4.603625e-05, "loss": 1.6359, "step": 63420 }, { "epoch": 0.09, "learning_rate": 4.6035625000000005e-05, "loss": 1.6863, "step": 63430 }, { "epoch": 0.09, "learning_rate": 4.6035e-05, "loss": 1.7414, "step": 63440 }, { "epoch": 0.09, "learning_rate": 4.6034375e-05, "loss": 1.632, "step": 63450 }, { "epoch": 0.09, "learning_rate": 4.603375e-05, "loss": 1.478, "step": 63460 }, { "epoch": 0.09, "learning_rate": 4.6033125000000007e-05, "loss": 1.6396, "step": 63470 }, { "epoch": 0.09, "learning_rate": 4.6032500000000004e-05, "loss": 1.6233, "step": 63480 }, { "epoch": 0.09, "learning_rate": 4.6031875e-05, "loss": 1.629, "step": 63490 }, { "epoch": 0.09, "learning_rate": 4.6031250000000004e-05, "loss": 1.4401, "step": 63500 }, { "epoch": 0.09, "learning_rate": 4.6030625e-05, "loss": 1.7875, "step": 63510 }, { "epoch": 0.09, "learning_rate": 4.603e-05, "loss": 1.6541, "step": 63520 }, { "epoch": 0.09, "learning_rate": 4.6029375e-05, "loss": 1.5605, "step": 63530 }, { "epoch": 0.09, "learning_rate": 4.6028750000000005e-05, "loss": 1.4193, "step": 63540 }, { "epoch": 0.09, "learning_rate": 4.6028125e-05, "loss": 1.6346, "step": 63550 }, { "epoch": 0.09, "learning_rate": 4.60275e-05, "loss": 1.666, "step": 63560 }, { "epoch": 0.09, "learning_rate": 4.6026875e-05, "loss": 1.4356, "step": 63570 }, { "epoch": 0.09, "learning_rate": 4.602625000000001e-05, "loss": 1.5007, "step": 63580 }, { "epoch": 0.09, "learning_rate": 4.6025625e-05, "loss": 1.5796, "step": 63590 }, { "epoch": 0.09, "learning_rate": 4.6025e-05, "loss": 1.5035, "step": 63600 }, { "epoch": 0.09, "learning_rate": 4.6024375000000004e-05, "loss": 1.7564, "step": 63610 }, { "epoch": 0.09, "learning_rate": 4.602375e-05, "loss": 1.7695, "step": 63620 }, { "epoch": 0.09, "learning_rate": 4.6023125e-05, "loss": 1.558, "step": 63630 }, { "epoch": 0.09, "learning_rate": 4.60225e-05, "loss": 1.606, "step": 63640 }, { "epoch": 0.09, "learning_rate": 4.6021875000000006e-05, "loss": 1.5224, "step": 63650 }, { "epoch": 0.09, "learning_rate": 4.602125e-05, "loss": 1.4193, "step": 63660 }, { "epoch": 0.09, "learning_rate": 4.6020625e-05, "loss": 1.5654, "step": 63670 }, { "epoch": 0.09, "learning_rate": 4.602e-05, "loss": 1.5887, "step": 63680 }, { "epoch": 0.09, "learning_rate": 4.6019375e-05, "loss": 1.6737, "step": 63690 }, { "epoch": 0.09, "learning_rate": 4.6018750000000004e-05, "loss": 1.5733, "step": 63700 }, { "epoch": 0.09, "learning_rate": 4.6018125e-05, "loss": 1.6722, "step": 63710 }, { "epoch": 0.09, "learning_rate": 4.6017500000000005e-05, "loss": 1.3102, "step": 63720 }, { "epoch": 0.09, "learning_rate": 4.6016875e-05, "loss": 1.6822, "step": 63730 }, { "epoch": 0.09, "learning_rate": 4.601625e-05, "loss": 1.5804, "step": 63740 }, { "epoch": 0.09, "learning_rate": 4.6015625e-05, "loss": 1.5004, "step": 63750 }, { "epoch": 0.09, "learning_rate": 4.6015000000000006e-05, "loss": 1.5536, "step": 63760 }, { "epoch": 0.09, "learning_rate": 4.6014375e-05, "loss": 1.6934, "step": 63770 }, { "epoch": 0.09, "learning_rate": 4.601375e-05, "loss": 1.7031, "step": 63780 }, { "epoch": 0.09, "learning_rate": 4.6013125000000003e-05, "loss": 1.6168, "step": 63790 }, { "epoch": 0.09, "learning_rate": 4.60125e-05, "loss": 1.8468, "step": 63800 }, { "epoch": 0.09, "learning_rate": 4.6011875000000004e-05, "loss": 1.6281, "step": 63810 }, { "epoch": 0.09, "learning_rate": 4.601125e-05, "loss": 1.7498, "step": 63820 }, { "epoch": 0.09, "learning_rate": 4.6010625000000005e-05, "loss": 1.4607, "step": 63830 }, { "epoch": 0.09, "learning_rate": 4.601e-05, "loss": 1.9863, "step": 63840 }, { "epoch": 0.09, "learning_rate": 4.6009375000000005e-05, "loss": 1.4951, "step": 63850 }, { "epoch": 0.09, "learning_rate": 4.600875e-05, "loss": 1.6974, "step": 63860 }, { "epoch": 0.09, "learning_rate": 4.6008125e-05, "loss": 1.5565, "step": 63870 }, { "epoch": 0.09, "learning_rate": 4.60075e-05, "loss": 1.6234, "step": 63880 }, { "epoch": 0.09, "learning_rate": 4.600687500000001e-05, "loss": 1.6345, "step": 63890 }, { "epoch": 0.09, "learning_rate": 4.6006250000000004e-05, "loss": 1.7684, "step": 63900 }, { "epoch": 0.09, "learning_rate": 4.6005625e-05, "loss": 1.7246, "step": 63910 }, { "epoch": 0.09, "learning_rate": 4.6005000000000004e-05, "loss": 1.5282, "step": 63920 }, { "epoch": 0.09, "learning_rate": 4.6004375e-05, "loss": 1.7097, "step": 63930 }, { "epoch": 0.09, "learning_rate": 4.600375e-05, "loss": 1.4545, "step": 63940 }, { "epoch": 0.09, "learning_rate": 4.6003125e-05, "loss": 1.5229, "step": 63950 }, { "epoch": 0.09, "learning_rate": 4.6002500000000006e-05, "loss": 1.8043, "step": 63960 }, { "epoch": 0.09, "learning_rate": 4.6001875e-05, "loss": 1.7249, "step": 63970 }, { "epoch": 0.09, "learning_rate": 4.600125e-05, "loss": 1.4603, "step": 63980 }, { "epoch": 0.09, "learning_rate": 4.6000625e-05, "loss": 1.6912, "step": 63990 }, { "epoch": 0.09, "learning_rate": 4.600000000000001e-05, "loss": 1.7811, "step": 64000 }, { "epoch": 0.09, "eval_loss": 1.555694580078125, "eval_runtime": 2385.8183, "eval_samples_per_second": 24.325, "eval_steps_per_second": 3.041, "step": 64000 }, { "epoch": 0.09, "learning_rate": 4.5999375e-05, "loss": 1.5855, "step": 64010 }, { "epoch": 0.09, "learning_rate": 4.599875e-05, "loss": 1.6917, "step": 64020 }, { "epoch": 0.09, "learning_rate": 4.5998125000000004e-05, "loss": 1.7417, "step": 64030 }, { "epoch": 0.09, "learning_rate": 4.59975e-05, "loss": 1.6192, "step": 64040 }, { "epoch": 0.09, "learning_rate": 4.5996875e-05, "loss": 1.6113, "step": 64050 }, { "epoch": 0.09, "learning_rate": 4.599625e-05, "loss": 1.5686, "step": 64060 }, { "epoch": 0.09, "learning_rate": 4.5995625000000006e-05, "loss": 1.6643, "step": 64070 }, { "epoch": 0.09, "learning_rate": 4.5995e-05, "loss": 1.5667, "step": 64080 }, { "epoch": 0.09, "learning_rate": 4.5994375e-05, "loss": 1.5188, "step": 64090 }, { "epoch": 0.09, "learning_rate": 4.599375e-05, "loss": 1.3652, "step": 64100 }, { "epoch": 0.09, "learning_rate": 4.5993125e-05, "loss": 1.5247, "step": 64110 }, { "epoch": 0.09, "learning_rate": 4.59925e-05, "loss": 1.7139, "step": 64120 }, { "epoch": 0.09, "learning_rate": 4.5991875e-05, "loss": 1.8769, "step": 64130 }, { "epoch": 0.09, "learning_rate": 4.5991250000000005e-05, "loss": 1.6544, "step": 64140 }, { "epoch": 0.09, "learning_rate": 4.5990625e-05, "loss": 1.5671, "step": 64150 }, { "epoch": 0.09, "learning_rate": 4.599e-05, "loss": 1.5266, "step": 64160 }, { "epoch": 0.09, "learning_rate": 4.5989375e-05, "loss": 1.4434, "step": 64170 }, { "epoch": 0.09, "learning_rate": 4.5988750000000006e-05, "loss": 1.7805, "step": 64180 }, { "epoch": 0.09, "learning_rate": 4.5988125e-05, "loss": 1.7467, "step": 64190 }, { "epoch": 0.09, "learning_rate": 4.59875e-05, "loss": 1.5896, "step": 64200 }, { "epoch": 0.09, "learning_rate": 4.5986875000000003e-05, "loss": 1.6126, "step": 64210 }, { "epoch": 0.09, "learning_rate": 4.598625e-05, "loss": 1.666, "step": 64220 }, { "epoch": 0.09, "learning_rate": 4.5985625000000004e-05, "loss": 1.526, "step": 64230 }, { "epoch": 0.09, "learning_rate": 4.5985e-05, "loss": 1.6314, "step": 64240 }, { "epoch": 0.09, "learning_rate": 4.5984375000000005e-05, "loss": 1.568, "step": 64250 }, { "epoch": 0.09, "learning_rate": 4.598375e-05, "loss": 1.6439, "step": 64260 }, { "epoch": 0.09, "learning_rate": 4.5983125000000005e-05, "loss": 1.8143, "step": 64270 }, { "epoch": 0.09, "learning_rate": 4.59825e-05, "loss": 1.6361, "step": 64280 }, { "epoch": 0.09, "learning_rate": 4.5981875e-05, "loss": 1.6161, "step": 64290 }, { "epoch": 0.09, "learning_rate": 4.598125e-05, "loss": 1.6539, "step": 64300 }, { "epoch": 0.09, "learning_rate": 4.598062500000001e-05, "loss": 1.4042, "step": 64310 }, { "epoch": 0.09, "learning_rate": 4.5980000000000004e-05, "loss": 1.41, "step": 64320 }, { "epoch": 0.09, "learning_rate": 4.5979375e-05, "loss": 1.675, "step": 64330 }, { "epoch": 0.09, "learning_rate": 4.5978750000000004e-05, "loss": 1.577, "step": 64340 }, { "epoch": 0.09, "learning_rate": 4.5978125e-05, "loss": 1.4868, "step": 64350 }, { "epoch": 0.09, "learning_rate": 4.59775e-05, "loss": 1.595, "step": 64360 }, { "epoch": 0.09, "learning_rate": 4.5976875e-05, "loss": 1.5024, "step": 64370 }, { "epoch": 0.09, "learning_rate": 4.5976250000000006e-05, "loss": 1.5956, "step": 64380 }, { "epoch": 0.09, "learning_rate": 4.5975625e-05, "loss": 1.6708, "step": 64390 }, { "epoch": 0.09, "learning_rate": 4.5975e-05, "loss": 1.4401, "step": 64400 }, { "epoch": 0.09, "learning_rate": 4.5974375e-05, "loss": 1.7233, "step": 64410 }, { "epoch": 0.09, "learning_rate": 4.597375000000001e-05, "loss": 1.4866, "step": 64420 }, { "epoch": 0.09, "learning_rate": 4.5973125e-05, "loss": 1.4947, "step": 64430 }, { "epoch": 0.09, "learning_rate": 4.59725e-05, "loss": 1.6656, "step": 64440 }, { "epoch": 0.09, "learning_rate": 4.5971875000000004e-05, "loss": 1.5891, "step": 64450 }, { "epoch": 0.09, "learning_rate": 4.597125e-05, "loss": 1.5128, "step": 64460 }, { "epoch": 0.09, "learning_rate": 4.5970625e-05, "loss": 1.5688, "step": 64470 }, { "epoch": 0.09, "learning_rate": 4.597e-05, "loss": 1.4733, "step": 64480 }, { "epoch": 0.09, "learning_rate": 4.5969375000000006e-05, "loss": 1.7408, "step": 64490 }, { "epoch": 0.09, "learning_rate": 4.596875e-05, "loss": 1.4972, "step": 64500 }, { "epoch": 0.09, "learning_rate": 4.5968125e-05, "loss": 1.4271, "step": 64510 }, { "epoch": 0.09, "learning_rate": 4.59675e-05, "loss": 1.6599, "step": 64520 }, { "epoch": 0.09, "learning_rate": 4.596687500000001e-05, "loss": 1.6649, "step": 64530 }, { "epoch": 0.09, "learning_rate": 4.596625e-05, "loss": 1.4676, "step": 64540 }, { "epoch": 0.09, "learning_rate": 4.5965625e-05, "loss": 1.4825, "step": 64550 }, { "epoch": 0.09, "learning_rate": 4.5965000000000005e-05, "loss": 1.7159, "step": 64560 }, { "epoch": 0.09, "learning_rate": 4.5964375e-05, "loss": 1.603, "step": 64570 }, { "epoch": 0.09, "learning_rate": 4.596375e-05, "loss": 1.6238, "step": 64580 }, { "epoch": 0.09, "learning_rate": 4.5963125e-05, "loss": 1.6302, "step": 64590 }, { "epoch": 0.09, "learning_rate": 4.5962500000000006e-05, "loss": 1.5428, "step": 64600 }, { "epoch": 0.09, "learning_rate": 4.5961875e-05, "loss": 1.666, "step": 64610 }, { "epoch": 0.09, "learning_rate": 4.596125e-05, "loss": 1.615, "step": 64620 }, { "epoch": 0.09, "learning_rate": 4.5960625000000003e-05, "loss": 1.6386, "step": 64630 }, { "epoch": 0.09, "learning_rate": 4.596e-05, "loss": 1.6641, "step": 64640 }, { "epoch": 0.09, "learning_rate": 4.5959375000000004e-05, "loss": 1.6717, "step": 64650 }, { "epoch": 0.09, "learning_rate": 4.595875e-05, "loss": 1.5933, "step": 64660 }, { "epoch": 0.09, "learning_rate": 4.5958125000000005e-05, "loss": 1.6658, "step": 64670 }, { "epoch": 0.09, "learning_rate": 4.59575e-05, "loss": 1.6501, "step": 64680 }, { "epoch": 0.09, "learning_rate": 4.5956875000000005e-05, "loss": 1.5604, "step": 64690 }, { "epoch": 0.09, "learning_rate": 4.595625e-05, "loss": 1.6084, "step": 64700 }, { "epoch": 0.09, "learning_rate": 4.5955625e-05, "loss": 1.6115, "step": 64710 }, { "epoch": 0.09, "learning_rate": 4.5955e-05, "loss": 1.6661, "step": 64720 }, { "epoch": 0.09, "learning_rate": 4.5954375e-05, "loss": 1.7826, "step": 64730 }, { "epoch": 0.09, "learning_rate": 4.5953750000000004e-05, "loss": 1.9099, "step": 64740 }, { "epoch": 0.09, "learning_rate": 4.5953125e-05, "loss": 1.9083, "step": 64750 }, { "epoch": 0.09, "learning_rate": 4.5952500000000004e-05, "loss": 1.6332, "step": 64760 }, { "epoch": 0.09, "learning_rate": 4.5951875e-05, "loss": 1.4635, "step": 64770 }, { "epoch": 0.09, "learning_rate": 4.595125e-05, "loss": 1.5143, "step": 64780 }, { "epoch": 0.09, "learning_rate": 4.5950625e-05, "loss": 1.437, "step": 64790 }, { "epoch": 0.09, "learning_rate": 4.5950000000000006e-05, "loss": 1.5293, "step": 64800 }, { "epoch": 0.09, "learning_rate": 4.5949375e-05, "loss": 1.6802, "step": 64810 }, { "epoch": 0.09, "learning_rate": 4.594875e-05, "loss": 1.714, "step": 64820 }, { "epoch": 0.09, "learning_rate": 4.5948125e-05, "loss": 1.5803, "step": 64830 }, { "epoch": 0.09, "learning_rate": 4.594750000000001e-05, "loss": 1.6376, "step": 64840 }, { "epoch": 0.09, "learning_rate": 4.5946875e-05, "loss": 1.587, "step": 64850 }, { "epoch": 0.09, "learning_rate": 4.594625e-05, "loss": 1.6924, "step": 64860 }, { "epoch": 0.09, "learning_rate": 4.5945625000000004e-05, "loss": 1.6374, "step": 64870 }, { "epoch": 0.09, "learning_rate": 4.5945e-05, "loss": 1.6222, "step": 64880 }, { "epoch": 0.09, "learning_rate": 4.5944375e-05, "loss": 1.6102, "step": 64890 }, { "epoch": 0.09, "learning_rate": 4.594375e-05, "loss": 1.6291, "step": 64900 }, { "epoch": 0.09, "learning_rate": 4.5943125000000006e-05, "loss": 1.5285, "step": 64910 }, { "epoch": 0.09, "learning_rate": 4.59425e-05, "loss": 1.5937, "step": 64920 }, { "epoch": 0.09, "learning_rate": 4.5941875e-05, "loss": 1.7554, "step": 64930 }, { "epoch": 0.09, "learning_rate": 4.594125e-05, "loss": 1.7356, "step": 64940 }, { "epoch": 0.09, "learning_rate": 4.594062500000001e-05, "loss": 1.4545, "step": 64950 }, { "epoch": 0.09, "learning_rate": 4.594e-05, "loss": 1.531, "step": 64960 }, { "epoch": 0.09, "learning_rate": 4.5939375e-05, "loss": 1.5173, "step": 64970 }, { "epoch": 0.09, "learning_rate": 4.5938750000000005e-05, "loss": 1.6421, "step": 64980 }, { "epoch": 0.09, "learning_rate": 4.5938125e-05, "loss": 1.6601, "step": 64990 }, { "epoch": 0.09, "learning_rate": 4.59375e-05, "loss": 1.6126, "step": 65000 }, { "epoch": 0.09, "eval_loss": 1.544957160949707, "eval_runtime": 1316.9754, "eval_samples_per_second": 44.068, "eval_steps_per_second": 5.509, "step": 65000 }, { "epoch": 0.09, "learning_rate": 4.5936875e-05, "loss": 1.7338, "step": 65010 }, { "epoch": 0.09, "learning_rate": 4.5936250000000006e-05, "loss": 1.6198, "step": 65020 }, { "epoch": 0.09, "learning_rate": 4.5935625e-05, "loss": 1.6532, "step": 65030 }, { "epoch": 0.09, "learning_rate": 4.5935e-05, "loss": 1.7005, "step": 65040 }, { "epoch": 0.09, "learning_rate": 4.5934375000000004e-05, "loss": 1.7757, "step": 65050 }, { "epoch": 0.09, "learning_rate": 4.593375e-05, "loss": 1.5824, "step": 65060 }, { "epoch": 0.09, "learning_rate": 4.5933125000000004e-05, "loss": 1.7731, "step": 65070 }, { "epoch": 0.09, "learning_rate": 4.59325e-05, "loss": 1.6127, "step": 65080 }, { "epoch": 0.09, "learning_rate": 4.5931875000000005e-05, "loss": 1.5999, "step": 65090 }, { "epoch": 0.09, "learning_rate": 4.593125e-05, "loss": 1.4692, "step": 65100 }, { "epoch": 0.09, "learning_rate": 4.5930625000000005e-05, "loss": 1.4865, "step": 65110 }, { "epoch": 0.09, "learning_rate": 4.593e-05, "loss": 1.7889, "step": 65120 }, { "epoch": 0.09, "learning_rate": 4.5929375e-05, "loss": 1.5198, "step": 65130 }, { "epoch": 0.09, "learning_rate": 4.592875e-05, "loss": 1.3767, "step": 65140 }, { "epoch": 0.09, "learning_rate": 4.5928125e-05, "loss": 1.718, "step": 65150 }, { "epoch": 0.09, "learning_rate": 4.5927500000000004e-05, "loss": 1.3928, "step": 65160 }, { "epoch": 0.09, "learning_rate": 4.5926875e-05, "loss": 1.6769, "step": 65170 }, { "epoch": 0.09, "learning_rate": 4.5926250000000004e-05, "loss": 1.6257, "step": 65180 }, { "epoch": 0.09, "learning_rate": 4.5925625e-05, "loss": 1.7899, "step": 65190 }, { "epoch": 0.09, "learning_rate": 4.5925e-05, "loss": 1.3581, "step": 65200 }, { "epoch": 0.09, "learning_rate": 4.5924375e-05, "loss": 1.6299, "step": 65210 }, { "epoch": 0.09, "learning_rate": 4.5923750000000006e-05, "loss": 1.5937, "step": 65220 }, { "epoch": 0.09, "learning_rate": 4.5923125e-05, "loss": 1.6677, "step": 65230 }, { "epoch": 0.09, "learning_rate": 4.59225e-05, "loss": 1.6228, "step": 65240 }, { "epoch": 0.09, "learning_rate": 4.5921875e-05, "loss": 1.551, "step": 65250 }, { "epoch": 0.09, "learning_rate": 4.592125000000001e-05, "loss": 1.8579, "step": 65260 }, { "epoch": 0.09, "learning_rate": 4.5920625e-05, "loss": 1.7306, "step": 65270 }, { "epoch": 0.09, "learning_rate": 4.592e-05, "loss": 1.474, "step": 65280 }, { "epoch": 0.09, "learning_rate": 4.5919375000000005e-05, "loss": 1.8471, "step": 65290 }, { "epoch": 0.09, "learning_rate": 4.591875e-05, "loss": 1.6202, "step": 65300 }, { "epoch": 0.09, "learning_rate": 4.5918125e-05, "loss": 1.702, "step": 65310 }, { "epoch": 0.09, "learning_rate": 4.59175e-05, "loss": 1.4192, "step": 65320 }, { "epoch": 0.09, "learning_rate": 4.5916875000000006e-05, "loss": 1.467, "step": 65330 }, { "epoch": 0.09, "learning_rate": 4.591625e-05, "loss": 1.5475, "step": 65340 }, { "epoch": 0.09, "learning_rate": 4.5915625e-05, "loss": 1.4637, "step": 65350 }, { "epoch": 0.09, "learning_rate": 4.5915000000000003e-05, "loss": 1.651, "step": 65360 }, { "epoch": 0.09, "learning_rate": 4.591437500000001e-05, "loss": 1.7493, "step": 65370 }, { "epoch": 0.09, "learning_rate": 4.591375e-05, "loss": 1.7119, "step": 65380 }, { "epoch": 0.09, "learning_rate": 4.5913125e-05, "loss": 1.4202, "step": 65390 }, { "epoch": 0.09, "learning_rate": 4.5912500000000005e-05, "loss": 1.7248, "step": 65400 }, { "epoch": 0.09, "learning_rate": 4.5911875e-05, "loss": 1.5986, "step": 65410 }, { "epoch": 0.09, "learning_rate": 4.591125e-05, "loss": 1.6332, "step": 65420 }, { "epoch": 0.09, "learning_rate": 4.5910625e-05, "loss": 1.4916, "step": 65430 }, { "epoch": 0.09, "learning_rate": 4.5910000000000006e-05, "loss": 1.6039, "step": 65440 }, { "epoch": 0.09, "learning_rate": 4.5909375e-05, "loss": 1.4126, "step": 65450 }, { "epoch": 0.09, "learning_rate": 4.590875e-05, "loss": 1.5268, "step": 65460 }, { "epoch": 0.09, "learning_rate": 4.5908125000000004e-05, "loss": 1.5142, "step": 65470 }, { "epoch": 0.09, "learning_rate": 4.59075e-05, "loss": 1.884, "step": 65480 }, { "epoch": 0.09, "learning_rate": 4.5906875000000004e-05, "loss": 1.4921, "step": 65490 }, { "epoch": 0.09, "learning_rate": 4.590625e-05, "loss": 1.5781, "step": 65500 }, { "epoch": 0.09, "learning_rate": 4.5905625000000005e-05, "loss": 1.37, "step": 65510 }, { "epoch": 0.09, "learning_rate": 4.5905e-05, "loss": 1.5039, "step": 65520 }, { "epoch": 0.09, "learning_rate": 4.5904375e-05, "loss": 1.5966, "step": 65530 }, { "epoch": 0.09, "learning_rate": 4.590375e-05, "loss": 1.7179, "step": 65540 }, { "epoch": 0.09, "learning_rate": 4.5903125e-05, "loss": 1.4165, "step": 65550 }, { "epoch": 0.09, "learning_rate": 4.59025e-05, "loss": 1.6352, "step": 65560 }, { "epoch": 0.09, "learning_rate": 4.5901875e-05, "loss": 1.3531, "step": 65570 }, { "epoch": 0.09, "learning_rate": 4.5901250000000004e-05, "loss": 1.7799, "step": 65580 }, { "epoch": 0.09, "learning_rate": 4.5900625e-05, "loss": 1.5917, "step": 65590 }, { "epoch": 0.09, "learning_rate": 4.5900000000000004e-05, "loss": 1.7172, "step": 65600 }, { "epoch": 0.09, "learning_rate": 4.5899375e-05, "loss": 1.5725, "step": 65610 }, { "epoch": 0.09, "learning_rate": 4.589875e-05, "loss": 1.5311, "step": 65620 }, { "epoch": 0.09, "learning_rate": 4.5898125e-05, "loss": 1.8198, "step": 65630 }, { "epoch": 0.09, "learning_rate": 4.5897500000000006e-05, "loss": 1.6093, "step": 65640 }, { "epoch": 0.09, "learning_rate": 4.5896875e-05, "loss": 1.4031, "step": 65650 }, { "epoch": 0.09, "learning_rate": 4.589625e-05, "loss": 1.5662, "step": 65660 }, { "epoch": 0.09, "learning_rate": 4.5895625e-05, "loss": 1.4478, "step": 65670 }, { "epoch": 0.09, "learning_rate": 4.589500000000001e-05, "loss": 1.7512, "step": 65680 }, { "epoch": 0.09, "learning_rate": 4.5894375000000004e-05, "loss": 1.5626, "step": 65690 }, { "epoch": 0.09, "learning_rate": 4.589375e-05, "loss": 1.7065, "step": 65700 }, { "epoch": 0.09, "learning_rate": 4.5893125000000005e-05, "loss": 1.6196, "step": 65710 }, { "epoch": 0.09, "learning_rate": 4.589250000000001e-05, "loss": 1.5087, "step": 65720 }, { "epoch": 0.09, "learning_rate": 4.5891875e-05, "loss": 1.6506, "step": 65730 }, { "epoch": 0.09, "learning_rate": 4.589125e-05, "loss": 1.4787, "step": 65740 }, { "epoch": 0.09, "learning_rate": 4.5890625000000006e-05, "loss": 1.6833, "step": 65750 }, { "epoch": 0.09, "learning_rate": 4.589e-05, "loss": 1.6768, "step": 65760 }, { "epoch": 0.09, "learning_rate": 4.5889375e-05, "loss": 1.5793, "step": 65770 }, { "epoch": 0.09, "learning_rate": 4.5888750000000003e-05, "loss": 1.6031, "step": 65780 }, { "epoch": 0.09, "learning_rate": 4.588812500000001e-05, "loss": 1.5017, "step": 65790 }, { "epoch": 0.09, "learning_rate": 4.58875e-05, "loss": 1.5416, "step": 65800 }, { "epoch": 0.09, "learning_rate": 4.5886875e-05, "loss": 1.4044, "step": 65810 }, { "epoch": 0.09, "learning_rate": 4.5886250000000005e-05, "loss": 1.4444, "step": 65820 }, { "epoch": 0.09, "learning_rate": 4.5885625e-05, "loss": 1.6866, "step": 65830 }, { "epoch": 0.09, "learning_rate": 4.5885e-05, "loss": 1.7182, "step": 65840 }, { "epoch": 0.09, "learning_rate": 4.5884375e-05, "loss": 1.5635, "step": 65850 }, { "epoch": 0.09, "learning_rate": 4.5883750000000006e-05, "loss": 1.6253, "step": 65860 }, { "epoch": 0.09, "learning_rate": 4.5883125e-05, "loss": 1.5799, "step": 65870 }, { "epoch": 0.09, "learning_rate": 4.58825e-05, "loss": 1.5709, "step": 65880 }, { "epoch": 0.09, "learning_rate": 4.5881875000000004e-05, "loss": 1.6073, "step": 65890 }, { "epoch": 0.09, "learning_rate": 4.588125e-05, "loss": 1.6353, "step": 65900 }, { "epoch": 0.09, "learning_rate": 4.5880625000000004e-05, "loss": 1.5856, "step": 65910 }, { "epoch": 0.09, "learning_rate": 4.588e-05, "loss": 1.5494, "step": 65920 }, { "epoch": 0.09, "learning_rate": 4.5879375000000005e-05, "loss": 1.6185, "step": 65930 }, { "epoch": 0.09, "learning_rate": 4.587875e-05, "loss": 1.79, "step": 65940 }, { "epoch": 0.09, "learning_rate": 4.5878125e-05, "loss": 1.8713, "step": 65950 }, { "epoch": 0.09, "learning_rate": 4.58775e-05, "loss": 1.3082, "step": 65960 }, { "epoch": 0.09, "learning_rate": 4.5876875e-05, "loss": 1.4109, "step": 65970 }, { "epoch": 0.09, "learning_rate": 4.587625e-05, "loss": 1.6726, "step": 65980 }, { "epoch": 0.09, "learning_rate": 4.5875625e-05, "loss": 1.8072, "step": 65990 }, { "epoch": 0.09, "learning_rate": 4.5875000000000004e-05, "loss": 1.6042, "step": 66000 }, { "epoch": 0.09, "eval_loss": 1.5366339683532715, "eval_runtime": 1310.5329, "eval_samples_per_second": 44.284, "eval_steps_per_second": 5.536, "step": 66000 }, { "epoch": 0.09, "learning_rate": 4.5874375e-05, "loss": 1.7236, "step": 66010 }, { "epoch": 0.09, "learning_rate": 4.5873750000000004e-05, "loss": 1.6576, "step": 66020 }, { "epoch": 0.09, "learning_rate": 4.5873125e-05, "loss": 1.6476, "step": 66030 }, { "epoch": 0.09, "learning_rate": 4.58725e-05, "loss": 1.4026, "step": 66040 }, { "epoch": 0.09, "learning_rate": 4.5871875e-05, "loss": 1.6979, "step": 66050 }, { "epoch": 0.09, "learning_rate": 4.5871250000000006e-05, "loss": 1.5541, "step": 66060 }, { "epoch": 0.09, "learning_rate": 4.5870625e-05, "loss": 1.4472, "step": 66070 }, { "epoch": 0.09, "learning_rate": 4.587e-05, "loss": 1.4373, "step": 66080 }, { "epoch": 0.09, "learning_rate": 4.5869375e-05, "loss": 1.7249, "step": 66090 }, { "epoch": 0.09, "learning_rate": 4.586875000000001e-05, "loss": 1.7263, "step": 66100 }, { "epoch": 0.09, "learning_rate": 4.5868125000000004e-05, "loss": 1.5244, "step": 66110 }, { "epoch": 0.09, "learning_rate": 4.58675e-05, "loss": 1.4213, "step": 66120 }, { "epoch": 0.09, "learning_rate": 4.5866875000000005e-05, "loss": 1.5074, "step": 66130 }, { "epoch": 0.09, "learning_rate": 4.586625e-05, "loss": 1.66, "step": 66140 }, { "epoch": 0.09, "learning_rate": 4.5865625e-05, "loss": 1.5973, "step": 66150 }, { "epoch": 0.09, "learning_rate": 4.5865e-05, "loss": 1.579, "step": 66160 }, { "epoch": 0.09, "learning_rate": 4.5864375000000006e-05, "loss": 1.5888, "step": 66170 }, { "epoch": 0.09, "learning_rate": 4.586375e-05, "loss": 1.4687, "step": 66180 }, { "epoch": 0.09, "learning_rate": 4.5863125e-05, "loss": 1.6205, "step": 66190 }, { "epoch": 0.09, "learning_rate": 4.5862500000000003e-05, "loss": 1.8076, "step": 66200 }, { "epoch": 0.09, "learning_rate": 4.586187500000001e-05, "loss": 1.5761, "step": 66210 }, { "epoch": 0.09, "learning_rate": 4.586125e-05, "loss": 1.7533, "step": 66220 }, { "epoch": 0.09, "learning_rate": 4.5860625e-05, "loss": 1.571, "step": 66230 }, { "epoch": 0.09, "learning_rate": 4.5860000000000005e-05, "loss": 1.7105, "step": 66240 }, { "epoch": 0.09, "learning_rate": 4.5859375e-05, "loss": 1.5264, "step": 66250 }, { "epoch": 0.09, "learning_rate": 4.585875e-05, "loss": 1.6975, "step": 66260 }, { "epoch": 0.09, "learning_rate": 4.5858125e-05, "loss": 1.5528, "step": 66270 }, { "epoch": 0.09, "learning_rate": 4.5857500000000006e-05, "loss": 1.5224, "step": 66280 }, { "epoch": 0.09, "learning_rate": 4.5856875e-05, "loss": 1.6807, "step": 66290 }, { "epoch": 0.09, "learning_rate": 4.585625e-05, "loss": 1.7038, "step": 66300 }, { "epoch": 0.09, "learning_rate": 4.5855625000000004e-05, "loss": 1.6492, "step": 66310 }, { "epoch": 0.09, "learning_rate": 4.5855e-05, "loss": 1.6336, "step": 66320 }, { "epoch": 0.09, "learning_rate": 4.5854375000000004e-05, "loss": 1.5151, "step": 66330 }, { "epoch": 0.09, "learning_rate": 4.585375e-05, "loss": 1.5942, "step": 66340 }, { "epoch": 0.09, "learning_rate": 4.5853125000000005e-05, "loss": 1.7056, "step": 66350 }, { "epoch": 0.09, "learning_rate": 4.58525e-05, "loss": 1.5212, "step": 66360 }, { "epoch": 0.09, "learning_rate": 4.5851875e-05, "loss": 1.537, "step": 66370 }, { "epoch": 0.09, "learning_rate": 4.585125e-05, "loss": 1.7835, "step": 66380 }, { "epoch": 0.09, "learning_rate": 4.5850625e-05, "loss": 1.6223, "step": 66390 }, { "epoch": 0.09, "learning_rate": 4.585e-05, "loss": 1.5921, "step": 66400 }, { "epoch": 0.09, "learning_rate": 4.5849375e-05, "loss": 1.3468, "step": 66410 }, { "epoch": 0.09, "learning_rate": 4.5848750000000004e-05, "loss": 1.6069, "step": 66420 }, { "epoch": 0.09, "learning_rate": 4.5848125e-05, "loss": 1.6845, "step": 66430 }, { "epoch": 0.09, "learning_rate": 4.5847500000000004e-05, "loss": 1.5694, "step": 66440 }, { "epoch": 0.09, "learning_rate": 4.5846875e-05, "loss": 1.7107, "step": 66450 }, { "epoch": 0.09, "learning_rate": 4.584625e-05, "loss": 1.8728, "step": 66460 }, { "epoch": 0.09, "learning_rate": 4.5845625e-05, "loss": 1.5476, "step": 66470 }, { "epoch": 0.09, "learning_rate": 4.5845000000000006e-05, "loss": 1.8015, "step": 66480 }, { "epoch": 0.09, "learning_rate": 4.5844375e-05, "loss": 1.7442, "step": 66490 }, { "epoch": 0.09, "learning_rate": 4.584375e-05, "loss": 1.7054, "step": 66500 }, { "epoch": 0.09, "learning_rate": 4.5843125e-05, "loss": 1.5497, "step": 66510 }, { "epoch": 0.09, "learning_rate": 4.584250000000001e-05, "loss": 1.4991, "step": 66520 }, { "epoch": 0.09, "learning_rate": 4.5841875000000004e-05, "loss": 1.7563, "step": 66530 }, { "epoch": 0.09, "learning_rate": 4.584125e-05, "loss": 1.7048, "step": 66540 }, { "epoch": 0.09, "learning_rate": 4.5840625000000005e-05, "loss": 1.5123, "step": 66550 }, { "epoch": 0.09, "learning_rate": 4.584e-05, "loss": 1.3566, "step": 66560 }, { "epoch": 0.09, "learning_rate": 4.5839375e-05, "loss": 1.5397, "step": 66570 }, { "epoch": 0.09, "learning_rate": 4.583875e-05, "loss": 1.5721, "step": 66580 }, { "epoch": 0.09, "learning_rate": 4.5838125000000006e-05, "loss": 1.5247, "step": 66590 }, { "epoch": 0.09, "learning_rate": 4.58375e-05, "loss": 1.6584, "step": 66600 }, { "epoch": 0.09, "learning_rate": 4.5836875e-05, "loss": 1.5696, "step": 66610 }, { "epoch": 0.09, "learning_rate": 4.5836250000000004e-05, "loss": 1.5847, "step": 66620 }, { "epoch": 0.09, "learning_rate": 4.583562500000001e-05, "loss": 1.534, "step": 66630 }, { "epoch": 0.09, "learning_rate": 4.5835e-05, "loss": 1.6224, "step": 66640 }, { "epoch": 0.09, "learning_rate": 4.5834375e-05, "loss": 1.561, "step": 66650 }, { "epoch": 0.09, "learning_rate": 4.5833750000000005e-05, "loss": 1.3932, "step": 66660 }, { "epoch": 0.09, "learning_rate": 4.5833125e-05, "loss": 1.6546, "step": 66670 }, { "epoch": 0.09, "learning_rate": 4.58325e-05, "loss": 1.5363, "step": 66680 }, { "epoch": 0.09, "learning_rate": 4.5831875e-05, "loss": 1.5326, "step": 66690 }, { "epoch": 0.09, "learning_rate": 4.5831250000000006e-05, "loss": 1.6868, "step": 66700 }, { "epoch": 0.09, "learning_rate": 4.5830625e-05, "loss": 1.577, "step": 66710 }, { "epoch": 0.09, "learning_rate": 4.583e-05, "loss": 1.5741, "step": 66720 }, { "epoch": 0.09, "learning_rate": 4.5829375000000004e-05, "loss": 1.6794, "step": 66730 }, { "epoch": 0.09, "learning_rate": 4.582875e-05, "loss": 1.7208, "step": 66740 }, { "epoch": 0.09, "learning_rate": 4.5828125e-05, "loss": 1.5915, "step": 66750 }, { "epoch": 0.09, "learning_rate": 4.58275e-05, "loss": 1.7503, "step": 66760 }, { "epoch": 0.09, "learning_rate": 4.5826875000000005e-05, "loss": 1.6332, "step": 66770 }, { "epoch": 0.09, "learning_rate": 4.582625e-05, "loss": 1.6006, "step": 66780 }, { "epoch": 0.09, "learning_rate": 4.5825625e-05, "loss": 1.6234, "step": 66790 }, { "epoch": 0.09, "learning_rate": 4.5825e-05, "loss": 1.4185, "step": 66800 }, { "epoch": 0.09, "learning_rate": 4.5824375e-05, "loss": 1.5125, "step": 66810 }, { "epoch": 0.09, "learning_rate": 4.582375e-05, "loss": 1.5035, "step": 66820 }, { "epoch": 0.09, "learning_rate": 4.5823125e-05, "loss": 1.6613, "step": 66830 }, { "epoch": 0.09, "learning_rate": 4.5822500000000004e-05, "loss": 1.6945, "step": 66840 }, { "epoch": 0.09, "learning_rate": 4.5821875e-05, "loss": 1.7488, "step": 66850 }, { "epoch": 0.09, "learning_rate": 4.5821250000000005e-05, "loss": 1.7801, "step": 66860 }, { "epoch": 0.09, "learning_rate": 4.5820625e-05, "loss": 1.446, "step": 66870 }, { "epoch": 0.09, "learning_rate": 4.5820000000000005e-05, "loss": 1.294, "step": 66880 }, { "epoch": 0.09, "learning_rate": 4.5819375e-05, "loss": 1.7395, "step": 66890 }, { "epoch": 0.09, "learning_rate": 4.5818750000000006e-05, "loss": 1.5526, "step": 66900 }, { "epoch": 0.09, "learning_rate": 4.5818125e-05, "loss": 1.5207, "step": 66910 }, { "epoch": 0.09, "learning_rate": 4.58175e-05, "loss": 1.6235, "step": 66920 }, { "epoch": 0.09, "learning_rate": 4.5816875e-05, "loss": 1.6678, "step": 66930 }, { "epoch": 0.09, "learning_rate": 4.581625e-05, "loss": 1.4124, "step": 66940 }, { "epoch": 0.09, "learning_rate": 4.5815625000000004e-05, "loss": 1.575, "step": 66950 }, { "epoch": 0.09, "learning_rate": 4.5815e-05, "loss": 1.4453, "step": 66960 }, { "epoch": 0.09, "learning_rate": 4.5814375000000005e-05, "loss": 1.3808, "step": 66970 }, { "epoch": 0.09, "learning_rate": 4.581375e-05, "loss": 1.4414, "step": 66980 }, { "epoch": 0.09, "learning_rate": 4.5813125e-05, "loss": 1.5029, "step": 66990 }, { "epoch": 0.09, "learning_rate": 4.58125e-05, "loss": 1.5869, "step": 67000 }, { "epoch": 0.09, "eval_loss": 1.5301387310028076, "eval_runtime": 1309.4565, "eval_samples_per_second": 44.321, "eval_steps_per_second": 5.54, "step": 67000 }, { "epoch": 0.09, "learning_rate": 4.5811875000000006e-05, "loss": 1.7115, "step": 67010 }, { "epoch": 0.09, "learning_rate": 4.581125e-05, "loss": 1.5321, "step": 67020 }, { "epoch": 0.09, "learning_rate": 4.5810625e-05, "loss": 1.4151, "step": 67030 }, { "epoch": 0.09, "learning_rate": 4.5810000000000004e-05, "loss": 1.6725, "step": 67040 }, { "epoch": 0.09, "learning_rate": 4.580937500000001e-05, "loss": 1.5619, "step": 67050 }, { "epoch": 0.09, "learning_rate": 4.580875e-05, "loss": 1.4664, "step": 67060 }, { "epoch": 0.09, "learning_rate": 4.5808125e-05, "loss": 1.6724, "step": 67070 }, { "epoch": 0.09, "learning_rate": 4.5807500000000005e-05, "loss": 1.502, "step": 67080 }, { "epoch": 0.09, "learning_rate": 4.5806875e-05, "loss": 1.7709, "step": 67090 }, { "epoch": 0.09, "learning_rate": 4.580625e-05, "loss": 1.6625, "step": 67100 }, { "epoch": 0.09, "learning_rate": 4.5805625e-05, "loss": 1.8001, "step": 67110 }, { "epoch": 0.09, "learning_rate": 4.5805000000000006e-05, "loss": 1.5655, "step": 67120 }, { "epoch": 0.09, "learning_rate": 4.5804375e-05, "loss": 1.5162, "step": 67130 }, { "epoch": 0.09, "learning_rate": 4.580375e-05, "loss": 1.5479, "step": 67140 }, { "epoch": 0.09, "learning_rate": 4.5803125000000004e-05, "loss": 1.4144, "step": 67150 }, { "epoch": 0.09, "learning_rate": 4.58025e-05, "loss": 1.6128, "step": 67160 }, { "epoch": 0.09, "learning_rate": 4.5801875e-05, "loss": 1.4257, "step": 67170 }, { "epoch": 0.09, "learning_rate": 4.580125e-05, "loss": 1.7036, "step": 67180 }, { "epoch": 0.09, "learning_rate": 4.5800625000000005e-05, "loss": 1.7983, "step": 67190 }, { "epoch": 0.09, "learning_rate": 4.58e-05, "loss": 1.4078, "step": 67200 }, { "epoch": 0.09, "learning_rate": 4.5799375e-05, "loss": 1.6074, "step": 67210 }, { "epoch": 0.09, "learning_rate": 4.579875e-05, "loss": 1.7917, "step": 67220 }, { "epoch": 0.09, "learning_rate": 4.5798125e-05, "loss": 1.6659, "step": 67230 }, { "epoch": 0.09, "learning_rate": 4.57975e-05, "loss": 1.8005, "step": 67240 }, { "epoch": 0.09, "learning_rate": 4.5796875e-05, "loss": 1.7392, "step": 67250 }, { "epoch": 0.09, "learning_rate": 4.5796250000000004e-05, "loss": 1.7778, "step": 67260 }, { "epoch": 0.09, "learning_rate": 4.5795625e-05, "loss": 1.5883, "step": 67270 }, { "epoch": 0.09, "learning_rate": 4.5795000000000005e-05, "loss": 1.7876, "step": 67280 }, { "epoch": 0.09, "learning_rate": 4.5794375e-05, "loss": 1.6751, "step": 67290 }, { "epoch": 0.09, "learning_rate": 4.5793750000000005e-05, "loss": 1.3919, "step": 67300 }, { "epoch": 0.09, "learning_rate": 4.5793125e-05, "loss": 1.6362, "step": 67310 }, { "epoch": 0.09, "learning_rate": 4.5792500000000006e-05, "loss": 1.366, "step": 67320 }, { "epoch": 0.09, "learning_rate": 4.5791875e-05, "loss": 1.5661, "step": 67330 }, { "epoch": 0.09, "learning_rate": 4.579125e-05, "loss": 1.4667, "step": 67340 }, { "epoch": 0.09, "learning_rate": 4.5790625000000003e-05, "loss": 1.6868, "step": 67350 }, { "epoch": 0.09, "learning_rate": 4.579e-05, "loss": 1.5492, "step": 67360 }, { "epoch": 0.09, "learning_rate": 4.5789375000000004e-05, "loss": 1.7393, "step": 67370 }, { "epoch": 0.09, "learning_rate": 4.578875e-05, "loss": 1.3818, "step": 67380 }, { "epoch": 0.09, "learning_rate": 4.5788125000000005e-05, "loss": 1.7473, "step": 67390 }, { "epoch": 0.09, "learning_rate": 4.57875e-05, "loss": 1.504, "step": 67400 }, { "epoch": 0.09, "learning_rate": 4.5786875e-05, "loss": 1.4893, "step": 67410 }, { "epoch": 0.09, "learning_rate": 4.578625e-05, "loss": 1.7246, "step": 67420 }, { "epoch": 0.09, "learning_rate": 4.5785625000000006e-05, "loss": 1.3939, "step": 67430 }, { "epoch": 0.09, "learning_rate": 4.5785e-05, "loss": 1.6196, "step": 67440 }, { "epoch": 0.09, "learning_rate": 4.5784375e-05, "loss": 1.5343, "step": 67450 }, { "epoch": 0.09, "learning_rate": 4.5783750000000004e-05, "loss": 1.4308, "step": 67460 }, { "epoch": 0.09, "learning_rate": 4.578312500000001e-05, "loss": 1.699, "step": 67470 }, { "epoch": 0.09, "learning_rate": 4.57825e-05, "loss": 1.624, "step": 67480 }, { "epoch": 0.09, "learning_rate": 4.5781875e-05, "loss": 1.4837, "step": 67490 }, { "epoch": 0.09, "learning_rate": 4.5781250000000005e-05, "loss": 1.5271, "step": 67500 }, { "epoch": 0.09, "learning_rate": 4.5780625e-05, "loss": 1.7576, "step": 67510 }, { "epoch": 0.09, "learning_rate": 4.578e-05, "loss": 1.6547, "step": 67520 }, { "epoch": 0.09, "learning_rate": 4.5779375e-05, "loss": 1.4063, "step": 67530 }, { "epoch": 0.09, "learning_rate": 4.5778750000000006e-05, "loss": 1.6414, "step": 67540 }, { "epoch": 0.09, "learning_rate": 4.5778124999999996e-05, "loss": 1.6368, "step": 67550 }, { "epoch": 0.09, "learning_rate": 4.57775e-05, "loss": 1.4862, "step": 67560 }, { "epoch": 0.09, "learning_rate": 4.5776875000000004e-05, "loss": 1.7104, "step": 67570 }, { "epoch": 0.09, "learning_rate": 4.577625e-05, "loss": 1.4074, "step": 67580 }, { "epoch": 0.09, "learning_rate": 4.5775625e-05, "loss": 1.5488, "step": 67590 }, { "epoch": 0.09, "learning_rate": 4.5775e-05, "loss": 1.5623, "step": 67600 }, { "epoch": 0.09, "learning_rate": 4.5774375000000005e-05, "loss": 1.6972, "step": 67610 }, { "epoch": 0.09, "learning_rate": 4.577375e-05, "loss": 1.7125, "step": 67620 }, { "epoch": 0.09, "learning_rate": 4.5773125e-05, "loss": 1.5006, "step": 67630 }, { "epoch": 0.09, "learning_rate": 4.57725e-05, "loss": 1.7329, "step": 67640 }, { "epoch": 0.09, "learning_rate": 4.5771875e-05, "loss": 1.4375, "step": 67650 }, { "epoch": 0.09, "learning_rate": 4.577125e-05, "loss": 1.6747, "step": 67660 }, { "epoch": 0.09, "learning_rate": 4.5770625e-05, "loss": 1.6389, "step": 67670 }, { "epoch": 0.09, "learning_rate": 4.5770000000000004e-05, "loss": 1.4624, "step": 67680 }, { "epoch": 0.09, "learning_rate": 4.5769375e-05, "loss": 1.6098, "step": 67690 }, { "epoch": 0.09, "learning_rate": 4.5768750000000005e-05, "loss": 1.5814, "step": 67700 }, { "epoch": 0.09, "learning_rate": 4.5768125e-05, "loss": 1.5551, "step": 67710 }, { "epoch": 0.09, "learning_rate": 4.5767500000000005e-05, "loss": 1.7617, "step": 67720 }, { "epoch": 0.09, "learning_rate": 4.5766875e-05, "loss": 1.5296, "step": 67730 }, { "epoch": 0.09, "learning_rate": 4.5766250000000006e-05, "loss": 1.4481, "step": 67740 }, { "epoch": 0.09, "learning_rate": 4.5765625e-05, "loss": 1.6389, "step": 67750 }, { "epoch": 0.09, "learning_rate": 4.5765e-05, "loss": 1.6066, "step": 67760 }, { "epoch": 0.09, "learning_rate": 4.5764375000000003e-05, "loss": 1.5664, "step": 67770 }, { "epoch": 0.09, "learning_rate": 4.576375e-05, "loss": 1.6395, "step": 67780 }, { "epoch": 0.09, "learning_rate": 4.5763125000000004e-05, "loss": 1.6689, "step": 67790 }, { "epoch": 0.09, "learning_rate": 4.57625e-05, "loss": 1.6756, "step": 67800 }, { "epoch": 0.09, "learning_rate": 4.5761875000000005e-05, "loss": 1.5786, "step": 67810 }, { "epoch": 0.09, "learning_rate": 4.576125e-05, "loss": 1.6166, "step": 67820 }, { "epoch": 0.09, "learning_rate": 4.5760625e-05, "loss": 1.5757, "step": 67830 }, { "epoch": 0.09, "learning_rate": 4.576e-05, "loss": 1.6328, "step": 67840 }, { "epoch": 0.09, "learning_rate": 4.5759375000000006e-05, "loss": 1.7446, "step": 67850 }, { "epoch": 0.09, "learning_rate": 4.575875e-05, "loss": 1.5071, "step": 67860 }, { "epoch": 0.09, "learning_rate": 4.5758125e-05, "loss": 1.5298, "step": 67870 }, { "epoch": 0.09, "learning_rate": 4.5757500000000004e-05, "loss": 1.4554, "step": 67880 }, { "epoch": 0.09, "learning_rate": 4.575687500000001e-05, "loss": 1.6997, "step": 67890 }, { "epoch": 0.09, "learning_rate": 4.575625e-05, "loss": 1.5675, "step": 67900 }, { "epoch": 0.09, "learning_rate": 4.5755625e-05, "loss": 1.524, "step": 67910 }, { "epoch": 0.09, "learning_rate": 4.5755000000000005e-05, "loss": 1.5397, "step": 67920 }, { "epoch": 0.09, "learning_rate": 4.5754375e-05, "loss": 1.7094, "step": 67930 }, { "epoch": 0.09, "learning_rate": 4.575375e-05, "loss": 1.5861, "step": 67940 }, { "epoch": 0.09, "learning_rate": 4.5753125e-05, "loss": 1.5327, "step": 67950 }, { "epoch": 0.09, "learning_rate": 4.5752500000000006e-05, "loss": 1.7562, "step": 67960 }, { "epoch": 0.09, "learning_rate": 4.5751874999999996e-05, "loss": 1.7949, "step": 67970 }, { "epoch": 0.09, "learning_rate": 4.575125e-05, "loss": 1.6452, "step": 67980 }, { "epoch": 0.09, "learning_rate": 4.5750625000000004e-05, "loss": 1.4615, "step": 67990 }, { "epoch": 0.09, "learning_rate": 4.575e-05, "loss": 1.4988, "step": 68000 }, { "epoch": 0.09, "eval_loss": 1.5233074426651, "eval_runtime": 1744.3554, "eval_samples_per_second": 33.271, "eval_steps_per_second": 4.159, "step": 68000 }, { "epoch": 0.09, "learning_rate": 4.5749375e-05, "loss": 1.6144, "step": 68010 }, { "epoch": 0.09, "learning_rate": 4.574875e-05, "loss": 1.7385, "step": 68020 }, { "epoch": 0.09, "learning_rate": 4.5748125000000005e-05, "loss": 1.5068, "step": 68030 }, { "epoch": 0.09, "learning_rate": 4.57475e-05, "loss": 1.3551, "step": 68040 }, { "epoch": 0.09, "learning_rate": 4.5746875e-05, "loss": 1.6281, "step": 68050 }, { "epoch": 0.09, "learning_rate": 4.574625e-05, "loss": 1.5651, "step": 68060 }, { "epoch": 0.09, "learning_rate": 4.5745625000000006e-05, "loss": 1.4033, "step": 68070 }, { "epoch": 0.09, "learning_rate": 4.5745e-05, "loss": 1.5867, "step": 68080 }, { "epoch": 0.09, "learning_rate": 4.5744375e-05, "loss": 1.6451, "step": 68090 }, { "epoch": 0.09, "learning_rate": 4.5743750000000004e-05, "loss": 1.6961, "step": 68100 }, { "epoch": 0.09, "learning_rate": 4.5743125e-05, "loss": 1.6454, "step": 68110 }, { "epoch": 0.09, "learning_rate": 4.5742500000000005e-05, "loss": 1.4565, "step": 68120 }, { "epoch": 0.09, "learning_rate": 4.5741875e-05, "loss": 1.7661, "step": 68130 }, { "epoch": 0.09, "learning_rate": 4.5741250000000005e-05, "loss": 1.4831, "step": 68140 }, { "epoch": 0.09, "learning_rate": 4.5740625e-05, "loss": 1.6126, "step": 68150 }, { "epoch": 0.09, "learning_rate": 4.574e-05, "loss": 1.7624, "step": 68160 }, { "epoch": 0.09, "learning_rate": 4.5739375e-05, "loss": 1.528, "step": 68170 }, { "epoch": 0.09, "learning_rate": 4.573875e-05, "loss": 1.54, "step": 68180 }, { "epoch": 0.09, "learning_rate": 4.5738125000000003e-05, "loss": 1.3745, "step": 68190 }, { "epoch": 0.09, "learning_rate": 4.57375e-05, "loss": 1.7461, "step": 68200 }, { "epoch": 0.09, "learning_rate": 4.5736875000000004e-05, "loss": 1.751, "step": 68210 }, { "epoch": 0.09, "learning_rate": 4.573625e-05, "loss": 1.5741, "step": 68220 }, { "epoch": 0.1, "learning_rate": 4.5735625000000005e-05, "loss": 1.7629, "step": 68230 }, { "epoch": 0.1, "learning_rate": 4.5735e-05, "loss": 1.6469, "step": 68240 }, { "epoch": 0.1, "learning_rate": 4.5734375e-05, "loss": 1.3349, "step": 68250 }, { "epoch": 0.1, "learning_rate": 4.573375e-05, "loss": 1.4447, "step": 68260 }, { "epoch": 0.1, "learning_rate": 4.5733125000000006e-05, "loss": 1.6384, "step": 68270 }, { "epoch": 0.1, "learning_rate": 4.57325e-05, "loss": 1.4, "step": 68280 }, { "epoch": 0.1, "learning_rate": 4.5731875e-05, "loss": 1.5545, "step": 68290 }, { "epoch": 0.1, "learning_rate": 4.5731250000000004e-05, "loss": 1.6394, "step": 68300 }, { "epoch": 0.1, "learning_rate": 4.573062500000001e-05, "loss": 1.7599, "step": 68310 }, { "epoch": 0.1, "learning_rate": 4.573e-05, "loss": 1.6985, "step": 68320 }, { "epoch": 0.1, "learning_rate": 4.5729375e-05, "loss": 1.646, "step": 68330 }, { "epoch": 0.1, "learning_rate": 4.5728750000000005e-05, "loss": 1.4116, "step": 68340 }, { "epoch": 0.1, "learning_rate": 4.5728125e-05, "loss": 1.5675, "step": 68350 }, { "epoch": 0.1, "learning_rate": 4.57275e-05, "loss": 1.5161, "step": 68360 }, { "epoch": 0.1, "learning_rate": 4.5726875e-05, "loss": 1.7197, "step": 68370 }, { "epoch": 0.1, "learning_rate": 4.5726250000000006e-05, "loss": 1.38, "step": 68380 }, { "epoch": 0.1, "learning_rate": 4.5725624999999996e-05, "loss": 1.5853, "step": 68390 }, { "epoch": 0.1, "learning_rate": 4.5725e-05, "loss": 1.6021, "step": 68400 }, { "epoch": 0.1, "learning_rate": 4.5724375000000004e-05, "loss": 1.3929, "step": 68410 }, { "epoch": 0.1, "learning_rate": 4.572375e-05, "loss": 1.58, "step": 68420 }, { "epoch": 0.1, "learning_rate": 4.5723125e-05, "loss": 1.6998, "step": 68430 }, { "epoch": 0.1, "learning_rate": 4.57225e-05, "loss": 1.6356, "step": 68440 }, { "epoch": 0.1, "learning_rate": 4.5721875000000005e-05, "loss": 1.5671, "step": 68450 }, { "epoch": 0.1, "learning_rate": 4.572125e-05, "loss": 1.2254, "step": 68460 }, { "epoch": 0.1, "learning_rate": 4.5720625e-05, "loss": 1.6604, "step": 68470 }, { "epoch": 0.1, "learning_rate": 4.572e-05, "loss": 1.6348, "step": 68480 }, { "epoch": 0.1, "learning_rate": 4.5719375000000006e-05, "loss": 1.685, "step": 68490 }, { "epoch": 0.1, "learning_rate": 4.571875e-05, "loss": 1.5199, "step": 68500 }, { "epoch": 0.1, "learning_rate": 4.5718125e-05, "loss": 1.6464, "step": 68510 }, { "epoch": 0.1, "learning_rate": 4.5717500000000004e-05, "loss": 1.4909, "step": 68520 }, { "epoch": 0.1, "learning_rate": 4.5716875e-05, "loss": 1.6331, "step": 68530 }, { "epoch": 0.1, "learning_rate": 4.5716250000000005e-05, "loss": 1.619, "step": 68540 }, { "epoch": 0.1, "learning_rate": 4.5715625e-05, "loss": 1.6855, "step": 68550 }, { "epoch": 0.1, "learning_rate": 4.5715000000000005e-05, "loss": 1.3329, "step": 68560 }, { "epoch": 0.1, "learning_rate": 4.5714375e-05, "loss": 1.6435, "step": 68570 }, { "epoch": 0.1, "learning_rate": 4.571375e-05, "loss": 1.4463, "step": 68580 }, { "epoch": 0.1, "learning_rate": 4.5713125e-05, "loss": 1.7651, "step": 68590 }, { "epoch": 0.1, "learning_rate": 4.57125e-05, "loss": 1.6402, "step": 68600 }, { "epoch": 0.1, "learning_rate": 4.5711875000000004e-05, "loss": 1.6377, "step": 68610 }, { "epoch": 0.1, "learning_rate": 4.571125e-05, "loss": 1.5313, "step": 68620 }, { "epoch": 0.1, "learning_rate": 4.5710625000000004e-05, "loss": 1.5444, "step": 68630 }, { "epoch": 0.1, "learning_rate": 4.571e-05, "loss": 1.4659, "step": 68640 }, { "epoch": 0.1, "learning_rate": 4.5709375000000005e-05, "loss": 1.6779, "step": 68650 }, { "epoch": 0.1, "learning_rate": 4.570875e-05, "loss": 1.4562, "step": 68660 }, { "epoch": 0.1, "learning_rate": 4.5708125e-05, "loss": 1.5642, "step": 68670 }, { "epoch": 0.1, "learning_rate": 4.57075e-05, "loss": 1.5532, "step": 68680 }, { "epoch": 0.1, "learning_rate": 4.5706875000000006e-05, "loss": 1.5245, "step": 68690 }, { "epoch": 0.1, "learning_rate": 4.570625e-05, "loss": 1.6445, "step": 68700 }, { "epoch": 0.1, "learning_rate": 4.5705625e-05, "loss": 1.5321, "step": 68710 }, { "epoch": 0.1, "learning_rate": 4.5705000000000004e-05, "loss": 1.4759, "step": 68720 }, { "epoch": 0.1, "learning_rate": 4.570437500000001e-05, "loss": 1.5748, "step": 68730 }, { "epoch": 0.1, "learning_rate": 4.570375e-05, "loss": 1.5893, "step": 68740 }, { "epoch": 0.1, "learning_rate": 4.5703125e-05, "loss": 1.7336, "step": 68750 }, { "epoch": 0.1, "learning_rate": 4.5702500000000005e-05, "loss": 1.6778, "step": 68760 }, { "epoch": 0.1, "learning_rate": 4.5701875e-05, "loss": 1.3919, "step": 68770 }, { "epoch": 0.1, "learning_rate": 4.570125e-05, "loss": 1.6096, "step": 68780 }, { "epoch": 0.1, "learning_rate": 4.5700625e-05, "loss": 1.5016, "step": 68790 }, { "epoch": 0.1, "learning_rate": 4.5700000000000006e-05, "loss": 1.7069, "step": 68800 }, { "epoch": 0.1, "learning_rate": 4.5699374999999996e-05, "loss": 1.5768, "step": 68810 }, { "epoch": 0.1, "learning_rate": 4.569875e-05, "loss": 1.7017, "step": 68820 }, { "epoch": 0.1, "learning_rate": 4.5698125000000004e-05, "loss": 1.5048, "step": 68830 }, { "epoch": 0.1, "learning_rate": 4.56975e-05, "loss": 1.5224, "step": 68840 }, { "epoch": 0.1, "learning_rate": 4.5696875e-05, "loss": 1.6678, "step": 68850 }, { "epoch": 0.1, "learning_rate": 4.569625e-05, "loss": 1.443, "step": 68860 }, { "epoch": 0.1, "learning_rate": 4.5695625000000005e-05, "loss": 1.4762, "step": 68870 }, { "epoch": 0.1, "learning_rate": 4.5695e-05, "loss": 1.4424, "step": 68880 }, { "epoch": 0.1, "learning_rate": 4.5694375e-05, "loss": 1.5477, "step": 68890 }, { "epoch": 0.1, "learning_rate": 4.569375e-05, "loss": 1.827, "step": 68900 }, { "epoch": 0.1, "learning_rate": 4.5693125000000006e-05, "loss": 1.6943, "step": 68910 }, { "epoch": 0.1, "learning_rate": 4.5692500000000003e-05, "loss": 1.6297, "step": 68920 }, { "epoch": 0.1, "learning_rate": 4.5691875e-05, "loss": 1.6007, "step": 68930 }, { "epoch": 0.1, "learning_rate": 4.5691250000000004e-05, "loss": 1.5433, "step": 68940 }, { "epoch": 0.1, "learning_rate": 4.5690625e-05, "loss": 1.522, "step": 68950 }, { "epoch": 0.1, "learning_rate": 4.569e-05, "loss": 1.7677, "step": 68960 }, { "epoch": 0.1, "learning_rate": 4.5689375e-05, "loss": 1.6899, "step": 68970 }, { "epoch": 0.1, "learning_rate": 4.5688750000000005e-05, "loss": 1.5391, "step": 68980 }, { "epoch": 0.1, "learning_rate": 4.5688125e-05, "loss": 1.5952, "step": 68990 }, { "epoch": 0.1, "learning_rate": 4.56875e-05, "loss": 1.4868, "step": 69000 }, { "epoch": 0.1, "eval_loss": 1.5202010869979858, "eval_runtime": 1830.5391, "eval_samples_per_second": 31.704, "eval_steps_per_second": 3.963, "step": 69000 }, { "epoch": 0.1, "learning_rate": 4.5686875e-05, "loss": 1.5864, "step": 69010 }, { "epoch": 0.1, "learning_rate": 4.568625e-05, "loss": 1.6634, "step": 69020 }, { "epoch": 0.1, "learning_rate": 4.5685625000000004e-05, "loss": 1.6619, "step": 69030 }, { "epoch": 0.1, "learning_rate": 4.5685e-05, "loss": 1.524, "step": 69040 }, { "epoch": 0.1, "learning_rate": 4.5684375000000004e-05, "loss": 1.6642, "step": 69050 }, { "epoch": 0.1, "learning_rate": 4.568375e-05, "loss": 1.6811, "step": 69060 }, { "epoch": 0.1, "learning_rate": 4.5683125000000005e-05, "loss": 1.5744, "step": 69070 }, { "epoch": 0.1, "learning_rate": 4.56825e-05, "loss": 1.4846, "step": 69080 }, { "epoch": 0.1, "learning_rate": 4.5681875e-05, "loss": 1.5142, "step": 69090 }, { "epoch": 0.1, "learning_rate": 4.568125e-05, "loss": 1.5704, "step": 69100 }, { "epoch": 0.1, "learning_rate": 4.5680625000000006e-05, "loss": 1.6364, "step": 69110 }, { "epoch": 0.1, "learning_rate": 4.568e-05, "loss": 1.9453, "step": 69120 }, { "epoch": 0.1, "learning_rate": 4.5679375e-05, "loss": 1.6224, "step": 69130 }, { "epoch": 0.1, "learning_rate": 4.5678750000000004e-05, "loss": 1.7002, "step": 69140 }, { "epoch": 0.1, "learning_rate": 4.567812500000001e-05, "loss": 1.695, "step": 69150 }, { "epoch": 0.1, "learning_rate": 4.56775e-05, "loss": 1.6578, "step": 69160 }, { "epoch": 0.1, "learning_rate": 4.5676875e-05, "loss": 1.6368, "step": 69170 }, { "epoch": 0.1, "learning_rate": 4.5676250000000005e-05, "loss": 1.547, "step": 69180 }, { "epoch": 0.1, "learning_rate": 4.5675625e-05, "loss": 1.848, "step": 69190 }, { "epoch": 0.1, "learning_rate": 4.5675e-05, "loss": 1.7744, "step": 69200 }, { "epoch": 0.1, "learning_rate": 4.5674375e-05, "loss": 1.6617, "step": 69210 }, { "epoch": 0.1, "learning_rate": 4.5673750000000006e-05, "loss": 1.5642, "step": 69220 }, { "epoch": 0.1, "learning_rate": 4.5673125e-05, "loss": 1.6174, "step": 69230 }, { "epoch": 0.1, "learning_rate": 4.56725e-05, "loss": 1.6277, "step": 69240 }, { "epoch": 0.1, "learning_rate": 4.5671875000000004e-05, "loss": 1.7598, "step": 69250 }, { "epoch": 0.1, "learning_rate": 4.567125000000001e-05, "loss": 1.3861, "step": 69260 }, { "epoch": 0.1, "learning_rate": 4.5670625e-05, "loss": 1.525, "step": 69270 }, { "epoch": 0.1, "learning_rate": 4.567e-05, "loss": 1.4381, "step": 69280 }, { "epoch": 0.1, "learning_rate": 4.5669375000000005e-05, "loss": 1.7199, "step": 69290 }, { "epoch": 0.1, "learning_rate": 4.566875e-05, "loss": 1.4755, "step": 69300 }, { "epoch": 0.1, "learning_rate": 4.5668125e-05, "loss": 1.4663, "step": 69310 }, { "epoch": 0.1, "learning_rate": 4.56675e-05, "loss": 1.8516, "step": 69320 }, { "epoch": 0.1, "learning_rate": 4.5666875000000007e-05, "loss": 1.5243, "step": 69330 }, { "epoch": 0.1, "learning_rate": 4.5666250000000003e-05, "loss": 1.6465, "step": 69340 }, { "epoch": 0.1, "learning_rate": 4.5665625e-05, "loss": 1.6993, "step": 69350 }, { "epoch": 0.1, "learning_rate": 4.5665000000000004e-05, "loss": 1.6211, "step": 69360 }, { "epoch": 0.1, "learning_rate": 4.5664375e-05, "loss": 1.7059, "step": 69370 }, { "epoch": 0.1, "learning_rate": 4.566375e-05, "loss": 1.4171, "step": 69380 }, { "epoch": 0.1, "learning_rate": 4.5663125e-05, "loss": 1.6473, "step": 69390 }, { "epoch": 0.1, "learning_rate": 4.5662500000000005e-05, "loss": 1.603, "step": 69400 }, { "epoch": 0.1, "learning_rate": 4.5661875e-05, "loss": 1.6384, "step": 69410 }, { "epoch": 0.1, "learning_rate": 4.566125e-05, "loss": 1.4466, "step": 69420 }, { "epoch": 0.1, "learning_rate": 4.5660625e-05, "loss": 1.6296, "step": 69430 }, { "epoch": 0.1, "learning_rate": 4.566e-05, "loss": 1.7224, "step": 69440 }, { "epoch": 0.1, "learning_rate": 4.5659375000000004e-05, "loss": 1.4607, "step": 69450 }, { "epoch": 0.1, "learning_rate": 4.565875e-05, "loss": 1.3005, "step": 69460 }, { "epoch": 0.1, "learning_rate": 4.5658125000000004e-05, "loss": 1.255, "step": 69470 }, { "epoch": 0.1, "learning_rate": 4.56575e-05, "loss": 1.3657, "step": 69480 }, { "epoch": 0.1, "learning_rate": 4.5656875000000005e-05, "loss": 1.7126, "step": 69490 }, { "epoch": 0.1, "learning_rate": 4.565625e-05, "loss": 1.8011, "step": 69500 }, { "epoch": 0.1, "learning_rate": 4.5655625e-05, "loss": 1.8176, "step": 69510 }, { "epoch": 0.1, "learning_rate": 4.5655e-05, "loss": 1.5965, "step": 69520 }, { "epoch": 0.1, "learning_rate": 4.5654375000000006e-05, "loss": 1.5428, "step": 69530 }, { "epoch": 0.1, "learning_rate": 4.565375e-05, "loss": 1.5855, "step": 69540 }, { "epoch": 0.1, "learning_rate": 4.5653125e-05, "loss": 1.7842, "step": 69550 }, { "epoch": 0.1, "learning_rate": 4.5652500000000004e-05, "loss": 1.6717, "step": 69560 }, { "epoch": 0.1, "learning_rate": 4.5651875e-05, "loss": 1.5358, "step": 69570 }, { "epoch": 0.1, "learning_rate": 4.565125e-05, "loss": 1.7049, "step": 69580 }, { "epoch": 0.1, "learning_rate": 4.5650625e-05, "loss": 1.5197, "step": 69590 }, { "epoch": 0.1, "learning_rate": 4.5650000000000005e-05, "loss": 1.6475, "step": 69600 }, { "epoch": 0.1, "learning_rate": 4.5649375e-05, "loss": 1.5104, "step": 69610 }, { "epoch": 0.1, "learning_rate": 4.564875e-05, "loss": 1.6533, "step": 69620 }, { "epoch": 0.1, "learning_rate": 4.5648125e-05, "loss": 1.5778, "step": 69630 }, { "epoch": 0.1, "learning_rate": 4.5647500000000006e-05, "loss": 1.4605, "step": 69640 }, { "epoch": 0.1, "learning_rate": 4.5646875e-05, "loss": 1.4, "step": 69650 }, { "epoch": 0.1, "learning_rate": 4.564625e-05, "loss": 1.4086, "step": 69660 }, { "epoch": 0.1, "learning_rate": 4.5645625000000004e-05, "loss": 1.8385, "step": 69670 }, { "epoch": 0.1, "learning_rate": 4.564500000000001e-05, "loss": 1.5779, "step": 69680 }, { "epoch": 0.1, "learning_rate": 4.5644375e-05, "loss": 1.598, "step": 69690 }, { "epoch": 0.1, "learning_rate": 4.564375e-05, "loss": 1.6654, "step": 69700 }, { "epoch": 0.1, "learning_rate": 4.5643125000000005e-05, "loss": 1.4034, "step": 69710 }, { "epoch": 0.1, "learning_rate": 4.56425e-05, "loss": 1.6776, "step": 69720 }, { "epoch": 0.1, "learning_rate": 4.5641875e-05, "loss": 1.5757, "step": 69730 }, { "epoch": 0.1, "learning_rate": 4.564125e-05, "loss": 1.4675, "step": 69740 }, { "epoch": 0.1, "learning_rate": 4.5640625000000007e-05, "loss": 1.6244, "step": 69750 }, { "epoch": 0.1, "learning_rate": 4.564e-05, "loss": 1.3364, "step": 69760 }, { "epoch": 0.1, "learning_rate": 4.5639375e-05, "loss": 1.4528, "step": 69770 }, { "epoch": 0.1, "learning_rate": 4.5638750000000004e-05, "loss": 1.525, "step": 69780 }, { "epoch": 0.1, "learning_rate": 4.5638125e-05, "loss": 1.3343, "step": 69790 }, { "epoch": 0.1, "learning_rate": 4.56375e-05, "loss": 1.7498, "step": 69800 }, { "epoch": 0.1, "learning_rate": 4.5636875e-05, "loss": 1.6657, "step": 69810 }, { "epoch": 0.1, "learning_rate": 4.5636250000000005e-05, "loss": 1.5022, "step": 69820 }, { "epoch": 0.1, "learning_rate": 4.5635625e-05, "loss": 1.5734, "step": 69830 }, { "epoch": 0.1, "learning_rate": 4.5635e-05, "loss": 1.5164, "step": 69840 }, { "epoch": 0.1, "learning_rate": 4.5634375e-05, "loss": 1.6554, "step": 69850 }, { "epoch": 0.1, "learning_rate": 4.563375e-05, "loss": 1.6982, "step": 69860 }, { "epoch": 0.1, "learning_rate": 4.5633125000000004e-05, "loss": 1.5387, "step": 69870 }, { "epoch": 0.1, "learning_rate": 4.56325e-05, "loss": 1.8223, "step": 69880 }, { "epoch": 0.1, "learning_rate": 4.5631875000000004e-05, "loss": 1.8096, "step": 69890 }, { "epoch": 0.1, "learning_rate": 4.563125e-05, "loss": 1.6202, "step": 69900 }, { "epoch": 0.1, "learning_rate": 4.5630625000000005e-05, "loss": 1.5653, "step": 69910 }, { "epoch": 0.1, "learning_rate": 4.563e-05, "loss": 1.5744, "step": 69920 }, { "epoch": 0.1, "learning_rate": 4.5629375e-05, "loss": 1.6972, "step": 69930 }, { "epoch": 0.1, "learning_rate": 4.562875e-05, "loss": 1.8361, "step": 69940 }, { "epoch": 0.1, "learning_rate": 4.5628125000000006e-05, "loss": 1.7209, "step": 69950 }, { "epoch": 0.1, "learning_rate": 4.56275e-05, "loss": 1.4772, "step": 69960 }, { "epoch": 0.1, "learning_rate": 4.5626875e-05, "loss": 1.6905, "step": 69970 }, { "epoch": 0.1, "learning_rate": 4.5626250000000004e-05, "loss": 1.5726, "step": 69980 }, { "epoch": 0.1, "learning_rate": 4.5625625e-05, "loss": 1.5148, "step": 69990 }, { "epoch": 0.1, "learning_rate": 4.5625e-05, "loss": 1.5597, "step": 70000 }, { "epoch": 0.1, "eval_loss": 1.512281894683838, "eval_runtime": 1963.9157, "eval_samples_per_second": 29.551, "eval_steps_per_second": 3.694, "step": 70000 }, { "epoch": 0.1, "learning_rate": 4.5624375e-05, "loss": 1.4973, "step": 70010 }, { "epoch": 0.1, "learning_rate": 4.5623750000000005e-05, "loss": 1.6827, "step": 70020 }, { "epoch": 0.1, "learning_rate": 4.5623125e-05, "loss": 1.4021, "step": 70030 }, { "epoch": 0.1, "learning_rate": 4.56225e-05, "loss": 1.6186, "step": 70040 }, { "epoch": 0.1, "learning_rate": 4.5621875e-05, "loss": 1.4356, "step": 70050 }, { "epoch": 0.1, "learning_rate": 4.5621250000000006e-05, "loss": 1.626, "step": 70060 }, { "epoch": 0.1, "learning_rate": 4.5620625e-05, "loss": 1.8528, "step": 70070 }, { "epoch": 0.1, "learning_rate": 4.562e-05, "loss": 1.601, "step": 70080 }, { "epoch": 0.1, "learning_rate": 4.5619375000000004e-05, "loss": 1.5956, "step": 70090 }, { "epoch": 0.1, "learning_rate": 4.561875000000001e-05, "loss": 1.6395, "step": 70100 }, { "epoch": 0.1, "learning_rate": 4.5618125e-05, "loss": 1.4414, "step": 70110 }, { "epoch": 0.1, "learning_rate": 4.56175e-05, "loss": 1.64, "step": 70120 }, { "epoch": 0.1, "learning_rate": 4.5616875000000005e-05, "loss": 1.7728, "step": 70130 }, { "epoch": 0.1, "learning_rate": 4.561625e-05, "loss": 1.4658, "step": 70140 }, { "epoch": 0.1, "learning_rate": 4.5615625e-05, "loss": 1.4828, "step": 70150 }, { "epoch": 0.1, "learning_rate": 4.5615e-05, "loss": 1.5977, "step": 70160 }, { "epoch": 0.1, "learning_rate": 4.5614375000000007e-05, "loss": 1.4844, "step": 70170 }, { "epoch": 0.1, "learning_rate": 4.561375e-05, "loss": 1.6771, "step": 70180 }, { "epoch": 0.1, "learning_rate": 4.5613125e-05, "loss": 1.8027, "step": 70190 }, { "epoch": 0.1, "learning_rate": 4.5612500000000004e-05, "loss": 1.5221, "step": 70200 }, { "epoch": 0.1, "learning_rate": 4.5611875e-05, "loss": 1.4701, "step": 70210 }, { "epoch": 0.1, "learning_rate": 4.561125e-05, "loss": 1.354, "step": 70220 }, { "epoch": 0.1, "learning_rate": 4.5610625e-05, "loss": 1.6168, "step": 70230 }, { "epoch": 0.1, "learning_rate": 4.5610000000000005e-05, "loss": 1.5148, "step": 70240 }, { "epoch": 0.1, "learning_rate": 4.5609375e-05, "loss": 1.3838, "step": 70250 }, { "epoch": 0.1, "learning_rate": 4.560875e-05, "loss": 1.4104, "step": 70260 }, { "epoch": 0.1, "learning_rate": 4.5608125e-05, "loss": 1.512, "step": 70270 }, { "epoch": 0.1, "learning_rate": 4.56075e-05, "loss": 1.6615, "step": 70280 }, { "epoch": 0.1, "learning_rate": 4.5606875000000004e-05, "loss": 1.6208, "step": 70290 }, { "epoch": 0.1, "learning_rate": 4.560625e-05, "loss": 1.6765, "step": 70300 }, { "epoch": 0.1, "learning_rate": 4.5605625000000004e-05, "loss": 1.4817, "step": 70310 }, { "epoch": 0.1, "learning_rate": 4.5605e-05, "loss": 1.559, "step": 70320 }, { "epoch": 0.1, "learning_rate": 4.5604375000000005e-05, "loss": 1.6204, "step": 70330 }, { "epoch": 0.1, "learning_rate": 4.560375e-05, "loss": 1.594, "step": 70340 }, { "epoch": 0.1, "learning_rate": 4.5603125e-05, "loss": 1.7843, "step": 70350 }, { "epoch": 0.1, "learning_rate": 4.56025e-05, "loss": 1.6762, "step": 70360 }, { "epoch": 0.1, "learning_rate": 4.5601875e-05, "loss": 1.3575, "step": 70370 }, { "epoch": 0.1, "learning_rate": 4.560125e-05, "loss": 1.4853, "step": 70380 }, { "epoch": 0.1, "learning_rate": 4.5600625e-05, "loss": 1.4955, "step": 70390 }, { "epoch": 0.1, "learning_rate": 4.5600000000000004e-05, "loss": 1.4748, "step": 70400 }, { "epoch": 0.1, "learning_rate": 4.5599375e-05, "loss": 1.7455, "step": 70410 }, { "epoch": 0.1, "learning_rate": 4.5598750000000005e-05, "loss": 1.4577, "step": 70420 }, { "epoch": 0.1, "learning_rate": 4.5598125e-05, "loss": 1.6213, "step": 70430 }, { "epoch": 0.1, "learning_rate": 4.5597500000000005e-05, "loss": 1.7075, "step": 70440 }, { "epoch": 0.1, "learning_rate": 4.5596875e-05, "loss": 1.5145, "step": 70450 }, { "epoch": 0.1, "learning_rate": 4.559625e-05, "loss": 1.4753, "step": 70460 }, { "epoch": 0.1, "learning_rate": 4.5595625e-05, "loss": 1.7206, "step": 70470 }, { "epoch": 0.1, "learning_rate": 4.5595000000000006e-05, "loss": 1.4969, "step": 70480 }, { "epoch": 0.1, "learning_rate": 4.5594375000000003e-05, "loss": 1.6263, "step": 70490 }, { "epoch": 0.1, "learning_rate": 4.559375e-05, "loss": 1.74, "step": 70500 }, { "epoch": 0.1, "learning_rate": 4.5593125000000004e-05, "loss": 1.4446, "step": 70510 }, { "epoch": 0.1, "learning_rate": 4.559250000000001e-05, "loss": 1.4771, "step": 70520 }, { "epoch": 0.1, "learning_rate": 4.5591875e-05, "loss": 1.6443, "step": 70530 }, { "epoch": 0.1, "learning_rate": 4.559125e-05, "loss": 1.4337, "step": 70540 }, { "epoch": 0.1, "learning_rate": 4.5590625000000005e-05, "loss": 1.737, "step": 70550 }, { "epoch": 0.1, "learning_rate": 4.559e-05, "loss": 1.4986, "step": 70560 }, { "epoch": 0.1, "learning_rate": 4.5589375e-05, "loss": 1.5865, "step": 70570 }, { "epoch": 0.1, "learning_rate": 4.558875e-05, "loss": 1.4805, "step": 70580 }, { "epoch": 0.1, "learning_rate": 4.558812500000001e-05, "loss": 1.5878, "step": 70590 }, { "epoch": 0.1, "learning_rate": 4.55875e-05, "loss": 1.7914, "step": 70600 }, { "epoch": 0.1, "learning_rate": 4.5586875e-05, "loss": 1.3618, "step": 70610 }, { "epoch": 0.1, "learning_rate": 4.5586250000000004e-05, "loss": 1.5099, "step": 70620 }, { "epoch": 0.1, "learning_rate": 4.5585625e-05, "loss": 1.4679, "step": 70630 }, { "epoch": 0.1, "learning_rate": 4.5585e-05, "loss": 1.5455, "step": 70640 }, { "epoch": 0.1, "learning_rate": 4.5584375e-05, "loss": 1.6607, "step": 70650 }, { "epoch": 0.1, "learning_rate": 4.5583750000000006e-05, "loss": 1.6431, "step": 70660 }, { "epoch": 0.1, "learning_rate": 4.5583125e-05, "loss": 1.4723, "step": 70670 }, { "epoch": 0.1, "learning_rate": 4.55825e-05, "loss": 1.6742, "step": 70680 }, { "epoch": 0.1, "learning_rate": 4.5581875e-05, "loss": 1.5269, "step": 70690 }, { "epoch": 0.1, "learning_rate": 4.558125e-05, "loss": 1.6687, "step": 70700 }, { "epoch": 0.1, "learning_rate": 4.5580625000000004e-05, "loss": 1.5779, "step": 70710 }, { "epoch": 0.1, "learning_rate": 4.558e-05, "loss": 1.5375, "step": 70720 }, { "epoch": 0.1, "learning_rate": 4.5579375000000004e-05, "loss": 1.6268, "step": 70730 }, { "epoch": 0.1, "learning_rate": 4.557875e-05, "loss": 1.5753, "step": 70740 }, { "epoch": 0.1, "learning_rate": 4.5578125000000005e-05, "loss": 1.5241, "step": 70750 }, { "epoch": 0.1, "learning_rate": 4.55775e-05, "loss": 1.5468, "step": 70760 }, { "epoch": 0.1, "learning_rate": 4.5576875e-05, "loss": 1.4601, "step": 70770 }, { "epoch": 0.1, "learning_rate": 4.557625e-05, "loss": 1.7585, "step": 70780 }, { "epoch": 0.1, "learning_rate": 4.5575625e-05, "loss": 1.4157, "step": 70790 }, { "epoch": 0.1, "learning_rate": 4.5575e-05, "loss": 1.513, "step": 70800 }, { "epoch": 0.1, "learning_rate": 4.5574375e-05, "loss": 1.4941, "step": 70810 }, { "epoch": 0.1, "learning_rate": 4.5573750000000004e-05, "loss": 1.5791, "step": 70820 }, { "epoch": 0.1, "learning_rate": 4.5573125e-05, "loss": 1.5098, "step": 70830 }, { "epoch": 0.1, "learning_rate": 4.5572500000000005e-05, "loss": 1.4973, "step": 70840 }, { "epoch": 0.1, "learning_rate": 4.5571875e-05, "loss": 1.5111, "step": 70850 }, { "epoch": 0.1, "learning_rate": 4.5571250000000005e-05, "loss": 1.5264, "step": 70860 }, { "epoch": 0.1, "learning_rate": 4.5570625e-05, "loss": 1.4493, "step": 70870 }, { "epoch": 0.1, "learning_rate": 4.557e-05, "loss": 1.5046, "step": 70880 }, { "epoch": 0.1, "learning_rate": 4.5569375e-05, "loss": 1.5825, "step": 70890 }, { "epoch": 0.1, "learning_rate": 4.5568750000000006e-05, "loss": 1.6204, "step": 70900 }, { "epoch": 0.1, "learning_rate": 4.5568125000000003e-05, "loss": 1.6057, "step": 70910 }, { "epoch": 0.1, "learning_rate": 4.55675e-05, "loss": 1.7094, "step": 70920 }, { "epoch": 0.1, "learning_rate": 4.5566875000000004e-05, "loss": 1.3091, "step": 70930 }, { "epoch": 0.1, "learning_rate": 4.556625000000001e-05, "loss": 1.518, "step": 70940 }, { "epoch": 0.1, "learning_rate": 4.5565625e-05, "loss": 1.6175, "step": 70950 }, { "epoch": 0.1, "learning_rate": 4.5565e-05, "loss": 1.6369, "step": 70960 }, { "epoch": 0.1, "learning_rate": 4.5564375000000005e-05, "loss": 1.7206, "step": 70970 }, { "epoch": 0.1, "learning_rate": 4.556375e-05, "loss": 1.5243, "step": 70980 }, { "epoch": 0.1, "learning_rate": 4.5563125e-05, "loss": 1.553, "step": 70990 }, { "epoch": 0.1, "learning_rate": 4.55625e-05, "loss": 1.2303, "step": 71000 }, { "epoch": 0.1, "eval_loss": 1.5156375169754028, "eval_runtime": 1913.2874, "eval_samples_per_second": 30.333, "eval_steps_per_second": 3.792, "step": 71000 }, { "epoch": 0.1, "learning_rate": 4.556187500000001e-05, "loss": 1.6997, "step": 71010 }, { "epoch": 0.1, "learning_rate": 4.556125e-05, "loss": 1.4675, "step": 71020 }, { "epoch": 0.1, "learning_rate": 4.5560625e-05, "loss": 1.6683, "step": 71030 }, { "epoch": 0.1, "learning_rate": 4.5560000000000004e-05, "loss": 1.6093, "step": 71040 }, { "epoch": 0.1, "learning_rate": 4.5559375e-05, "loss": 1.6038, "step": 71050 }, { "epoch": 0.1, "learning_rate": 4.555875e-05, "loss": 1.7998, "step": 71060 }, { "epoch": 0.1, "learning_rate": 4.5558125e-05, "loss": 1.7953, "step": 71070 }, { "epoch": 0.1, "learning_rate": 4.5557500000000006e-05, "loss": 1.5272, "step": 71080 }, { "epoch": 0.1, "learning_rate": 4.5556875e-05, "loss": 1.6853, "step": 71090 }, { "epoch": 0.1, "learning_rate": 4.555625e-05, "loss": 1.5049, "step": 71100 }, { "epoch": 0.1, "learning_rate": 4.5555625e-05, "loss": 1.6332, "step": 71110 }, { "epoch": 0.1, "learning_rate": 4.5555e-05, "loss": 1.64, "step": 71120 }, { "epoch": 0.1, "learning_rate": 4.5554375000000004e-05, "loss": 1.3638, "step": 71130 }, { "epoch": 0.1, "learning_rate": 4.555375e-05, "loss": 1.5707, "step": 71140 }, { "epoch": 0.1, "learning_rate": 4.5553125000000004e-05, "loss": 1.5888, "step": 71150 }, { "epoch": 0.1, "learning_rate": 4.55525e-05, "loss": 1.5317, "step": 71160 }, { "epoch": 0.1, "learning_rate": 4.5551875e-05, "loss": 1.7076, "step": 71170 }, { "epoch": 0.1, "learning_rate": 4.555125e-05, "loss": 1.6227, "step": 71180 }, { "epoch": 0.1, "learning_rate": 4.5550625e-05, "loss": 1.6251, "step": 71190 }, { "epoch": 0.1, "learning_rate": 4.555e-05, "loss": 1.5709, "step": 71200 }, { "epoch": 0.1, "learning_rate": 4.5549375e-05, "loss": 1.6527, "step": 71210 }, { "epoch": 0.1, "learning_rate": 4.554875e-05, "loss": 1.6801, "step": 71220 }, { "epoch": 0.1, "learning_rate": 4.5548125e-05, "loss": 1.5202, "step": 71230 }, { "epoch": 0.1, "learning_rate": 4.5547500000000004e-05, "loss": 1.5517, "step": 71240 }, { "epoch": 0.1, "learning_rate": 4.5546875e-05, "loss": 1.4531, "step": 71250 }, { "epoch": 0.1, "learning_rate": 4.5546250000000005e-05, "loss": 1.6193, "step": 71260 }, { "epoch": 0.1, "learning_rate": 4.5545625e-05, "loss": 1.4471, "step": 71270 }, { "epoch": 0.1, "learning_rate": 4.5545000000000005e-05, "loss": 1.6182, "step": 71280 }, { "epoch": 0.1, "learning_rate": 4.5544375e-05, "loss": 1.4481, "step": 71290 }, { "epoch": 0.1, "learning_rate": 4.554375e-05, "loss": 1.4928, "step": 71300 }, { "epoch": 0.1, "learning_rate": 4.5543125e-05, "loss": 1.3273, "step": 71310 }, { "epoch": 0.1, "learning_rate": 4.5542500000000007e-05, "loss": 1.65, "step": 71320 }, { "epoch": 0.1, "learning_rate": 4.5541875000000003e-05, "loss": 1.6386, "step": 71330 }, { "epoch": 0.1, "learning_rate": 4.554125e-05, "loss": 1.5721, "step": 71340 }, { "epoch": 0.1, "learning_rate": 4.5540625000000004e-05, "loss": 1.6472, "step": 71350 }, { "epoch": 0.1, "learning_rate": 4.554000000000001e-05, "loss": 1.4734, "step": 71360 }, { "epoch": 0.1, "learning_rate": 4.5539375e-05, "loss": 1.5197, "step": 71370 }, { "epoch": 0.1, "learning_rate": 4.553875e-05, "loss": 1.6938, "step": 71380 }, { "epoch": 0.1, "learning_rate": 4.5538125000000005e-05, "loss": 1.635, "step": 71390 }, { "epoch": 0.1, "learning_rate": 4.55375e-05, "loss": 1.5946, "step": 71400 }, { "epoch": 0.1, "learning_rate": 4.5536875e-05, "loss": 1.6425, "step": 71410 }, { "epoch": 0.1, "learning_rate": 4.553625e-05, "loss": 1.4159, "step": 71420 }, { "epoch": 0.1, "learning_rate": 4.553562500000001e-05, "loss": 1.6523, "step": 71430 }, { "epoch": 0.1, "learning_rate": 4.5535e-05, "loss": 1.5597, "step": 71440 }, { "epoch": 0.1, "learning_rate": 4.5534375e-05, "loss": 1.6297, "step": 71450 }, { "epoch": 0.1, "learning_rate": 4.5533750000000004e-05, "loss": 1.4665, "step": 71460 }, { "epoch": 0.1, "learning_rate": 4.5533125e-05, "loss": 1.6509, "step": 71470 }, { "epoch": 0.1, "learning_rate": 4.55325e-05, "loss": 1.4791, "step": 71480 }, { "epoch": 0.1, "learning_rate": 4.5531875e-05, "loss": 1.3696, "step": 71490 }, { "epoch": 0.1, "learning_rate": 4.5531250000000006e-05, "loss": 1.5212, "step": 71500 }, { "epoch": 0.1, "learning_rate": 4.5530625e-05, "loss": 1.4285, "step": 71510 }, { "epoch": 0.1, "learning_rate": 4.553e-05, "loss": 1.5117, "step": 71520 }, { "epoch": 0.1, "learning_rate": 4.5529375e-05, "loss": 1.6445, "step": 71530 }, { "epoch": 0.1, "learning_rate": 4.552875e-05, "loss": 1.5652, "step": 71540 }, { "epoch": 0.1, "learning_rate": 4.5528125000000004e-05, "loss": 1.5139, "step": 71550 }, { "epoch": 0.1, "learning_rate": 4.55275e-05, "loss": 1.49, "step": 71560 }, { "epoch": 0.1, "learning_rate": 4.5526875000000004e-05, "loss": 1.2643, "step": 71570 }, { "epoch": 0.1, "learning_rate": 4.552625e-05, "loss": 1.5951, "step": 71580 }, { "epoch": 0.1, "learning_rate": 4.5525625e-05, "loss": 1.6278, "step": 71590 }, { "epoch": 0.1, "learning_rate": 4.5525e-05, "loss": 1.7272, "step": 71600 }, { "epoch": 0.1, "learning_rate": 4.5524375000000006e-05, "loss": 1.4457, "step": 71610 }, { "epoch": 0.1, "learning_rate": 4.552375e-05, "loss": 1.6161, "step": 71620 }, { "epoch": 0.1, "learning_rate": 4.5523125e-05, "loss": 1.5015, "step": 71630 }, { "epoch": 0.1, "learning_rate": 4.55225e-05, "loss": 1.8231, "step": 71640 }, { "epoch": 0.1, "learning_rate": 4.5521875e-05, "loss": 1.5597, "step": 71650 }, { "epoch": 0.1, "learning_rate": 4.5521250000000004e-05, "loss": 1.5777, "step": 71660 }, { "epoch": 0.1, "learning_rate": 4.5520625e-05, "loss": 1.6805, "step": 71670 }, { "epoch": 0.1, "learning_rate": 4.5520000000000005e-05, "loss": 1.4503, "step": 71680 }, { "epoch": 0.1, "learning_rate": 4.5519375e-05, "loss": 1.4697, "step": 71690 }, { "epoch": 0.1, "learning_rate": 4.5518750000000005e-05, "loss": 1.5177, "step": 71700 }, { "epoch": 0.1, "learning_rate": 4.5518125e-05, "loss": 1.4809, "step": 71710 }, { "epoch": 0.1, "learning_rate": 4.55175e-05, "loss": 1.7588, "step": 71720 }, { "epoch": 0.1, "learning_rate": 4.5516875e-05, "loss": 1.6972, "step": 71730 }, { "epoch": 0.1, "learning_rate": 4.5516250000000007e-05, "loss": 1.6238, "step": 71740 }, { "epoch": 0.1, "learning_rate": 4.5515625000000004e-05, "loss": 1.5283, "step": 71750 }, { "epoch": 0.1, "learning_rate": 4.5515e-05, "loss": 1.7892, "step": 71760 }, { "epoch": 0.1, "learning_rate": 4.5514375000000004e-05, "loss": 1.4124, "step": 71770 }, { "epoch": 0.1, "learning_rate": 4.551375e-05, "loss": 1.5866, "step": 71780 }, { "epoch": 0.1, "learning_rate": 4.5513125e-05, "loss": 1.9239, "step": 71790 }, { "epoch": 0.1, "learning_rate": 4.55125e-05, "loss": 1.5778, "step": 71800 }, { "epoch": 0.1, "learning_rate": 4.5511875000000005e-05, "loss": 1.4329, "step": 71810 }, { "epoch": 0.1, "learning_rate": 4.551125e-05, "loss": 1.3624, "step": 71820 }, { "epoch": 0.1, "learning_rate": 4.5510625e-05, "loss": 1.5144, "step": 71830 }, { "epoch": 0.1, "learning_rate": 4.551e-05, "loss": 1.5674, "step": 71840 }, { "epoch": 0.1, "learning_rate": 4.550937500000001e-05, "loss": 1.591, "step": 71850 }, { "epoch": 0.1, "learning_rate": 4.550875e-05, "loss": 1.5779, "step": 71860 }, { "epoch": 0.1, "learning_rate": 4.5508125e-05, "loss": 1.541, "step": 71870 }, { "epoch": 0.1, "learning_rate": 4.5507500000000004e-05, "loss": 1.5949, "step": 71880 }, { "epoch": 0.1, "learning_rate": 4.5506875e-05, "loss": 1.5123, "step": 71890 }, { "epoch": 0.1, "learning_rate": 4.550625e-05, "loss": 1.7369, "step": 71900 }, { "epoch": 0.1, "learning_rate": 4.5505625e-05, "loss": 1.3471, "step": 71910 }, { "epoch": 0.1, "learning_rate": 4.5505000000000006e-05, "loss": 1.4403, "step": 71920 }, { "epoch": 0.1, "learning_rate": 4.5504375e-05, "loss": 1.4443, "step": 71930 }, { "epoch": 0.1, "learning_rate": 4.550375e-05, "loss": 1.3498, "step": 71940 }, { "epoch": 0.1, "learning_rate": 4.5503125e-05, "loss": 1.7496, "step": 71950 }, { "epoch": 0.1, "learning_rate": 4.55025e-05, "loss": 1.629, "step": 71960 }, { "epoch": 0.1, "learning_rate": 4.5501875000000004e-05, "loss": 1.6058, "step": 71970 }, { "epoch": 0.1, "learning_rate": 4.550125e-05, "loss": 1.6476, "step": 71980 }, { "epoch": 0.1, "learning_rate": 4.5500625000000004e-05, "loss": 1.379, "step": 71990 }, { "epoch": 0.1, "learning_rate": 4.55e-05, "loss": 1.2526, "step": 72000 }, { "epoch": 0.1, "eval_loss": 1.5058029890060425, "eval_runtime": 1307.7035, "eval_samples_per_second": 44.38, "eval_steps_per_second": 5.548, "step": 72000 }, { "epoch": 0.1, "learning_rate": 4.5499375e-05, "loss": 1.6584, "step": 72010 }, { "epoch": 0.1, "learning_rate": 4.549875e-05, "loss": 1.3807, "step": 72020 }, { "epoch": 0.1, "learning_rate": 4.5498125000000006e-05, "loss": 1.6589, "step": 72030 }, { "epoch": 0.1, "learning_rate": 4.54975e-05, "loss": 1.6172, "step": 72040 }, { "epoch": 0.1, "learning_rate": 4.5496875e-05, "loss": 1.6601, "step": 72050 }, { "epoch": 0.1, "learning_rate": 4.549625e-05, "loss": 1.5768, "step": 72060 }, { "epoch": 0.1, "learning_rate": 4.5495625e-05, "loss": 1.7525, "step": 72070 }, { "epoch": 0.1, "learning_rate": 4.5495000000000004e-05, "loss": 1.7943, "step": 72080 }, { "epoch": 0.1, "learning_rate": 4.5494375e-05, "loss": 1.4854, "step": 72090 }, { "epoch": 0.1, "learning_rate": 4.5493750000000005e-05, "loss": 1.6586, "step": 72100 }, { "epoch": 0.1, "learning_rate": 4.5493125e-05, "loss": 1.6672, "step": 72110 }, { "epoch": 0.1, "learning_rate": 4.5492500000000005e-05, "loss": 1.5614, "step": 72120 }, { "epoch": 0.1, "learning_rate": 4.5491875e-05, "loss": 1.5934, "step": 72130 }, { "epoch": 0.1, "learning_rate": 4.549125e-05, "loss": 1.5161, "step": 72140 }, { "epoch": 0.1, "learning_rate": 4.5490625e-05, "loss": 1.4884, "step": 72150 }, { "epoch": 0.1, "learning_rate": 4.549000000000001e-05, "loss": 1.5748, "step": 72160 }, { "epoch": 0.1, "learning_rate": 4.5489375000000004e-05, "loss": 1.3896, "step": 72170 }, { "epoch": 0.1, "learning_rate": 4.548875e-05, "loss": 1.5748, "step": 72180 }, { "epoch": 0.1, "learning_rate": 4.5488125000000004e-05, "loss": 1.61, "step": 72190 }, { "epoch": 0.1, "learning_rate": 4.54875e-05, "loss": 1.6323, "step": 72200 }, { "epoch": 0.1, "learning_rate": 4.5486875e-05, "loss": 1.4249, "step": 72210 }, { "epoch": 0.1, "learning_rate": 4.548625e-05, "loss": 1.5566, "step": 72220 }, { "epoch": 0.1, "learning_rate": 4.5485625000000005e-05, "loss": 1.2227, "step": 72230 }, { "epoch": 0.1, "learning_rate": 4.5485e-05, "loss": 1.6823, "step": 72240 }, { "epoch": 0.1, "learning_rate": 4.5484375e-05, "loss": 1.6574, "step": 72250 }, { "epoch": 0.1, "learning_rate": 4.548375e-05, "loss": 1.6315, "step": 72260 }, { "epoch": 0.1, "learning_rate": 4.548312500000001e-05, "loss": 1.5348, "step": 72270 }, { "epoch": 0.1, "learning_rate": 4.54825e-05, "loss": 1.7658, "step": 72280 }, { "epoch": 0.1, "learning_rate": 4.5481875e-05, "loss": 1.558, "step": 72290 }, { "epoch": 0.1, "learning_rate": 4.5481250000000004e-05, "loss": 1.6138, "step": 72300 }, { "epoch": 0.1, "learning_rate": 4.5480625e-05, "loss": 1.3842, "step": 72310 }, { "epoch": 0.1, "learning_rate": 4.548e-05, "loss": 1.5409, "step": 72320 }, { "epoch": 0.1, "learning_rate": 4.5479375e-05, "loss": 1.7248, "step": 72330 }, { "epoch": 0.1, "learning_rate": 4.5478750000000006e-05, "loss": 1.6224, "step": 72340 }, { "epoch": 0.1, "learning_rate": 4.5478125e-05, "loss": 1.7116, "step": 72350 }, { "epoch": 0.1, "learning_rate": 4.54775e-05, "loss": 1.5079, "step": 72360 }, { "epoch": 0.1, "learning_rate": 4.5476875e-05, "loss": 1.6776, "step": 72370 }, { "epoch": 0.1, "learning_rate": 4.547625000000001e-05, "loss": 1.604, "step": 72380 }, { "epoch": 0.1, "learning_rate": 4.5475625e-05, "loss": 1.4765, "step": 72390 }, { "epoch": 0.1, "learning_rate": 4.5475e-05, "loss": 1.6767, "step": 72400 }, { "epoch": 0.1, "learning_rate": 4.5474375000000005e-05, "loss": 1.6342, "step": 72410 }, { "epoch": 0.1, "learning_rate": 4.547375e-05, "loss": 1.6775, "step": 72420 }, { "epoch": 0.1, "learning_rate": 4.5473125e-05, "loss": 1.5809, "step": 72430 }, { "epoch": 0.1, "learning_rate": 4.54725e-05, "loss": 1.588, "step": 72440 }, { "epoch": 0.1, "learning_rate": 4.5471875000000006e-05, "loss": 1.6264, "step": 72450 }, { "epoch": 0.1, "learning_rate": 4.547125e-05, "loss": 1.6304, "step": 72460 }, { "epoch": 0.1, "learning_rate": 4.5470625e-05, "loss": 1.6156, "step": 72470 }, { "epoch": 0.1, "learning_rate": 4.5470000000000003e-05, "loss": 1.386, "step": 72480 }, { "epoch": 0.1, "learning_rate": 4.5469375e-05, "loss": 1.5552, "step": 72490 }, { "epoch": 0.1, "learning_rate": 4.5468750000000004e-05, "loss": 1.5799, "step": 72500 }, { "epoch": 0.1, "learning_rate": 4.5468125e-05, "loss": 1.4967, "step": 72510 }, { "epoch": 0.1, "learning_rate": 4.5467500000000005e-05, "loss": 1.4882, "step": 72520 }, { "epoch": 0.1, "learning_rate": 4.5466875e-05, "loss": 1.6499, "step": 72530 }, { "epoch": 0.1, "learning_rate": 4.5466250000000005e-05, "loss": 1.6758, "step": 72540 }, { "epoch": 0.1, "learning_rate": 4.5465625e-05, "loss": 1.5863, "step": 72550 }, { "epoch": 0.1, "learning_rate": 4.5465e-05, "loss": 1.664, "step": 72560 }, { "epoch": 0.1, "learning_rate": 4.5464375e-05, "loss": 1.6902, "step": 72570 }, { "epoch": 0.1, "learning_rate": 4.546375000000001e-05, "loss": 1.5729, "step": 72580 }, { "epoch": 0.1, "learning_rate": 4.5463125000000004e-05, "loss": 1.5406, "step": 72590 }, { "epoch": 0.1, "learning_rate": 4.54625e-05, "loss": 1.6851, "step": 72600 }, { "epoch": 0.1, "learning_rate": 4.5461875000000004e-05, "loss": 1.3705, "step": 72610 }, { "epoch": 0.1, "learning_rate": 4.546125e-05, "loss": 1.7238, "step": 72620 }, { "epoch": 0.1, "learning_rate": 4.5460625e-05, "loss": 1.5173, "step": 72630 }, { "epoch": 0.1, "learning_rate": 4.546e-05, "loss": 1.7012, "step": 72640 }, { "epoch": 0.1, "learning_rate": 4.5459375000000006e-05, "loss": 1.5484, "step": 72650 }, { "epoch": 0.1, "learning_rate": 4.545875e-05, "loss": 1.5563, "step": 72660 }, { "epoch": 0.1, "learning_rate": 4.5458125e-05, "loss": 1.5128, "step": 72670 }, { "epoch": 0.1, "learning_rate": 4.54575e-05, "loss": 1.8026, "step": 72680 }, { "epoch": 0.1, "learning_rate": 4.545687500000001e-05, "loss": 1.4612, "step": 72690 }, { "epoch": 0.1, "learning_rate": 4.545625e-05, "loss": 1.6205, "step": 72700 }, { "epoch": 0.1, "learning_rate": 4.5455625e-05, "loss": 1.6497, "step": 72710 }, { "epoch": 0.1, "learning_rate": 4.5455000000000004e-05, "loss": 1.4197, "step": 72720 }, { "epoch": 0.1, "learning_rate": 4.5454375e-05, "loss": 1.5346, "step": 72730 }, { "epoch": 0.1, "learning_rate": 4.545375e-05, "loss": 1.6116, "step": 72740 }, { "epoch": 0.1, "learning_rate": 4.5453125e-05, "loss": 1.6901, "step": 72750 }, { "epoch": 0.1, "learning_rate": 4.5452500000000006e-05, "loss": 1.6746, "step": 72760 }, { "epoch": 0.1, "learning_rate": 4.5451875e-05, "loss": 1.6845, "step": 72770 }, { "epoch": 0.1, "learning_rate": 4.545125e-05, "loss": 1.4919, "step": 72780 }, { "epoch": 0.1, "learning_rate": 4.5450625e-05, "loss": 1.554, "step": 72790 }, { "epoch": 0.1, "learning_rate": 4.545000000000001e-05, "loss": 1.5952, "step": 72800 }, { "epoch": 0.1, "learning_rate": 4.5449375e-05, "loss": 1.3897, "step": 72810 }, { "epoch": 0.1, "learning_rate": 4.544875e-05, "loss": 1.5179, "step": 72820 }, { "epoch": 0.1, "learning_rate": 4.5448125000000005e-05, "loss": 1.6775, "step": 72830 }, { "epoch": 0.1, "learning_rate": 4.54475e-05, "loss": 1.4919, "step": 72840 }, { "epoch": 0.1, "learning_rate": 4.5446875e-05, "loss": 1.4626, "step": 72850 }, { "epoch": 0.1, "learning_rate": 4.544625e-05, "loss": 1.491, "step": 72860 }, { "epoch": 0.1, "learning_rate": 4.5445625000000006e-05, "loss": 1.7496, "step": 72870 }, { "epoch": 0.1, "learning_rate": 4.5445e-05, "loss": 1.4896, "step": 72880 }, { "epoch": 0.1, "learning_rate": 4.5444375e-05, "loss": 1.5786, "step": 72890 }, { "epoch": 0.1, "learning_rate": 4.5443750000000003e-05, "loss": 1.4572, "step": 72900 }, { "epoch": 0.1, "learning_rate": 4.5443125e-05, "loss": 1.4199, "step": 72910 }, { "epoch": 0.1, "learning_rate": 4.5442500000000004e-05, "loss": 1.5432, "step": 72920 }, { "epoch": 0.1, "learning_rate": 4.5441875e-05, "loss": 1.5624, "step": 72930 }, { "epoch": 0.1, "learning_rate": 4.5441250000000005e-05, "loss": 1.4201, "step": 72940 }, { "epoch": 0.1, "learning_rate": 4.5440625e-05, "loss": 1.624, "step": 72950 }, { "epoch": 0.1, "learning_rate": 4.5440000000000005e-05, "loss": 1.459, "step": 72960 }, { "epoch": 0.1, "learning_rate": 4.5439375e-05, "loss": 1.4219, "step": 72970 }, { "epoch": 0.1, "learning_rate": 4.543875e-05, "loss": 1.462, "step": 72980 }, { "epoch": 0.1, "learning_rate": 4.5438125e-05, "loss": 1.6068, "step": 72990 }, { "epoch": 0.1, "learning_rate": 4.54375e-05, "loss": 1.7231, "step": 73000 }, { "epoch": 0.1, "eval_loss": 1.496966004371643, "eval_runtime": 1322.1122, "eval_samples_per_second": 43.896, "eval_steps_per_second": 5.487, "step": 73000 }, { "epoch": 0.1, "learning_rate": 4.5436875000000004e-05, "loss": 1.5985, "step": 73010 }, { "epoch": 0.1, "learning_rate": 4.543625e-05, "loss": 1.5928, "step": 73020 }, { "epoch": 0.1, "learning_rate": 4.5435625000000004e-05, "loss": 1.4028, "step": 73030 }, { "epoch": 0.1, "learning_rate": 4.5435e-05, "loss": 1.588, "step": 73040 }, { "epoch": 0.1, "learning_rate": 4.5434375e-05, "loss": 1.5167, "step": 73050 }, { "epoch": 0.1, "learning_rate": 4.543375e-05, "loss": 1.652, "step": 73060 }, { "epoch": 0.1, "learning_rate": 4.5433125000000006e-05, "loss": 1.5478, "step": 73070 }, { "epoch": 0.1, "learning_rate": 4.54325e-05, "loss": 1.5438, "step": 73080 }, { "epoch": 0.1, "learning_rate": 4.5431875e-05, "loss": 1.4806, "step": 73090 }, { "epoch": 0.1, "learning_rate": 4.543125e-05, "loss": 1.5553, "step": 73100 }, { "epoch": 0.1, "learning_rate": 4.543062500000001e-05, "loss": 1.7263, "step": 73110 }, { "epoch": 0.1, "learning_rate": 4.543e-05, "loss": 1.4866, "step": 73120 }, { "epoch": 0.1, "learning_rate": 4.5429375e-05, "loss": 1.6659, "step": 73130 }, { "epoch": 0.1, "learning_rate": 4.5428750000000004e-05, "loss": 1.5294, "step": 73140 }, { "epoch": 0.1, "learning_rate": 4.5428125e-05, "loss": 1.4646, "step": 73150 }, { "epoch": 0.1, "learning_rate": 4.54275e-05, "loss": 1.5287, "step": 73160 }, { "epoch": 0.1, "learning_rate": 4.5426875e-05, "loss": 1.7409, "step": 73170 }, { "epoch": 0.1, "learning_rate": 4.5426250000000006e-05, "loss": 1.7476, "step": 73180 }, { "epoch": 0.1, "learning_rate": 4.5425625e-05, "loss": 1.3622, "step": 73190 }, { "epoch": 0.1, "learning_rate": 4.5425e-05, "loss": 1.5594, "step": 73200 }, { "epoch": 0.1, "learning_rate": 4.5424375e-05, "loss": 1.8683, "step": 73210 }, { "epoch": 0.1, "learning_rate": 4.542375000000001e-05, "loss": 1.7691, "step": 73220 }, { "epoch": 0.1, "learning_rate": 4.5423125e-05, "loss": 1.6642, "step": 73230 }, { "epoch": 0.1, "learning_rate": 4.54225e-05, "loss": 1.5737, "step": 73240 }, { "epoch": 0.1, "learning_rate": 4.5421875000000005e-05, "loss": 1.5852, "step": 73250 }, { "epoch": 0.1, "learning_rate": 4.542125e-05, "loss": 1.4958, "step": 73260 }, { "epoch": 0.1, "learning_rate": 4.5420625e-05, "loss": 1.5097, "step": 73270 }, { "epoch": 0.1, "learning_rate": 4.542e-05, "loss": 1.6872, "step": 73280 }, { "epoch": 0.1, "learning_rate": 4.5419375000000006e-05, "loss": 1.647, "step": 73290 }, { "epoch": 0.1, "learning_rate": 4.541875e-05, "loss": 1.4011, "step": 73300 }, { "epoch": 0.1, "learning_rate": 4.5418125e-05, "loss": 1.4084, "step": 73310 }, { "epoch": 0.1, "learning_rate": 4.5417500000000003e-05, "loss": 1.5603, "step": 73320 }, { "epoch": 0.1, "learning_rate": 4.5416875e-05, "loss": 1.761, "step": 73330 }, { "epoch": 0.1, "learning_rate": 4.5416250000000004e-05, "loss": 1.5426, "step": 73340 }, { "epoch": 0.1, "learning_rate": 4.5415625e-05, "loss": 1.4379, "step": 73350 }, { "epoch": 0.1, "learning_rate": 4.5415000000000005e-05, "loss": 1.466, "step": 73360 }, { "epoch": 0.1, "learning_rate": 4.5414375e-05, "loss": 1.5308, "step": 73370 }, { "epoch": 0.1, "learning_rate": 4.5413750000000005e-05, "loss": 1.6223, "step": 73380 }, { "epoch": 0.1, "learning_rate": 4.5413125e-05, "loss": 1.6069, "step": 73390 }, { "epoch": 0.1, "learning_rate": 4.54125e-05, "loss": 1.5596, "step": 73400 }, { "epoch": 0.1, "learning_rate": 4.5411875e-05, "loss": 1.6116, "step": 73410 }, { "epoch": 0.1, "learning_rate": 4.541125e-05, "loss": 1.2996, "step": 73420 }, { "epoch": 0.1, "learning_rate": 4.5410625000000004e-05, "loss": 1.6513, "step": 73430 }, { "epoch": 0.1, "learning_rate": 4.541e-05, "loss": 1.5007, "step": 73440 }, { "epoch": 0.1, "learning_rate": 4.5409375000000004e-05, "loss": 1.7432, "step": 73450 }, { "epoch": 0.1, "learning_rate": 4.540875e-05, "loss": 1.4836, "step": 73460 }, { "epoch": 0.1, "learning_rate": 4.5408125e-05, "loss": 1.4587, "step": 73470 }, { "epoch": 0.1, "learning_rate": 4.54075e-05, "loss": 1.389, "step": 73480 }, { "epoch": 0.1, "learning_rate": 4.5406875000000006e-05, "loss": 1.7704, "step": 73490 }, { "epoch": 0.1, "learning_rate": 4.540625e-05, "loss": 1.5807, "step": 73500 }, { "epoch": 0.1, "learning_rate": 4.5405625e-05, "loss": 1.5972, "step": 73510 }, { "epoch": 0.1, "learning_rate": 4.5405e-05, "loss": 1.4613, "step": 73520 }, { "epoch": 0.1, "learning_rate": 4.540437500000001e-05, "loss": 1.5064, "step": 73530 }, { "epoch": 0.1, "learning_rate": 4.5403750000000004e-05, "loss": 1.433, "step": 73540 }, { "epoch": 0.1, "learning_rate": 4.5403125e-05, "loss": 1.4642, "step": 73550 }, { "epoch": 0.1, "learning_rate": 4.5402500000000004e-05, "loss": 1.6599, "step": 73560 }, { "epoch": 0.1, "learning_rate": 4.540187500000001e-05, "loss": 1.7526, "step": 73570 }, { "epoch": 0.1, "learning_rate": 4.540125e-05, "loss": 1.4966, "step": 73580 }, { "epoch": 0.1, "learning_rate": 4.5400625e-05, "loss": 1.4913, "step": 73590 }, { "epoch": 0.1, "learning_rate": 4.5400000000000006e-05, "loss": 1.5807, "step": 73600 }, { "epoch": 0.1, "learning_rate": 4.5399375e-05, "loss": 1.4273, "step": 73610 }, { "epoch": 0.1, "learning_rate": 4.539875e-05, "loss": 1.7403, "step": 73620 }, { "epoch": 0.1, "learning_rate": 4.5398125e-05, "loss": 1.2729, "step": 73630 }, { "epoch": 0.1, "learning_rate": 4.539750000000001e-05, "loss": 1.7469, "step": 73640 }, { "epoch": 0.1, "learning_rate": 4.5396875e-05, "loss": 1.7865, "step": 73650 }, { "epoch": 0.1, "learning_rate": 4.539625e-05, "loss": 1.5592, "step": 73660 }, { "epoch": 0.1, "learning_rate": 4.5395625000000005e-05, "loss": 1.5754, "step": 73670 }, { "epoch": 0.1, "learning_rate": 4.5395e-05, "loss": 1.5904, "step": 73680 }, { "epoch": 0.1, "learning_rate": 4.5394375e-05, "loss": 1.7246, "step": 73690 }, { "epoch": 0.1, "learning_rate": 4.539375e-05, "loss": 1.7028, "step": 73700 }, { "epoch": 0.1, "learning_rate": 4.5393125000000006e-05, "loss": 1.7066, "step": 73710 }, { "epoch": 0.1, "learning_rate": 4.53925e-05, "loss": 1.7326, "step": 73720 }, { "epoch": 0.1, "learning_rate": 4.5391875e-05, "loss": 1.5134, "step": 73730 }, { "epoch": 0.1, "learning_rate": 4.5391250000000004e-05, "loss": 1.3979, "step": 73740 }, { "epoch": 0.1, "learning_rate": 4.5390625e-05, "loss": 1.523, "step": 73750 }, { "epoch": 0.1, "learning_rate": 4.5390000000000004e-05, "loss": 1.6037, "step": 73760 }, { "epoch": 0.1, "learning_rate": 4.5389375e-05, "loss": 1.4379, "step": 73770 }, { "epoch": 0.1, "learning_rate": 4.5388750000000005e-05, "loss": 1.4179, "step": 73780 }, { "epoch": 0.1, "learning_rate": 4.5388125e-05, "loss": 1.3947, "step": 73790 }, { "epoch": 0.1, "learning_rate": 4.53875e-05, "loss": 1.8091, "step": 73800 }, { "epoch": 0.1, "learning_rate": 4.5386875e-05, "loss": 1.3795, "step": 73810 }, { "epoch": 0.1, "learning_rate": 4.538625e-05, "loss": 1.9347, "step": 73820 }, { "epoch": 0.1, "learning_rate": 4.5385625e-05, "loss": 1.4845, "step": 73830 }, { "epoch": 0.1, "learning_rate": 4.5385e-05, "loss": 1.85, "step": 73840 }, { "epoch": 0.1, "learning_rate": 4.5384375000000004e-05, "loss": 1.385, "step": 73850 }, { "epoch": 0.1, "learning_rate": 4.538375e-05, "loss": 1.5252, "step": 73860 }, { "epoch": 0.1, "learning_rate": 4.5383125000000004e-05, "loss": 1.7967, "step": 73870 }, { "epoch": 0.1, "learning_rate": 4.53825e-05, "loss": 1.6011, "step": 73880 }, { "epoch": 0.1, "learning_rate": 4.5381875e-05, "loss": 1.5302, "step": 73890 }, { "epoch": 0.1, "learning_rate": 4.538125e-05, "loss": 1.505, "step": 73900 }, { "epoch": 0.1, "learning_rate": 4.5380625000000006e-05, "loss": 1.5224, "step": 73910 }, { "epoch": 0.1, "learning_rate": 4.538e-05, "loss": 1.731, "step": 73920 }, { "epoch": 0.1, "learning_rate": 4.5379375e-05, "loss": 1.4907, "step": 73930 }, { "epoch": 0.1, "learning_rate": 4.537875e-05, "loss": 1.6339, "step": 73940 }, { "epoch": 0.1, "learning_rate": 4.537812500000001e-05, "loss": 1.3324, "step": 73950 }, { "epoch": 0.1, "learning_rate": 4.5377500000000004e-05, "loss": 1.5996, "step": 73960 }, { "epoch": 0.1, "learning_rate": 4.5376875e-05, "loss": 1.396, "step": 73970 }, { "epoch": 0.1, "learning_rate": 4.5376250000000005e-05, "loss": 1.7177, "step": 73980 }, { "epoch": 0.1, "learning_rate": 4.537562500000001e-05, "loss": 1.3896, "step": 73990 }, { "epoch": 0.1, "learning_rate": 4.5375e-05, "loss": 1.4594, "step": 74000 }, { "epoch": 0.1, "eval_loss": 1.489876389503479, "eval_runtime": 1331.9723, "eval_samples_per_second": 43.571, "eval_steps_per_second": 5.447, "step": 74000 }, { "epoch": 0.1, "learning_rate": 4.5374375e-05, "loss": 1.471, "step": 74010 }, { "epoch": 0.1, "learning_rate": 4.5373750000000006e-05, "loss": 1.4358, "step": 74020 }, { "epoch": 0.1, "learning_rate": 4.5373125e-05, "loss": 1.4671, "step": 74030 }, { "epoch": 0.1, "learning_rate": 4.53725e-05, "loss": 1.6227, "step": 74040 }, { "epoch": 0.1, "learning_rate": 4.5371875000000003e-05, "loss": 1.4653, "step": 74050 }, { "epoch": 0.1, "learning_rate": 4.537125000000001e-05, "loss": 1.5244, "step": 74060 }, { "epoch": 0.1, "learning_rate": 4.5370625e-05, "loss": 1.411, "step": 74070 }, { "epoch": 0.1, "learning_rate": 4.537e-05, "loss": 1.7056, "step": 74080 }, { "epoch": 0.1, "learning_rate": 4.5369375000000005e-05, "loss": 1.3035, "step": 74090 }, { "epoch": 0.1, "learning_rate": 4.536875e-05, "loss": 1.6755, "step": 74100 }, { "epoch": 0.1, "learning_rate": 4.5368125e-05, "loss": 1.5598, "step": 74110 }, { "epoch": 0.1, "learning_rate": 4.53675e-05, "loss": 1.5038, "step": 74120 }, { "epoch": 0.1, "learning_rate": 4.5366875000000006e-05, "loss": 1.455, "step": 74130 }, { "epoch": 0.1, "learning_rate": 4.536625e-05, "loss": 1.6034, "step": 74140 }, { "epoch": 0.1, "learning_rate": 4.5365625e-05, "loss": 1.4468, "step": 74150 }, { "epoch": 0.1, "learning_rate": 4.5365000000000004e-05, "loss": 1.4699, "step": 74160 }, { "epoch": 0.1, "learning_rate": 4.5364375e-05, "loss": 1.6632, "step": 74170 }, { "epoch": 0.1, "learning_rate": 4.5363750000000004e-05, "loss": 1.5819, "step": 74180 }, { "epoch": 0.1, "learning_rate": 4.5363125e-05, "loss": 1.4896, "step": 74190 }, { "epoch": 0.1, "learning_rate": 4.5362500000000005e-05, "loss": 1.4644, "step": 74200 }, { "epoch": 0.1, "learning_rate": 4.5361875e-05, "loss": 1.7401, "step": 74210 }, { "epoch": 0.1, "learning_rate": 4.536125e-05, "loss": 1.5952, "step": 74220 }, { "epoch": 0.1, "learning_rate": 4.5360625e-05, "loss": 1.6352, "step": 74230 }, { "epoch": 0.1, "learning_rate": 4.536e-05, "loss": 1.3518, "step": 74240 }, { "epoch": 0.1, "learning_rate": 4.5359375e-05, "loss": 1.315, "step": 74250 }, { "epoch": 0.1, "learning_rate": 4.535875e-05, "loss": 1.2869, "step": 74260 }, { "epoch": 0.1, "learning_rate": 4.5358125000000004e-05, "loss": 1.6775, "step": 74270 }, { "epoch": 0.1, "learning_rate": 4.53575e-05, "loss": 1.3972, "step": 74280 }, { "epoch": 0.1, "learning_rate": 4.5356875000000004e-05, "loss": 1.3834, "step": 74290 }, { "epoch": 0.1, "learning_rate": 4.535625e-05, "loss": 1.5637, "step": 74300 }, { "epoch": 0.1, "learning_rate": 4.5355625e-05, "loss": 1.6876, "step": 74310 }, { "epoch": 0.1, "learning_rate": 4.5355e-05, "loss": 1.3153, "step": 74320 }, { "epoch": 0.1, "learning_rate": 4.5354375000000006e-05, "loss": 1.5828, "step": 74330 }, { "epoch": 0.1, "learning_rate": 4.535375e-05, "loss": 1.658, "step": 74340 }, { "epoch": 0.1, "learning_rate": 4.5353125e-05, "loss": 1.5423, "step": 74350 }, { "epoch": 0.1, "learning_rate": 4.53525e-05, "loss": 1.6162, "step": 74360 }, { "epoch": 0.1, "learning_rate": 4.535187500000001e-05, "loss": 1.6284, "step": 74370 }, { "epoch": 0.1, "learning_rate": 4.5351250000000004e-05, "loss": 1.3214, "step": 74380 }, { "epoch": 0.1, "learning_rate": 4.5350625e-05, "loss": 1.7219, "step": 74390 }, { "epoch": 0.1, "learning_rate": 4.5350000000000005e-05, "loss": 1.4012, "step": 74400 }, { "epoch": 0.1, "learning_rate": 4.5349375e-05, "loss": 1.4109, "step": 74410 }, { "epoch": 0.1, "learning_rate": 4.534875e-05, "loss": 1.5848, "step": 74420 }, { "epoch": 0.1, "learning_rate": 4.5348125e-05, "loss": 1.3425, "step": 74430 }, { "epoch": 0.1, "learning_rate": 4.5347500000000006e-05, "loss": 1.7949, "step": 74440 }, { "epoch": 0.1, "learning_rate": 4.5346875e-05, "loss": 1.4912, "step": 74450 }, { "epoch": 0.1, "learning_rate": 4.534625e-05, "loss": 1.7568, "step": 74460 }, { "epoch": 0.1, "learning_rate": 4.5345625000000003e-05, "loss": 1.3854, "step": 74470 }, { "epoch": 0.1, "learning_rate": 4.534500000000001e-05, "loss": 1.7295, "step": 74480 }, { "epoch": 0.1, "learning_rate": 4.5344375e-05, "loss": 1.678, "step": 74490 }, { "epoch": 0.1, "learning_rate": 4.534375e-05, "loss": 1.5209, "step": 74500 }, { "epoch": 0.1, "learning_rate": 4.5343125000000005e-05, "loss": 1.5202, "step": 74510 }, { "epoch": 0.1, "learning_rate": 4.53425e-05, "loss": 1.6458, "step": 74520 }, { "epoch": 0.1, "learning_rate": 4.5341875e-05, "loss": 1.5794, "step": 74530 }, { "epoch": 0.1, "learning_rate": 4.534125e-05, "loss": 1.4303, "step": 74540 }, { "epoch": 0.1, "learning_rate": 4.5340625000000006e-05, "loss": 1.5302, "step": 74550 }, { "epoch": 0.1, "learning_rate": 4.534e-05, "loss": 1.7599, "step": 74560 }, { "epoch": 0.1, "learning_rate": 4.5339375e-05, "loss": 1.5436, "step": 74570 }, { "epoch": 0.1, "learning_rate": 4.5338750000000004e-05, "loss": 1.4779, "step": 74580 }, { "epoch": 0.1, "learning_rate": 4.5338125e-05, "loss": 1.7043, "step": 74590 }, { "epoch": 0.1, "learning_rate": 4.53375e-05, "loss": 1.6229, "step": 74600 }, { "epoch": 0.1, "learning_rate": 4.5336875e-05, "loss": 1.5726, "step": 74610 }, { "epoch": 0.1, "learning_rate": 4.5336250000000005e-05, "loss": 1.5738, "step": 74620 }, { "epoch": 0.1, "learning_rate": 4.5335625e-05, "loss": 1.4927, "step": 74630 }, { "epoch": 0.1, "learning_rate": 4.5335e-05, "loss": 1.5426, "step": 74640 }, { "epoch": 0.1, "learning_rate": 4.5334375e-05, "loss": 1.5709, "step": 74650 }, { "epoch": 0.1, "learning_rate": 4.533375e-05, "loss": 1.3577, "step": 74660 }, { "epoch": 0.1, "learning_rate": 4.5333125e-05, "loss": 1.7743, "step": 74670 }, { "epoch": 0.1, "learning_rate": 4.53325e-05, "loss": 1.3636, "step": 74680 }, { "epoch": 0.1, "learning_rate": 4.5331875000000004e-05, "loss": 1.5563, "step": 74690 }, { "epoch": 0.1, "learning_rate": 4.533125e-05, "loss": 1.6814, "step": 74700 }, { "epoch": 0.1, "learning_rate": 4.5330625000000004e-05, "loss": 1.4098, "step": 74710 }, { "epoch": 0.1, "learning_rate": 4.533e-05, "loss": 1.3797, "step": 74720 }, { "epoch": 0.1, "learning_rate": 4.5329375000000005e-05, "loss": 1.6614, "step": 74730 }, { "epoch": 0.1, "learning_rate": 4.532875e-05, "loss": 1.4114, "step": 74740 }, { "epoch": 0.1, "learning_rate": 4.5328125000000006e-05, "loss": 1.6019, "step": 74750 }, { "epoch": 0.1, "learning_rate": 4.53275e-05, "loss": 1.455, "step": 74760 }, { "epoch": 0.1, "learning_rate": 4.5326875e-05, "loss": 1.4505, "step": 74770 }, { "epoch": 0.1, "learning_rate": 4.532625e-05, "loss": 1.5942, "step": 74780 }, { "epoch": 0.1, "learning_rate": 4.532562500000001e-05, "loss": 1.7348, "step": 74790 }, { "epoch": 0.1, "learning_rate": 4.5325000000000004e-05, "loss": 1.7133, "step": 74800 }, { "epoch": 0.1, "learning_rate": 4.5324375e-05, "loss": 1.6827, "step": 74810 }, { "epoch": 0.1, "learning_rate": 4.5323750000000005e-05, "loss": 1.4858, "step": 74820 }, { "epoch": 0.1, "learning_rate": 4.5323125e-05, "loss": 1.3853, "step": 74830 }, { "epoch": 0.1, "learning_rate": 4.53225e-05, "loss": 1.4434, "step": 74840 }, { "epoch": 0.1, "learning_rate": 4.5321875e-05, "loss": 1.1669, "step": 74850 }, { "epoch": 0.1, "learning_rate": 4.5321250000000006e-05, "loss": 1.5788, "step": 74860 }, { "epoch": 0.1, "learning_rate": 4.5320625e-05, "loss": 1.5014, "step": 74870 }, { "epoch": 0.1, "learning_rate": 4.532e-05, "loss": 1.5629, "step": 74880 }, { "epoch": 0.1, "learning_rate": 4.5319375000000003e-05, "loss": 1.6506, "step": 74890 }, { "epoch": 0.1, "learning_rate": 4.531875000000001e-05, "loss": 1.4576, "step": 74900 }, { "epoch": 0.1, "learning_rate": 4.5318125e-05, "loss": 1.5194, "step": 74910 }, { "epoch": 0.1, "learning_rate": 4.53175e-05, "loss": 1.7932, "step": 74920 }, { "epoch": 0.1, "learning_rate": 4.5316875000000005e-05, "loss": 1.6678, "step": 74930 }, { "epoch": 0.1, "learning_rate": 4.531625e-05, "loss": 1.4316, "step": 74940 }, { "epoch": 0.1, "learning_rate": 4.5315625e-05, "loss": 1.3737, "step": 74950 }, { "epoch": 0.1, "learning_rate": 4.5315e-05, "loss": 1.6283, "step": 74960 }, { "epoch": 0.1, "learning_rate": 4.5314375000000006e-05, "loss": 1.5722, "step": 74970 }, { "epoch": 0.1, "learning_rate": 4.531375e-05, "loss": 1.4739, "step": 74980 }, { "epoch": 0.1, "learning_rate": 4.5313125e-05, "loss": 1.6173, "step": 74990 }, { "epoch": 0.1, "learning_rate": 4.5312500000000004e-05, "loss": 1.4377, "step": 75000 }, { "epoch": 0.1, "eval_loss": 1.4931584596633911, "eval_runtime": 1293.0519, "eval_samples_per_second": 44.883, "eval_steps_per_second": 5.611, "step": 75000 }, { "epoch": 0.1, "learning_rate": 4.5311875e-05, "loss": 1.4699, "step": 75010 }, { "epoch": 0.1, "learning_rate": 4.531125e-05, "loss": 1.6224, "step": 75020 }, { "epoch": 0.1, "learning_rate": 4.5310625e-05, "loss": 1.5171, "step": 75030 }, { "epoch": 0.1, "learning_rate": 4.5310000000000005e-05, "loss": 1.5114, "step": 75040 }, { "epoch": 0.1, "learning_rate": 4.5309375e-05, "loss": 1.6217, "step": 75050 }, { "epoch": 0.1, "learning_rate": 4.530875e-05, "loss": 1.6084, "step": 75060 }, { "epoch": 0.1, "learning_rate": 4.5308125e-05, "loss": 1.8606, "step": 75070 }, { "epoch": 0.1, "learning_rate": 4.53075e-05, "loss": 1.6282, "step": 75080 }, { "epoch": 0.1, "learning_rate": 4.5306875e-05, "loss": 1.5298, "step": 75090 }, { "epoch": 0.1, "learning_rate": 4.530625e-05, "loss": 1.7361, "step": 75100 }, { "epoch": 0.1, "learning_rate": 4.5305625000000004e-05, "loss": 1.4468, "step": 75110 }, { "epoch": 0.1, "learning_rate": 4.5305e-05, "loss": 1.5765, "step": 75120 }, { "epoch": 0.1, "learning_rate": 4.5304375000000004e-05, "loss": 1.5352, "step": 75130 }, { "epoch": 0.1, "learning_rate": 4.530375e-05, "loss": 1.4281, "step": 75140 }, { "epoch": 0.1, "learning_rate": 4.5303125000000005e-05, "loss": 1.5058, "step": 75150 }, { "epoch": 0.1, "learning_rate": 4.53025e-05, "loss": 1.5199, "step": 75160 }, { "epoch": 0.1, "learning_rate": 4.5301875000000006e-05, "loss": 1.6237, "step": 75170 }, { "epoch": 0.1, "learning_rate": 4.530125e-05, "loss": 1.7839, "step": 75180 }, { "epoch": 0.1, "learning_rate": 4.5300625e-05, "loss": 1.9657, "step": 75190 }, { "epoch": 0.1, "learning_rate": 4.53e-05, "loss": 1.8391, "step": 75200 }, { "epoch": 0.1, "learning_rate": 4.5299375e-05, "loss": 1.3603, "step": 75210 }, { "epoch": 0.1, "learning_rate": 4.5298750000000004e-05, "loss": 1.5816, "step": 75220 }, { "epoch": 0.1, "learning_rate": 4.5298125e-05, "loss": 1.6436, "step": 75230 }, { "epoch": 0.1, "learning_rate": 4.5297500000000005e-05, "loss": 1.4479, "step": 75240 }, { "epoch": 0.1, "learning_rate": 4.5296875e-05, "loss": 1.6983, "step": 75250 }, { "epoch": 0.1, "learning_rate": 4.529625e-05, "loss": 1.2616, "step": 75260 }, { "epoch": 0.1, "learning_rate": 4.5295625e-05, "loss": 1.2814, "step": 75270 }, { "epoch": 0.1, "learning_rate": 4.5295000000000006e-05, "loss": 1.3255, "step": 75280 }, { "epoch": 0.1, "learning_rate": 4.5294375e-05, "loss": 1.5893, "step": 75290 }, { "epoch": 0.1, "learning_rate": 4.529375e-05, "loss": 1.6481, "step": 75300 }, { "epoch": 0.1, "learning_rate": 4.5293125000000004e-05, "loss": 1.6683, "step": 75310 }, { "epoch": 0.1, "learning_rate": 4.529250000000001e-05, "loss": 1.686, "step": 75320 }, { "epoch": 0.1, "learning_rate": 4.5291875e-05, "loss": 1.457, "step": 75330 }, { "epoch": 0.1, "learning_rate": 4.529125e-05, "loss": 1.5528, "step": 75340 }, { "epoch": 0.1, "learning_rate": 4.5290625000000005e-05, "loss": 1.495, "step": 75350 }, { "epoch": 0.1, "learning_rate": 4.529e-05, "loss": 1.3349, "step": 75360 }, { "epoch": 0.1, "learning_rate": 4.5289375e-05, "loss": 1.4552, "step": 75370 }, { "epoch": 0.1, "learning_rate": 4.528875e-05, "loss": 1.6194, "step": 75380 }, { "epoch": 0.1, "learning_rate": 4.5288125000000006e-05, "loss": 1.7911, "step": 75390 }, { "epoch": 0.1, "learning_rate": 4.52875e-05, "loss": 1.4719, "step": 75400 }, { "epoch": 0.11, "learning_rate": 4.5286875e-05, "loss": 1.4212, "step": 75410 }, { "epoch": 0.11, "learning_rate": 4.5286250000000004e-05, "loss": 1.4591, "step": 75420 }, { "epoch": 0.11, "learning_rate": 4.5285625e-05, "loss": 1.5616, "step": 75430 }, { "epoch": 0.11, "learning_rate": 4.5285e-05, "loss": 1.4841, "step": 75440 }, { "epoch": 0.11, "learning_rate": 4.5284375e-05, "loss": 1.5498, "step": 75450 }, { "epoch": 0.11, "learning_rate": 4.5283750000000005e-05, "loss": 1.4914, "step": 75460 }, { "epoch": 0.11, "learning_rate": 4.5283125e-05, "loss": 1.6396, "step": 75470 }, { "epoch": 0.11, "learning_rate": 4.52825e-05, "loss": 1.6419, "step": 75480 }, { "epoch": 0.11, "learning_rate": 4.5281875e-05, "loss": 1.4054, "step": 75490 }, { "epoch": 0.11, "learning_rate": 4.528125e-05, "loss": 1.3728, "step": 75500 }, { "epoch": 0.11, "learning_rate": 4.5280625e-05, "loss": 1.7303, "step": 75510 }, { "epoch": 0.11, "learning_rate": 4.528e-05, "loss": 1.6645, "step": 75520 }, { "epoch": 0.11, "learning_rate": 4.5279375000000004e-05, "loss": 1.6299, "step": 75530 }, { "epoch": 0.11, "learning_rate": 4.527875e-05, "loss": 1.6828, "step": 75540 }, { "epoch": 0.11, "learning_rate": 4.5278125000000005e-05, "loss": 1.7203, "step": 75550 }, { "epoch": 0.11, "learning_rate": 4.52775e-05, "loss": 1.6014, "step": 75560 }, { "epoch": 0.11, "learning_rate": 4.5276875000000005e-05, "loss": 1.4565, "step": 75570 }, { "epoch": 0.11, "learning_rate": 4.527625e-05, "loss": 1.397, "step": 75580 }, { "epoch": 0.11, "learning_rate": 4.5275625000000006e-05, "loss": 1.6578, "step": 75590 }, { "epoch": 0.11, "learning_rate": 4.5275e-05, "loss": 1.5253, "step": 75600 }, { "epoch": 0.11, "learning_rate": 4.5274375e-05, "loss": 1.7021, "step": 75610 }, { "epoch": 0.11, "learning_rate": 4.5273750000000003e-05, "loss": 1.429, "step": 75620 }, { "epoch": 0.11, "learning_rate": 4.5273125e-05, "loss": 1.5568, "step": 75630 }, { "epoch": 0.11, "learning_rate": 4.5272500000000004e-05, "loss": 1.5443, "step": 75640 }, { "epoch": 0.11, "learning_rate": 4.5271875e-05, "loss": 1.5104, "step": 75650 }, { "epoch": 0.11, "learning_rate": 4.5271250000000005e-05, "loss": 1.4289, "step": 75660 }, { "epoch": 0.11, "learning_rate": 4.5270625e-05, "loss": 1.5201, "step": 75670 }, { "epoch": 0.11, "learning_rate": 4.527e-05, "loss": 1.4312, "step": 75680 }, { "epoch": 0.11, "learning_rate": 4.5269375e-05, "loss": 1.4184, "step": 75690 }, { "epoch": 0.11, "learning_rate": 4.5268750000000006e-05, "loss": 1.6401, "step": 75700 }, { "epoch": 0.11, "learning_rate": 4.5268125e-05, "loss": 1.623, "step": 75710 }, { "epoch": 0.11, "learning_rate": 4.52675e-05, "loss": 1.3239, "step": 75720 }, { "epoch": 0.11, "learning_rate": 4.5266875000000004e-05, "loss": 1.4291, "step": 75730 }, { "epoch": 0.11, "learning_rate": 4.526625000000001e-05, "loss": 1.5762, "step": 75740 }, { "epoch": 0.11, "learning_rate": 4.5265625e-05, "loss": 1.7294, "step": 75750 }, { "epoch": 0.11, "learning_rate": 4.5265e-05, "loss": 1.7006, "step": 75760 }, { "epoch": 0.11, "learning_rate": 4.5264375000000005e-05, "loss": 1.5424, "step": 75770 }, { "epoch": 0.11, "learning_rate": 4.526375e-05, "loss": 1.5721, "step": 75780 }, { "epoch": 0.11, "learning_rate": 4.5263125e-05, "loss": 1.49, "step": 75790 }, { "epoch": 0.11, "learning_rate": 4.52625e-05, "loss": 1.5769, "step": 75800 }, { "epoch": 0.11, "learning_rate": 4.5261875000000006e-05, "loss": 1.5405, "step": 75810 }, { "epoch": 0.11, "learning_rate": 4.5261249999999996e-05, "loss": 1.6946, "step": 75820 }, { "epoch": 0.11, "learning_rate": 4.5260625e-05, "loss": 1.7455, "step": 75830 }, { "epoch": 0.11, "learning_rate": 4.5260000000000004e-05, "loss": 1.377, "step": 75840 }, { "epoch": 0.11, "learning_rate": 4.5259375e-05, "loss": 1.7167, "step": 75850 }, { "epoch": 0.11, "learning_rate": 4.525875e-05, "loss": 1.3799, "step": 75860 }, { "epoch": 0.11, "learning_rate": 4.5258125e-05, "loss": 1.5871, "step": 75870 }, { "epoch": 0.11, "learning_rate": 4.5257500000000005e-05, "loss": 1.5229, "step": 75880 }, { "epoch": 0.11, "learning_rate": 4.5256875e-05, "loss": 1.3697, "step": 75890 }, { "epoch": 0.11, "learning_rate": 4.525625e-05, "loss": 1.4628, "step": 75900 }, { "epoch": 0.11, "learning_rate": 4.5255625e-05, "loss": 1.4211, "step": 75910 }, { "epoch": 0.11, "learning_rate": 4.5255000000000006e-05, "loss": 1.7544, "step": 75920 }, { "epoch": 0.11, "learning_rate": 4.5254375e-05, "loss": 1.4932, "step": 75930 }, { "epoch": 0.11, "learning_rate": 4.525375e-05, "loss": 1.5622, "step": 75940 }, { "epoch": 0.11, "learning_rate": 4.5253125000000004e-05, "loss": 1.536, "step": 75950 }, { "epoch": 0.11, "learning_rate": 4.52525e-05, "loss": 1.696, "step": 75960 }, { "epoch": 0.11, "learning_rate": 4.5251875000000005e-05, "loss": 1.596, "step": 75970 }, { "epoch": 0.11, "learning_rate": 4.525125e-05, "loss": 1.4995, "step": 75980 }, { "epoch": 0.11, "learning_rate": 4.5250625000000005e-05, "loss": 1.552, "step": 75990 }, { "epoch": 0.11, "learning_rate": 4.525e-05, "loss": 1.5265, "step": 76000 }, { "epoch": 0.11, "eval_loss": 1.485759973526001, "eval_runtime": 1351.2006, "eval_samples_per_second": 42.951, "eval_steps_per_second": 5.369, "step": 76000 }, { "epoch": 0.11, "learning_rate": 4.5249375e-05, "loss": 1.5598, "step": 76010 }, { "epoch": 0.11, "learning_rate": 4.524875e-05, "loss": 1.5487, "step": 76020 }, { "epoch": 0.11, "learning_rate": 4.5248125e-05, "loss": 1.426, "step": 76030 }, { "epoch": 0.11, "learning_rate": 4.5247500000000003e-05, "loss": 1.5897, "step": 76040 }, { "epoch": 0.11, "learning_rate": 4.5246875e-05, "loss": 1.4591, "step": 76050 }, { "epoch": 0.11, "learning_rate": 4.5246250000000004e-05, "loss": 1.5382, "step": 76060 }, { "epoch": 0.11, "learning_rate": 4.5245625e-05, "loss": 1.48, "step": 76070 }, { "epoch": 0.11, "learning_rate": 4.5245000000000005e-05, "loss": 1.5638, "step": 76080 }, { "epoch": 0.11, "learning_rate": 4.5244375e-05, "loss": 1.5981, "step": 76090 }, { "epoch": 0.11, "learning_rate": 4.524375e-05, "loss": 1.5286, "step": 76100 }, { "epoch": 0.11, "learning_rate": 4.5243125e-05, "loss": 1.44, "step": 76110 }, { "epoch": 0.11, "learning_rate": 4.5242500000000006e-05, "loss": 1.4975, "step": 76120 }, { "epoch": 0.11, "learning_rate": 4.5241875e-05, "loss": 1.2789, "step": 76130 }, { "epoch": 0.11, "learning_rate": 4.524125e-05, "loss": 1.3639, "step": 76140 }, { "epoch": 0.11, "learning_rate": 4.5240625000000004e-05, "loss": 1.811, "step": 76150 }, { "epoch": 0.11, "learning_rate": 4.524000000000001e-05, "loss": 1.5694, "step": 76160 }, { "epoch": 0.11, "learning_rate": 4.5239375e-05, "loss": 1.6747, "step": 76170 }, { "epoch": 0.11, "learning_rate": 4.523875e-05, "loss": 1.7481, "step": 76180 }, { "epoch": 0.11, "learning_rate": 4.5238125000000005e-05, "loss": 1.6183, "step": 76190 }, { "epoch": 0.11, "learning_rate": 4.52375e-05, "loss": 1.646, "step": 76200 }, { "epoch": 0.11, "learning_rate": 4.5236875e-05, "loss": 1.3972, "step": 76210 }, { "epoch": 0.11, "learning_rate": 4.523625e-05, "loss": 1.4897, "step": 76220 }, { "epoch": 0.11, "learning_rate": 4.5235625000000006e-05, "loss": 1.6536, "step": 76230 }, { "epoch": 0.11, "learning_rate": 4.5234999999999996e-05, "loss": 1.5243, "step": 76240 }, { "epoch": 0.11, "learning_rate": 4.5234375e-05, "loss": 1.4177, "step": 76250 }, { "epoch": 0.11, "learning_rate": 4.5233750000000004e-05, "loss": 1.6819, "step": 76260 }, { "epoch": 0.11, "learning_rate": 4.5233125e-05, "loss": 1.4829, "step": 76270 }, { "epoch": 0.11, "learning_rate": 4.52325e-05, "loss": 1.4025, "step": 76280 }, { "epoch": 0.11, "learning_rate": 4.5231875e-05, "loss": 1.4986, "step": 76290 }, { "epoch": 0.11, "learning_rate": 4.5231250000000005e-05, "loss": 1.6641, "step": 76300 }, { "epoch": 0.11, "learning_rate": 4.5230625e-05, "loss": 1.4998, "step": 76310 }, { "epoch": 0.11, "learning_rate": 4.523e-05, "loss": 1.5252, "step": 76320 }, { "epoch": 0.11, "learning_rate": 4.5229375e-05, "loss": 1.5023, "step": 76330 }, { "epoch": 0.11, "learning_rate": 4.5228750000000006e-05, "loss": 1.6164, "step": 76340 }, { "epoch": 0.11, "learning_rate": 4.5228125e-05, "loss": 1.5575, "step": 76350 }, { "epoch": 0.11, "learning_rate": 4.52275e-05, "loss": 1.6185, "step": 76360 }, { "epoch": 0.11, "learning_rate": 4.5226875000000004e-05, "loss": 1.6431, "step": 76370 }, { "epoch": 0.11, "learning_rate": 4.522625e-05, "loss": 1.6238, "step": 76380 }, { "epoch": 0.11, "learning_rate": 4.5225625000000005e-05, "loss": 1.7328, "step": 76390 }, { "epoch": 0.11, "learning_rate": 4.5225e-05, "loss": 1.402, "step": 76400 }, { "epoch": 0.11, "learning_rate": 4.5224375000000005e-05, "loss": 1.7904, "step": 76410 }, { "epoch": 0.11, "learning_rate": 4.522375e-05, "loss": 1.4618, "step": 76420 }, { "epoch": 0.11, "learning_rate": 4.5223125e-05, "loss": 1.7296, "step": 76430 }, { "epoch": 0.11, "learning_rate": 4.52225e-05, "loss": 1.4069, "step": 76440 }, { "epoch": 0.11, "learning_rate": 4.5221875e-05, "loss": 1.5214, "step": 76450 }, { "epoch": 0.11, "learning_rate": 4.5221250000000003e-05, "loss": 1.4813, "step": 76460 }, { "epoch": 0.11, "learning_rate": 4.5220625e-05, "loss": 1.6543, "step": 76470 }, { "epoch": 0.11, "learning_rate": 4.5220000000000004e-05, "loss": 1.5524, "step": 76480 }, { "epoch": 0.11, "learning_rate": 4.5219375e-05, "loss": 1.5451, "step": 76490 }, { "epoch": 0.11, "learning_rate": 4.5218750000000005e-05, "loss": 1.619, "step": 76500 }, { "epoch": 0.11, "learning_rate": 4.5218125e-05, "loss": 1.568, "step": 76510 }, { "epoch": 0.11, "learning_rate": 4.52175e-05, "loss": 1.4957, "step": 76520 }, { "epoch": 0.11, "learning_rate": 4.5216875e-05, "loss": 1.4834, "step": 76530 }, { "epoch": 0.11, "learning_rate": 4.5216250000000006e-05, "loss": 1.5652, "step": 76540 }, { "epoch": 0.11, "learning_rate": 4.5215625e-05, "loss": 1.6004, "step": 76550 }, { "epoch": 0.11, "learning_rate": 4.5215e-05, "loss": 1.3999, "step": 76560 }, { "epoch": 0.11, "learning_rate": 4.5214375000000004e-05, "loss": 1.4767, "step": 76570 }, { "epoch": 0.11, "learning_rate": 4.521375000000001e-05, "loss": 1.5357, "step": 76580 }, { "epoch": 0.11, "learning_rate": 4.5213125e-05, "loss": 1.4899, "step": 76590 }, { "epoch": 0.11, "learning_rate": 4.52125e-05, "loss": 1.7949, "step": 76600 }, { "epoch": 0.11, "learning_rate": 4.5211875000000005e-05, "loss": 1.712, "step": 76610 }, { "epoch": 0.11, "learning_rate": 4.521125e-05, "loss": 1.4269, "step": 76620 }, { "epoch": 0.11, "learning_rate": 4.5210625e-05, "loss": 1.6044, "step": 76630 }, { "epoch": 0.11, "learning_rate": 4.521e-05, "loss": 1.6433, "step": 76640 }, { "epoch": 0.11, "learning_rate": 4.5209375000000006e-05, "loss": 1.438, "step": 76650 }, { "epoch": 0.11, "learning_rate": 4.5208749999999996e-05, "loss": 1.4917, "step": 76660 }, { "epoch": 0.11, "learning_rate": 4.5208125e-05, "loss": 1.5938, "step": 76670 }, { "epoch": 0.11, "learning_rate": 4.5207500000000004e-05, "loss": 1.4557, "step": 76680 }, { "epoch": 0.11, "learning_rate": 4.5206875e-05, "loss": 1.6333, "step": 76690 }, { "epoch": 0.11, "learning_rate": 4.520625e-05, "loss": 1.6149, "step": 76700 }, { "epoch": 0.11, "learning_rate": 4.5205625e-05, "loss": 1.5128, "step": 76710 }, { "epoch": 0.11, "learning_rate": 4.5205000000000005e-05, "loss": 1.5525, "step": 76720 }, { "epoch": 0.11, "learning_rate": 4.5204375e-05, "loss": 1.5498, "step": 76730 }, { "epoch": 0.11, "learning_rate": 4.520375e-05, "loss": 1.4698, "step": 76740 }, { "epoch": 0.11, "learning_rate": 4.5203125e-05, "loss": 1.5633, "step": 76750 }, { "epoch": 0.11, "learning_rate": 4.5202500000000006e-05, "loss": 1.5962, "step": 76760 }, { "epoch": 0.11, "learning_rate": 4.5201875e-05, "loss": 1.5472, "step": 76770 }, { "epoch": 0.11, "learning_rate": 4.520125e-05, "loss": 1.6432, "step": 76780 }, { "epoch": 0.11, "learning_rate": 4.5200625000000004e-05, "loss": 1.4655, "step": 76790 }, { "epoch": 0.11, "learning_rate": 4.52e-05, "loss": 1.6629, "step": 76800 }, { "epoch": 0.11, "learning_rate": 4.5199375000000005e-05, "loss": 1.5373, "step": 76810 }, { "epoch": 0.11, "learning_rate": 4.519875e-05, "loss": 1.6267, "step": 76820 }, { "epoch": 0.11, "learning_rate": 4.5198125000000005e-05, "loss": 1.6006, "step": 76830 }, { "epoch": 0.11, "learning_rate": 4.51975e-05, "loss": 1.4494, "step": 76840 }, { "epoch": 0.11, "learning_rate": 4.5196875e-05, "loss": 1.6212, "step": 76850 }, { "epoch": 0.11, "learning_rate": 4.519625e-05, "loss": 1.6062, "step": 76860 }, { "epoch": 0.11, "learning_rate": 4.5195625e-05, "loss": 1.479, "step": 76870 }, { "epoch": 0.11, "learning_rate": 4.5195000000000004e-05, "loss": 1.5021, "step": 76880 }, { "epoch": 0.11, "learning_rate": 4.5194375e-05, "loss": 1.5572, "step": 76890 }, { "epoch": 0.11, "learning_rate": 4.5193750000000004e-05, "loss": 1.4102, "step": 76900 }, { "epoch": 0.11, "learning_rate": 4.5193125e-05, "loss": 1.4715, "step": 76910 }, { "epoch": 0.11, "learning_rate": 4.5192500000000005e-05, "loss": 1.6072, "step": 76920 }, { "epoch": 0.11, "learning_rate": 4.5191875e-05, "loss": 1.566, "step": 76930 }, { "epoch": 0.11, "learning_rate": 4.519125e-05, "loss": 1.6733, "step": 76940 }, { "epoch": 0.11, "learning_rate": 4.5190625e-05, "loss": 1.471, "step": 76950 }, { "epoch": 0.11, "learning_rate": 4.5190000000000006e-05, "loss": 1.5441, "step": 76960 }, { "epoch": 0.11, "learning_rate": 4.5189375e-05, "loss": 1.6494, "step": 76970 }, { "epoch": 0.11, "learning_rate": 4.518875e-05, "loss": 1.664, "step": 76980 }, { "epoch": 0.11, "learning_rate": 4.5188125000000004e-05, "loss": 1.4867, "step": 76990 }, { "epoch": 0.11, "learning_rate": 4.518750000000001e-05, "loss": 1.5169, "step": 77000 }, { "epoch": 0.11, "eval_loss": 1.4865235090255737, "eval_runtime": 1299.2226, "eval_samples_per_second": 44.67, "eval_steps_per_second": 5.584, "step": 77000 }, { "epoch": 0.11, "learning_rate": 4.5186875e-05, "loss": 1.6225, "step": 77010 }, { "epoch": 0.11, "learning_rate": 4.518625e-05, "loss": 1.5822, "step": 77020 }, { "epoch": 0.11, "learning_rate": 4.5185625000000005e-05, "loss": 1.6436, "step": 77030 }, { "epoch": 0.11, "learning_rate": 4.5185e-05, "loss": 1.4813, "step": 77040 }, { "epoch": 0.11, "learning_rate": 4.5184375e-05, "loss": 1.6149, "step": 77050 }, { "epoch": 0.11, "learning_rate": 4.518375e-05, "loss": 1.5696, "step": 77060 }, { "epoch": 0.11, "learning_rate": 4.5183125000000006e-05, "loss": 1.4922, "step": 77070 }, { "epoch": 0.11, "learning_rate": 4.51825e-05, "loss": 1.5413, "step": 77080 }, { "epoch": 0.11, "learning_rate": 4.5181875e-05, "loss": 1.5195, "step": 77090 }, { "epoch": 0.11, "learning_rate": 4.5181250000000004e-05, "loss": 1.7513, "step": 77100 }, { "epoch": 0.11, "learning_rate": 4.518062500000001e-05, "loss": 1.4232, "step": 77110 }, { "epoch": 0.11, "learning_rate": 4.518e-05, "loss": 1.4846, "step": 77120 }, { "epoch": 0.11, "learning_rate": 4.5179375e-05, "loss": 1.2985, "step": 77130 }, { "epoch": 0.11, "learning_rate": 4.5178750000000005e-05, "loss": 1.3909, "step": 77140 }, { "epoch": 0.11, "learning_rate": 4.5178125e-05, "loss": 1.6648, "step": 77150 }, { "epoch": 0.11, "learning_rate": 4.51775e-05, "loss": 1.5601, "step": 77160 }, { "epoch": 0.11, "learning_rate": 4.5176875e-05, "loss": 1.5136, "step": 77170 }, { "epoch": 0.11, "learning_rate": 4.5176250000000006e-05, "loss": 1.5255, "step": 77180 }, { "epoch": 0.11, "learning_rate": 4.5175625e-05, "loss": 1.5524, "step": 77190 }, { "epoch": 0.11, "learning_rate": 4.5175e-05, "loss": 1.5032, "step": 77200 }, { "epoch": 0.11, "learning_rate": 4.5174375000000004e-05, "loss": 1.4828, "step": 77210 }, { "epoch": 0.11, "learning_rate": 4.517375e-05, "loss": 1.3803, "step": 77220 }, { "epoch": 0.11, "learning_rate": 4.5173125e-05, "loss": 1.7339, "step": 77230 }, { "epoch": 0.11, "learning_rate": 4.51725e-05, "loss": 1.7287, "step": 77240 }, { "epoch": 0.11, "learning_rate": 4.5171875000000005e-05, "loss": 1.5512, "step": 77250 }, { "epoch": 0.11, "learning_rate": 4.517125e-05, "loss": 1.3324, "step": 77260 }, { "epoch": 0.11, "learning_rate": 4.5170625e-05, "loss": 1.5226, "step": 77270 }, { "epoch": 0.11, "learning_rate": 4.517e-05, "loss": 1.519, "step": 77280 }, { "epoch": 0.11, "learning_rate": 4.5169375e-05, "loss": 1.5167, "step": 77290 }, { "epoch": 0.11, "learning_rate": 4.5168750000000004e-05, "loss": 1.5265, "step": 77300 }, { "epoch": 0.11, "learning_rate": 4.5168125e-05, "loss": 1.5503, "step": 77310 }, { "epoch": 0.11, "learning_rate": 4.5167500000000004e-05, "loss": 1.2883, "step": 77320 }, { "epoch": 0.11, "learning_rate": 4.5166875e-05, "loss": 1.5241, "step": 77330 }, { "epoch": 0.11, "learning_rate": 4.5166250000000005e-05, "loss": 1.6402, "step": 77340 }, { "epoch": 0.11, "learning_rate": 4.5165625e-05, "loss": 1.5472, "step": 77350 }, { "epoch": 0.11, "learning_rate": 4.5165e-05, "loss": 1.4136, "step": 77360 }, { "epoch": 0.11, "learning_rate": 4.5164375e-05, "loss": 1.5459, "step": 77370 }, { "epoch": 0.11, "learning_rate": 4.5163750000000006e-05, "loss": 1.4961, "step": 77380 }, { "epoch": 0.11, "learning_rate": 4.5163125e-05, "loss": 1.4124, "step": 77390 }, { "epoch": 0.11, "learning_rate": 4.51625e-05, "loss": 1.6959, "step": 77400 }, { "epoch": 0.11, "learning_rate": 4.5161875000000004e-05, "loss": 1.3851, "step": 77410 }, { "epoch": 0.11, "learning_rate": 4.516125e-05, "loss": 1.6709, "step": 77420 }, { "epoch": 0.11, "learning_rate": 4.5160625e-05, "loss": 1.5267, "step": 77430 }, { "epoch": 0.11, "learning_rate": 4.516e-05, "loss": 1.6933, "step": 77440 }, { "epoch": 0.11, "learning_rate": 4.5159375000000005e-05, "loss": 1.5495, "step": 77450 }, { "epoch": 0.11, "learning_rate": 4.515875e-05, "loss": 1.6607, "step": 77460 }, { "epoch": 0.11, "learning_rate": 4.5158125e-05, "loss": 1.6767, "step": 77470 }, { "epoch": 0.11, "learning_rate": 4.51575e-05, "loss": 1.6427, "step": 77480 }, { "epoch": 0.11, "learning_rate": 4.5156875000000006e-05, "loss": 1.5327, "step": 77490 }, { "epoch": 0.11, "learning_rate": 4.515625e-05, "loss": 1.3439, "step": 77500 }, { "epoch": 0.11, "learning_rate": 4.5155625e-05, "loss": 1.6045, "step": 77510 }, { "epoch": 0.11, "learning_rate": 4.5155000000000004e-05, "loss": 1.566, "step": 77520 }, { "epoch": 0.11, "learning_rate": 4.515437500000001e-05, "loss": 1.5045, "step": 77530 }, { "epoch": 0.11, "learning_rate": 4.515375e-05, "loss": 1.462, "step": 77540 }, { "epoch": 0.11, "learning_rate": 4.5153125e-05, "loss": 1.4168, "step": 77550 }, { "epoch": 0.11, "learning_rate": 4.5152500000000005e-05, "loss": 1.4904, "step": 77560 }, { "epoch": 0.11, "learning_rate": 4.5151875e-05, "loss": 1.5449, "step": 77570 }, { "epoch": 0.11, "learning_rate": 4.515125e-05, "loss": 1.613, "step": 77580 }, { "epoch": 0.11, "learning_rate": 4.5150625e-05, "loss": 1.7562, "step": 77590 }, { "epoch": 0.11, "learning_rate": 4.5150000000000006e-05, "loss": 1.5423, "step": 77600 }, { "epoch": 0.11, "learning_rate": 4.5149375000000003e-05, "loss": 1.5656, "step": 77610 }, { "epoch": 0.11, "learning_rate": 4.514875e-05, "loss": 1.4567, "step": 77620 }, { "epoch": 0.11, "learning_rate": 4.5148125000000004e-05, "loss": 1.5306, "step": 77630 }, { "epoch": 0.11, "learning_rate": 4.51475e-05, "loss": 1.6924, "step": 77640 }, { "epoch": 0.11, "learning_rate": 4.5146875e-05, "loss": 1.5137, "step": 77650 }, { "epoch": 0.11, "learning_rate": 4.514625e-05, "loss": 1.5516, "step": 77660 }, { "epoch": 0.11, "learning_rate": 4.5145625000000005e-05, "loss": 1.4249, "step": 77670 }, { "epoch": 0.11, "learning_rate": 4.5145e-05, "loss": 1.4294, "step": 77680 }, { "epoch": 0.11, "learning_rate": 4.5144375e-05, "loss": 1.3788, "step": 77690 }, { "epoch": 0.11, "learning_rate": 4.514375e-05, "loss": 1.5257, "step": 77700 }, { "epoch": 0.11, "learning_rate": 4.5143125e-05, "loss": 1.448, "step": 77710 }, { "epoch": 0.11, "learning_rate": 4.5142500000000004e-05, "loss": 1.6649, "step": 77720 }, { "epoch": 0.11, "learning_rate": 4.5141875e-05, "loss": 1.4955, "step": 77730 }, { "epoch": 0.11, "learning_rate": 4.5141250000000004e-05, "loss": 1.6246, "step": 77740 }, { "epoch": 0.11, "learning_rate": 4.5140625e-05, "loss": 1.4759, "step": 77750 }, { "epoch": 0.11, "learning_rate": 4.5140000000000005e-05, "loss": 1.4723, "step": 77760 }, { "epoch": 0.11, "learning_rate": 4.5139375e-05, "loss": 1.5959, "step": 77770 }, { "epoch": 0.11, "learning_rate": 4.513875e-05, "loss": 1.4663, "step": 77780 }, { "epoch": 0.11, "learning_rate": 4.5138125e-05, "loss": 1.4238, "step": 77790 }, { "epoch": 0.11, "learning_rate": 4.5137500000000006e-05, "loss": 1.5456, "step": 77800 }, { "epoch": 0.11, "learning_rate": 4.5136875e-05, "loss": 1.4765, "step": 77810 }, { "epoch": 0.11, "learning_rate": 4.513625e-05, "loss": 1.8261, "step": 77820 }, { "epoch": 0.11, "learning_rate": 4.5135625000000004e-05, "loss": 1.3463, "step": 77830 }, { "epoch": 0.11, "learning_rate": 4.5135e-05, "loss": 1.5902, "step": 77840 }, { "epoch": 0.11, "learning_rate": 4.5134375e-05, "loss": 1.503, "step": 77850 }, { "epoch": 0.11, "learning_rate": 4.513375e-05, "loss": 1.5713, "step": 77860 }, { "epoch": 0.11, "learning_rate": 4.5133125000000005e-05, "loss": 1.6723, "step": 77870 }, { "epoch": 0.11, "learning_rate": 4.51325e-05, "loss": 1.4021, "step": 77880 }, { "epoch": 0.11, "learning_rate": 4.5131875e-05, "loss": 1.7202, "step": 77890 }, { "epoch": 0.11, "learning_rate": 4.513125e-05, "loss": 1.3809, "step": 77900 }, { "epoch": 0.11, "learning_rate": 4.5130625000000006e-05, "loss": 1.6642, "step": 77910 }, { "epoch": 0.11, "learning_rate": 4.513e-05, "loss": 1.5613, "step": 77920 }, { "epoch": 0.11, "learning_rate": 4.5129375e-05, "loss": 1.6353, "step": 77930 }, { "epoch": 0.11, "learning_rate": 4.5128750000000004e-05, "loss": 1.5045, "step": 77940 }, { "epoch": 0.11, "learning_rate": 4.512812500000001e-05, "loss": 1.4795, "step": 77950 }, { "epoch": 0.11, "learning_rate": 4.51275e-05, "loss": 1.3539, "step": 77960 }, { "epoch": 0.11, "learning_rate": 4.5126875e-05, "loss": 1.4623, "step": 77970 }, { "epoch": 0.11, "learning_rate": 4.5126250000000005e-05, "loss": 1.5382, "step": 77980 }, { "epoch": 0.11, "learning_rate": 4.5125625e-05, "loss": 1.4082, "step": 77990 }, { "epoch": 0.11, "learning_rate": 4.5125e-05, "loss": 1.5409, "step": 78000 }, { "epoch": 0.11, "eval_loss": 1.474303126335144, "eval_runtime": 1299.5726, "eval_samples_per_second": 44.658, "eval_steps_per_second": 5.583, "step": 78000 }, { "epoch": 0.11, "learning_rate": 4.5124375e-05, "loss": 1.3661, "step": 78010 }, { "epoch": 0.11, "learning_rate": 4.5123750000000007e-05, "loss": 1.5056, "step": 78020 }, { "epoch": 0.11, "learning_rate": 4.5123125e-05, "loss": 1.6904, "step": 78030 }, { "epoch": 0.11, "learning_rate": 4.51225e-05, "loss": 1.6585, "step": 78040 }, { "epoch": 0.11, "learning_rate": 4.5121875000000004e-05, "loss": 1.3555, "step": 78050 }, { "epoch": 0.11, "learning_rate": 4.512125e-05, "loss": 1.4569, "step": 78060 }, { "epoch": 0.11, "learning_rate": 4.5120625e-05, "loss": 1.5331, "step": 78070 }, { "epoch": 0.11, "learning_rate": 4.512e-05, "loss": 1.5793, "step": 78080 }, { "epoch": 0.11, "learning_rate": 4.5119375000000005e-05, "loss": 1.5932, "step": 78090 }, { "epoch": 0.11, "learning_rate": 4.511875e-05, "loss": 1.5164, "step": 78100 }, { "epoch": 0.11, "learning_rate": 4.5118125e-05, "loss": 1.6169, "step": 78110 }, { "epoch": 0.11, "learning_rate": 4.51175e-05, "loss": 1.7864, "step": 78120 }, { "epoch": 0.11, "learning_rate": 4.5116875e-05, "loss": 1.6083, "step": 78130 }, { "epoch": 0.11, "learning_rate": 4.5116250000000004e-05, "loss": 1.4767, "step": 78140 }, { "epoch": 0.11, "learning_rate": 4.5115625e-05, "loss": 1.7557, "step": 78150 }, { "epoch": 0.11, "learning_rate": 4.5115000000000004e-05, "loss": 1.453, "step": 78160 }, { "epoch": 0.11, "learning_rate": 4.5114375e-05, "loss": 1.6679, "step": 78170 }, { "epoch": 0.11, "learning_rate": 4.5113750000000005e-05, "loss": 1.6899, "step": 78180 }, { "epoch": 0.11, "learning_rate": 4.5113125e-05, "loss": 1.4596, "step": 78190 }, { "epoch": 0.11, "learning_rate": 4.51125e-05, "loss": 1.4928, "step": 78200 }, { "epoch": 0.11, "learning_rate": 4.5111875e-05, "loss": 1.4498, "step": 78210 }, { "epoch": 0.11, "learning_rate": 4.5111250000000006e-05, "loss": 1.4052, "step": 78220 }, { "epoch": 0.11, "learning_rate": 4.5110625e-05, "loss": 1.3461, "step": 78230 }, { "epoch": 0.11, "learning_rate": 4.511e-05, "loss": 1.641, "step": 78240 }, { "epoch": 0.11, "learning_rate": 4.5109375000000004e-05, "loss": 1.6385, "step": 78250 }, { "epoch": 0.11, "learning_rate": 4.510875e-05, "loss": 1.6539, "step": 78260 }, { "epoch": 0.11, "learning_rate": 4.5108125000000004e-05, "loss": 1.425, "step": 78270 }, { "epoch": 0.11, "learning_rate": 4.51075e-05, "loss": 1.6154, "step": 78280 }, { "epoch": 0.11, "learning_rate": 4.5106875000000005e-05, "loss": 1.4981, "step": 78290 }, { "epoch": 0.11, "learning_rate": 4.510625e-05, "loss": 1.3501, "step": 78300 }, { "epoch": 0.11, "learning_rate": 4.5105625e-05, "loss": 1.419, "step": 78310 }, { "epoch": 0.11, "learning_rate": 4.5105e-05, "loss": 1.3479, "step": 78320 }, { "epoch": 0.11, "learning_rate": 4.5104375000000006e-05, "loss": 1.5313, "step": 78330 }, { "epoch": 0.11, "learning_rate": 4.510375e-05, "loss": 1.6705, "step": 78340 }, { "epoch": 0.11, "learning_rate": 4.5103125e-05, "loss": 1.3902, "step": 78350 }, { "epoch": 0.11, "learning_rate": 4.5102500000000004e-05, "loss": 1.5553, "step": 78360 }, { "epoch": 0.11, "learning_rate": 4.510187500000001e-05, "loss": 1.4663, "step": 78370 }, { "epoch": 0.11, "learning_rate": 4.510125e-05, "loss": 1.2457, "step": 78380 }, { "epoch": 0.11, "learning_rate": 4.5100625e-05, "loss": 1.4366, "step": 78390 }, { "epoch": 0.11, "learning_rate": 4.5100000000000005e-05, "loss": 1.5032, "step": 78400 }, { "epoch": 0.11, "learning_rate": 4.5099375e-05, "loss": 1.6401, "step": 78410 }, { "epoch": 0.11, "learning_rate": 4.509875e-05, "loss": 1.4908, "step": 78420 }, { "epoch": 0.11, "learning_rate": 4.5098125e-05, "loss": 1.3424, "step": 78430 }, { "epoch": 0.11, "learning_rate": 4.5097500000000007e-05, "loss": 1.5225, "step": 78440 }, { "epoch": 0.11, "learning_rate": 4.5096875e-05, "loss": 1.5659, "step": 78450 }, { "epoch": 0.11, "learning_rate": 4.509625e-05, "loss": 1.6069, "step": 78460 }, { "epoch": 0.11, "learning_rate": 4.5095625000000004e-05, "loss": 1.4137, "step": 78470 }, { "epoch": 0.11, "learning_rate": 4.5095e-05, "loss": 1.402, "step": 78480 }, { "epoch": 0.11, "learning_rate": 4.5094375e-05, "loss": 1.5993, "step": 78490 }, { "epoch": 0.11, "learning_rate": 4.509375e-05, "loss": 1.6121, "step": 78500 }, { "epoch": 0.11, "learning_rate": 4.5093125000000005e-05, "loss": 1.5426, "step": 78510 }, { "epoch": 0.11, "learning_rate": 4.50925e-05, "loss": 1.5369, "step": 78520 }, { "epoch": 0.11, "learning_rate": 4.5091875e-05, "loss": 1.7936, "step": 78530 }, { "epoch": 0.11, "learning_rate": 4.509125e-05, "loss": 1.4898, "step": 78540 }, { "epoch": 0.11, "learning_rate": 4.5090625e-05, "loss": 1.5925, "step": 78550 }, { "epoch": 0.11, "learning_rate": 4.5090000000000004e-05, "loss": 1.4858, "step": 78560 }, { "epoch": 0.11, "learning_rate": 4.5089375e-05, "loss": 1.5506, "step": 78570 }, { "epoch": 0.11, "learning_rate": 4.5088750000000004e-05, "loss": 1.5887, "step": 78580 }, { "epoch": 0.11, "learning_rate": 4.5088125e-05, "loss": 1.5907, "step": 78590 }, { "epoch": 0.11, "learning_rate": 4.5087500000000005e-05, "loss": 1.3836, "step": 78600 }, { "epoch": 0.11, "learning_rate": 4.5086875e-05, "loss": 1.4994, "step": 78610 }, { "epoch": 0.11, "learning_rate": 4.508625e-05, "loss": 1.5242, "step": 78620 }, { "epoch": 0.11, "learning_rate": 4.5085625e-05, "loss": 1.5747, "step": 78630 }, { "epoch": 0.11, "learning_rate": 4.5085e-05, "loss": 1.6782, "step": 78640 }, { "epoch": 0.11, "learning_rate": 4.5084375e-05, "loss": 1.7419, "step": 78650 }, { "epoch": 0.11, "learning_rate": 4.508375e-05, "loss": 1.5549, "step": 78660 }, { "epoch": 0.11, "learning_rate": 4.5083125000000004e-05, "loss": 1.4118, "step": 78670 }, { "epoch": 0.11, "learning_rate": 4.50825e-05, "loss": 1.5472, "step": 78680 }, { "epoch": 0.11, "learning_rate": 4.5081875000000004e-05, "loss": 1.7784, "step": 78690 }, { "epoch": 0.11, "learning_rate": 4.508125e-05, "loss": 1.3678, "step": 78700 }, { "epoch": 0.11, "learning_rate": 4.5080625000000005e-05, "loss": 1.4958, "step": 78710 }, { "epoch": 0.11, "learning_rate": 4.508e-05, "loss": 1.426, "step": 78720 }, { "epoch": 0.11, "learning_rate": 4.5079375e-05, "loss": 1.755, "step": 78730 }, { "epoch": 0.11, "learning_rate": 4.507875e-05, "loss": 1.4926, "step": 78740 }, { "epoch": 0.11, "learning_rate": 4.5078125000000006e-05, "loss": 1.6691, "step": 78750 }, { "epoch": 0.11, "learning_rate": 4.50775e-05, "loss": 1.5793, "step": 78760 }, { "epoch": 0.11, "learning_rate": 4.5076875e-05, "loss": 1.539, "step": 78770 }, { "epoch": 0.11, "learning_rate": 4.5076250000000004e-05, "loss": 1.3786, "step": 78780 }, { "epoch": 0.11, "learning_rate": 4.507562500000001e-05, "loss": 1.2603, "step": 78790 }, { "epoch": 0.11, "learning_rate": 4.5075e-05, "loss": 1.5474, "step": 78800 }, { "epoch": 0.11, "learning_rate": 4.5074375e-05, "loss": 1.5484, "step": 78810 }, { "epoch": 0.11, "learning_rate": 4.5073750000000005e-05, "loss": 1.5887, "step": 78820 }, { "epoch": 0.11, "learning_rate": 4.5073125e-05, "loss": 1.584, "step": 78830 }, { "epoch": 0.11, "learning_rate": 4.50725e-05, "loss": 1.403, "step": 78840 }, { "epoch": 0.11, "learning_rate": 4.5071875e-05, "loss": 1.6452, "step": 78850 }, { "epoch": 0.11, "learning_rate": 4.5071250000000007e-05, "loss": 1.4949, "step": 78860 }, { "epoch": 0.11, "learning_rate": 4.5070625e-05, "loss": 1.8507, "step": 78870 }, { "epoch": 0.11, "learning_rate": 4.507e-05, "loss": 1.509, "step": 78880 }, { "epoch": 0.11, "learning_rate": 4.5069375000000004e-05, "loss": 1.4263, "step": 78890 }, { "epoch": 0.11, "learning_rate": 4.506875e-05, "loss": 1.7913, "step": 78900 }, { "epoch": 0.11, "learning_rate": 4.5068125e-05, "loss": 1.6018, "step": 78910 }, { "epoch": 0.11, "learning_rate": 4.50675e-05, "loss": 1.6211, "step": 78920 }, { "epoch": 0.11, "learning_rate": 4.5066875000000005e-05, "loss": 1.4514, "step": 78930 }, { "epoch": 0.11, "learning_rate": 4.506625e-05, "loss": 1.3848, "step": 78940 }, { "epoch": 0.11, "learning_rate": 4.5065625e-05, "loss": 1.5007, "step": 78950 }, { "epoch": 0.11, "learning_rate": 4.5065e-05, "loss": 1.4584, "step": 78960 }, { "epoch": 0.11, "learning_rate": 4.5064375e-05, "loss": 1.5168, "step": 78970 }, { "epoch": 0.11, "learning_rate": 4.5063750000000004e-05, "loss": 1.4994, "step": 78980 }, { "epoch": 0.11, "learning_rate": 4.5063125e-05, "loss": 1.6765, "step": 78990 }, { "epoch": 0.11, "learning_rate": 4.5062500000000004e-05, "loss": 1.5622, "step": 79000 }, { "epoch": 0.11, "eval_loss": 1.4721444845199585, "eval_runtime": 1295.1463, "eval_samples_per_second": 44.81, "eval_steps_per_second": 5.602, "step": 79000 }, { "epoch": 0.11, "learning_rate": 4.5061875e-05, "loss": 1.4431, "step": 79010 }, { "epoch": 0.11, "learning_rate": 4.5061250000000005e-05, "loss": 1.6291, "step": 79020 }, { "epoch": 0.11, "learning_rate": 4.5060625e-05, "loss": 1.7727, "step": 79030 }, { "epoch": 0.11, "learning_rate": 4.506e-05, "loss": 1.3042, "step": 79040 }, { "epoch": 0.11, "learning_rate": 4.5059375e-05, "loss": 1.5859, "step": 79050 }, { "epoch": 0.11, "learning_rate": 4.505875e-05, "loss": 1.5611, "step": 79060 }, { "epoch": 0.11, "learning_rate": 4.5058125e-05, "loss": 1.6644, "step": 79070 }, { "epoch": 0.11, "learning_rate": 4.50575e-05, "loss": 1.5174, "step": 79080 }, { "epoch": 0.11, "learning_rate": 4.5056875000000004e-05, "loss": 1.7385, "step": 79090 }, { "epoch": 0.11, "learning_rate": 4.505625e-05, "loss": 1.6751, "step": 79100 }, { "epoch": 0.11, "learning_rate": 4.5055625000000005e-05, "loss": 1.5624, "step": 79110 }, { "epoch": 0.11, "learning_rate": 4.5055e-05, "loss": 1.3132, "step": 79120 }, { "epoch": 0.11, "learning_rate": 4.5054375000000005e-05, "loss": 1.6695, "step": 79130 }, { "epoch": 0.11, "learning_rate": 4.505375e-05, "loss": 1.4772, "step": 79140 }, { "epoch": 0.11, "learning_rate": 4.5053125e-05, "loss": 1.32, "step": 79150 }, { "epoch": 0.11, "learning_rate": 4.50525e-05, "loss": 1.5859, "step": 79160 }, { "epoch": 0.11, "learning_rate": 4.5051875000000006e-05, "loss": 1.6473, "step": 79170 }, { "epoch": 0.11, "learning_rate": 4.5051250000000003e-05, "loss": 1.4498, "step": 79180 }, { "epoch": 0.11, "learning_rate": 4.5050625e-05, "loss": 1.6119, "step": 79190 }, { "epoch": 0.11, "learning_rate": 4.5050000000000004e-05, "loss": 1.4954, "step": 79200 }, { "epoch": 0.11, "learning_rate": 4.504937500000001e-05, "loss": 1.5052, "step": 79210 }, { "epoch": 0.11, "learning_rate": 4.504875e-05, "loss": 1.7635, "step": 79220 }, { "epoch": 0.11, "learning_rate": 4.5048125e-05, "loss": 1.6228, "step": 79230 }, { "epoch": 0.11, "learning_rate": 4.5047500000000005e-05, "loss": 1.6009, "step": 79240 }, { "epoch": 0.11, "learning_rate": 4.5046875e-05, "loss": 1.4144, "step": 79250 }, { "epoch": 0.11, "learning_rate": 4.504625e-05, "loss": 1.6409, "step": 79260 }, { "epoch": 0.11, "learning_rate": 4.5045625e-05, "loss": 1.4843, "step": 79270 }, { "epoch": 0.11, "learning_rate": 4.504500000000001e-05, "loss": 1.3895, "step": 79280 }, { "epoch": 0.11, "learning_rate": 4.5044375e-05, "loss": 1.4603, "step": 79290 }, { "epoch": 0.11, "learning_rate": 4.504375e-05, "loss": 1.4934, "step": 79300 }, { "epoch": 0.11, "learning_rate": 4.5043125000000004e-05, "loss": 1.485, "step": 79310 }, { "epoch": 0.11, "learning_rate": 4.50425e-05, "loss": 1.7202, "step": 79320 }, { "epoch": 0.11, "learning_rate": 4.5041875e-05, "loss": 1.5799, "step": 79330 }, { "epoch": 0.11, "learning_rate": 4.504125e-05, "loss": 1.6429, "step": 79340 }, { "epoch": 0.11, "learning_rate": 4.5040625000000006e-05, "loss": 1.5581, "step": 79350 }, { "epoch": 0.11, "learning_rate": 4.504e-05, "loss": 1.5928, "step": 79360 }, { "epoch": 0.11, "learning_rate": 4.5039375e-05, "loss": 1.6253, "step": 79370 }, { "epoch": 0.11, "learning_rate": 4.503875e-05, "loss": 1.642, "step": 79380 }, { "epoch": 0.11, "learning_rate": 4.5038125e-05, "loss": 1.5165, "step": 79390 }, { "epoch": 0.11, "learning_rate": 4.5037500000000004e-05, "loss": 1.6451, "step": 79400 }, { "epoch": 0.11, "learning_rate": 4.5036875e-05, "loss": 1.6018, "step": 79410 }, { "epoch": 0.11, "learning_rate": 4.5036250000000004e-05, "loss": 1.4609, "step": 79420 }, { "epoch": 0.11, "learning_rate": 4.5035625e-05, "loss": 1.6351, "step": 79430 }, { "epoch": 0.11, "learning_rate": 4.5035e-05, "loss": 1.5561, "step": 79440 }, { "epoch": 0.11, "learning_rate": 4.5034375e-05, "loss": 1.1956, "step": 79450 }, { "epoch": 0.11, "learning_rate": 4.5033750000000006e-05, "loss": 1.5436, "step": 79460 }, { "epoch": 0.11, "learning_rate": 4.5033125e-05, "loss": 1.4033, "step": 79470 }, { "epoch": 0.11, "learning_rate": 4.50325e-05, "loss": 1.7053, "step": 79480 }, { "epoch": 0.11, "learning_rate": 4.5031875e-05, "loss": 1.4664, "step": 79490 }, { "epoch": 0.11, "learning_rate": 4.503125e-05, "loss": 1.6271, "step": 79500 }, { "epoch": 0.11, "learning_rate": 4.5030625000000004e-05, "loss": 1.7451, "step": 79510 }, { "epoch": 0.11, "learning_rate": 4.503e-05, "loss": 1.6291, "step": 79520 }, { "epoch": 0.11, "learning_rate": 4.5029375000000005e-05, "loss": 1.787, "step": 79530 }, { "epoch": 0.11, "learning_rate": 4.502875e-05, "loss": 1.6339, "step": 79540 }, { "epoch": 0.11, "learning_rate": 4.5028125000000005e-05, "loss": 1.6569, "step": 79550 }, { "epoch": 0.11, "learning_rate": 4.50275e-05, "loss": 1.6311, "step": 79560 }, { "epoch": 0.11, "learning_rate": 4.5026875e-05, "loss": 1.3842, "step": 79570 }, { "epoch": 0.11, "learning_rate": 4.502625e-05, "loss": 1.4542, "step": 79580 }, { "epoch": 0.11, "learning_rate": 4.5025625000000007e-05, "loss": 1.3103, "step": 79590 }, { "epoch": 0.11, "learning_rate": 4.5025000000000003e-05, "loss": 1.898, "step": 79600 }, { "epoch": 0.11, "learning_rate": 4.5024375e-05, "loss": 1.5857, "step": 79610 }, { "epoch": 0.11, "learning_rate": 4.5023750000000004e-05, "loss": 1.6196, "step": 79620 }, { "epoch": 0.11, "learning_rate": 4.502312500000001e-05, "loss": 1.5921, "step": 79630 }, { "epoch": 0.11, "learning_rate": 4.50225e-05, "loss": 1.4319, "step": 79640 }, { "epoch": 0.11, "learning_rate": 4.5021875e-05, "loss": 1.6472, "step": 79650 }, { "epoch": 0.11, "learning_rate": 4.5021250000000005e-05, "loss": 1.6232, "step": 79660 }, { "epoch": 0.11, "learning_rate": 4.5020625e-05, "loss": 1.6955, "step": 79670 }, { "epoch": 0.11, "learning_rate": 4.502e-05, "loss": 1.6691, "step": 79680 }, { "epoch": 0.11, "learning_rate": 4.5019375e-05, "loss": 1.6158, "step": 79690 }, { "epoch": 0.11, "learning_rate": 4.501875000000001e-05, "loss": 1.611, "step": 79700 }, { "epoch": 0.11, "learning_rate": 4.5018125e-05, "loss": 1.5919, "step": 79710 }, { "epoch": 0.11, "learning_rate": 4.50175e-05, "loss": 1.4469, "step": 79720 }, { "epoch": 0.11, "learning_rate": 4.5016875000000004e-05, "loss": 1.5067, "step": 79730 }, { "epoch": 0.11, "learning_rate": 4.501625e-05, "loss": 1.6278, "step": 79740 }, { "epoch": 0.11, "learning_rate": 4.5015625e-05, "loss": 1.462, "step": 79750 }, { "epoch": 0.11, "learning_rate": 4.5015e-05, "loss": 1.4776, "step": 79760 }, { "epoch": 0.11, "learning_rate": 4.5014375000000006e-05, "loss": 1.388, "step": 79770 }, { "epoch": 0.11, "learning_rate": 4.501375e-05, "loss": 1.411, "step": 79780 }, { "epoch": 0.11, "learning_rate": 4.5013125e-05, "loss": 1.4758, "step": 79790 }, { "epoch": 0.11, "learning_rate": 4.50125e-05, "loss": 1.5998, "step": 79800 }, { "epoch": 0.11, "learning_rate": 4.5011875e-05, "loss": 1.5278, "step": 79810 }, { "epoch": 0.11, "learning_rate": 4.5011250000000004e-05, "loss": 1.4769, "step": 79820 }, { "epoch": 0.11, "learning_rate": 4.5010625e-05, "loss": 1.5183, "step": 79830 }, { "epoch": 0.11, "learning_rate": 4.5010000000000004e-05, "loss": 1.5957, "step": 79840 }, { "epoch": 0.11, "learning_rate": 4.5009375e-05, "loss": 1.4875, "step": 79850 }, { "epoch": 0.11, "learning_rate": 4.500875e-05, "loss": 1.6298, "step": 79860 }, { "epoch": 0.11, "learning_rate": 4.5008125e-05, "loss": 1.5774, "step": 79870 }, { "epoch": 0.11, "learning_rate": 4.5007500000000006e-05, "loss": 1.3408, "step": 79880 }, { "epoch": 0.11, "learning_rate": 4.5006875e-05, "loss": 1.6043, "step": 79890 }, { "epoch": 0.11, "learning_rate": 4.500625e-05, "loss": 1.6242, "step": 79900 }, { "epoch": 0.11, "learning_rate": 4.5005625e-05, "loss": 1.617, "step": 79910 }, { "epoch": 0.11, "learning_rate": 4.5005e-05, "loss": 1.6918, "step": 79920 }, { "epoch": 0.11, "learning_rate": 4.5004375000000004e-05, "loss": 1.5469, "step": 79930 }, { "epoch": 0.11, "learning_rate": 4.500375e-05, "loss": 1.4018, "step": 79940 }, { "epoch": 0.11, "learning_rate": 4.5003125000000005e-05, "loss": 1.6075, "step": 79950 }, { "epoch": 0.11, "learning_rate": 4.50025e-05, "loss": 1.6396, "step": 79960 }, { "epoch": 0.11, "learning_rate": 4.5001875000000005e-05, "loss": 1.4785, "step": 79970 }, { "epoch": 0.11, "learning_rate": 4.500125e-05, "loss": 1.4519, "step": 79980 }, { "epoch": 0.11, "learning_rate": 4.5000625e-05, "loss": 1.4207, "step": 79990 }, { "epoch": 0.11, "learning_rate": 4.5e-05, "loss": 1.5089, "step": 80000 }, { "epoch": 0.11, "eval_loss": 1.4642947912216187, "eval_runtime": 1298.9103, "eval_samples_per_second": 44.681, "eval_steps_per_second": 5.585, "step": 80000 }, { "epoch": 0.11, "learning_rate": 4.4999375000000007e-05, "loss": 1.5459, "step": 80010 }, { "epoch": 0.11, "learning_rate": 4.4998750000000003e-05, "loss": 1.6546, "step": 80020 }, { "epoch": 0.11, "learning_rate": 4.4998125e-05, "loss": 1.6281, "step": 80030 }, { "epoch": 0.11, "learning_rate": 4.4997500000000004e-05, "loss": 1.416, "step": 80040 }, { "epoch": 0.11, "learning_rate": 4.4996875e-05, "loss": 1.5169, "step": 80050 }, { "epoch": 0.11, "learning_rate": 4.499625e-05, "loss": 1.6105, "step": 80060 }, { "epoch": 0.11, "learning_rate": 4.4995625e-05, "loss": 1.3291, "step": 80070 }, { "epoch": 0.11, "learning_rate": 4.4995000000000005e-05, "loss": 1.4621, "step": 80080 }, { "epoch": 0.11, "learning_rate": 4.4994375e-05, "loss": 1.7131, "step": 80090 }, { "epoch": 0.11, "learning_rate": 4.499375e-05, "loss": 1.4414, "step": 80100 }, { "epoch": 0.11, "learning_rate": 4.4993125e-05, "loss": 1.5683, "step": 80110 }, { "epoch": 0.11, "learning_rate": 4.499250000000001e-05, "loss": 1.5388, "step": 80120 }, { "epoch": 0.11, "learning_rate": 4.4991875e-05, "loss": 1.3969, "step": 80130 }, { "epoch": 0.11, "learning_rate": 4.499125e-05, "loss": 1.6866, "step": 80140 }, { "epoch": 0.11, "learning_rate": 4.4990625000000004e-05, "loss": 1.4895, "step": 80150 }, { "epoch": 0.11, "learning_rate": 4.499e-05, "loss": 1.5851, "step": 80160 }, { "epoch": 0.11, "learning_rate": 4.4989375e-05, "loss": 1.6955, "step": 80170 }, { "epoch": 0.11, "learning_rate": 4.498875e-05, "loss": 1.6243, "step": 80180 }, { "epoch": 0.11, "learning_rate": 4.4988125000000006e-05, "loss": 1.351, "step": 80190 }, { "epoch": 0.11, "learning_rate": 4.49875e-05, "loss": 1.4819, "step": 80200 }, { "epoch": 0.11, "learning_rate": 4.4986875e-05, "loss": 1.4294, "step": 80210 }, { "epoch": 0.11, "learning_rate": 4.498625e-05, "loss": 1.4808, "step": 80220 }, { "epoch": 0.11, "learning_rate": 4.4985625e-05, "loss": 1.6889, "step": 80230 }, { "epoch": 0.11, "learning_rate": 4.4985000000000004e-05, "loss": 1.6341, "step": 80240 }, { "epoch": 0.11, "learning_rate": 4.4984375e-05, "loss": 1.3216, "step": 80250 }, { "epoch": 0.11, "learning_rate": 4.4983750000000004e-05, "loss": 1.4935, "step": 80260 }, { "epoch": 0.11, "learning_rate": 4.4983125e-05, "loss": 1.6422, "step": 80270 }, { "epoch": 0.11, "learning_rate": 4.49825e-05, "loss": 1.755, "step": 80280 }, { "epoch": 0.11, "learning_rate": 4.4981875e-05, "loss": 1.6309, "step": 80290 }, { "epoch": 0.11, "learning_rate": 4.4981250000000006e-05, "loss": 1.3515, "step": 80300 }, { "epoch": 0.11, "learning_rate": 4.4980625e-05, "loss": 1.4862, "step": 80310 }, { "epoch": 0.11, "learning_rate": 4.498e-05, "loss": 1.3545, "step": 80320 }, { "epoch": 0.11, "learning_rate": 4.4979375e-05, "loss": 1.5718, "step": 80330 }, { "epoch": 0.11, "learning_rate": 4.497875e-05, "loss": 1.5085, "step": 80340 }, { "epoch": 0.11, "learning_rate": 4.4978125000000004e-05, "loss": 1.7062, "step": 80350 }, { "epoch": 0.11, "learning_rate": 4.49775e-05, "loss": 1.7326, "step": 80360 }, { "epoch": 0.11, "learning_rate": 4.4976875000000005e-05, "loss": 1.6444, "step": 80370 }, { "epoch": 0.11, "learning_rate": 4.497625e-05, "loss": 1.6626, "step": 80380 }, { "epoch": 0.11, "learning_rate": 4.4975625000000005e-05, "loss": 1.6681, "step": 80390 }, { "epoch": 0.11, "learning_rate": 4.4975e-05, "loss": 1.5325, "step": 80400 }, { "epoch": 0.11, "learning_rate": 4.4974375e-05, "loss": 1.7446, "step": 80410 }, { "epoch": 0.11, "learning_rate": 4.497375e-05, "loss": 1.6049, "step": 80420 }, { "epoch": 0.11, "learning_rate": 4.4973125000000007e-05, "loss": 1.6645, "step": 80430 }, { "epoch": 0.11, "learning_rate": 4.4972500000000004e-05, "loss": 1.2873, "step": 80440 }, { "epoch": 0.11, "learning_rate": 4.4971875e-05, "loss": 1.3969, "step": 80450 }, { "epoch": 0.11, "learning_rate": 4.4971250000000004e-05, "loss": 1.5536, "step": 80460 }, { "epoch": 0.11, "learning_rate": 4.4970625e-05, "loss": 1.4292, "step": 80470 }, { "epoch": 0.11, "learning_rate": 4.497e-05, "loss": 1.6978, "step": 80480 }, { "epoch": 0.11, "learning_rate": 4.4969375e-05, "loss": 1.5726, "step": 80490 }, { "epoch": 0.11, "learning_rate": 4.4968750000000005e-05, "loss": 1.5322, "step": 80500 }, { "epoch": 0.11, "learning_rate": 4.4968125e-05, "loss": 1.4014, "step": 80510 }, { "epoch": 0.11, "learning_rate": 4.49675e-05, "loss": 1.6544, "step": 80520 }, { "epoch": 0.11, "learning_rate": 4.4966875e-05, "loss": 1.4722, "step": 80530 }, { "epoch": 0.11, "learning_rate": 4.496625000000001e-05, "loss": 1.7566, "step": 80540 }, { "epoch": 0.11, "learning_rate": 4.4965625e-05, "loss": 1.407, "step": 80550 }, { "epoch": 0.11, "learning_rate": 4.4965e-05, "loss": 1.8065, "step": 80560 }, { "epoch": 0.11, "learning_rate": 4.4964375000000004e-05, "loss": 1.6957, "step": 80570 }, { "epoch": 0.11, "learning_rate": 4.496375e-05, "loss": 1.5561, "step": 80580 }, { "epoch": 0.11, "learning_rate": 4.4963125e-05, "loss": 1.4682, "step": 80590 }, { "epoch": 0.11, "learning_rate": 4.49625e-05, "loss": 1.6209, "step": 80600 }, { "epoch": 0.11, "learning_rate": 4.4961875000000006e-05, "loss": 1.5479, "step": 80610 }, { "epoch": 0.11, "learning_rate": 4.496125e-05, "loss": 1.445, "step": 80620 }, { "epoch": 0.11, "learning_rate": 4.4960625e-05, "loss": 1.6379, "step": 80630 }, { "epoch": 0.11, "learning_rate": 4.496e-05, "loss": 1.4954, "step": 80640 }, { "epoch": 0.11, "learning_rate": 4.495937500000001e-05, "loss": 1.6, "step": 80650 }, { "epoch": 0.11, "learning_rate": 4.495875e-05, "loss": 1.606, "step": 80660 }, { "epoch": 0.11, "learning_rate": 4.4958125e-05, "loss": 1.7511, "step": 80670 }, { "epoch": 0.11, "learning_rate": 4.4957500000000005e-05, "loss": 1.4722, "step": 80680 }, { "epoch": 0.11, "learning_rate": 4.4956875e-05, "loss": 1.8352, "step": 80690 }, { "epoch": 0.11, "learning_rate": 4.495625e-05, "loss": 1.5609, "step": 80700 }, { "epoch": 0.11, "learning_rate": 4.4955625e-05, "loss": 1.5933, "step": 80710 }, { "epoch": 0.11, "learning_rate": 4.4955000000000006e-05, "loss": 1.2999, "step": 80720 }, { "epoch": 0.11, "learning_rate": 4.4954375e-05, "loss": 1.5127, "step": 80730 }, { "epoch": 0.11, "learning_rate": 4.495375e-05, "loss": 1.5268, "step": 80740 }, { "epoch": 0.11, "learning_rate": 4.4953125000000003e-05, "loss": 1.5758, "step": 80750 }, { "epoch": 0.11, "learning_rate": 4.49525e-05, "loss": 1.4623, "step": 80760 }, { "epoch": 0.11, "learning_rate": 4.4951875000000004e-05, "loss": 1.2712, "step": 80770 }, { "epoch": 0.11, "learning_rate": 4.495125e-05, "loss": 1.363, "step": 80780 }, { "epoch": 0.11, "learning_rate": 4.4950625000000005e-05, "loss": 1.3846, "step": 80790 }, { "epoch": 0.11, "learning_rate": 4.495e-05, "loss": 1.7138, "step": 80800 }, { "epoch": 0.11, "learning_rate": 4.4949375000000005e-05, "loss": 1.3411, "step": 80810 }, { "epoch": 0.11, "learning_rate": 4.494875e-05, "loss": 1.515, "step": 80820 }, { "epoch": 0.11, "learning_rate": 4.4948125e-05, "loss": 1.6456, "step": 80830 }, { "epoch": 0.11, "learning_rate": 4.49475e-05, "loss": 1.4295, "step": 80840 }, { "epoch": 0.11, "learning_rate": 4.4946875e-05, "loss": 1.5926, "step": 80850 }, { "epoch": 0.11, "learning_rate": 4.4946250000000004e-05, "loss": 1.4313, "step": 80860 }, { "epoch": 0.11, "learning_rate": 4.4945625e-05, "loss": 1.6065, "step": 80870 }, { "epoch": 0.11, "learning_rate": 4.4945000000000004e-05, "loss": 1.5811, "step": 80880 }, { "epoch": 0.11, "learning_rate": 4.4944375e-05, "loss": 1.3823, "step": 80890 }, { "epoch": 0.11, "learning_rate": 4.494375e-05, "loss": 1.4378, "step": 80900 }, { "epoch": 0.11, "learning_rate": 4.4943125e-05, "loss": 1.4384, "step": 80910 }, { "epoch": 0.11, "learning_rate": 4.4942500000000006e-05, "loss": 1.6553, "step": 80920 }, { "epoch": 0.11, "learning_rate": 4.4941875e-05, "loss": 1.5387, "step": 80930 }, { "epoch": 0.11, "learning_rate": 4.494125e-05, "loss": 1.5914, "step": 80940 }, { "epoch": 0.11, "learning_rate": 4.4940625e-05, "loss": 1.4153, "step": 80950 }, { "epoch": 0.11, "learning_rate": 4.494000000000001e-05, "loss": 1.6199, "step": 80960 }, { "epoch": 0.11, "learning_rate": 4.4939375e-05, "loss": 1.354, "step": 80970 }, { "epoch": 0.11, "learning_rate": 4.493875e-05, "loss": 1.5408, "step": 80980 }, { "epoch": 0.11, "learning_rate": 4.4938125000000004e-05, "loss": 1.4285, "step": 80990 }, { "epoch": 0.11, "learning_rate": 4.49375e-05, "loss": 1.5133, "step": 81000 }, { "epoch": 0.11, "eval_loss": 1.4578566551208496, "eval_runtime": 1308.9831, "eval_samples_per_second": 44.337, "eval_steps_per_second": 5.542, "step": 81000 }, { "epoch": 0.11, "learning_rate": 4.4936875e-05, "loss": 1.4778, "step": 81010 }, { "epoch": 0.11, "learning_rate": 4.493625e-05, "loss": 1.4272, "step": 81020 }, { "epoch": 0.11, "learning_rate": 4.4935625000000006e-05, "loss": 1.5396, "step": 81030 }, { "epoch": 0.11, "learning_rate": 4.4935e-05, "loss": 1.4714, "step": 81040 }, { "epoch": 0.11, "learning_rate": 4.4934375e-05, "loss": 1.5595, "step": 81050 }, { "epoch": 0.11, "learning_rate": 4.493375e-05, "loss": 1.3687, "step": 81060 }, { "epoch": 0.11, "learning_rate": 4.493312500000001e-05, "loss": 1.3674, "step": 81070 }, { "epoch": 0.11, "learning_rate": 4.49325e-05, "loss": 1.4386, "step": 81080 }, { "epoch": 0.11, "learning_rate": 4.4931875e-05, "loss": 1.5011, "step": 81090 }, { "epoch": 0.11, "learning_rate": 4.4931250000000005e-05, "loss": 1.4677, "step": 81100 }, { "epoch": 0.11, "learning_rate": 4.4930625e-05, "loss": 1.4991, "step": 81110 }, { "epoch": 0.11, "learning_rate": 4.493e-05, "loss": 1.6322, "step": 81120 }, { "epoch": 0.11, "learning_rate": 4.4929375e-05, "loss": 1.7004, "step": 81130 }, { "epoch": 0.11, "learning_rate": 4.4928750000000006e-05, "loss": 1.5537, "step": 81140 }, { "epoch": 0.11, "learning_rate": 4.4928125e-05, "loss": 1.6326, "step": 81150 }, { "epoch": 0.11, "learning_rate": 4.49275e-05, "loss": 1.5251, "step": 81160 }, { "epoch": 0.11, "learning_rate": 4.4926875000000003e-05, "loss": 1.5479, "step": 81170 }, { "epoch": 0.11, "learning_rate": 4.492625e-05, "loss": 1.3853, "step": 81180 }, { "epoch": 0.11, "learning_rate": 4.4925625000000004e-05, "loss": 1.4939, "step": 81190 }, { "epoch": 0.11, "learning_rate": 4.4925e-05, "loss": 1.6487, "step": 81200 }, { "epoch": 0.11, "learning_rate": 4.4924375000000005e-05, "loss": 1.7519, "step": 81210 }, { "epoch": 0.11, "learning_rate": 4.492375e-05, "loss": 1.6222, "step": 81220 }, { "epoch": 0.11, "learning_rate": 4.4923125000000005e-05, "loss": 1.4574, "step": 81230 }, { "epoch": 0.11, "learning_rate": 4.49225e-05, "loss": 1.5576, "step": 81240 }, { "epoch": 0.11, "learning_rate": 4.4921875e-05, "loss": 1.8184, "step": 81250 }, { "epoch": 0.11, "learning_rate": 4.492125e-05, "loss": 1.5203, "step": 81260 }, { "epoch": 0.11, "learning_rate": 4.4920625e-05, "loss": 1.492, "step": 81270 }, { "epoch": 0.11, "learning_rate": 4.4920000000000004e-05, "loss": 1.5299, "step": 81280 }, { "epoch": 0.11, "learning_rate": 4.4919375e-05, "loss": 1.4157, "step": 81290 }, { "epoch": 0.11, "learning_rate": 4.4918750000000004e-05, "loss": 1.5047, "step": 81300 }, { "epoch": 0.11, "learning_rate": 4.4918125e-05, "loss": 1.5088, "step": 81310 }, { "epoch": 0.11, "learning_rate": 4.49175e-05, "loss": 1.367, "step": 81320 }, { "epoch": 0.11, "learning_rate": 4.4916875e-05, "loss": 1.5188, "step": 81330 }, { "epoch": 0.11, "learning_rate": 4.4916250000000006e-05, "loss": 1.5787, "step": 81340 }, { "epoch": 0.11, "learning_rate": 4.4915625e-05, "loss": 1.5472, "step": 81350 }, { "epoch": 0.11, "learning_rate": 4.4915e-05, "loss": 1.4959, "step": 81360 }, { "epoch": 0.11, "learning_rate": 4.4914375e-05, "loss": 1.5694, "step": 81370 }, { "epoch": 0.11, "learning_rate": 4.491375000000001e-05, "loss": 1.49, "step": 81380 }, { "epoch": 0.11, "learning_rate": 4.4913125e-05, "loss": 1.4291, "step": 81390 }, { "epoch": 0.11, "learning_rate": 4.49125e-05, "loss": 1.7563, "step": 81400 }, { "epoch": 0.11, "learning_rate": 4.4911875000000004e-05, "loss": 1.4116, "step": 81410 }, { "epoch": 0.11, "learning_rate": 4.491125e-05, "loss": 1.3652, "step": 81420 }, { "epoch": 0.11, "learning_rate": 4.4910625e-05, "loss": 1.4529, "step": 81430 }, { "epoch": 0.11, "learning_rate": 4.491e-05, "loss": 1.581, "step": 81440 }, { "epoch": 0.11, "learning_rate": 4.4909375000000006e-05, "loss": 1.3685, "step": 81450 }, { "epoch": 0.11, "learning_rate": 4.490875e-05, "loss": 1.8168, "step": 81460 }, { "epoch": 0.11, "learning_rate": 4.4908125e-05, "loss": 1.4185, "step": 81470 }, { "epoch": 0.11, "learning_rate": 4.49075e-05, "loss": 1.5482, "step": 81480 }, { "epoch": 0.11, "learning_rate": 4.490687500000001e-05, "loss": 1.5437, "step": 81490 }, { "epoch": 0.11, "learning_rate": 4.490625e-05, "loss": 1.6345, "step": 81500 }, { "epoch": 0.11, "learning_rate": 4.4905625e-05, "loss": 1.5206, "step": 81510 }, { "epoch": 0.11, "learning_rate": 4.4905000000000005e-05, "loss": 1.394, "step": 81520 }, { "epoch": 0.11, "learning_rate": 4.4904375e-05, "loss": 1.5165, "step": 81530 }, { "epoch": 0.11, "learning_rate": 4.490375e-05, "loss": 1.8237, "step": 81540 }, { "epoch": 0.11, "learning_rate": 4.4903125e-05, "loss": 1.4613, "step": 81550 }, { "epoch": 0.11, "learning_rate": 4.4902500000000006e-05, "loss": 1.3982, "step": 81560 }, { "epoch": 0.11, "learning_rate": 4.4901875e-05, "loss": 1.5577, "step": 81570 }, { "epoch": 0.11, "learning_rate": 4.490125e-05, "loss": 1.4037, "step": 81580 }, { "epoch": 0.11, "learning_rate": 4.4900625000000003e-05, "loss": 1.6153, "step": 81590 }, { "epoch": 0.11, "learning_rate": 4.49e-05, "loss": 1.5612, "step": 81600 }, { "epoch": 0.11, "learning_rate": 4.4899375000000004e-05, "loss": 1.5287, "step": 81610 }, { "epoch": 0.11, "learning_rate": 4.489875e-05, "loss": 1.7267, "step": 81620 }, { "epoch": 0.11, "learning_rate": 4.4898125000000005e-05, "loss": 1.4755, "step": 81630 }, { "epoch": 0.11, "learning_rate": 4.48975e-05, "loss": 1.4347, "step": 81640 }, { "epoch": 0.11, "learning_rate": 4.4896875000000005e-05, "loss": 1.4205, "step": 81650 }, { "epoch": 0.11, "learning_rate": 4.489625e-05, "loss": 1.2559, "step": 81660 }, { "epoch": 0.11, "learning_rate": 4.4895625e-05, "loss": 1.5167, "step": 81670 }, { "epoch": 0.11, "learning_rate": 4.4895e-05, "loss": 1.6417, "step": 81680 }, { "epoch": 0.11, "learning_rate": 4.4894375e-05, "loss": 1.474, "step": 81690 }, { "epoch": 0.11, "learning_rate": 4.4893750000000004e-05, "loss": 1.6938, "step": 81700 }, { "epoch": 0.11, "learning_rate": 4.4893125e-05, "loss": 1.4112, "step": 81710 }, { "epoch": 0.11, "learning_rate": 4.4892500000000004e-05, "loss": 1.5277, "step": 81720 }, { "epoch": 0.11, "learning_rate": 4.4891875e-05, "loss": 1.5207, "step": 81730 }, { "epoch": 0.11, "learning_rate": 4.489125e-05, "loss": 1.5672, "step": 81740 }, { "epoch": 0.11, "learning_rate": 4.4890625e-05, "loss": 1.5822, "step": 81750 }, { "epoch": 0.11, "learning_rate": 4.4890000000000006e-05, "loss": 1.3729, "step": 81760 }, { "epoch": 0.11, "learning_rate": 4.4889375e-05, "loss": 1.5915, "step": 81770 }, { "epoch": 0.11, "learning_rate": 4.488875e-05, "loss": 1.6106, "step": 81780 }, { "epoch": 0.11, "learning_rate": 4.4888125e-05, "loss": 1.4608, "step": 81790 }, { "epoch": 0.11, "learning_rate": 4.488750000000001e-05, "loss": 1.5852, "step": 81800 }, { "epoch": 0.11, "learning_rate": 4.4886875000000004e-05, "loss": 1.4333, "step": 81810 }, { "epoch": 0.11, "learning_rate": 4.488625e-05, "loss": 1.349, "step": 81820 }, { "epoch": 0.11, "learning_rate": 4.4885625000000004e-05, "loss": 1.4074, "step": 81830 }, { "epoch": 0.11, "learning_rate": 4.488500000000001e-05, "loss": 1.5906, "step": 81840 }, { "epoch": 0.11, "learning_rate": 4.4884375e-05, "loss": 1.2205, "step": 81850 }, { "epoch": 0.11, "learning_rate": 4.488375e-05, "loss": 1.5751, "step": 81860 }, { "epoch": 0.11, "learning_rate": 4.4883125000000006e-05, "loss": 1.5417, "step": 81870 }, { "epoch": 0.11, "learning_rate": 4.48825e-05, "loss": 1.6088, "step": 81880 }, { "epoch": 0.11, "learning_rate": 4.4881875e-05, "loss": 1.6914, "step": 81890 }, { "epoch": 0.11, "learning_rate": 4.488125e-05, "loss": 1.5087, "step": 81900 }, { "epoch": 0.11, "learning_rate": 4.488062500000001e-05, "loss": 1.4768, "step": 81910 }, { "epoch": 0.11, "learning_rate": 4.488e-05, "loss": 1.5113, "step": 81920 }, { "epoch": 0.11, "learning_rate": 4.4879375e-05, "loss": 1.5293, "step": 81930 }, { "epoch": 0.11, "learning_rate": 4.4878750000000005e-05, "loss": 1.3835, "step": 81940 }, { "epoch": 0.11, "learning_rate": 4.4878125e-05, "loss": 1.488, "step": 81950 }, { "epoch": 0.11, "learning_rate": 4.48775e-05, "loss": 1.6122, "step": 81960 }, { "epoch": 0.11, "learning_rate": 4.4876875e-05, "loss": 1.3178, "step": 81970 }, { "epoch": 0.11, "learning_rate": 4.4876250000000006e-05, "loss": 1.3667, "step": 81980 }, { "epoch": 0.11, "learning_rate": 4.4875625e-05, "loss": 1.4644, "step": 81990 }, { "epoch": 0.11, "learning_rate": 4.4875e-05, "loss": 1.4248, "step": 82000 }, { "epoch": 0.11, "eval_loss": 1.4510304927825928, "eval_runtime": 1319.3434, "eval_samples_per_second": 43.989, "eval_steps_per_second": 5.499, "step": 82000 }, { "epoch": 0.11, "learning_rate": 4.4874375000000004e-05, "loss": 1.5627, "step": 82010 }, { "epoch": 0.11, "learning_rate": 4.487375e-05, "loss": 1.4978, "step": 82020 }, { "epoch": 0.11, "learning_rate": 4.4873125000000004e-05, "loss": 1.4589, "step": 82030 }, { "epoch": 0.11, "learning_rate": 4.48725e-05, "loss": 1.4345, "step": 82040 }, { "epoch": 0.11, "learning_rate": 4.4871875000000005e-05, "loss": 1.6444, "step": 82050 }, { "epoch": 0.11, "learning_rate": 4.487125e-05, "loss": 1.553, "step": 82060 }, { "epoch": 0.11, "learning_rate": 4.4870625e-05, "loss": 1.5145, "step": 82070 }, { "epoch": 0.11, "learning_rate": 4.487e-05, "loss": 1.4595, "step": 82080 }, { "epoch": 0.11, "learning_rate": 4.4869375e-05, "loss": 1.5049, "step": 82090 }, { "epoch": 0.11, "learning_rate": 4.486875e-05, "loss": 1.6399, "step": 82100 }, { "epoch": 0.11, "learning_rate": 4.4868125e-05, "loss": 1.7478, "step": 82110 }, { "epoch": 0.11, "learning_rate": 4.4867500000000004e-05, "loss": 1.5002, "step": 82120 }, { "epoch": 0.11, "learning_rate": 4.4866875e-05, "loss": 1.3999, "step": 82130 }, { "epoch": 0.11, "learning_rate": 4.4866250000000004e-05, "loss": 1.6421, "step": 82140 }, { "epoch": 0.11, "learning_rate": 4.4865625e-05, "loss": 1.4186, "step": 82150 }, { "epoch": 0.11, "learning_rate": 4.4865e-05, "loss": 1.6635, "step": 82160 }, { "epoch": 0.11, "learning_rate": 4.4864375e-05, "loss": 1.6083, "step": 82170 }, { "epoch": 0.11, "learning_rate": 4.4863750000000006e-05, "loss": 1.3229, "step": 82180 }, { "epoch": 0.11, "learning_rate": 4.4863125e-05, "loss": 1.3811, "step": 82190 }, { "epoch": 0.11, "learning_rate": 4.48625e-05, "loss": 1.5313, "step": 82200 }, { "epoch": 0.11, "learning_rate": 4.4861875e-05, "loss": 1.3627, "step": 82210 }, { "epoch": 0.11, "learning_rate": 4.486125000000001e-05, "loss": 1.6532, "step": 82220 }, { "epoch": 0.11, "learning_rate": 4.4860625000000004e-05, "loss": 1.6613, "step": 82230 }, { "epoch": 0.11, "learning_rate": 4.486e-05, "loss": 1.6563, "step": 82240 }, { "epoch": 0.11, "learning_rate": 4.4859375000000004e-05, "loss": 1.2829, "step": 82250 }, { "epoch": 0.11, "learning_rate": 4.485875e-05, "loss": 1.5437, "step": 82260 }, { "epoch": 0.11, "learning_rate": 4.4858125e-05, "loss": 1.7486, "step": 82270 }, { "epoch": 0.11, "learning_rate": 4.48575e-05, "loss": 1.526, "step": 82280 }, { "epoch": 0.11, "learning_rate": 4.4856875000000006e-05, "loss": 1.3492, "step": 82290 }, { "epoch": 0.11, "learning_rate": 4.485625e-05, "loss": 1.5503, "step": 82300 }, { "epoch": 0.11, "learning_rate": 4.4855625e-05, "loss": 1.4401, "step": 82310 }, { "epoch": 0.11, "learning_rate": 4.4855e-05, "loss": 1.5127, "step": 82320 }, { "epoch": 0.11, "learning_rate": 4.485437500000001e-05, "loss": 1.3392, "step": 82330 }, { "epoch": 0.11, "learning_rate": 4.485375e-05, "loss": 1.5977, "step": 82340 }, { "epoch": 0.11, "learning_rate": 4.4853125e-05, "loss": 1.6007, "step": 82350 }, { "epoch": 0.11, "learning_rate": 4.4852500000000005e-05, "loss": 1.5114, "step": 82360 }, { "epoch": 0.11, "learning_rate": 4.4851875e-05, "loss": 1.6473, "step": 82370 }, { "epoch": 0.11, "learning_rate": 4.485125e-05, "loss": 1.5066, "step": 82380 }, { "epoch": 0.11, "learning_rate": 4.4850625e-05, "loss": 1.4111, "step": 82390 }, { "epoch": 0.11, "learning_rate": 4.4850000000000006e-05, "loss": 1.5392, "step": 82400 }, { "epoch": 0.11, "learning_rate": 4.4849375e-05, "loss": 1.7211, "step": 82410 }, { "epoch": 0.11, "learning_rate": 4.484875e-05, "loss": 1.5555, "step": 82420 }, { "epoch": 0.11, "learning_rate": 4.4848125000000004e-05, "loss": 1.355, "step": 82430 }, { "epoch": 0.11, "learning_rate": 4.48475e-05, "loss": 1.5103, "step": 82440 }, { "epoch": 0.11, "learning_rate": 4.4846875000000004e-05, "loss": 1.6527, "step": 82450 }, { "epoch": 0.11, "learning_rate": 4.484625e-05, "loss": 1.6118, "step": 82460 }, { "epoch": 0.11, "learning_rate": 4.4845625000000005e-05, "loss": 1.277, "step": 82470 }, { "epoch": 0.11, "learning_rate": 4.4845e-05, "loss": 1.5735, "step": 82480 }, { "epoch": 0.11, "learning_rate": 4.4844375e-05, "loss": 1.4339, "step": 82490 }, { "epoch": 0.11, "learning_rate": 4.484375e-05, "loss": 1.3338, "step": 82500 }, { "epoch": 0.11, "learning_rate": 4.4843125e-05, "loss": 1.5315, "step": 82510 }, { "epoch": 0.11, "learning_rate": 4.48425e-05, "loss": 1.5746, "step": 82520 }, { "epoch": 0.11, "learning_rate": 4.4841875e-05, "loss": 1.2171, "step": 82530 }, { "epoch": 0.11, "learning_rate": 4.4841250000000004e-05, "loss": 1.3223, "step": 82540 }, { "epoch": 0.11, "learning_rate": 4.4840625e-05, "loss": 1.4328, "step": 82550 }, { "epoch": 0.11, "learning_rate": 4.4840000000000004e-05, "loss": 1.6117, "step": 82560 }, { "epoch": 0.11, "learning_rate": 4.4839375e-05, "loss": 1.5989, "step": 82570 }, { "epoch": 0.11, "learning_rate": 4.483875e-05, "loss": 1.4765, "step": 82580 }, { "epoch": 0.11, "learning_rate": 4.4838125e-05, "loss": 1.648, "step": 82590 }, { "epoch": 0.12, "learning_rate": 4.4837500000000006e-05, "loss": 1.5719, "step": 82600 }, { "epoch": 0.12, "learning_rate": 4.4836875e-05, "loss": 1.4712, "step": 82610 }, { "epoch": 0.12, "learning_rate": 4.483625e-05, "loss": 1.5268, "step": 82620 }, { "epoch": 0.12, "learning_rate": 4.4835625e-05, "loss": 1.2698, "step": 82630 }, { "epoch": 0.12, "learning_rate": 4.483500000000001e-05, "loss": 1.5869, "step": 82640 }, { "epoch": 0.12, "learning_rate": 4.4834375000000004e-05, "loss": 1.6523, "step": 82650 }, { "epoch": 0.12, "learning_rate": 4.483375e-05, "loss": 1.3633, "step": 82660 }, { "epoch": 0.12, "learning_rate": 4.4833125000000005e-05, "loss": 1.5404, "step": 82670 }, { "epoch": 0.12, "learning_rate": 4.48325e-05, "loss": 1.5701, "step": 82680 }, { "epoch": 0.12, "learning_rate": 4.4831875e-05, "loss": 1.626, "step": 82690 }, { "epoch": 0.12, "learning_rate": 4.483125e-05, "loss": 1.4947, "step": 82700 }, { "epoch": 0.12, "learning_rate": 4.4830625000000006e-05, "loss": 1.6299, "step": 82710 }, { "epoch": 0.12, "learning_rate": 4.483e-05, "loss": 1.4362, "step": 82720 }, { "epoch": 0.12, "learning_rate": 4.4829375e-05, "loss": 1.4627, "step": 82730 }, { "epoch": 0.12, "learning_rate": 4.4828750000000003e-05, "loss": 1.5763, "step": 82740 }, { "epoch": 0.12, "learning_rate": 4.482812500000001e-05, "loss": 1.2443, "step": 82750 }, { "epoch": 0.12, "learning_rate": 4.48275e-05, "loss": 1.6674, "step": 82760 }, { "epoch": 0.12, "learning_rate": 4.4826875e-05, "loss": 1.6488, "step": 82770 }, { "epoch": 0.12, "learning_rate": 4.4826250000000005e-05, "loss": 1.5309, "step": 82780 }, { "epoch": 0.12, "learning_rate": 4.4825625e-05, "loss": 1.3235, "step": 82790 }, { "epoch": 0.12, "learning_rate": 4.4825e-05, "loss": 1.6175, "step": 82800 }, { "epoch": 0.12, "learning_rate": 4.4824375e-05, "loss": 1.3674, "step": 82810 }, { "epoch": 0.12, "learning_rate": 4.4823750000000006e-05, "loss": 1.6464, "step": 82820 }, { "epoch": 0.12, "learning_rate": 4.4823125e-05, "loss": 1.3286, "step": 82830 }, { "epoch": 0.12, "learning_rate": 4.48225e-05, "loss": 1.5554, "step": 82840 }, { "epoch": 0.12, "learning_rate": 4.4821875000000004e-05, "loss": 1.64, "step": 82850 }, { "epoch": 0.12, "learning_rate": 4.482125e-05, "loss": 1.5028, "step": 82860 }, { "epoch": 0.12, "learning_rate": 4.4820625e-05, "loss": 1.4802, "step": 82870 }, { "epoch": 0.12, "learning_rate": 4.482e-05, "loss": 1.4672, "step": 82880 }, { "epoch": 0.12, "learning_rate": 4.4819375000000005e-05, "loss": 1.4932, "step": 82890 }, { "epoch": 0.12, "learning_rate": 4.481875e-05, "loss": 1.7124, "step": 82900 }, { "epoch": 0.12, "learning_rate": 4.4818125e-05, "loss": 1.4375, "step": 82910 }, { "epoch": 0.12, "learning_rate": 4.48175e-05, "loss": 1.6678, "step": 82920 }, { "epoch": 0.12, "learning_rate": 4.4816875e-05, "loss": 1.6212, "step": 82930 }, { "epoch": 0.12, "learning_rate": 4.481625e-05, "loss": 1.542, "step": 82940 }, { "epoch": 0.12, "learning_rate": 4.4815625e-05, "loss": 1.5036, "step": 82950 }, { "epoch": 0.12, "learning_rate": 4.4815000000000004e-05, "loss": 1.4547, "step": 82960 }, { "epoch": 0.12, "learning_rate": 4.4814375e-05, "loss": 1.2087, "step": 82970 }, { "epoch": 0.12, "learning_rate": 4.4813750000000004e-05, "loss": 1.464, "step": 82980 }, { "epoch": 0.12, "learning_rate": 4.4813125e-05, "loss": 1.7552, "step": 82990 }, { "epoch": 0.12, "learning_rate": 4.4812500000000005e-05, "loss": 1.3078, "step": 83000 }, { "epoch": 0.12, "eval_loss": 1.4462056159973145, "eval_runtime": 1316.0015, "eval_samples_per_second": 44.1, "eval_steps_per_second": 5.513, "step": 83000 }, { "epoch": 0.12, "learning_rate": 4.4811875e-05, "loss": 1.5859, "step": 83010 }, { "epoch": 0.12, "learning_rate": 4.4811250000000006e-05, "loss": 1.3814, "step": 83020 }, { "epoch": 0.12, "learning_rate": 4.4810625e-05, "loss": 1.4878, "step": 83030 }, { "epoch": 0.12, "learning_rate": 4.481e-05, "loss": 1.2532, "step": 83040 }, { "epoch": 0.12, "learning_rate": 4.4809375e-05, "loss": 1.6157, "step": 83050 }, { "epoch": 0.12, "learning_rate": 4.480875000000001e-05, "loss": 1.4454, "step": 83060 }, { "epoch": 0.12, "learning_rate": 4.4808125000000004e-05, "loss": 1.4651, "step": 83070 }, { "epoch": 0.12, "learning_rate": 4.48075e-05, "loss": 1.4749, "step": 83080 }, { "epoch": 0.12, "learning_rate": 4.4806875000000005e-05, "loss": 1.6538, "step": 83090 }, { "epoch": 0.12, "learning_rate": 4.480625e-05, "loss": 1.4807, "step": 83100 }, { "epoch": 0.12, "learning_rate": 4.4805625e-05, "loss": 1.5258, "step": 83110 }, { "epoch": 0.12, "learning_rate": 4.4805e-05, "loss": 1.4385, "step": 83120 }, { "epoch": 0.12, "learning_rate": 4.4804375000000006e-05, "loss": 1.5038, "step": 83130 }, { "epoch": 0.12, "learning_rate": 4.480375e-05, "loss": 1.3973, "step": 83140 }, { "epoch": 0.12, "learning_rate": 4.4803125e-05, "loss": 1.4399, "step": 83150 }, { "epoch": 0.12, "learning_rate": 4.4802500000000003e-05, "loss": 1.5618, "step": 83160 }, { "epoch": 0.12, "learning_rate": 4.480187500000001e-05, "loss": 1.6071, "step": 83170 }, { "epoch": 0.12, "learning_rate": 4.480125e-05, "loss": 1.6341, "step": 83180 }, { "epoch": 0.12, "learning_rate": 4.4800625e-05, "loss": 1.7721, "step": 83190 }, { "epoch": 0.12, "learning_rate": 4.4800000000000005e-05, "loss": 1.5304, "step": 83200 }, { "epoch": 0.12, "learning_rate": 4.4799375e-05, "loss": 1.473, "step": 83210 }, { "epoch": 0.12, "learning_rate": 4.479875e-05, "loss": 1.6972, "step": 83220 }, { "epoch": 0.12, "learning_rate": 4.4798125e-05, "loss": 1.5866, "step": 83230 }, { "epoch": 0.12, "learning_rate": 4.4797500000000006e-05, "loss": 1.3994, "step": 83240 }, { "epoch": 0.12, "learning_rate": 4.4796875e-05, "loss": 1.4615, "step": 83250 }, { "epoch": 0.12, "learning_rate": 4.479625e-05, "loss": 1.6267, "step": 83260 }, { "epoch": 0.12, "learning_rate": 4.4795625000000004e-05, "loss": 1.675, "step": 83270 }, { "epoch": 0.12, "learning_rate": 4.4795e-05, "loss": 1.4286, "step": 83280 }, { "epoch": 0.12, "learning_rate": 4.4794375e-05, "loss": 1.503, "step": 83290 }, { "epoch": 0.12, "learning_rate": 4.479375e-05, "loss": 1.6209, "step": 83300 }, { "epoch": 0.12, "learning_rate": 4.4793125000000005e-05, "loss": 1.6292, "step": 83310 }, { "epoch": 0.12, "learning_rate": 4.47925e-05, "loss": 1.4949, "step": 83320 }, { "epoch": 0.12, "learning_rate": 4.4791875e-05, "loss": 1.5867, "step": 83330 }, { "epoch": 0.12, "learning_rate": 4.479125e-05, "loss": 1.3359, "step": 83340 }, { "epoch": 0.12, "learning_rate": 4.4790625e-05, "loss": 1.5656, "step": 83350 }, { "epoch": 0.12, "learning_rate": 4.479e-05, "loss": 1.7793, "step": 83360 }, { "epoch": 0.12, "learning_rate": 4.4789375e-05, "loss": 1.587, "step": 83370 }, { "epoch": 0.12, "learning_rate": 4.4788750000000004e-05, "loss": 1.5443, "step": 83380 }, { "epoch": 0.12, "learning_rate": 4.4788125e-05, "loss": 1.3219, "step": 83390 }, { "epoch": 0.12, "learning_rate": 4.4787500000000004e-05, "loss": 1.5633, "step": 83400 }, { "epoch": 0.12, "learning_rate": 4.4786875e-05, "loss": 1.4544, "step": 83410 }, { "epoch": 0.12, "learning_rate": 4.4786250000000005e-05, "loss": 1.5027, "step": 83420 }, { "epoch": 0.12, "learning_rate": 4.4785625e-05, "loss": 1.4668, "step": 83430 }, { "epoch": 0.12, "learning_rate": 4.4785000000000006e-05, "loss": 1.7024, "step": 83440 }, { "epoch": 0.12, "learning_rate": 4.4784375e-05, "loss": 1.6962, "step": 83450 }, { "epoch": 0.12, "learning_rate": 4.478375e-05, "loss": 1.706, "step": 83460 }, { "epoch": 0.12, "learning_rate": 4.4783125e-05, "loss": 1.4803, "step": 83470 }, { "epoch": 0.12, "learning_rate": 4.47825e-05, "loss": 1.5953, "step": 83480 }, { "epoch": 0.12, "learning_rate": 4.4781875000000004e-05, "loss": 1.5355, "step": 83490 }, { "epoch": 0.12, "learning_rate": 4.478125e-05, "loss": 1.5975, "step": 83500 }, { "epoch": 0.12, "learning_rate": 4.4780625000000005e-05, "loss": 1.664, "step": 83510 }, { "epoch": 0.12, "learning_rate": 4.478e-05, "loss": 1.4866, "step": 83520 }, { "epoch": 0.12, "learning_rate": 4.4779375e-05, "loss": 1.5094, "step": 83530 }, { "epoch": 0.12, "learning_rate": 4.477875e-05, "loss": 1.3458, "step": 83540 }, { "epoch": 0.12, "learning_rate": 4.4778125000000006e-05, "loss": 1.4763, "step": 83550 }, { "epoch": 0.12, "learning_rate": 4.47775e-05, "loss": 1.5102, "step": 83560 }, { "epoch": 0.12, "learning_rate": 4.4776875e-05, "loss": 1.4693, "step": 83570 }, { "epoch": 0.12, "learning_rate": 4.4776250000000003e-05, "loss": 1.3566, "step": 83580 }, { "epoch": 0.12, "learning_rate": 4.477562500000001e-05, "loss": 1.6605, "step": 83590 }, { "epoch": 0.12, "learning_rate": 4.4775e-05, "loss": 1.7624, "step": 83600 }, { "epoch": 0.12, "learning_rate": 4.4774375e-05, "loss": 1.7529, "step": 83610 }, { "epoch": 0.12, "learning_rate": 4.4773750000000005e-05, "loss": 1.6253, "step": 83620 }, { "epoch": 0.12, "learning_rate": 4.4773125e-05, "loss": 1.7321, "step": 83630 }, { "epoch": 0.12, "learning_rate": 4.47725e-05, "loss": 1.5803, "step": 83640 }, { "epoch": 0.12, "learning_rate": 4.4771875e-05, "loss": 1.3841, "step": 83650 }, { "epoch": 0.12, "learning_rate": 4.4771250000000006e-05, "loss": 1.6681, "step": 83660 }, { "epoch": 0.12, "learning_rate": 4.4770624999999996e-05, "loss": 1.543, "step": 83670 }, { "epoch": 0.12, "learning_rate": 4.477e-05, "loss": 1.5242, "step": 83680 }, { "epoch": 0.12, "learning_rate": 4.4769375000000004e-05, "loss": 1.6547, "step": 83690 }, { "epoch": 0.12, "learning_rate": 4.476875e-05, "loss": 1.7509, "step": 83700 }, { "epoch": 0.12, "learning_rate": 4.4768125e-05, "loss": 1.3346, "step": 83710 }, { "epoch": 0.12, "learning_rate": 4.47675e-05, "loss": 1.7491, "step": 83720 }, { "epoch": 0.12, "learning_rate": 4.4766875000000005e-05, "loss": 1.4958, "step": 83730 }, { "epoch": 0.12, "learning_rate": 4.476625e-05, "loss": 1.4004, "step": 83740 }, { "epoch": 0.12, "learning_rate": 4.4765625e-05, "loss": 1.4114, "step": 83750 }, { "epoch": 0.12, "learning_rate": 4.4765e-05, "loss": 1.4403, "step": 83760 }, { "epoch": 0.12, "learning_rate": 4.4764375e-05, "loss": 1.4935, "step": 83770 }, { "epoch": 0.12, "learning_rate": 4.476375e-05, "loss": 1.5457, "step": 83780 }, { "epoch": 0.12, "learning_rate": 4.4763125e-05, "loss": 1.5993, "step": 83790 }, { "epoch": 0.12, "learning_rate": 4.4762500000000004e-05, "loss": 1.4288, "step": 83800 }, { "epoch": 0.12, "learning_rate": 4.4761875e-05, "loss": 1.5356, "step": 83810 }, { "epoch": 0.12, "learning_rate": 4.4761250000000004e-05, "loss": 1.552, "step": 83820 }, { "epoch": 0.12, "learning_rate": 4.4760625e-05, "loss": 1.4646, "step": 83830 }, { "epoch": 0.12, "learning_rate": 4.4760000000000005e-05, "loss": 1.6723, "step": 83840 }, { "epoch": 0.12, "learning_rate": 4.4759375e-05, "loss": 1.4663, "step": 83850 }, { "epoch": 0.12, "learning_rate": 4.4758750000000006e-05, "loss": 1.4368, "step": 83860 }, { "epoch": 0.12, "learning_rate": 4.4758125e-05, "loss": 1.6256, "step": 83870 }, { "epoch": 0.12, "learning_rate": 4.47575e-05, "loss": 1.4338, "step": 83880 }, { "epoch": 0.12, "learning_rate": 4.4756875e-05, "loss": 1.4472, "step": 83890 }, { "epoch": 0.12, "learning_rate": 4.475625e-05, "loss": 1.6263, "step": 83900 }, { "epoch": 0.12, "learning_rate": 4.4755625000000004e-05, "loss": 1.6082, "step": 83910 }, { "epoch": 0.12, "learning_rate": 4.4755e-05, "loss": 1.3827, "step": 83920 }, { "epoch": 0.12, "learning_rate": 4.4754375000000005e-05, "loss": 1.3243, "step": 83930 }, { "epoch": 0.12, "learning_rate": 4.475375e-05, "loss": 1.638, "step": 83940 }, { "epoch": 0.12, "learning_rate": 4.4753125e-05, "loss": 1.7032, "step": 83950 }, { "epoch": 0.12, "learning_rate": 4.47525e-05, "loss": 1.6971, "step": 83960 }, { "epoch": 0.12, "learning_rate": 4.4751875000000006e-05, "loss": 1.3423, "step": 83970 }, { "epoch": 0.12, "learning_rate": 4.475125e-05, "loss": 1.7011, "step": 83980 }, { "epoch": 0.12, "learning_rate": 4.4750625e-05, "loss": 1.6616, "step": 83990 }, { "epoch": 0.12, "learning_rate": 4.4750000000000004e-05, "loss": 1.4493, "step": 84000 }, { "epoch": 0.12, "eval_loss": 1.4369900226593018, "eval_runtime": 1328.1934, "eval_samples_per_second": 43.695, "eval_steps_per_second": 5.462, "step": 84000 }, { "epoch": 0.12, "learning_rate": 4.474937500000001e-05, "loss": 1.4403, "step": 84010 }, { "epoch": 0.12, "learning_rate": 4.474875e-05, "loss": 1.4748, "step": 84020 }, { "epoch": 0.12, "learning_rate": 4.4748125e-05, "loss": 1.476, "step": 84030 }, { "epoch": 0.12, "learning_rate": 4.4747500000000005e-05, "loss": 1.5095, "step": 84040 }, { "epoch": 0.12, "learning_rate": 4.4746875e-05, "loss": 1.6167, "step": 84050 }, { "epoch": 0.12, "learning_rate": 4.474625e-05, "loss": 1.543, "step": 84060 }, { "epoch": 0.12, "learning_rate": 4.4745625e-05, "loss": 1.7204, "step": 84070 }, { "epoch": 0.12, "learning_rate": 4.4745000000000006e-05, "loss": 1.3553, "step": 84080 }, { "epoch": 0.12, "learning_rate": 4.4744374999999996e-05, "loss": 1.5904, "step": 84090 }, { "epoch": 0.12, "learning_rate": 4.474375e-05, "loss": 1.6537, "step": 84100 }, { "epoch": 0.12, "learning_rate": 4.4743125000000004e-05, "loss": 1.594, "step": 84110 }, { "epoch": 0.12, "learning_rate": 4.47425e-05, "loss": 1.3896, "step": 84120 }, { "epoch": 0.12, "learning_rate": 4.4741875e-05, "loss": 1.5567, "step": 84130 }, { "epoch": 0.12, "learning_rate": 4.474125e-05, "loss": 1.2513, "step": 84140 }, { "epoch": 0.12, "learning_rate": 4.4740625000000005e-05, "loss": 1.6185, "step": 84150 }, { "epoch": 0.12, "learning_rate": 4.474e-05, "loss": 1.6154, "step": 84160 }, { "epoch": 0.12, "learning_rate": 4.4739375e-05, "loss": 1.5861, "step": 84170 }, { "epoch": 0.12, "learning_rate": 4.473875e-05, "loss": 1.3465, "step": 84180 }, { "epoch": 0.12, "learning_rate": 4.4738125000000006e-05, "loss": 1.3497, "step": 84190 }, { "epoch": 0.12, "learning_rate": 4.47375e-05, "loss": 1.2957, "step": 84200 }, { "epoch": 0.12, "learning_rate": 4.4736875e-05, "loss": 1.3841, "step": 84210 }, { "epoch": 0.12, "learning_rate": 4.4736250000000004e-05, "loss": 1.4903, "step": 84220 }, { "epoch": 0.12, "learning_rate": 4.4735625e-05, "loss": 1.3496, "step": 84230 }, { "epoch": 0.12, "learning_rate": 4.4735000000000005e-05, "loss": 1.4785, "step": 84240 }, { "epoch": 0.12, "learning_rate": 4.4734375e-05, "loss": 1.3538, "step": 84250 }, { "epoch": 0.12, "learning_rate": 4.4733750000000005e-05, "loss": 1.5292, "step": 84260 }, { "epoch": 0.12, "learning_rate": 4.4733125e-05, "loss": 1.7829, "step": 84270 }, { "epoch": 0.12, "learning_rate": 4.47325e-05, "loss": 1.3971, "step": 84280 }, { "epoch": 0.12, "learning_rate": 4.4731875e-05, "loss": 1.4666, "step": 84290 }, { "epoch": 0.12, "learning_rate": 4.473125e-05, "loss": 1.6111, "step": 84300 }, { "epoch": 0.12, "learning_rate": 4.4730625000000003e-05, "loss": 1.7708, "step": 84310 }, { "epoch": 0.12, "learning_rate": 4.473e-05, "loss": 1.4573, "step": 84320 }, { "epoch": 0.12, "learning_rate": 4.4729375000000004e-05, "loss": 1.6421, "step": 84330 }, { "epoch": 0.12, "learning_rate": 4.472875e-05, "loss": 1.7238, "step": 84340 }, { "epoch": 0.12, "learning_rate": 4.4728125000000005e-05, "loss": 1.4124, "step": 84350 }, { "epoch": 0.12, "learning_rate": 4.47275e-05, "loss": 1.6742, "step": 84360 }, { "epoch": 0.12, "learning_rate": 4.4726875e-05, "loss": 1.5486, "step": 84370 }, { "epoch": 0.12, "learning_rate": 4.472625e-05, "loss": 1.4528, "step": 84380 }, { "epoch": 0.12, "learning_rate": 4.4725625000000006e-05, "loss": 1.3731, "step": 84390 }, { "epoch": 0.12, "learning_rate": 4.4725e-05, "loss": 1.5211, "step": 84400 }, { "epoch": 0.12, "learning_rate": 4.4724375e-05, "loss": 1.5392, "step": 84410 }, { "epoch": 0.12, "learning_rate": 4.4723750000000004e-05, "loss": 1.3402, "step": 84420 }, { "epoch": 0.12, "learning_rate": 4.472312500000001e-05, "loss": 1.2538, "step": 84430 }, { "epoch": 0.12, "learning_rate": 4.47225e-05, "loss": 1.4871, "step": 84440 }, { "epoch": 0.12, "learning_rate": 4.4721875e-05, "loss": 1.3854, "step": 84450 }, { "epoch": 0.12, "learning_rate": 4.4721250000000005e-05, "loss": 1.3816, "step": 84460 }, { "epoch": 0.12, "learning_rate": 4.4720625e-05, "loss": 1.5613, "step": 84470 }, { "epoch": 0.12, "learning_rate": 4.472e-05, "loss": 1.4094, "step": 84480 }, { "epoch": 0.12, "learning_rate": 4.4719375e-05, "loss": 1.7006, "step": 84490 }, { "epoch": 0.12, "learning_rate": 4.4718750000000006e-05, "loss": 1.4399, "step": 84500 }, { "epoch": 0.12, "learning_rate": 4.4718124999999996e-05, "loss": 1.3459, "step": 84510 }, { "epoch": 0.12, "learning_rate": 4.47175e-05, "loss": 1.4204, "step": 84520 }, { "epoch": 0.12, "learning_rate": 4.4716875000000004e-05, "loss": 1.5997, "step": 84530 }, { "epoch": 0.12, "learning_rate": 4.471625e-05, "loss": 1.4467, "step": 84540 }, { "epoch": 0.12, "learning_rate": 4.4715625e-05, "loss": 1.5208, "step": 84550 }, { "epoch": 0.12, "learning_rate": 4.4715e-05, "loss": 1.5345, "step": 84560 }, { "epoch": 0.12, "learning_rate": 4.4714375000000005e-05, "loss": 1.5077, "step": 84570 }, { "epoch": 0.12, "learning_rate": 4.471375e-05, "loss": 1.6534, "step": 84580 }, { "epoch": 0.12, "learning_rate": 4.4713125e-05, "loss": 1.5433, "step": 84590 }, { "epoch": 0.12, "learning_rate": 4.47125e-05, "loss": 1.4648, "step": 84600 }, { "epoch": 0.12, "learning_rate": 4.4711875000000006e-05, "loss": 1.5153, "step": 84610 }, { "epoch": 0.12, "learning_rate": 4.471125e-05, "loss": 1.5831, "step": 84620 }, { "epoch": 0.12, "learning_rate": 4.4710625e-05, "loss": 1.5631, "step": 84630 }, { "epoch": 0.12, "learning_rate": 4.4710000000000004e-05, "loss": 1.4648, "step": 84640 }, { "epoch": 0.12, "learning_rate": 4.4709375e-05, "loss": 1.4129, "step": 84650 }, { "epoch": 0.12, "learning_rate": 4.4708750000000005e-05, "loss": 1.5087, "step": 84660 }, { "epoch": 0.12, "learning_rate": 4.4708125e-05, "loss": 1.6395, "step": 84670 }, { "epoch": 0.12, "learning_rate": 4.4707500000000005e-05, "loss": 1.4557, "step": 84680 }, { "epoch": 0.12, "learning_rate": 4.4706875e-05, "loss": 1.5051, "step": 84690 }, { "epoch": 0.12, "learning_rate": 4.470625e-05, "loss": 1.5027, "step": 84700 }, { "epoch": 0.12, "learning_rate": 4.4705625e-05, "loss": 1.4823, "step": 84710 }, { "epoch": 0.12, "learning_rate": 4.4705e-05, "loss": 1.8266, "step": 84720 }, { "epoch": 0.12, "learning_rate": 4.4704375000000003e-05, "loss": 1.3692, "step": 84730 }, { "epoch": 0.12, "learning_rate": 4.470375e-05, "loss": 1.4546, "step": 84740 }, { "epoch": 0.12, "learning_rate": 4.4703125000000004e-05, "loss": 1.5259, "step": 84750 }, { "epoch": 0.12, "learning_rate": 4.47025e-05, "loss": 1.221, "step": 84760 }, { "epoch": 0.12, "learning_rate": 4.4701875000000005e-05, "loss": 1.5446, "step": 84770 }, { "epoch": 0.12, "learning_rate": 4.470125e-05, "loss": 1.4724, "step": 84780 }, { "epoch": 0.12, "learning_rate": 4.4700625e-05, "loss": 1.5056, "step": 84790 }, { "epoch": 0.12, "learning_rate": 4.47e-05, "loss": 1.4448, "step": 84800 }, { "epoch": 0.12, "learning_rate": 4.4699375000000006e-05, "loss": 1.6608, "step": 84810 }, { "epoch": 0.12, "learning_rate": 4.469875e-05, "loss": 1.3469, "step": 84820 }, { "epoch": 0.12, "learning_rate": 4.4698125e-05, "loss": 1.673, "step": 84830 }, { "epoch": 0.12, "learning_rate": 4.4697500000000004e-05, "loss": 1.6917, "step": 84840 }, { "epoch": 0.12, "learning_rate": 4.469687500000001e-05, "loss": 1.4308, "step": 84850 }, { "epoch": 0.12, "learning_rate": 4.469625e-05, "loss": 1.489, "step": 84860 }, { "epoch": 0.12, "learning_rate": 4.4695625e-05, "loss": 1.3757, "step": 84870 }, { "epoch": 0.12, "learning_rate": 4.4695000000000005e-05, "loss": 1.6002, "step": 84880 }, { "epoch": 0.12, "learning_rate": 4.4694375e-05, "loss": 1.4726, "step": 84890 }, { "epoch": 0.12, "learning_rate": 4.469375e-05, "loss": 1.2639, "step": 84900 }, { "epoch": 0.12, "learning_rate": 4.4693125e-05, "loss": 1.445, "step": 84910 }, { "epoch": 0.12, "learning_rate": 4.4692500000000006e-05, "loss": 1.3912, "step": 84920 }, { "epoch": 0.12, "learning_rate": 4.4691874999999996e-05, "loss": 1.547, "step": 84930 }, { "epoch": 0.12, "learning_rate": 4.469125e-05, "loss": 1.5091, "step": 84940 }, { "epoch": 0.12, "learning_rate": 4.4690625000000004e-05, "loss": 1.4797, "step": 84950 }, { "epoch": 0.12, "learning_rate": 4.469e-05, "loss": 1.6101, "step": 84960 }, { "epoch": 0.12, "learning_rate": 4.4689375e-05, "loss": 1.4285, "step": 84970 }, { "epoch": 0.12, "learning_rate": 4.468875e-05, "loss": 1.326, "step": 84980 }, { "epoch": 0.12, "learning_rate": 4.4688125000000005e-05, "loss": 1.7581, "step": 84990 }, { "epoch": 0.12, "learning_rate": 4.46875e-05, "loss": 1.2991, "step": 85000 }, { "epoch": 0.12, "eval_loss": 1.4417284727096558, "eval_runtime": 1312.1135, "eval_samples_per_second": 44.231, "eval_steps_per_second": 5.529, "step": 85000 }, { "epoch": 0.12, "learning_rate": 4.4686875e-05, "loss": 1.4935, "step": 85010 }, { "epoch": 0.12, "learning_rate": 4.468625e-05, "loss": 1.537, "step": 85020 }, { "epoch": 0.12, "learning_rate": 4.4685625000000006e-05, "loss": 1.5093, "step": 85030 }, { "epoch": 0.12, "learning_rate": 4.4685e-05, "loss": 1.7035, "step": 85040 }, { "epoch": 0.12, "learning_rate": 4.4684375e-05, "loss": 1.5164, "step": 85050 }, { "epoch": 0.12, "learning_rate": 4.4683750000000004e-05, "loss": 1.3748, "step": 85060 }, { "epoch": 0.12, "learning_rate": 4.4683125e-05, "loss": 1.3466, "step": 85070 }, { "epoch": 0.12, "learning_rate": 4.4682500000000005e-05, "loss": 1.3475, "step": 85080 }, { "epoch": 0.12, "learning_rate": 4.4681875e-05, "loss": 1.568, "step": 85090 }, { "epoch": 0.12, "learning_rate": 4.4681250000000005e-05, "loss": 1.6135, "step": 85100 }, { "epoch": 0.12, "learning_rate": 4.4680625e-05, "loss": 1.5003, "step": 85110 }, { "epoch": 0.12, "learning_rate": 4.468e-05, "loss": 1.6942, "step": 85120 }, { "epoch": 0.12, "learning_rate": 4.4679375e-05, "loss": 1.3928, "step": 85130 }, { "epoch": 0.12, "learning_rate": 4.467875e-05, "loss": 1.3957, "step": 85140 }, { "epoch": 0.12, "learning_rate": 4.4678125000000003e-05, "loss": 1.5436, "step": 85150 }, { "epoch": 0.12, "learning_rate": 4.46775e-05, "loss": 1.5695, "step": 85160 }, { "epoch": 0.12, "learning_rate": 4.4676875000000004e-05, "loss": 1.6585, "step": 85170 }, { "epoch": 0.12, "learning_rate": 4.467625e-05, "loss": 1.3796, "step": 85180 }, { "epoch": 0.12, "learning_rate": 4.4675625000000005e-05, "loss": 1.6963, "step": 85190 }, { "epoch": 0.12, "learning_rate": 4.4675e-05, "loss": 1.4618, "step": 85200 }, { "epoch": 0.12, "learning_rate": 4.4674375e-05, "loss": 1.6431, "step": 85210 }, { "epoch": 0.12, "learning_rate": 4.467375e-05, "loss": 1.4175, "step": 85220 }, { "epoch": 0.12, "learning_rate": 4.4673125000000006e-05, "loss": 1.5511, "step": 85230 }, { "epoch": 0.12, "learning_rate": 4.46725e-05, "loss": 1.3984, "step": 85240 }, { "epoch": 0.12, "learning_rate": 4.4671875e-05, "loss": 1.4237, "step": 85250 }, { "epoch": 0.12, "learning_rate": 4.4671250000000004e-05, "loss": 1.5254, "step": 85260 }, { "epoch": 0.12, "learning_rate": 4.467062500000001e-05, "loss": 1.6165, "step": 85270 }, { "epoch": 0.12, "learning_rate": 4.467e-05, "loss": 1.609, "step": 85280 }, { "epoch": 0.12, "learning_rate": 4.4669375e-05, "loss": 1.3991, "step": 85290 }, { "epoch": 0.12, "learning_rate": 4.4668750000000005e-05, "loss": 1.4116, "step": 85300 }, { "epoch": 0.12, "learning_rate": 4.4668125e-05, "loss": 1.5596, "step": 85310 }, { "epoch": 0.12, "learning_rate": 4.46675e-05, "loss": 1.3957, "step": 85320 }, { "epoch": 0.12, "learning_rate": 4.4666875e-05, "loss": 1.3374, "step": 85330 }, { "epoch": 0.12, "learning_rate": 4.4666250000000006e-05, "loss": 1.5063, "step": 85340 }, { "epoch": 0.12, "learning_rate": 4.4665625e-05, "loss": 1.6629, "step": 85350 }, { "epoch": 0.12, "learning_rate": 4.4665e-05, "loss": 1.4901, "step": 85360 }, { "epoch": 0.12, "learning_rate": 4.4664375000000004e-05, "loss": 1.4627, "step": 85370 }, { "epoch": 0.12, "learning_rate": 4.466375000000001e-05, "loss": 1.4272, "step": 85380 }, { "epoch": 0.12, "learning_rate": 4.4663125e-05, "loss": 1.5283, "step": 85390 }, { "epoch": 0.12, "learning_rate": 4.46625e-05, "loss": 1.4644, "step": 85400 }, { "epoch": 0.12, "learning_rate": 4.4661875000000005e-05, "loss": 1.5684, "step": 85410 }, { "epoch": 0.12, "learning_rate": 4.466125e-05, "loss": 1.4883, "step": 85420 }, { "epoch": 0.12, "learning_rate": 4.4660625e-05, "loss": 1.5926, "step": 85430 }, { "epoch": 0.12, "learning_rate": 4.466e-05, "loss": 1.6682, "step": 85440 }, { "epoch": 0.12, "learning_rate": 4.4659375000000006e-05, "loss": 1.5285, "step": 85450 }, { "epoch": 0.12, "learning_rate": 4.465875e-05, "loss": 1.3116, "step": 85460 }, { "epoch": 0.12, "learning_rate": 4.4658125e-05, "loss": 1.621, "step": 85470 }, { "epoch": 0.12, "learning_rate": 4.4657500000000004e-05, "loss": 1.4759, "step": 85480 }, { "epoch": 0.12, "learning_rate": 4.4656875e-05, "loss": 1.4965, "step": 85490 }, { "epoch": 0.12, "learning_rate": 4.465625e-05, "loss": 1.5521, "step": 85500 }, { "epoch": 0.12, "learning_rate": 4.4655625e-05, "loss": 1.8429, "step": 85510 }, { "epoch": 0.12, "learning_rate": 4.4655000000000005e-05, "loss": 1.4264, "step": 85520 }, { "epoch": 0.12, "learning_rate": 4.4654375e-05, "loss": 1.3422, "step": 85530 }, { "epoch": 0.12, "learning_rate": 4.465375e-05, "loss": 1.5413, "step": 85540 }, { "epoch": 0.12, "learning_rate": 4.4653125e-05, "loss": 1.5139, "step": 85550 }, { "epoch": 0.12, "learning_rate": 4.46525e-05, "loss": 1.4131, "step": 85560 }, { "epoch": 0.12, "learning_rate": 4.4651875000000004e-05, "loss": 1.5166, "step": 85570 }, { "epoch": 0.12, "learning_rate": 4.465125e-05, "loss": 1.45, "step": 85580 }, { "epoch": 0.12, "learning_rate": 4.4650625000000004e-05, "loss": 1.5626, "step": 85590 }, { "epoch": 0.12, "learning_rate": 4.465e-05, "loss": 1.5026, "step": 85600 }, { "epoch": 0.12, "learning_rate": 4.4649375000000005e-05, "loss": 1.4229, "step": 85610 }, { "epoch": 0.12, "learning_rate": 4.464875e-05, "loss": 1.4614, "step": 85620 }, { "epoch": 0.12, "learning_rate": 4.4648125e-05, "loss": 1.3803, "step": 85630 }, { "epoch": 0.12, "learning_rate": 4.46475e-05, "loss": 1.3859, "step": 85640 }, { "epoch": 0.12, "learning_rate": 4.4646875000000006e-05, "loss": 1.452, "step": 85650 }, { "epoch": 0.12, "learning_rate": 4.464625e-05, "loss": 1.4627, "step": 85660 }, { "epoch": 0.12, "learning_rate": 4.4645625e-05, "loss": 1.4705, "step": 85670 }, { "epoch": 0.12, "learning_rate": 4.4645000000000004e-05, "loss": 1.5148, "step": 85680 }, { "epoch": 0.12, "learning_rate": 4.4644375e-05, "loss": 1.359, "step": 85690 }, { "epoch": 0.12, "learning_rate": 4.464375e-05, "loss": 1.4362, "step": 85700 }, { "epoch": 0.12, "learning_rate": 4.4643125e-05, "loss": 1.6526, "step": 85710 }, { "epoch": 0.12, "learning_rate": 4.4642500000000005e-05, "loss": 1.6857, "step": 85720 }, { "epoch": 0.12, "learning_rate": 4.4641875e-05, "loss": 1.4436, "step": 85730 }, { "epoch": 0.12, "learning_rate": 4.464125e-05, "loss": 1.8214, "step": 85740 }, { "epoch": 0.12, "learning_rate": 4.4640625e-05, "loss": 1.4863, "step": 85750 }, { "epoch": 0.12, "learning_rate": 4.4640000000000006e-05, "loss": 1.5939, "step": 85760 }, { "epoch": 0.12, "learning_rate": 4.4639375e-05, "loss": 1.4058, "step": 85770 }, { "epoch": 0.12, "learning_rate": 4.463875e-05, "loss": 1.5174, "step": 85780 }, { "epoch": 0.12, "learning_rate": 4.4638125000000004e-05, "loss": 1.6163, "step": 85790 }, { "epoch": 0.12, "learning_rate": 4.463750000000001e-05, "loss": 1.487, "step": 85800 }, { "epoch": 0.12, "learning_rate": 4.4636875e-05, "loss": 1.2519, "step": 85810 }, { "epoch": 0.12, "learning_rate": 4.463625e-05, "loss": 1.6143, "step": 85820 }, { "epoch": 0.12, "learning_rate": 4.4635625000000005e-05, "loss": 1.5178, "step": 85830 }, { "epoch": 0.12, "learning_rate": 4.4635e-05, "loss": 1.5439, "step": 85840 }, { "epoch": 0.12, "learning_rate": 4.4634375e-05, "loss": 1.5651, "step": 85850 }, { "epoch": 0.12, "learning_rate": 4.463375e-05, "loss": 1.3875, "step": 85860 }, { "epoch": 0.12, "learning_rate": 4.4633125000000006e-05, "loss": 1.7231, "step": 85870 }, { "epoch": 0.12, "learning_rate": 4.46325e-05, "loss": 1.3953, "step": 85880 }, { "epoch": 0.12, "learning_rate": 4.4631875e-05, "loss": 1.5026, "step": 85890 }, { "epoch": 0.12, "learning_rate": 4.4631250000000004e-05, "loss": 1.5158, "step": 85900 }, { "epoch": 0.12, "learning_rate": 4.4630625e-05, "loss": 1.5342, "step": 85910 }, { "epoch": 0.12, "learning_rate": 4.463e-05, "loss": 1.6596, "step": 85920 }, { "epoch": 0.12, "learning_rate": 4.4629375e-05, "loss": 1.4748, "step": 85930 }, { "epoch": 0.12, "learning_rate": 4.4628750000000005e-05, "loss": 1.5232, "step": 85940 }, { "epoch": 0.12, "learning_rate": 4.4628125e-05, "loss": 1.4674, "step": 85950 }, { "epoch": 0.12, "learning_rate": 4.46275e-05, "loss": 1.4499, "step": 85960 }, { "epoch": 0.12, "learning_rate": 4.4626875e-05, "loss": 1.3628, "step": 85970 }, { "epoch": 0.12, "learning_rate": 4.462625e-05, "loss": 1.268, "step": 85980 }, { "epoch": 0.12, "learning_rate": 4.4625625000000004e-05, "loss": 1.3458, "step": 85990 }, { "epoch": 0.12, "learning_rate": 4.4625e-05, "loss": 1.4783, "step": 86000 }, { "epoch": 0.12, "eval_loss": 1.428999900817871, "eval_runtime": 1729.068, "eval_samples_per_second": 33.565, "eval_steps_per_second": 4.196, "step": 86000 }, { "epoch": 0.12, "learning_rate": 4.4624375000000004e-05, "loss": 1.5125, "step": 86010 }, { "epoch": 0.12, "learning_rate": 4.462375e-05, "loss": 1.4829, "step": 86020 }, { "epoch": 0.12, "learning_rate": 4.4623125000000005e-05, "loss": 1.4894, "step": 86030 }, { "epoch": 0.12, "learning_rate": 4.46225e-05, "loss": 1.4293, "step": 86040 }, { "epoch": 0.12, "learning_rate": 4.4621875e-05, "loss": 1.5504, "step": 86050 }, { "epoch": 0.12, "learning_rate": 4.462125e-05, "loss": 1.7119, "step": 86060 }, { "epoch": 0.12, "learning_rate": 4.4620625000000006e-05, "loss": 1.5899, "step": 86070 }, { "epoch": 0.12, "learning_rate": 4.462e-05, "loss": 1.3503, "step": 86080 }, { "epoch": 0.12, "learning_rate": 4.4619375e-05, "loss": 1.475, "step": 86090 }, { "epoch": 0.12, "learning_rate": 4.4618750000000004e-05, "loss": 1.5083, "step": 86100 }, { "epoch": 0.12, "learning_rate": 4.4618125e-05, "loss": 1.3787, "step": 86110 }, { "epoch": 0.12, "learning_rate": 4.46175e-05, "loss": 1.4891, "step": 86120 }, { "epoch": 0.12, "learning_rate": 4.4616875e-05, "loss": 1.6537, "step": 86130 }, { "epoch": 0.12, "learning_rate": 4.4616250000000005e-05, "loss": 1.4884, "step": 86140 }, { "epoch": 0.12, "learning_rate": 4.4615625e-05, "loss": 1.5534, "step": 86150 }, { "epoch": 0.12, "learning_rate": 4.4615e-05, "loss": 1.5189, "step": 86160 }, { "epoch": 0.12, "learning_rate": 4.4614375e-05, "loss": 1.4045, "step": 86170 }, { "epoch": 0.12, "learning_rate": 4.4613750000000006e-05, "loss": 1.3365, "step": 86180 }, { "epoch": 0.12, "learning_rate": 4.4613125e-05, "loss": 1.3282, "step": 86190 }, { "epoch": 0.12, "learning_rate": 4.46125e-05, "loss": 1.732, "step": 86200 }, { "epoch": 0.12, "learning_rate": 4.4611875000000004e-05, "loss": 1.4706, "step": 86210 }, { "epoch": 0.12, "learning_rate": 4.461125000000001e-05, "loss": 1.5558, "step": 86220 }, { "epoch": 0.12, "learning_rate": 4.4610625e-05, "loss": 1.6804, "step": 86230 }, { "epoch": 0.12, "learning_rate": 4.461e-05, "loss": 1.3207, "step": 86240 }, { "epoch": 0.12, "learning_rate": 4.4609375000000005e-05, "loss": 1.4183, "step": 86250 }, { "epoch": 0.12, "learning_rate": 4.460875e-05, "loss": 1.4494, "step": 86260 }, { "epoch": 0.12, "learning_rate": 4.4608125e-05, "loss": 1.6188, "step": 86270 }, { "epoch": 0.12, "learning_rate": 4.46075e-05, "loss": 1.7503, "step": 86280 }, { "epoch": 0.12, "learning_rate": 4.4606875000000006e-05, "loss": 1.6721, "step": 86290 }, { "epoch": 0.12, "learning_rate": 4.460625e-05, "loss": 1.5214, "step": 86300 }, { "epoch": 0.12, "learning_rate": 4.4605625e-05, "loss": 1.4248, "step": 86310 }, { "epoch": 0.12, "learning_rate": 4.4605000000000004e-05, "loss": 1.5937, "step": 86320 }, { "epoch": 0.12, "learning_rate": 4.4604375e-05, "loss": 1.4767, "step": 86330 }, { "epoch": 0.12, "learning_rate": 4.460375e-05, "loss": 1.3523, "step": 86340 }, { "epoch": 0.12, "learning_rate": 4.4603125e-05, "loss": 1.5996, "step": 86350 }, { "epoch": 0.12, "learning_rate": 4.4602500000000005e-05, "loss": 1.4952, "step": 86360 }, { "epoch": 0.12, "learning_rate": 4.4601875e-05, "loss": 1.5362, "step": 86370 }, { "epoch": 0.12, "learning_rate": 4.460125e-05, "loss": 1.5394, "step": 86380 }, { "epoch": 0.12, "learning_rate": 4.4600625e-05, "loss": 1.6004, "step": 86390 }, { "epoch": 0.12, "learning_rate": 4.46e-05, "loss": 1.6299, "step": 86400 }, { "epoch": 0.12, "learning_rate": 4.4599375000000004e-05, "loss": 1.5198, "step": 86410 }, { "epoch": 0.12, "learning_rate": 4.459875e-05, "loss": 1.3414, "step": 86420 }, { "epoch": 0.12, "learning_rate": 4.4598125000000004e-05, "loss": 1.364, "step": 86430 }, { "epoch": 0.12, "learning_rate": 4.45975e-05, "loss": 1.5362, "step": 86440 }, { "epoch": 0.12, "learning_rate": 4.4596875000000005e-05, "loss": 1.3873, "step": 86450 }, { "epoch": 0.12, "learning_rate": 4.459625e-05, "loss": 1.3886, "step": 86460 }, { "epoch": 0.12, "learning_rate": 4.4595625e-05, "loss": 1.6766, "step": 86470 }, { "epoch": 0.12, "learning_rate": 4.4595e-05, "loss": 1.4028, "step": 86480 }, { "epoch": 0.12, "learning_rate": 4.4594375000000006e-05, "loss": 1.7467, "step": 86490 }, { "epoch": 0.12, "learning_rate": 4.459375e-05, "loss": 1.5332, "step": 86500 }, { "epoch": 0.12, "learning_rate": 4.4593125e-05, "loss": 1.4217, "step": 86510 }, { "epoch": 0.12, "learning_rate": 4.4592500000000004e-05, "loss": 1.587, "step": 86520 }, { "epoch": 0.12, "learning_rate": 4.4591875e-05, "loss": 1.4215, "step": 86530 }, { "epoch": 0.12, "learning_rate": 4.4591250000000004e-05, "loss": 1.6153, "step": 86540 }, { "epoch": 0.12, "learning_rate": 4.4590625e-05, "loss": 1.4087, "step": 86550 }, { "epoch": 0.12, "learning_rate": 4.4590000000000005e-05, "loss": 1.4799, "step": 86560 }, { "epoch": 0.12, "learning_rate": 4.4589375e-05, "loss": 1.5093, "step": 86570 }, { "epoch": 0.12, "learning_rate": 4.458875e-05, "loss": 1.4195, "step": 86580 }, { "epoch": 0.12, "learning_rate": 4.4588125e-05, "loss": 1.5863, "step": 86590 }, { "epoch": 0.12, "learning_rate": 4.4587500000000006e-05, "loss": 1.5965, "step": 86600 }, { "epoch": 0.12, "learning_rate": 4.4586875e-05, "loss": 1.4131, "step": 86610 }, { "epoch": 0.12, "learning_rate": 4.458625e-05, "loss": 1.469, "step": 86620 }, { "epoch": 0.12, "learning_rate": 4.4585625000000004e-05, "loss": 1.5468, "step": 86630 }, { "epoch": 0.12, "learning_rate": 4.458500000000001e-05, "loss": 1.4247, "step": 86640 }, { "epoch": 0.12, "learning_rate": 4.4584375e-05, "loss": 1.3583, "step": 86650 }, { "epoch": 0.12, "learning_rate": 4.458375e-05, "loss": 1.4487, "step": 86660 }, { "epoch": 0.12, "learning_rate": 4.4583125000000005e-05, "loss": 1.4305, "step": 86670 }, { "epoch": 0.12, "learning_rate": 4.45825e-05, "loss": 1.5321, "step": 86680 }, { "epoch": 0.12, "learning_rate": 4.4581875e-05, "loss": 1.5266, "step": 86690 }, { "epoch": 0.12, "learning_rate": 4.458125e-05, "loss": 1.4765, "step": 86700 }, { "epoch": 0.12, "learning_rate": 4.4580625000000007e-05, "loss": 1.4329, "step": 86710 }, { "epoch": 0.12, "learning_rate": 4.458e-05, "loss": 1.4963, "step": 86720 }, { "epoch": 0.12, "learning_rate": 4.4579375e-05, "loss": 1.5191, "step": 86730 }, { "epoch": 0.12, "learning_rate": 4.4578750000000004e-05, "loss": 1.3249, "step": 86740 }, { "epoch": 0.12, "learning_rate": 4.4578125e-05, "loss": 1.5893, "step": 86750 }, { "epoch": 0.12, "learning_rate": 4.45775e-05, "loss": 1.4137, "step": 86760 }, { "epoch": 0.12, "learning_rate": 4.4576875e-05, "loss": 1.5176, "step": 86770 }, { "epoch": 0.12, "learning_rate": 4.4576250000000005e-05, "loss": 1.7564, "step": 86780 }, { "epoch": 0.12, "learning_rate": 4.4575625e-05, "loss": 1.3778, "step": 86790 }, { "epoch": 0.12, "learning_rate": 4.4575e-05, "loss": 1.2685, "step": 86800 }, { "epoch": 0.12, "learning_rate": 4.4574375e-05, "loss": 1.524, "step": 86810 }, { "epoch": 0.12, "learning_rate": 4.457375e-05, "loss": 1.3661, "step": 86820 }, { "epoch": 0.12, "learning_rate": 4.4573125000000004e-05, "loss": 1.4523, "step": 86830 }, { "epoch": 0.12, "learning_rate": 4.45725e-05, "loss": 1.7043, "step": 86840 }, { "epoch": 0.12, "learning_rate": 4.4571875000000004e-05, "loss": 1.5962, "step": 86850 }, { "epoch": 0.12, "learning_rate": 4.457125e-05, "loss": 1.5959, "step": 86860 }, { "epoch": 0.12, "learning_rate": 4.4570625000000005e-05, "loss": 1.405, "step": 86870 }, { "epoch": 0.12, "learning_rate": 4.457e-05, "loss": 1.482, "step": 86880 }, { "epoch": 0.12, "learning_rate": 4.4569375e-05, "loss": 1.4251, "step": 86890 }, { "epoch": 0.12, "learning_rate": 4.456875e-05, "loss": 1.4401, "step": 86900 }, { "epoch": 0.12, "learning_rate": 4.4568125e-05, "loss": 1.5329, "step": 86910 }, { "epoch": 0.12, "learning_rate": 4.45675e-05, "loss": 1.3932, "step": 86920 }, { "epoch": 0.12, "learning_rate": 4.4566875e-05, "loss": 1.5057, "step": 86930 }, { "epoch": 0.12, "learning_rate": 4.4566250000000004e-05, "loss": 1.5163, "step": 86940 }, { "epoch": 0.12, "learning_rate": 4.4565625e-05, "loss": 1.4614, "step": 86950 }, { "epoch": 0.12, "learning_rate": 4.4565000000000004e-05, "loss": 1.569, "step": 86960 }, { "epoch": 0.12, "learning_rate": 4.4564375e-05, "loss": 1.638, "step": 86970 }, { "epoch": 0.12, "learning_rate": 4.4563750000000005e-05, "loss": 1.5598, "step": 86980 }, { "epoch": 0.12, "learning_rate": 4.4563125e-05, "loss": 1.3175, "step": 86990 }, { "epoch": 0.12, "learning_rate": 4.45625e-05, "loss": 1.6781, "step": 87000 }, { "epoch": 0.12, "eval_loss": 1.4281865358352661, "eval_runtime": 1294.5964, "eval_samples_per_second": 44.829, "eval_steps_per_second": 5.604, "step": 87000 }, { "epoch": 0.12, "learning_rate": 4.4561875e-05, "loss": 1.5103, "step": 87010 }, { "epoch": 0.12, "learning_rate": 4.4561250000000006e-05, "loss": 1.3669, "step": 87020 }, { "epoch": 0.12, "learning_rate": 4.4560625e-05, "loss": 1.3827, "step": 87030 }, { "epoch": 0.12, "learning_rate": 4.456e-05, "loss": 1.4205, "step": 87040 }, { "epoch": 0.12, "learning_rate": 4.4559375000000004e-05, "loss": 1.4861, "step": 87050 }, { "epoch": 0.12, "learning_rate": 4.455875000000001e-05, "loss": 1.5037, "step": 87060 }, { "epoch": 0.12, "learning_rate": 4.4558125e-05, "loss": 1.3657, "step": 87070 }, { "epoch": 0.12, "learning_rate": 4.45575e-05, "loss": 1.3165, "step": 87080 }, { "epoch": 0.12, "learning_rate": 4.4556875000000005e-05, "loss": 1.5128, "step": 87090 }, { "epoch": 0.12, "learning_rate": 4.455625e-05, "loss": 1.3025, "step": 87100 }, { "epoch": 0.12, "learning_rate": 4.4555625e-05, "loss": 1.5297, "step": 87110 }, { "epoch": 0.12, "learning_rate": 4.4555e-05, "loss": 1.4424, "step": 87120 }, { "epoch": 0.12, "learning_rate": 4.4554375000000007e-05, "loss": 1.4943, "step": 87130 }, { "epoch": 0.12, "learning_rate": 4.455375e-05, "loss": 1.4949, "step": 87140 }, { "epoch": 0.12, "learning_rate": 4.4553125e-05, "loss": 1.3194, "step": 87150 }, { "epoch": 0.12, "learning_rate": 4.4552500000000004e-05, "loss": 1.6403, "step": 87160 }, { "epoch": 0.12, "learning_rate": 4.4551875e-05, "loss": 1.5581, "step": 87170 }, { "epoch": 0.12, "learning_rate": 4.455125e-05, "loss": 1.4282, "step": 87180 }, { "epoch": 0.12, "learning_rate": 4.4550625e-05, "loss": 1.7395, "step": 87190 }, { "epoch": 0.12, "learning_rate": 4.4550000000000005e-05, "loss": 1.6565, "step": 87200 }, { "epoch": 0.12, "learning_rate": 4.4549375e-05, "loss": 1.4767, "step": 87210 }, { "epoch": 0.12, "learning_rate": 4.454875e-05, "loss": 1.5192, "step": 87220 }, { "epoch": 0.12, "learning_rate": 4.4548125e-05, "loss": 1.5892, "step": 87230 }, { "epoch": 0.12, "learning_rate": 4.45475e-05, "loss": 1.5558, "step": 87240 }, { "epoch": 0.12, "learning_rate": 4.4546875000000004e-05, "loss": 1.3337, "step": 87250 }, { "epoch": 0.12, "learning_rate": 4.454625e-05, "loss": 1.6173, "step": 87260 }, { "epoch": 0.12, "learning_rate": 4.4545625000000004e-05, "loss": 1.3854, "step": 87270 }, { "epoch": 0.12, "learning_rate": 4.4545e-05, "loss": 1.7052, "step": 87280 }, { "epoch": 0.12, "learning_rate": 4.4544375000000005e-05, "loss": 1.5103, "step": 87290 }, { "epoch": 0.12, "learning_rate": 4.454375e-05, "loss": 1.6925, "step": 87300 }, { "epoch": 0.12, "learning_rate": 4.4543125e-05, "loss": 1.5945, "step": 87310 }, { "epoch": 0.12, "learning_rate": 4.45425e-05, "loss": 1.4288, "step": 87320 }, { "epoch": 0.12, "learning_rate": 4.4541875e-05, "loss": 1.3974, "step": 87330 }, { "epoch": 0.12, "learning_rate": 4.454125e-05, "loss": 1.3534, "step": 87340 }, { "epoch": 0.12, "learning_rate": 4.4540625e-05, "loss": 1.6705, "step": 87350 }, { "epoch": 0.12, "learning_rate": 4.4540000000000004e-05, "loss": 1.4552, "step": 87360 }, { "epoch": 0.12, "learning_rate": 4.4539375e-05, "loss": 1.6636, "step": 87370 }, { "epoch": 0.12, "learning_rate": 4.4538750000000004e-05, "loss": 1.3962, "step": 87380 }, { "epoch": 0.12, "learning_rate": 4.4538125e-05, "loss": 1.339, "step": 87390 }, { "epoch": 0.12, "learning_rate": 4.4537500000000005e-05, "loss": 1.3964, "step": 87400 }, { "epoch": 0.12, "learning_rate": 4.4536875e-05, "loss": 1.4978, "step": 87410 }, { "epoch": 0.12, "learning_rate": 4.453625e-05, "loss": 1.5295, "step": 87420 }, { "epoch": 0.12, "learning_rate": 4.4535625e-05, "loss": 1.459, "step": 87430 }, { "epoch": 0.12, "learning_rate": 4.4535000000000006e-05, "loss": 1.5155, "step": 87440 }, { "epoch": 0.12, "learning_rate": 4.4534375e-05, "loss": 1.4435, "step": 87450 }, { "epoch": 0.12, "learning_rate": 4.453375e-05, "loss": 1.4853, "step": 87460 }, { "epoch": 0.12, "learning_rate": 4.4533125000000004e-05, "loss": 1.4588, "step": 87470 }, { "epoch": 0.12, "learning_rate": 4.453250000000001e-05, "loss": 1.3925, "step": 87480 }, { "epoch": 0.12, "learning_rate": 4.4531875e-05, "loss": 1.5785, "step": 87490 }, { "epoch": 0.12, "learning_rate": 4.453125e-05, "loss": 1.442, "step": 87500 }, { "epoch": 0.12, "learning_rate": 4.4530625000000005e-05, "loss": 1.4331, "step": 87510 }, { "epoch": 0.12, "learning_rate": 4.453e-05, "loss": 1.3544, "step": 87520 }, { "epoch": 0.12, "learning_rate": 4.4529375e-05, "loss": 1.3698, "step": 87530 }, { "epoch": 0.12, "learning_rate": 4.452875e-05, "loss": 1.5276, "step": 87540 }, { "epoch": 0.12, "learning_rate": 4.4528125000000007e-05, "loss": 1.4835, "step": 87550 }, { "epoch": 0.12, "learning_rate": 4.45275e-05, "loss": 1.3868, "step": 87560 }, { "epoch": 0.12, "learning_rate": 4.4526875e-05, "loss": 1.3649, "step": 87570 }, { "epoch": 0.12, "learning_rate": 4.4526250000000004e-05, "loss": 1.3825, "step": 87580 }, { "epoch": 0.12, "learning_rate": 4.4525625e-05, "loss": 1.5302, "step": 87590 }, { "epoch": 0.12, "learning_rate": 4.4525e-05, "loss": 1.5429, "step": 87600 }, { "epoch": 0.12, "learning_rate": 4.4524375e-05, "loss": 1.2997, "step": 87610 }, { "epoch": 0.12, "learning_rate": 4.4523750000000005e-05, "loss": 1.501, "step": 87620 }, { "epoch": 0.12, "learning_rate": 4.4523125e-05, "loss": 1.5378, "step": 87630 }, { "epoch": 0.12, "learning_rate": 4.45225e-05, "loss": 1.4362, "step": 87640 }, { "epoch": 0.12, "learning_rate": 4.4521875e-05, "loss": 1.4671, "step": 87650 }, { "epoch": 0.12, "learning_rate": 4.452125e-05, "loss": 1.3447, "step": 87660 }, { "epoch": 0.12, "learning_rate": 4.4520625000000004e-05, "loss": 1.3836, "step": 87670 }, { "epoch": 0.12, "learning_rate": 4.452e-05, "loss": 1.3675, "step": 87680 }, { "epoch": 0.12, "learning_rate": 4.4519375000000004e-05, "loss": 1.3947, "step": 87690 }, { "epoch": 0.12, "learning_rate": 4.451875e-05, "loss": 1.4093, "step": 87700 }, { "epoch": 0.12, "learning_rate": 4.4518125e-05, "loss": 1.3186, "step": 87710 }, { "epoch": 0.12, "learning_rate": 4.45175e-05, "loss": 1.4387, "step": 87720 }, { "epoch": 0.12, "learning_rate": 4.4516875000000006e-05, "loss": 1.5223, "step": 87730 }, { "epoch": 0.12, "learning_rate": 4.451625e-05, "loss": 1.5655, "step": 87740 }, { "epoch": 0.12, "learning_rate": 4.4515625e-05, "loss": 1.5602, "step": 87750 }, { "epoch": 0.12, "learning_rate": 4.4515e-05, "loss": 1.3669, "step": 87760 }, { "epoch": 0.12, "learning_rate": 4.4514375e-05, "loss": 1.4777, "step": 87770 }, { "epoch": 0.12, "learning_rate": 4.4513750000000004e-05, "loss": 1.4578, "step": 87780 }, { "epoch": 0.12, "learning_rate": 4.4513125e-05, "loss": 1.669, "step": 87790 }, { "epoch": 0.12, "learning_rate": 4.4512500000000005e-05, "loss": 1.4979, "step": 87800 }, { "epoch": 0.12, "learning_rate": 4.4511875e-05, "loss": 1.6064, "step": 87810 }, { "epoch": 0.12, "learning_rate": 4.4511250000000005e-05, "loss": 1.4462, "step": 87820 }, { "epoch": 0.12, "learning_rate": 4.4510625e-05, "loss": 1.3276, "step": 87830 }, { "epoch": 0.12, "learning_rate": 4.451e-05, "loss": 1.3904, "step": 87840 }, { "epoch": 0.12, "learning_rate": 4.4509375e-05, "loss": 1.6361, "step": 87850 }, { "epoch": 0.12, "learning_rate": 4.4508750000000006e-05, "loss": 1.568, "step": 87860 }, { "epoch": 0.12, "learning_rate": 4.4508125000000003e-05, "loss": 1.4423, "step": 87870 }, { "epoch": 0.12, "learning_rate": 4.45075e-05, "loss": 1.5946, "step": 87880 }, { "epoch": 0.12, "learning_rate": 4.4506875000000004e-05, "loss": 1.5399, "step": 87890 }, { "epoch": 0.12, "learning_rate": 4.450625000000001e-05, "loss": 1.7509, "step": 87900 }, { "epoch": 0.12, "learning_rate": 4.4505625e-05, "loss": 1.4505, "step": 87910 }, { "epoch": 0.12, "learning_rate": 4.4505e-05, "loss": 1.5848, "step": 87920 }, { "epoch": 0.12, "learning_rate": 4.4504375000000005e-05, "loss": 1.4078, "step": 87930 }, { "epoch": 0.12, "learning_rate": 4.450375e-05, "loss": 1.8506, "step": 87940 }, { "epoch": 0.12, "learning_rate": 4.4503125e-05, "loss": 1.2909, "step": 87950 }, { "epoch": 0.12, "learning_rate": 4.45025e-05, "loss": 1.6296, "step": 87960 }, { "epoch": 0.12, "learning_rate": 4.450187500000001e-05, "loss": 1.2706, "step": 87970 }, { "epoch": 0.12, "learning_rate": 4.450125e-05, "loss": 1.5197, "step": 87980 }, { "epoch": 0.12, "learning_rate": 4.4500625e-05, "loss": 1.3135, "step": 87990 }, { "epoch": 0.12, "learning_rate": 4.4500000000000004e-05, "loss": 1.4441, "step": 88000 }, { "epoch": 0.12, "eval_loss": 1.4187402725219727, "eval_runtime": 1505.2007, "eval_samples_per_second": 38.557, "eval_steps_per_second": 4.82, "step": 88000 }, { "epoch": 0.12, "learning_rate": 4.4499375e-05, "loss": 1.3907, "step": 88010 }, { "epoch": 0.12, "learning_rate": 4.449875e-05, "loss": 1.4357, "step": 88020 }, { "epoch": 0.12, "learning_rate": 4.4498125e-05, "loss": 1.6574, "step": 88030 }, { "epoch": 0.12, "learning_rate": 4.4497500000000006e-05, "loss": 1.776, "step": 88040 }, { "epoch": 0.12, "learning_rate": 4.4496875e-05, "loss": 1.3813, "step": 88050 }, { "epoch": 0.12, "learning_rate": 4.449625e-05, "loss": 1.4516, "step": 88060 }, { "epoch": 0.12, "learning_rate": 4.4495625e-05, "loss": 1.5372, "step": 88070 }, { "epoch": 0.12, "learning_rate": 4.4495e-05, "loss": 1.3608, "step": 88080 }, { "epoch": 0.12, "learning_rate": 4.4494375000000004e-05, "loss": 1.474, "step": 88090 }, { "epoch": 0.12, "learning_rate": 4.449375e-05, "loss": 1.6308, "step": 88100 }, { "epoch": 0.12, "learning_rate": 4.4493125000000004e-05, "loss": 1.5693, "step": 88110 }, { "epoch": 0.12, "learning_rate": 4.44925e-05, "loss": 1.4052, "step": 88120 }, { "epoch": 0.12, "learning_rate": 4.4491875e-05, "loss": 1.3475, "step": 88130 }, { "epoch": 0.12, "learning_rate": 4.449125e-05, "loss": 1.4396, "step": 88140 }, { "epoch": 0.12, "learning_rate": 4.4490625000000006e-05, "loss": 1.444, "step": 88150 }, { "epoch": 0.12, "learning_rate": 4.449e-05, "loss": 1.4952, "step": 88160 }, { "epoch": 0.12, "learning_rate": 4.4489375e-05, "loss": 1.3876, "step": 88170 }, { "epoch": 0.12, "learning_rate": 4.448875e-05, "loss": 1.5245, "step": 88180 }, { "epoch": 0.12, "learning_rate": 4.4488125e-05, "loss": 1.6683, "step": 88190 }, { "epoch": 0.12, "learning_rate": 4.4487500000000004e-05, "loss": 1.5723, "step": 88200 }, { "epoch": 0.12, "learning_rate": 4.4486875e-05, "loss": 1.5206, "step": 88210 }, { "epoch": 0.12, "learning_rate": 4.4486250000000005e-05, "loss": 1.7063, "step": 88220 }, { "epoch": 0.12, "learning_rate": 4.4485625e-05, "loss": 1.4067, "step": 88230 }, { "epoch": 0.12, "learning_rate": 4.4485000000000005e-05, "loss": 1.5842, "step": 88240 }, { "epoch": 0.12, "learning_rate": 4.4484375e-05, "loss": 1.5267, "step": 88250 }, { "epoch": 0.12, "learning_rate": 4.448375e-05, "loss": 1.4042, "step": 88260 }, { "epoch": 0.12, "learning_rate": 4.4483125e-05, "loss": 1.5295, "step": 88270 }, { "epoch": 0.12, "learning_rate": 4.4482500000000007e-05, "loss": 1.332, "step": 88280 }, { "epoch": 0.12, "learning_rate": 4.4481875000000003e-05, "loss": 1.5869, "step": 88290 }, { "epoch": 0.12, "learning_rate": 4.448125e-05, "loss": 1.4891, "step": 88300 }, { "epoch": 0.12, "learning_rate": 4.4480625000000004e-05, "loss": 1.3401, "step": 88310 }, { "epoch": 0.12, "learning_rate": 4.448e-05, "loss": 1.6258, "step": 88320 }, { "epoch": 0.12, "learning_rate": 4.4479375e-05, "loss": 1.5085, "step": 88330 }, { "epoch": 0.12, "learning_rate": 4.447875e-05, "loss": 1.6419, "step": 88340 }, { "epoch": 0.12, "learning_rate": 4.4478125000000005e-05, "loss": 1.4732, "step": 88350 }, { "epoch": 0.12, "learning_rate": 4.44775e-05, "loss": 1.4753, "step": 88360 }, { "epoch": 0.12, "learning_rate": 4.4476875e-05, "loss": 1.8484, "step": 88370 }, { "epoch": 0.12, "learning_rate": 4.447625e-05, "loss": 1.5389, "step": 88380 }, { "epoch": 0.12, "learning_rate": 4.447562500000001e-05, "loss": 1.6967, "step": 88390 }, { "epoch": 0.12, "learning_rate": 4.4475e-05, "loss": 1.5734, "step": 88400 }, { "epoch": 0.12, "learning_rate": 4.4474375e-05, "loss": 1.3761, "step": 88410 }, { "epoch": 0.12, "learning_rate": 4.4473750000000004e-05, "loss": 1.3121, "step": 88420 }, { "epoch": 0.12, "learning_rate": 4.4473125e-05, "loss": 1.4538, "step": 88430 }, { "epoch": 0.12, "learning_rate": 4.44725e-05, "loss": 1.6387, "step": 88440 }, { "epoch": 0.12, "learning_rate": 4.4471875e-05, "loss": 1.3474, "step": 88450 }, { "epoch": 0.12, "learning_rate": 4.4471250000000006e-05, "loss": 1.5008, "step": 88460 }, { "epoch": 0.12, "learning_rate": 4.4470625e-05, "loss": 1.3191, "step": 88470 }, { "epoch": 0.12, "learning_rate": 4.447e-05, "loss": 1.6932, "step": 88480 }, { "epoch": 0.12, "learning_rate": 4.4469375e-05, "loss": 1.3939, "step": 88490 }, { "epoch": 0.12, "learning_rate": 4.446875e-05, "loss": 1.5977, "step": 88500 }, { "epoch": 0.12, "learning_rate": 4.4468125e-05, "loss": 1.4616, "step": 88510 }, { "epoch": 0.12, "learning_rate": 4.44675e-05, "loss": 1.4021, "step": 88520 }, { "epoch": 0.12, "learning_rate": 4.4466875000000004e-05, "loss": 1.3796, "step": 88530 }, { "epoch": 0.12, "learning_rate": 4.446625e-05, "loss": 1.3624, "step": 88540 }, { "epoch": 0.12, "learning_rate": 4.4465625e-05, "loss": 1.256, "step": 88550 }, { "epoch": 0.12, "learning_rate": 4.4465e-05, "loss": 1.657, "step": 88560 }, { "epoch": 0.12, "learning_rate": 4.4464375000000006e-05, "loss": 1.4707, "step": 88570 }, { "epoch": 0.12, "learning_rate": 4.446375e-05, "loss": 1.4179, "step": 88580 }, { "epoch": 0.12, "learning_rate": 4.4463125e-05, "loss": 1.5088, "step": 88590 }, { "epoch": 0.12, "learning_rate": 4.44625e-05, "loss": 1.2809, "step": 88600 }, { "epoch": 0.12, "learning_rate": 4.4461875e-05, "loss": 1.3041, "step": 88610 }, { "epoch": 0.12, "learning_rate": 4.4461250000000004e-05, "loss": 1.4141, "step": 88620 }, { "epoch": 0.12, "learning_rate": 4.4460625e-05, "loss": 1.4311, "step": 88630 }, { "epoch": 0.12, "learning_rate": 4.4460000000000005e-05, "loss": 1.4875, "step": 88640 }, { "epoch": 0.12, "learning_rate": 4.4459375e-05, "loss": 1.5446, "step": 88650 }, { "epoch": 0.12, "learning_rate": 4.4458750000000005e-05, "loss": 1.7331, "step": 88660 }, { "epoch": 0.12, "learning_rate": 4.4458125e-05, "loss": 1.4881, "step": 88670 }, { "epoch": 0.12, "learning_rate": 4.44575e-05, "loss": 1.6057, "step": 88680 }, { "epoch": 0.12, "learning_rate": 4.4456875e-05, "loss": 1.2904, "step": 88690 }, { "epoch": 0.12, "learning_rate": 4.4456250000000007e-05, "loss": 1.573, "step": 88700 }, { "epoch": 0.12, "learning_rate": 4.4455625000000003e-05, "loss": 1.6171, "step": 88710 }, { "epoch": 0.12, "learning_rate": 4.4455e-05, "loss": 1.5112, "step": 88720 }, { "epoch": 0.12, "learning_rate": 4.4454375000000004e-05, "loss": 1.3458, "step": 88730 }, { "epoch": 0.12, "learning_rate": 4.445375e-05, "loss": 1.482, "step": 88740 }, { "epoch": 0.12, "learning_rate": 4.4453125e-05, "loss": 1.4622, "step": 88750 }, { "epoch": 0.12, "learning_rate": 4.44525e-05, "loss": 1.5392, "step": 88760 }, { "epoch": 0.12, "learning_rate": 4.4451875000000005e-05, "loss": 1.3341, "step": 88770 }, { "epoch": 0.12, "learning_rate": 4.445125e-05, "loss": 1.47, "step": 88780 }, { "epoch": 0.12, "learning_rate": 4.4450625e-05, "loss": 1.4413, "step": 88790 }, { "epoch": 0.12, "learning_rate": 4.445e-05, "loss": 1.5586, "step": 88800 }, { "epoch": 0.12, "learning_rate": 4.444937500000001e-05, "loss": 1.6157, "step": 88810 }, { "epoch": 0.12, "learning_rate": 4.444875e-05, "loss": 1.3946, "step": 88820 }, { "epoch": 0.12, "learning_rate": 4.4448125e-05, "loss": 1.4757, "step": 88830 }, { "epoch": 0.12, "learning_rate": 4.4447500000000004e-05, "loss": 1.4339, "step": 88840 }, { "epoch": 0.12, "learning_rate": 4.4446875e-05, "loss": 1.4733, "step": 88850 }, { "epoch": 0.12, "learning_rate": 4.444625e-05, "loss": 1.4588, "step": 88860 }, { "epoch": 0.12, "learning_rate": 4.4445625e-05, "loss": 1.6041, "step": 88870 }, { "epoch": 0.12, "learning_rate": 4.4445000000000006e-05, "loss": 1.5276, "step": 88880 }, { "epoch": 0.12, "learning_rate": 4.4444375e-05, "loss": 1.4446, "step": 88890 }, { "epoch": 0.12, "learning_rate": 4.444375e-05, "loss": 1.6616, "step": 88900 }, { "epoch": 0.12, "learning_rate": 4.4443125e-05, "loss": 1.5687, "step": 88910 }, { "epoch": 0.12, "learning_rate": 4.444250000000001e-05, "loss": 1.5145, "step": 88920 }, { "epoch": 0.12, "learning_rate": 4.4441875e-05, "loss": 1.2199, "step": 88930 }, { "epoch": 0.12, "learning_rate": 4.444125e-05, "loss": 1.4212, "step": 88940 }, { "epoch": 0.12, "learning_rate": 4.4440625000000004e-05, "loss": 1.5451, "step": 88950 }, { "epoch": 0.12, "learning_rate": 4.444e-05, "loss": 1.4244, "step": 88960 }, { "epoch": 0.12, "learning_rate": 4.4439375e-05, "loss": 1.673, "step": 88970 }, { "epoch": 0.12, "learning_rate": 4.443875e-05, "loss": 1.4465, "step": 88980 }, { "epoch": 0.12, "learning_rate": 4.4438125000000006e-05, "loss": 1.5687, "step": 88990 }, { "epoch": 0.12, "learning_rate": 4.44375e-05, "loss": 1.5811, "step": 89000 }, { "epoch": 0.12, "eval_loss": 1.4123581647872925, "eval_runtime": 1315.3125, "eval_samples_per_second": 44.123, "eval_steps_per_second": 5.516, "step": 89000 }, { "epoch": 0.12, "learning_rate": 4.4436875e-05, "loss": 1.753, "step": 89010 }, { "epoch": 0.12, "learning_rate": 4.443625e-05, "loss": 1.6926, "step": 89020 }, { "epoch": 0.12, "learning_rate": 4.4435625e-05, "loss": 1.4493, "step": 89030 }, { "epoch": 0.12, "learning_rate": 4.4435000000000004e-05, "loss": 1.5175, "step": 89040 }, { "epoch": 0.12, "learning_rate": 4.4434375e-05, "loss": 1.5162, "step": 89050 }, { "epoch": 0.12, "learning_rate": 4.4433750000000005e-05, "loss": 1.4629, "step": 89060 }, { "epoch": 0.12, "learning_rate": 4.4433125e-05, "loss": 1.5842, "step": 89070 }, { "epoch": 0.12, "learning_rate": 4.4432500000000005e-05, "loss": 1.7883, "step": 89080 }, { "epoch": 0.12, "learning_rate": 4.4431875e-05, "loss": 1.5639, "step": 89090 }, { "epoch": 0.12, "learning_rate": 4.443125e-05, "loss": 1.284, "step": 89100 }, { "epoch": 0.12, "learning_rate": 4.4430625e-05, "loss": 1.4488, "step": 89110 }, { "epoch": 0.12, "learning_rate": 4.443e-05, "loss": 1.4092, "step": 89120 }, { "epoch": 0.12, "learning_rate": 4.4429375000000004e-05, "loss": 1.7392, "step": 89130 }, { "epoch": 0.12, "learning_rate": 4.442875e-05, "loss": 1.3568, "step": 89140 }, { "epoch": 0.12, "learning_rate": 4.4428125000000004e-05, "loss": 1.4357, "step": 89150 }, { "epoch": 0.12, "learning_rate": 4.44275e-05, "loss": 1.3978, "step": 89160 }, { "epoch": 0.12, "learning_rate": 4.4426875e-05, "loss": 1.5165, "step": 89170 }, { "epoch": 0.12, "learning_rate": 4.442625e-05, "loss": 1.5499, "step": 89180 }, { "epoch": 0.12, "learning_rate": 4.4425625000000005e-05, "loss": 1.4679, "step": 89190 }, { "epoch": 0.12, "learning_rate": 4.4425e-05, "loss": 1.1411, "step": 89200 }, { "epoch": 0.12, "learning_rate": 4.4424375e-05, "loss": 1.6463, "step": 89210 }, { "epoch": 0.12, "learning_rate": 4.442375e-05, "loss": 1.6863, "step": 89220 }, { "epoch": 0.12, "learning_rate": 4.442312500000001e-05, "loss": 1.4475, "step": 89230 }, { "epoch": 0.12, "learning_rate": 4.44225e-05, "loss": 1.4298, "step": 89240 }, { "epoch": 0.12, "learning_rate": 4.4421875e-05, "loss": 1.4909, "step": 89250 }, { "epoch": 0.12, "learning_rate": 4.4421250000000004e-05, "loss": 1.6372, "step": 89260 }, { "epoch": 0.12, "learning_rate": 4.4420625e-05, "loss": 1.6506, "step": 89270 }, { "epoch": 0.12, "learning_rate": 4.442e-05, "loss": 1.4978, "step": 89280 }, { "epoch": 0.12, "learning_rate": 4.4419375e-05, "loss": 1.3579, "step": 89290 }, { "epoch": 0.12, "learning_rate": 4.4418750000000006e-05, "loss": 1.2374, "step": 89300 }, { "epoch": 0.12, "learning_rate": 4.4418125e-05, "loss": 1.5579, "step": 89310 }, { "epoch": 0.12, "learning_rate": 4.44175e-05, "loss": 1.695, "step": 89320 }, { "epoch": 0.12, "learning_rate": 4.4416875e-05, "loss": 1.386, "step": 89330 }, { "epoch": 0.12, "learning_rate": 4.441625000000001e-05, "loss": 1.3999, "step": 89340 }, { "epoch": 0.12, "learning_rate": 4.4415625e-05, "loss": 1.3863, "step": 89350 }, { "epoch": 0.12, "learning_rate": 4.4415e-05, "loss": 1.391, "step": 89360 }, { "epoch": 0.12, "learning_rate": 4.4414375000000005e-05, "loss": 1.4035, "step": 89370 }, { "epoch": 0.12, "learning_rate": 4.441375e-05, "loss": 1.3271, "step": 89380 }, { "epoch": 0.12, "learning_rate": 4.4413125e-05, "loss": 1.5079, "step": 89390 }, { "epoch": 0.12, "learning_rate": 4.44125e-05, "loss": 1.4096, "step": 89400 }, { "epoch": 0.12, "learning_rate": 4.4411875000000006e-05, "loss": 1.6621, "step": 89410 }, { "epoch": 0.12, "learning_rate": 4.441125e-05, "loss": 1.4461, "step": 89420 }, { "epoch": 0.12, "learning_rate": 4.4410625e-05, "loss": 1.363, "step": 89430 }, { "epoch": 0.12, "learning_rate": 4.4410000000000003e-05, "loss": 1.6558, "step": 89440 }, { "epoch": 0.12, "learning_rate": 4.4409375e-05, "loss": 1.2421, "step": 89450 }, { "epoch": 0.12, "learning_rate": 4.4408750000000004e-05, "loss": 1.5736, "step": 89460 }, { "epoch": 0.12, "learning_rate": 4.4408125e-05, "loss": 1.4989, "step": 89470 }, { "epoch": 0.12, "learning_rate": 4.4407500000000005e-05, "loss": 1.4755, "step": 89480 }, { "epoch": 0.12, "learning_rate": 4.4406875e-05, "loss": 1.6006, "step": 89490 }, { "epoch": 0.12, "learning_rate": 4.4406250000000005e-05, "loss": 1.4128, "step": 89500 }, { "epoch": 0.12, "learning_rate": 4.4405625e-05, "loss": 1.2407, "step": 89510 }, { "epoch": 0.12, "learning_rate": 4.4405e-05, "loss": 1.4514, "step": 89520 }, { "epoch": 0.12, "learning_rate": 4.4404375e-05, "loss": 1.4463, "step": 89530 }, { "epoch": 0.12, "learning_rate": 4.440375e-05, "loss": 1.5212, "step": 89540 }, { "epoch": 0.12, "learning_rate": 4.4403125000000004e-05, "loss": 1.4424, "step": 89550 }, { "epoch": 0.12, "learning_rate": 4.44025e-05, "loss": 1.5035, "step": 89560 }, { "epoch": 0.12, "learning_rate": 4.4401875000000004e-05, "loss": 1.4911, "step": 89570 }, { "epoch": 0.12, "learning_rate": 4.440125e-05, "loss": 1.5117, "step": 89580 }, { "epoch": 0.12, "learning_rate": 4.4400625e-05, "loss": 1.5344, "step": 89590 }, { "epoch": 0.12, "learning_rate": 4.44e-05, "loss": 1.5461, "step": 89600 }, { "epoch": 0.12, "learning_rate": 4.4399375000000006e-05, "loss": 1.4056, "step": 89610 }, { "epoch": 0.12, "learning_rate": 4.439875e-05, "loss": 1.5581, "step": 89620 }, { "epoch": 0.12, "learning_rate": 4.4398125e-05, "loss": 1.4882, "step": 89630 }, { "epoch": 0.12, "learning_rate": 4.43975e-05, "loss": 1.429, "step": 89640 }, { "epoch": 0.12, "learning_rate": 4.439687500000001e-05, "loss": 1.4544, "step": 89650 }, { "epoch": 0.12, "learning_rate": 4.439625e-05, "loss": 1.3679, "step": 89660 }, { "epoch": 0.12, "learning_rate": 4.4395625e-05, "loss": 1.6065, "step": 89670 }, { "epoch": 0.12, "learning_rate": 4.4395000000000004e-05, "loss": 1.442, "step": 89680 }, { "epoch": 0.12, "learning_rate": 4.4394375e-05, "loss": 1.3226, "step": 89690 }, { "epoch": 0.12, "learning_rate": 4.439375e-05, "loss": 1.5318, "step": 89700 }, { "epoch": 0.12, "learning_rate": 4.4393125e-05, "loss": 1.4726, "step": 89710 }, { "epoch": 0.12, "learning_rate": 4.4392500000000006e-05, "loss": 1.6151, "step": 89720 }, { "epoch": 0.12, "learning_rate": 4.4391875e-05, "loss": 1.6315, "step": 89730 }, { "epoch": 0.12, "learning_rate": 4.439125e-05, "loss": 1.5127, "step": 89740 }, { "epoch": 0.12, "learning_rate": 4.4390625e-05, "loss": 1.599, "step": 89750 }, { "epoch": 0.12, "learning_rate": 4.439000000000001e-05, "loss": 1.4767, "step": 89760 }, { "epoch": 0.12, "learning_rate": 4.4389375e-05, "loss": 1.7559, "step": 89770 }, { "epoch": 0.13, "learning_rate": 4.438875e-05, "loss": 1.5389, "step": 89780 }, { "epoch": 0.13, "learning_rate": 4.4388125000000005e-05, "loss": 1.661, "step": 89790 }, { "epoch": 0.13, "learning_rate": 4.43875e-05, "loss": 1.5083, "step": 89800 }, { "epoch": 0.13, "learning_rate": 4.4386875e-05, "loss": 1.5152, "step": 89810 }, { "epoch": 0.13, "learning_rate": 4.438625e-05, "loss": 1.5836, "step": 89820 }, { "epoch": 0.13, "learning_rate": 4.4385625000000006e-05, "loss": 1.5768, "step": 89830 }, { "epoch": 0.13, "learning_rate": 4.4385e-05, "loss": 1.3769, "step": 89840 }, { "epoch": 0.13, "learning_rate": 4.4384375e-05, "loss": 1.4585, "step": 89850 }, { "epoch": 0.13, "learning_rate": 4.4383750000000003e-05, "loss": 1.4283, "step": 89860 }, { "epoch": 0.13, "learning_rate": 4.4383125e-05, "loss": 1.3954, "step": 89870 }, { "epoch": 0.13, "learning_rate": 4.4382500000000004e-05, "loss": 1.3672, "step": 89880 }, { "epoch": 0.13, "learning_rate": 4.4381875e-05, "loss": 1.6213, "step": 89890 }, { "epoch": 0.13, "learning_rate": 4.4381250000000005e-05, "loss": 1.4095, "step": 89900 }, { "epoch": 0.13, "learning_rate": 4.4380625e-05, "loss": 1.5336, "step": 89910 }, { "epoch": 0.13, "learning_rate": 4.438e-05, "loss": 1.6044, "step": 89920 }, { "epoch": 0.13, "learning_rate": 4.4379375e-05, "loss": 1.4324, "step": 89930 }, { "epoch": 0.13, "learning_rate": 4.437875e-05, "loss": 1.4095, "step": 89940 }, { "epoch": 0.13, "learning_rate": 4.4378125e-05, "loss": 1.4378, "step": 89950 }, { "epoch": 0.13, "learning_rate": 4.43775e-05, "loss": 1.3757, "step": 89960 }, { "epoch": 0.13, "learning_rate": 4.4376875000000004e-05, "loss": 1.4675, "step": 89970 }, { "epoch": 0.13, "learning_rate": 4.437625e-05, "loss": 1.5264, "step": 89980 }, { "epoch": 0.13, "learning_rate": 4.4375625000000004e-05, "loss": 1.3369, "step": 89990 }, { "epoch": 0.13, "learning_rate": 4.4375e-05, "loss": 1.3706, "step": 90000 }, { "epoch": 0.13, "eval_loss": 1.415058970451355, "eval_runtime": 1320.5494, "eval_samples_per_second": 43.948, "eval_steps_per_second": 5.494, "step": 90000 }, { "epoch": 0.13, "learning_rate": 4.4374375e-05, "loss": 1.6619, "step": 90010 }, { "epoch": 0.13, "learning_rate": 4.437375e-05, "loss": 1.4421, "step": 90020 }, { "epoch": 0.13, "learning_rate": 4.4373125000000006e-05, "loss": 1.3975, "step": 90030 }, { "epoch": 0.13, "learning_rate": 4.43725e-05, "loss": 1.6207, "step": 90040 }, { "epoch": 0.13, "learning_rate": 4.4371875e-05, "loss": 1.8397, "step": 90050 }, { "epoch": 0.13, "learning_rate": 4.437125e-05, "loss": 1.54, "step": 90060 }, { "epoch": 0.13, "learning_rate": 4.437062500000001e-05, "loss": 1.2957, "step": 90070 }, { "epoch": 0.13, "learning_rate": 4.4370000000000004e-05, "loss": 1.3068, "step": 90080 }, { "epoch": 0.13, "learning_rate": 4.4369375e-05, "loss": 1.6056, "step": 90090 }, { "epoch": 0.13, "learning_rate": 4.4368750000000004e-05, "loss": 1.4307, "step": 90100 }, { "epoch": 0.13, "learning_rate": 4.436812500000001e-05, "loss": 1.3369, "step": 90110 }, { "epoch": 0.13, "learning_rate": 4.43675e-05, "loss": 1.4963, "step": 90120 }, { "epoch": 0.13, "learning_rate": 4.4366875e-05, "loss": 1.5234, "step": 90130 }, { "epoch": 0.13, "learning_rate": 4.4366250000000006e-05, "loss": 1.543, "step": 90140 }, { "epoch": 0.13, "learning_rate": 4.4365625e-05, "loss": 1.4073, "step": 90150 }, { "epoch": 0.13, "learning_rate": 4.4365e-05, "loss": 1.3606, "step": 90160 }, { "epoch": 0.13, "learning_rate": 4.4364375e-05, "loss": 1.622, "step": 90170 }, { "epoch": 0.13, "learning_rate": 4.436375000000001e-05, "loss": 1.3122, "step": 90180 }, { "epoch": 0.13, "learning_rate": 4.4363125e-05, "loss": 1.4778, "step": 90190 }, { "epoch": 0.13, "learning_rate": 4.43625e-05, "loss": 1.4197, "step": 90200 }, { "epoch": 0.13, "learning_rate": 4.4361875000000005e-05, "loss": 1.7013, "step": 90210 }, { "epoch": 0.13, "learning_rate": 4.436125e-05, "loss": 1.7372, "step": 90220 }, { "epoch": 0.13, "learning_rate": 4.4360625e-05, "loss": 1.4485, "step": 90230 }, { "epoch": 0.13, "learning_rate": 4.436e-05, "loss": 1.4651, "step": 90240 }, { "epoch": 0.13, "learning_rate": 4.4359375000000006e-05, "loss": 1.3401, "step": 90250 }, { "epoch": 0.13, "learning_rate": 4.435875e-05, "loss": 1.5563, "step": 90260 }, { "epoch": 0.13, "learning_rate": 4.4358125e-05, "loss": 1.4382, "step": 90270 }, { "epoch": 0.13, "learning_rate": 4.4357500000000003e-05, "loss": 1.4152, "step": 90280 }, { "epoch": 0.13, "learning_rate": 4.4356875e-05, "loss": 1.2727, "step": 90290 }, { "epoch": 0.13, "learning_rate": 4.4356250000000004e-05, "loss": 1.5911, "step": 90300 }, { "epoch": 0.13, "learning_rate": 4.4355625e-05, "loss": 1.5228, "step": 90310 }, { "epoch": 0.13, "learning_rate": 4.4355000000000005e-05, "loss": 1.1963, "step": 90320 }, { "epoch": 0.13, "learning_rate": 4.4354375e-05, "loss": 1.3475, "step": 90330 }, { "epoch": 0.13, "learning_rate": 4.435375e-05, "loss": 1.5858, "step": 90340 }, { "epoch": 0.13, "learning_rate": 4.4353125e-05, "loss": 1.5951, "step": 90350 }, { "epoch": 0.13, "learning_rate": 4.43525e-05, "loss": 1.4106, "step": 90360 }, { "epoch": 0.13, "learning_rate": 4.4351875e-05, "loss": 1.5659, "step": 90370 }, { "epoch": 0.13, "learning_rate": 4.435125e-05, "loss": 1.6912, "step": 90380 }, { "epoch": 0.13, "learning_rate": 4.4350625000000004e-05, "loss": 1.5849, "step": 90390 }, { "epoch": 0.13, "learning_rate": 4.435e-05, "loss": 1.6085, "step": 90400 }, { "epoch": 0.13, "learning_rate": 4.4349375000000004e-05, "loss": 1.3117, "step": 90410 }, { "epoch": 0.13, "learning_rate": 4.434875e-05, "loss": 1.6461, "step": 90420 }, { "epoch": 0.13, "learning_rate": 4.4348125e-05, "loss": 1.413, "step": 90430 }, { "epoch": 0.13, "learning_rate": 4.43475e-05, "loss": 1.347, "step": 90440 }, { "epoch": 0.13, "learning_rate": 4.4346875000000006e-05, "loss": 1.4528, "step": 90450 }, { "epoch": 0.13, "learning_rate": 4.434625e-05, "loss": 1.5452, "step": 90460 }, { "epoch": 0.13, "learning_rate": 4.4345625e-05, "loss": 1.4121, "step": 90470 }, { "epoch": 0.13, "learning_rate": 4.4345e-05, "loss": 1.4728, "step": 90480 }, { "epoch": 0.13, "learning_rate": 4.434437500000001e-05, "loss": 1.7773, "step": 90490 }, { "epoch": 0.13, "learning_rate": 4.4343750000000004e-05, "loss": 1.5913, "step": 90500 }, { "epoch": 0.13, "learning_rate": 4.4343125e-05, "loss": 1.4544, "step": 90510 }, { "epoch": 0.13, "learning_rate": 4.4342500000000004e-05, "loss": 1.4153, "step": 90520 }, { "epoch": 0.13, "learning_rate": 4.4341875e-05, "loss": 1.4605, "step": 90530 }, { "epoch": 0.13, "learning_rate": 4.434125e-05, "loss": 1.6811, "step": 90540 }, { "epoch": 0.13, "learning_rate": 4.4340625e-05, "loss": 1.4544, "step": 90550 }, { "epoch": 0.13, "learning_rate": 4.4340000000000006e-05, "loss": 1.445, "step": 90560 }, { "epoch": 0.13, "learning_rate": 4.4339375e-05, "loss": 1.5326, "step": 90570 }, { "epoch": 0.13, "learning_rate": 4.433875e-05, "loss": 1.5128, "step": 90580 }, { "epoch": 0.13, "learning_rate": 4.4338125e-05, "loss": 1.5351, "step": 90590 }, { "epoch": 0.13, "learning_rate": 4.433750000000001e-05, "loss": 1.5238, "step": 90600 }, { "epoch": 0.13, "learning_rate": 4.4336875e-05, "loss": 1.4144, "step": 90610 }, { "epoch": 0.13, "learning_rate": 4.433625e-05, "loss": 1.5372, "step": 90620 }, { "epoch": 0.13, "learning_rate": 4.4335625000000005e-05, "loss": 1.5085, "step": 90630 }, { "epoch": 0.13, "learning_rate": 4.4335e-05, "loss": 1.4433, "step": 90640 }, { "epoch": 0.13, "learning_rate": 4.4334375e-05, "loss": 1.5082, "step": 90650 }, { "epoch": 0.13, "learning_rate": 4.433375e-05, "loss": 1.5836, "step": 90660 }, { "epoch": 0.13, "learning_rate": 4.4333125000000006e-05, "loss": 1.2005, "step": 90670 }, { "epoch": 0.13, "learning_rate": 4.43325e-05, "loss": 1.5317, "step": 90680 }, { "epoch": 0.13, "learning_rate": 4.4331875e-05, "loss": 1.7543, "step": 90690 }, { "epoch": 0.13, "learning_rate": 4.4331250000000004e-05, "loss": 1.5415, "step": 90700 }, { "epoch": 0.13, "learning_rate": 4.4330625e-05, "loss": 1.6168, "step": 90710 }, { "epoch": 0.13, "learning_rate": 4.4330000000000004e-05, "loss": 1.34, "step": 90720 }, { "epoch": 0.13, "learning_rate": 4.4329375e-05, "loss": 1.6297, "step": 90730 }, { "epoch": 0.13, "learning_rate": 4.4328750000000005e-05, "loss": 1.4847, "step": 90740 }, { "epoch": 0.13, "learning_rate": 4.4328125e-05, "loss": 1.5239, "step": 90750 }, { "epoch": 0.13, "learning_rate": 4.43275e-05, "loss": 1.4955, "step": 90760 }, { "epoch": 0.13, "learning_rate": 4.4326875e-05, "loss": 1.612, "step": 90770 }, { "epoch": 0.13, "learning_rate": 4.432625e-05, "loss": 1.6248, "step": 90780 }, { "epoch": 0.13, "learning_rate": 4.4325625e-05, "loss": 1.6069, "step": 90790 }, { "epoch": 0.13, "learning_rate": 4.4325e-05, "loss": 1.2107, "step": 90800 }, { "epoch": 0.13, "learning_rate": 4.4324375000000004e-05, "loss": 1.2613, "step": 90810 }, { "epoch": 0.13, "learning_rate": 4.432375e-05, "loss": 1.5827, "step": 90820 }, { "epoch": 0.13, "learning_rate": 4.4323125000000004e-05, "loss": 1.5283, "step": 90830 }, { "epoch": 0.13, "learning_rate": 4.43225e-05, "loss": 1.4547, "step": 90840 }, { "epoch": 0.13, "learning_rate": 4.4321875e-05, "loss": 1.2964, "step": 90850 }, { "epoch": 0.13, "learning_rate": 4.432125e-05, "loss": 1.4987, "step": 90860 }, { "epoch": 0.13, "learning_rate": 4.4320625000000006e-05, "loss": 1.4031, "step": 90870 }, { "epoch": 0.13, "learning_rate": 4.432e-05, "loss": 1.7277, "step": 90880 }, { "epoch": 0.13, "learning_rate": 4.4319375e-05, "loss": 1.6884, "step": 90890 }, { "epoch": 0.13, "learning_rate": 4.431875e-05, "loss": 1.5997, "step": 90900 }, { "epoch": 0.13, "learning_rate": 4.431812500000001e-05, "loss": 1.2413, "step": 90910 }, { "epoch": 0.13, "learning_rate": 4.4317500000000004e-05, "loss": 1.4966, "step": 90920 }, { "epoch": 0.13, "learning_rate": 4.4316875e-05, "loss": 1.3851, "step": 90930 }, { "epoch": 0.13, "learning_rate": 4.4316250000000005e-05, "loss": 1.4199, "step": 90940 }, { "epoch": 0.13, "learning_rate": 4.4315625e-05, "loss": 1.3271, "step": 90950 }, { "epoch": 0.13, "learning_rate": 4.4315e-05, "loss": 1.3996, "step": 90960 }, { "epoch": 0.13, "learning_rate": 4.4314375e-05, "loss": 1.5648, "step": 90970 }, { "epoch": 0.13, "learning_rate": 4.4313750000000006e-05, "loss": 1.4997, "step": 90980 }, { "epoch": 0.13, "learning_rate": 4.4313125e-05, "loss": 1.3149, "step": 90990 }, { "epoch": 0.13, "learning_rate": 4.43125e-05, "loss": 1.3722, "step": 91000 }, { "epoch": 0.13, "eval_loss": 1.401999592781067, "eval_runtime": 1321.5081, "eval_samples_per_second": 43.916, "eval_steps_per_second": 5.49, "step": 91000 }, { "epoch": 0.13, "learning_rate": 4.4311875e-05, "loss": 1.5359, "step": 91010 }, { "epoch": 0.13, "learning_rate": 4.431125000000001e-05, "loss": 1.4345, "step": 91020 }, { "epoch": 0.13, "learning_rate": 4.4310625e-05, "loss": 1.5056, "step": 91030 }, { "epoch": 0.13, "learning_rate": 4.431e-05, "loss": 1.5741, "step": 91040 }, { "epoch": 0.13, "learning_rate": 4.4309375000000005e-05, "loss": 1.5413, "step": 91050 }, { "epoch": 0.13, "learning_rate": 4.430875e-05, "loss": 1.5098, "step": 91060 }, { "epoch": 0.13, "learning_rate": 4.4308125e-05, "loss": 1.48, "step": 91070 }, { "epoch": 0.13, "learning_rate": 4.43075e-05, "loss": 1.3503, "step": 91080 }, { "epoch": 0.13, "learning_rate": 4.4306875000000006e-05, "loss": 1.6556, "step": 91090 }, { "epoch": 0.13, "learning_rate": 4.430625e-05, "loss": 1.3214, "step": 91100 }, { "epoch": 0.13, "learning_rate": 4.4305625e-05, "loss": 1.4181, "step": 91110 }, { "epoch": 0.13, "learning_rate": 4.4305000000000004e-05, "loss": 1.4371, "step": 91120 }, { "epoch": 0.13, "learning_rate": 4.4304375e-05, "loss": 1.7875, "step": 91130 }, { "epoch": 0.13, "learning_rate": 4.430375e-05, "loss": 1.3745, "step": 91140 }, { "epoch": 0.13, "learning_rate": 4.4303125e-05, "loss": 1.3937, "step": 91150 }, { "epoch": 0.13, "learning_rate": 4.4302500000000005e-05, "loss": 1.54, "step": 91160 }, { "epoch": 0.13, "learning_rate": 4.4301875e-05, "loss": 1.447, "step": 91170 }, { "epoch": 0.13, "learning_rate": 4.430125e-05, "loss": 1.3049, "step": 91180 }, { "epoch": 0.13, "learning_rate": 4.4300625e-05, "loss": 1.6525, "step": 91190 }, { "epoch": 0.13, "learning_rate": 4.43e-05, "loss": 1.3386, "step": 91200 }, { "epoch": 0.13, "learning_rate": 4.4299375e-05, "loss": 1.5091, "step": 91210 }, { "epoch": 0.13, "learning_rate": 4.429875e-05, "loss": 1.5629, "step": 91220 }, { "epoch": 0.13, "learning_rate": 4.4298125000000004e-05, "loss": 1.6453, "step": 91230 }, { "epoch": 0.13, "learning_rate": 4.42975e-05, "loss": 1.4141, "step": 91240 }, { "epoch": 0.13, "learning_rate": 4.4296875000000004e-05, "loss": 1.5576, "step": 91250 }, { "epoch": 0.13, "learning_rate": 4.429625e-05, "loss": 1.5325, "step": 91260 }, { "epoch": 0.13, "learning_rate": 4.4295625000000005e-05, "loss": 1.4386, "step": 91270 }, { "epoch": 0.13, "learning_rate": 4.4295e-05, "loss": 1.2932, "step": 91280 }, { "epoch": 0.13, "learning_rate": 4.4294375000000006e-05, "loss": 1.5189, "step": 91290 }, { "epoch": 0.13, "learning_rate": 4.429375e-05, "loss": 1.472, "step": 91300 }, { "epoch": 0.13, "learning_rate": 4.4293125e-05, "loss": 1.4931, "step": 91310 }, { "epoch": 0.13, "learning_rate": 4.42925e-05, "loss": 1.4488, "step": 91320 }, { "epoch": 0.13, "learning_rate": 4.429187500000001e-05, "loss": 1.4812, "step": 91330 }, { "epoch": 0.13, "learning_rate": 4.4291250000000004e-05, "loss": 1.4051, "step": 91340 }, { "epoch": 0.13, "learning_rate": 4.4290625e-05, "loss": 1.5116, "step": 91350 }, { "epoch": 0.13, "learning_rate": 4.4290000000000005e-05, "loss": 1.5465, "step": 91360 }, { "epoch": 0.13, "learning_rate": 4.4289375e-05, "loss": 1.4206, "step": 91370 }, { "epoch": 0.13, "learning_rate": 4.428875e-05, "loss": 1.4482, "step": 91380 }, { "epoch": 0.13, "learning_rate": 4.4288125e-05, "loss": 1.3533, "step": 91390 }, { "epoch": 0.13, "learning_rate": 4.4287500000000006e-05, "loss": 1.4525, "step": 91400 }, { "epoch": 0.13, "learning_rate": 4.4286875e-05, "loss": 1.2373, "step": 91410 }, { "epoch": 0.13, "learning_rate": 4.428625e-05, "loss": 1.5145, "step": 91420 }, { "epoch": 0.13, "learning_rate": 4.4285625000000003e-05, "loss": 1.5204, "step": 91430 }, { "epoch": 0.13, "learning_rate": 4.428500000000001e-05, "loss": 1.4165, "step": 91440 }, { "epoch": 0.13, "learning_rate": 4.4284375e-05, "loss": 1.5833, "step": 91450 }, { "epoch": 0.13, "learning_rate": 4.428375e-05, "loss": 1.5308, "step": 91460 }, { "epoch": 0.13, "learning_rate": 4.4283125000000005e-05, "loss": 1.0168, "step": 91470 }, { "epoch": 0.13, "learning_rate": 4.42825e-05, "loss": 1.3917, "step": 91480 }, { "epoch": 0.13, "learning_rate": 4.4281875e-05, "loss": 1.6584, "step": 91490 }, { "epoch": 0.13, "learning_rate": 4.428125e-05, "loss": 1.4544, "step": 91500 }, { "epoch": 0.13, "learning_rate": 4.4280625000000006e-05, "loss": 1.4617, "step": 91510 }, { "epoch": 0.13, "learning_rate": 4.428e-05, "loss": 1.4409, "step": 91520 }, { "epoch": 0.13, "learning_rate": 4.4279375e-05, "loss": 1.6102, "step": 91530 }, { "epoch": 0.13, "learning_rate": 4.4278750000000004e-05, "loss": 1.5205, "step": 91540 }, { "epoch": 0.13, "learning_rate": 4.4278125e-05, "loss": 1.4549, "step": 91550 }, { "epoch": 0.13, "learning_rate": 4.42775e-05, "loss": 1.4779, "step": 91560 }, { "epoch": 0.13, "learning_rate": 4.4276875e-05, "loss": 1.3769, "step": 91570 }, { "epoch": 0.13, "learning_rate": 4.4276250000000005e-05, "loss": 1.4629, "step": 91580 }, { "epoch": 0.13, "learning_rate": 4.4275625e-05, "loss": 1.4007, "step": 91590 }, { "epoch": 0.13, "learning_rate": 4.4275e-05, "loss": 1.487, "step": 91600 }, { "epoch": 0.13, "learning_rate": 4.4274375e-05, "loss": 1.3381, "step": 91610 }, { "epoch": 0.13, "learning_rate": 4.427375e-05, "loss": 1.4336, "step": 91620 }, { "epoch": 0.13, "learning_rate": 4.4273125e-05, "loss": 1.3803, "step": 91630 }, { "epoch": 0.13, "learning_rate": 4.42725e-05, "loss": 1.4677, "step": 91640 }, { "epoch": 0.13, "learning_rate": 4.4271875000000004e-05, "loss": 1.4519, "step": 91650 }, { "epoch": 0.13, "learning_rate": 4.427125e-05, "loss": 1.5204, "step": 91660 }, { "epoch": 0.13, "learning_rate": 4.4270625000000004e-05, "loss": 1.4439, "step": 91670 }, { "epoch": 0.13, "learning_rate": 4.427e-05, "loss": 1.4704, "step": 91680 }, { "epoch": 0.13, "learning_rate": 4.4269375000000005e-05, "loss": 1.7086, "step": 91690 }, { "epoch": 0.13, "learning_rate": 4.426875e-05, "loss": 1.6793, "step": 91700 }, { "epoch": 0.13, "learning_rate": 4.4268125000000006e-05, "loss": 1.4669, "step": 91710 }, { "epoch": 0.13, "learning_rate": 4.42675e-05, "loss": 1.4112, "step": 91720 }, { "epoch": 0.13, "learning_rate": 4.4266875e-05, "loss": 1.5732, "step": 91730 }, { "epoch": 0.13, "learning_rate": 4.426625e-05, "loss": 1.404, "step": 91740 }, { "epoch": 0.13, "learning_rate": 4.4265625e-05, "loss": 1.6557, "step": 91750 }, { "epoch": 0.13, "learning_rate": 4.4265000000000004e-05, "loss": 1.2931, "step": 91760 }, { "epoch": 0.13, "learning_rate": 4.4264375e-05, "loss": 1.4779, "step": 91770 }, { "epoch": 0.13, "learning_rate": 4.4263750000000005e-05, "loss": 1.3959, "step": 91780 }, { "epoch": 0.13, "learning_rate": 4.4263125e-05, "loss": 1.3818, "step": 91790 }, { "epoch": 0.13, "learning_rate": 4.42625e-05, "loss": 1.4757, "step": 91800 }, { "epoch": 0.13, "learning_rate": 4.4261875e-05, "loss": 1.4894, "step": 91810 }, { "epoch": 0.13, "learning_rate": 4.4261250000000006e-05, "loss": 1.49, "step": 91820 }, { "epoch": 0.13, "learning_rate": 4.4260625e-05, "loss": 1.4146, "step": 91830 }, { "epoch": 0.13, "learning_rate": 4.426e-05, "loss": 1.6378, "step": 91840 }, { "epoch": 0.13, "learning_rate": 4.4259375000000003e-05, "loss": 1.4491, "step": 91850 }, { "epoch": 0.13, "learning_rate": 4.425875000000001e-05, "loss": 1.3664, "step": 91860 }, { "epoch": 0.13, "learning_rate": 4.4258125e-05, "loss": 1.5655, "step": 91870 }, { "epoch": 0.13, "learning_rate": 4.42575e-05, "loss": 1.5627, "step": 91880 }, { "epoch": 0.13, "learning_rate": 4.4256875000000005e-05, "loss": 1.3654, "step": 91890 }, { "epoch": 0.13, "learning_rate": 4.425625e-05, "loss": 1.4241, "step": 91900 }, { "epoch": 0.13, "learning_rate": 4.4255625e-05, "loss": 1.4689, "step": 91910 }, { "epoch": 0.13, "learning_rate": 4.4255e-05, "loss": 1.3715, "step": 91920 }, { "epoch": 0.13, "learning_rate": 4.4254375000000006e-05, "loss": 1.2758, "step": 91930 }, { "epoch": 0.13, "learning_rate": 4.4253749999999996e-05, "loss": 1.4968, "step": 91940 }, { "epoch": 0.13, "learning_rate": 4.4253125e-05, "loss": 1.5067, "step": 91950 }, { "epoch": 0.13, "learning_rate": 4.4252500000000004e-05, "loss": 1.3632, "step": 91960 }, { "epoch": 0.13, "learning_rate": 4.4251875e-05, "loss": 1.2775, "step": 91970 }, { "epoch": 0.13, "learning_rate": 4.425125e-05, "loss": 1.4322, "step": 91980 }, { "epoch": 0.13, "learning_rate": 4.4250625e-05, "loss": 1.5634, "step": 91990 }, { "epoch": 0.13, "learning_rate": 4.4250000000000005e-05, "loss": 1.505, "step": 92000 }, { "epoch": 0.13, "eval_loss": 1.3937928676605225, "eval_runtime": 1297.06, "eval_samples_per_second": 44.744, "eval_steps_per_second": 5.593, "step": 92000 }, { "epoch": 0.13, "learning_rate": 4.4249375e-05, "loss": 1.5998, "step": 92010 }, { "epoch": 0.13, "learning_rate": 4.424875e-05, "loss": 1.5886, "step": 92020 }, { "epoch": 0.13, "learning_rate": 4.4248125e-05, "loss": 1.4034, "step": 92030 }, { "epoch": 0.13, "learning_rate": 4.4247500000000006e-05, "loss": 1.4329, "step": 92040 }, { "epoch": 0.13, "learning_rate": 4.4246875e-05, "loss": 1.4211, "step": 92050 }, { "epoch": 0.13, "learning_rate": 4.424625e-05, "loss": 1.4751, "step": 92060 }, { "epoch": 0.13, "learning_rate": 4.4245625000000004e-05, "loss": 1.3212, "step": 92070 }, { "epoch": 0.13, "learning_rate": 4.4245e-05, "loss": 1.4645, "step": 92080 }, { "epoch": 0.13, "learning_rate": 4.4244375000000004e-05, "loss": 1.5206, "step": 92090 }, { "epoch": 0.13, "learning_rate": 4.424375e-05, "loss": 1.3781, "step": 92100 }, { "epoch": 0.13, "learning_rate": 4.4243125000000005e-05, "loss": 1.3718, "step": 92110 }, { "epoch": 0.13, "learning_rate": 4.42425e-05, "loss": 1.4432, "step": 92120 }, { "epoch": 0.13, "learning_rate": 4.4241875000000006e-05, "loss": 1.4945, "step": 92130 }, { "epoch": 0.13, "learning_rate": 4.424125e-05, "loss": 1.3393, "step": 92140 }, { "epoch": 0.13, "learning_rate": 4.4240625e-05, "loss": 1.543, "step": 92150 }, { "epoch": 0.13, "learning_rate": 4.424e-05, "loss": 1.4447, "step": 92160 }, { "epoch": 0.13, "learning_rate": 4.4239375e-05, "loss": 1.5005, "step": 92170 }, { "epoch": 0.13, "learning_rate": 4.4238750000000004e-05, "loss": 1.4959, "step": 92180 }, { "epoch": 0.13, "learning_rate": 4.4238125e-05, "loss": 1.4258, "step": 92190 }, { "epoch": 0.13, "learning_rate": 4.4237500000000005e-05, "loss": 1.5363, "step": 92200 }, { "epoch": 0.13, "learning_rate": 4.4236875e-05, "loss": 1.3906, "step": 92210 }, { "epoch": 0.13, "learning_rate": 4.423625e-05, "loss": 1.4394, "step": 92220 }, { "epoch": 0.13, "learning_rate": 4.4235625e-05, "loss": 1.5407, "step": 92230 }, { "epoch": 0.13, "learning_rate": 4.4235000000000006e-05, "loss": 1.4437, "step": 92240 }, { "epoch": 0.13, "learning_rate": 4.4234375e-05, "loss": 1.3791, "step": 92250 }, { "epoch": 0.13, "learning_rate": 4.423375e-05, "loss": 1.3458, "step": 92260 }, { "epoch": 0.13, "learning_rate": 4.4233125000000004e-05, "loss": 1.3638, "step": 92270 }, { "epoch": 0.13, "learning_rate": 4.423250000000001e-05, "loss": 1.5939, "step": 92280 }, { "epoch": 0.13, "learning_rate": 4.4231875e-05, "loss": 1.462, "step": 92290 }, { "epoch": 0.13, "learning_rate": 4.423125e-05, "loss": 1.3019, "step": 92300 }, { "epoch": 0.13, "learning_rate": 4.4230625000000005e-05, "loss": 1.4178, "step": 92310 }, { "epoch": 0.13, "learning_rate": 4.423e-05, "loss": 1.5894, "step": 92320 }, { "epoch": 0.13, "learning_rate": 4.4229375e-05, "loss": 1.4515, "step": 92330 }, { "epoch": 0.13, "learning_rate": 4.422875e-05, "loss": 1.3231, "step": 92340 }, { "epoch": 0.13, "learning_rate": 4.4228125000000006e-05, "loss": 1.4516, "step": 92350 }, { "epoch": 0.13, "learning_rate": 4.4227499999999996e-05, "loss": 1.5243, "step": 92360 }, { "epoch": 0.13, "learning_rate": 4.4226875e-05, "loss": 1.4533, "step": 92370 }, { "epoch": 0.13, "learning_rate": 4.4226250000000004e-05, "loss": 1.4589, "step": 92380 }, { "epoch": 0.13, "learning_rate": 4.4225625e-05, "loss": 1.5887, "step": 92390 }, { "epoch": 0.13, "learning_rate": 4.4225e-05, "loss": 1.3818, "step": 92400 }, { "epoch": 0.13, "learning_rate": 4.4224375e-05, "loss": 1.541, "step": 92410 }, { "epoch": 0.13, "learning_rate": 4.4223750000000005e-05, "loss": 1.5486, "step": 92420 }, { "epoch": 0.13, "learning_rate": 4.4223125e-05, "loss": 1.535, "step": 92430 }, { "epoch": 0.13, "learning_rate": 4.42225e-05, "loss": 1.6964, "step": 92440 }, { "epoch": 0.13, "learning_rate": 4.4221875e-05, "loss": 1.4484, "step": 92450 }, { "epoch": 0.13, "learning_rate": 4.4221250000000006e-05, "loss": 1.456, "step": 92460 }, { "epoch": 0.13, "learning_rate": 4.4220625e-05, "loss": 1.5691, "step": 92470 }, { "epoch": 0.13, "learning_rate": 4.422e-05, "loss": 1.3833, "step": 92480 }, { "epoch": 0.13, "learning_rate": 4.4219375000000004e-05, "loss": 1.561, "step": 92490 }, { "epoch": 0.13, "learning_rate": 4.421875e-05, "loss": 1.6856, "step": 92500 }, { "epoch": 0.13, "learning_rate": 4.4218125000000004e-05, "loss": 1.4586, "step": 92510 }, { "epoch": 0.13, "learning_rate": 4.42175e-05, "loss": 1.4817, "step": 92520 }, { "epoch": 0.13, "learning_rate": 4.4216875000000005e-05, "loss": 1.4659, "step": 92530 }, { "epoch": 0.13, "learning_rate": 4.421625e-05, "loss": 1.3783, "step": 92540 }, { "epoch": 0.13, "learning_rate": 4.4215625e-05, "loss": 1.285, "step": 92550 }, { "epoch": 0.13, "learning_rate": 4.4215e-05, "loss": 1.407, "step": 92560 }, { "epoch": 0.13, "learning_rate": 4.4214375e-05, "loss": 1.4774, "step": 92570 }, { "epoch": 0.13, "learning_rate": 4.421375e-05, "loss": 1.4802, "step": 92580 }, { "epoch": 0.13, "learning_rate": 4.4213125e-05, "loss": 1.4839, "step": 92590 }, { "epoch": 0.13, "learning_rate": 4.4212500000000004e-05, "loss": 1.3847, "step": 92600 }, { "epoch": 0.13, "learning_rate": 4.4211875e-05, "loss": 1.3802, "step": 92610 }, { "epoch": 0.13, "learning_rate": 4.4211250000000005e-05, "loss": 1.3825, "step": 92620 }, { "epoch": 0.13, "learning_rate": 4.4210625e-05, "loss": 1.3071, "step": 92630 }, { "epoch": 0.13, "learning_rate": 4.421e-05, "loss": 1.3888, "step": 92640 }, { "epoch": 0.13, "learning_rate": 4.4209375e-05, "loss": 1.3557, "step": 92650 }, { "epoch": 0.13, "learning_rate": 4.4208750000000006e-05, "loss": 1.4123, "step": 92660 }, { "epoch": 0.13, "learning_rate": 4.4208125e-05, "loss": 1.5104, "step": 92670 }, { "epoch": 0.13, "learning_rate": 4.42075e-05, "loss": 1.6167, "step": 92680 }, { "epoch": 0.13, "learning_rate": 4.4206875000000004e-05, "loss": 1.3632, "step": 92690 }, { "epoch": 0.13, "learning_rate": 4.420625000000001e-05, "loss": 1.3367, "step": 92700 }, { "epoch": 0.13, "learning_rate": 4.4205625e-05, "loss": 1.5105, "step": 92710 }, { "epoch": 0.13, "learning_rate": 4.4205e-05, "loss": 1.3713, "step": 92720 }, { "epoch": 0.13, "learning_rate": 4.4204375000000005e-05, "loss": 1.5443, "step": 92730 }, { "epoch": 0.13, "learning_rate": 4.420375e-05, "loss": 1.6332, "step": 92740 }, { "epoch": 0.13, "learning_rate": 4.4203125e-05, "loss": 1.4022, "step": 92750 }, { "epoch": 0.13, "learning_rate": 4.42025e-05, "loss": 1.4221, "step": 92760 }, { "epoch": 0.13, "learning_rate": 4.4201875000000006e-05, "loss": 1.4385, "step": 92770 }, { "epoch": 0.13, "learning_rate": 4.4201249999999996e-05, "loss": 1.6706, "step": 92780 }, { "epoch": 0.13, "learning_rate": 4.4200625e-05, "loss": 1.2581, "step": 92790 }, { "epoch": 0.13, "learning_rate": 4.4200000000000004e-05, "loss": 1.4015, "step": 92800 }, { "epoch": 0.13, "learning_rate": 4.4199375e-05, "loss": 1.4002, "step": 92810 }, { "epoch": 0.13, "learning_rate": 4.419875e-05, "loss": 1.4639, "step": 92820 }, { "epoch": 0.13, "learning_rate": 4.4198125e-05, "loss": 1.5329, "step": 92830 }, { "epoch": 0.13, "learning_rate": 4.4197500000000005e-05, "loss": 1.2438, "step": 92840 }, { "epoch": 0.13, "learning_rate": 4.4196875e-05, "loss": 1.3407, "step": 92850 }, { "epoch": 0.13, "learning_rate": 4.419625e-05, "loss": 1.5478, "step": 92860 }, { "epoch": 0.13, "learning_rate": 4.4195625e-05, "loss": 1.6165, "step": 92870 }, { "epoch": 0.13, "learning_rate": 4.4195000000000006e-05, "loss": 1.5897, "step": 92880 }, { "epoch": 0.13, "learning_rate": 4.4194375e-05, "loss": 1.5448, "step": 92890 }, { "epoch": 0.13, "learning_rate": 4.419375e-05, "loss": 1.6484, "step": 92900 }, { "epoch": 0.13, "learning_rate": 4.4193125000000004e-05, "loss": 1.4357, "step": 92910 }, { "epoch": 0.13, "learning_rate": 4.41925e-05, "loss": 1.5394, "step": 92920 }, { "epoch": 0.13, "learning_rate": 4.4191875000000005e-05, "loss": 1.3945, "step": 92930 }, { "epoch": 0.13, "learning_rate": 4.419125e-05, "loss": 1.4224, "step": 92940 }, { "epoch": 0.13, "learning_rate": 4.4190625000000005e-05, "loss": 1.4635, "step": 92950 }, { "epoch": 0.13, "learning_rate": 4.419e-05, "loss": 1.5295, "step": 92960 }, { "epoch": 0.13, "learning_rate": 4.4189375e-05, "loss": 1.65, "step": 92970 }, { "epoch": 0.13, "learning_rate": 4.418875e-05, "loss": 1.6375, "step": 92980 }, { "epoch": 0.13, "learning_rate": 4.4188125e-05, "loss": 1.4624, "step": 92990 }, { "epoch": 0.13, "learning_rate": 4.4187500000000003e-05, "loss": 1.4821, "step": 93000 }, { "epoch": 0.13, "eval_loss": 1.3924627304077148, "eval_runtime": 1300.0819, "eval_samples_per_second": 44.64, "eval_steps_per_second": 5.58, "step": 93000 }, { "epoch": 0.13, "learning_rate": 4.4186875e-05, "loss": 1.5875, "step": 93010 }, { "epoch": 0.13, "learning_rate": 4.4186250000000004e-05, "loss": 1.5831, "step": 93020 }, { "epoch": 0.13, "learning_rate": 4.4185625e-05, "loss": 1.3239, "step": 93030 }, { "epoch": 0.13, "learning_rate": 4.4185000000000005e-05, "loss": 1.5432, "step": 93040 }, { "epoch": 0.13, "learning_rate": 4.4184375e-05, "loss": 1.3281, "step": 93050 }, { "epoch": 0.13, "learning_rate": 4.418375e-05, "loss": 1.6422, "step": 93060 }, { "epoch": 0.13, "learning_rate": 4.4183125e-05, "loss": 1.2736, "step": 93070 }, { "epoch": 0.13, "learning_rate": 4.4182500000000006e-05, "loss": 1.4744, "step": 93080 }, { "epoch": 0.13, "learning_rate": 4.4181875e-05, "loss": 1.4328, "step": 93090 }, { "epoch": 0.13, "learning_rate": 4.418125e-05, "loss": 1.639, "step": 93100 }, { "epoch": 0.13, "learning_rate": 4.4180625000000004e-05, "loss": 1.4145, "step": 93110 }, { "epoch": 0.13, "learning_rate": 4.418000000000001e-05, "loss": 1.5364, "step": 93120 }, { "epoch": 0.13, "learning_rate": 4.4179375e-05, "loss": 1.5228, "step": 93130 }, { "epoch": 0.13, "learning_rate": 4.417875e-05, "loss": 1.3787, "step": 93140 }, { "epoch": 0.13, "learning_rate": 4.4178125000000005e-05, "loss": 1.3967, "step": 93150 }, { "epoch": 0.13, "learning_rate": 4.41775e-05, "loss": 1.4955, "step": 93160 }, { "epoch": 0.13, "learning_rate": 4.4176875e-05, "loss": 1.4563, "step": 93170 }, { "epoch": 0.13, "learning_rate": 4.417625e-05, "loss": 1.3664, "step": 93180 }, { "epoch": 0.13, "learning_rate": 4.4175625000000006e-05, "loss": 1.417, "step": 93190 }, { "epoch": 0.13, "learning_rate": 4.4174999999999996e-05, "loss": 1.562, "step": 93200 }, { "epoch": 0.13, "learning_rate": 4.4174375e-05, "loss": 1.3083, "step": 93210 }, { "epoch": 0.13, "learning_rate": 4.4173750000000004e-05, "loss": 1.5685, "step": 93220 }, { "epoch": 0.13, "learning_rate": 4.417312500000001e-05, "loss": 1.4871, "step": 93230 }, { "epoch": 0.13, "learning_rate": 4.41725e-05, "loss": 1.47, "step": 93240 }, { "epoch": 0.13, "learning_rate": 4.4171875e-05, "loss": 1.7269, "step": 93250 }, { "epoch": 0.13, "learning_rate": 4.4171250000000005e-05, "loss": 1.3179, "step": 93260 }, { "epoch": 0.13, "learning_rate": 4.4170625e-05, "loss": 1.3086, "step": 93270 }, { "epoch": 0.13, "learning_rate": 4.417e-05, "loss": 1.4124, "step": 93280 }, { "epoch": 0.13, "learning_rate": 4.4169375e-05, "loss": 1.4418, "step": 93290 }, { "epoch": 0.13, "learning_rate": 4.4168750000000006e-05, "loss": 1.3355, "step": 93300 }, { "epoch": 0.13, "learning_rate": 4.4168125e-05, "loss": 1.4922, "step": 93310 }, { "epoch": 0.13, "learning_rate": 4.41675e-05, "loss": 1.5626, "step": 93320 }, { "epoch": 0.13, "learning_rate": 4.4166875000000004e-05, "loss": 1.5254, "step": 93330 }, { "epoch": 0.13, "learning_rate": 4.416625e-05, "loss": 1.283, "step": 93340 }, { "epoch": 0.13, "learning_rate": 4.4165625e-05, "loss": 1.6646, "step": 93350 }, { "epoch": 0.13, "learning_rate": 4.4165e-05, "loss": 1.54, "step": 93360 }, { "epoch": 0.13, "learning_rate": 4.4164375000000005e-05, "loss": 1.6281, "step": 93370 }, { "epoch": 0.13, "learning_rate": 4.416375e-05, "loss": 1.6437, "step": 93380 }, { "epoch": 0.13, "learning_rate": 4.4163125e-05, "loss": 1.4972, "step": 93390 }, { "epoch": 0.13, "learning_rate": 4.41625e-05, "loss": 1.2602, "step": 93400 }, { "epoch": 0.13, "learning_rate": 4.4161875e-05, "loss": 1.5194, "step": 93410 }, { "epoch": 0.13, "learning_rate": 4.4161250000000003e-05, "loss": 1.5409, "step": 93420 }, { "epoch": 0.13, "learning_rate": 4.4160625e-05, "loss": 1.5909, "step": 93430 }, { "epoch": 0.13, "learning_rate": 4.4160000000000004e-05, "loss": 1.4201, "step": 93440 }, { "epoch": 0.13, "learning_rate": 4.4159375e-05, "loss": 1.2653, "step": 93450 }, { "epoch": 0.13, "learning_rate": 4.4158750000000005e-05, "loss": 1.5401, "step": 93460 }, { "epoch": 0.13, "learning_rate": 4.4158125e-05, "loss": 1.3577, "step": 93470 }, { "epoch": 0.13, "learning_rate": 4.41575e-05, "loss": 1.4913, "step": 93480 }, { "epoch": 0.13, "learning_rate": 4.4156875e-05, "loss": 1.4121, "step": 93490 }, { "epoch": 0.13, "learning_rate": 4.4156250000000006e-05, "loss": 1.5176, "step": 93500 }, { "epoch": 0.13, "learning_rate": 4.4155625e-05, "loss": 1.2458, "step": 93510 }, { "epoch": 0.13, "learning_rate": 4.4155e-05, "loss": 1.5848, "step": 93520 }, { "epoch": 0.13, "learning_rate": 4.4154375000000004e-05, "loss": 1.3613, "step": 93530 }, { "epoch": 0.13, "learning_rate": 4.415375000000001e-05, "loss": 1.4281, "step": 93540 }, { "epoch": 0.13, "learning_rate": 4.4153125e-05, "loss": 1.402, "step": 93550 }, { "epoch": 0.13, "learning_rate": 4.41525e-05, "loss": 1.4395, "step": 93560 }, { "epoch": 0.13, "learning_rate": 4.4151875000000005e-05, "loss": 1.4891, "step": 93570 }, { "epoch": 0.13, "learning_rate": 4.415125e-05, "loss": 1.2814, "step": 93580 }, { "epoch": 0.13, "learning_rate": 4.4150625e-05, "loss": 1.3591, "step": 93590 }, { "epoch": 0.13, "learning_rate": 4.415e-05, "loss": 1.441, "step": 93600 }, { "epoch": 0.13, "learning_rate": 4.4149375000000006e-05, "loss": 1.4377, "step": 93610 }, { "epoch": 0.13, "learning_rate": 4.414875e-05, "loss": 1.2947, "step": 93620 }, { "epoch": 0.13, "learning_rate": 4.4148125e-05, "loss": 1.6573, "step": 93630 }, { "epoch": 0.13, "learning_rate": 4.4147500000000004e-05, "loss": 1.4338, "step": 93640 }, { "epoch": 0.13, "learning_rate": 4.414687500000001e-05, "loss": 1.6001, "step": 93650 }, { "epoch": 0.13, "learning_rate": 4.414625e-05, "loss": 1.3198, "step": 93660 }, { "epoch": 0.13, "learning_rate": 4.4145625e-05, "loss": 1.3632, "step": 93670 }, { "epoch": 0.13, "learning_rate": 4.4145000000000005e-05, "loss": 1.5119, "step": 93680 }, { "epoch": 0.13, "learning_rate": 4.4144375e-05, "loss": 1.5481, "step": 93690 }, { "epoch": 0.13, "learning_rate": 4.414375e-05, "loss": 1.3468, "step": 93700 }, { "epoch": 0.13, "learning_rate": 4.4143125e-05, "loss": 1.738, "step": 93710 }, { "epoch": 0.13, "learning_rate": 4.4142500000000006e-05, "loss": 1.4566, "step": 93720 }, { "epoch": 0.13, "learning_rate": 4.4141875e-05, "loss": 1.4051, "step": 93730 }, { "epoch": 0.13, "learning_rate": 4.414125e-05, "loss": 1.5627, "step": 93740 }, { "epoch": 0.13, "learning_rate": 4.4140625000000004e-05, "loss": 1.3953, "step": 93750 }, { "epoch": 0.13, "learning_rate": 4.414e-05, "loss": 1.3201, "step": 93760 }, { "epoch": 0.13, "learning_rate": 4.4139375e-05, "loss": 1.506, "step": 93770 }, { "epoch": 0.13, "learning_rate": 4.413875e-05, "loss": 1.3508, "step": 93780 }, { "epoch": 0.13, "learning_rate": 4.4138125000000005e-05, "loss": 1.4497, "step": 93790 }, { "epoch": 0.13, "learning_rate": 4.41375e-05, "loss": 1.2502, "step": 93800 }, { "epoch": 0.13, "learning_rate": 4.4136875e-05, "loss": 1.4727, "step": 93810 }, { "epoch": 0.13, "learning_rate": 4.413625e-05, "loss": 1.6484, "step": 93820 }, { "epoch": 0.13, "learning_rate": 4.4135625e-05, "loss": 1.3733, "step": 93830 }, { "epoch": 0.13, "learning_rate": 4.4135000000000003e-05, "loss": 1.3745, "step": 93840 }, { "epoch": 0.13, "learning_rate": 4.4134375e-05, "loss": 1.3694, "step": 93850 }, { "epoch": 0.13, "learning_rate": 4.4133750000000004e-05, "loss": 1.4902, "step": 93860 }, { "epoch": 0.13, "learning_rate": 4.4133125e-05, "loss": 1.5401, "step": 93870 }, { "epoch": 0.13, "learning_rate": 4.4132500000000005e-05, "loss": 1.4357, "step": 93880 }, { "epoch": 0.13, "learning_rate": 4.4131875e-05, "loss": 1.4095, "step": 93890 }, { "epoch": 0.13, "learning_rate": 4.413125e-05, "loss": 1.3113, "step": 93900 }, { "epoch": 0.13, "learning_rate": 4.4130625e-05, "loss": 1.4795, "step": 93910 }, { "epoch": 0.13, "learning_rate": 4.4130000000000006e-05, "loss": 1.552, "step": 93920 }, { "epoch": 0.13, "learning_rate": 4.4129375e-05, "loss": 1.3453, "step": 93930 }, { "epoch": 0.13, "learning_rate": 4.412875e-05, "loss": 1.5058, "step": 93940 }, { "epoch": 0.13, "learning_rate": 4.4128125000000004e-05, "loss": 1.5047, "step": 93950 }, { "epoch": 0.13, "learning_rate": 4.41275e-05, "loss": 1.3819, "step": 93960 }, { "epoch": 0.13, "learning_rate": 4.4126875e-05, "loss": 1.7528, "step": 93970 }, { "epoch": 0.13, "learning_rate": 4.412625e-05, "loss": 1.5317, "step": 93980 }, { "epoch": 0.13, "learning_rate": 4.4125625000000005e-05, "loss": 1.5251, "step": 93990 }, { "epoch": 0.13, "learning_rate": 4.4125e-05, "loss": 1.5949, "step": 94000 }, { "epoch": 0.13, "eval_loss": 1.379119634628296, "eval_runtime": 1306.5693, "eval_samples_per_second": 44.419, "eval_steps_per_second": 5.553, "step": 94000 }, { "epoch": 0.13, "learning_rate": 4.4124375e-05, "loss": 1.3598, "step": 94010 }, { "epoch": 0.13, "learning_rate": 4.412375e-05, "loss": 1.5432, "step": 94020 }, { "epoch": 0.13, "learning_rate": 4.4123125000000006e-05, "loss": 1.545, "step": 94030 }, { "epoch": 0.13, "learning_rate": 4.41225e-05, "loss": 1.4129, "step": 94040 }, { "epoch": 0.13, "learning_rate": 4.4121875e-05, "loss": 1.4153, "step": 94050 }, { "epoch": 0.13, "learning_rate": 4.4121250000000004e-05, "loss": 1.5925, "step": 94060 }, { "epoch": 0.13, "learning_rate": 4.412062500000001e-05, "loss": 1.5341, "step": 94070 }, { "epoch": 0.13, "learning_rate": 4.412e-05, "loss": 1.3336, "step": 94080 }, { "epoch": 0.13, "learning_rate": 4.4119375e-05, "loss": 1.5199, "step": 94090 }, { "epoch": 0.13, "learning_rate": 4.4118750000000005e-05, "loss": 1.4445, "step": 94100 }, { "epoch": 0.13, "learning_rate": 4.4118125e-05, "loss": 1.5536, "step": 94110 }, { "epoch": 0.13, "learning_rate": 4.41175e-05, "loss": 1.2227, "step": 94120 }, { "epoch": 0.13, "learning_rate": 4.4116875e-05, "loss": 1.6894, "step": 94130 }, { "epoch": 0.13, "learning_rate": 4.4116250000000006e-05, "loss": 1.2653, "step": 94140 }, { "epoch": 0.13, "learning_rate": 4.4115625e-05, "loss": 1.4906, "step": 94150 }, { "epoch": 0.13, "learning_rate": 4.4115e-05, "loss": 1.471, "step": 94160 }, { "epoch": 0.13, "learning_rate": 4.4114375000000004e-05, "loss": 1.5987, "step": 94170 }, { "epoch": 0.13, "learning_rate": 4.411375e-05, "loss": 1.4071, "step": 94180 }, { "epoch": 0.13, "learning_rate": 4.4113125e-05, "loss": 1.4207, "step": 94190 }, { "epoch": 0.13, "learning_rate": 4.41125e-05, "loss": 1.4811, "step": 94200 }, { "epoch": 0.13, "learning_rate": 4.4111875000000005e-05, "loss": 1.6731, "step": 94210 }, { "epoch": 0.13, "learning_rate": 4.411125e-05, "loss": 1.3617, "step": 94220 }, { "epoch": 0.13, "learning_rate": 4.4110625e-05, "loss": 1.2801, "step": 94230 }, { "epoch": 0.13, "learning_rate": 4.411e-05, "loss": 1.3821, "step": 94240 }, { "epoch": 0.13, "learning_rate": 4.4109375e-05, "loss": 1.5481, "step": 94250 }, { "epoch": 0.13, "learning_rate": 4.4108750000000004e-05, "loss": 1.6044, "step": 94260 }, { "epoch": 0.13, "learning_rate": 4.4108125e-05, "loss": 1.5609, "step": 94270 }, { "epoch": 0.13, "learning_rate": 4.4107500000000004e-05, "loss": 1.419, "step": 94280 }, { "epoch": 0.13, "learning_rate": 4.4106875e-05, "loss": 1.523, "step": 94290 }, { "epoch": 0.13, "learning_rate": 4.4106250000000005e-05, "loss": 1.6427, "step": 94300 }, { "epoch": 0.13, "learning_rate": 4.4105625e-05, "loss": 1.4743, "step": 94310 }, { "epoch": 0.13, "learning_rate": 4.4105e-05, "loss": 1.4662, "step": 94320 }, { "epoch": 0.13, "learning_rate": 4.4104375e-05, "loss": 1.4208, "step": 94330 }, { "epoch": 0.13, "learning_rate": 4.4103750000000006e-05, "loss": 1.5989, "step": 94340 }, { "epoch": 0.13, "learning_rate": 4.4103125e-05, "loss": 1.4726, "step": 94350 }, { "epoch": 0.13, "learning_rate": 4.41025e-05, "loss": 1.5648, "step": 94360 }, { "epoch": 0.13, "learning_rate": 4.4101875000000004e-05, "loss": 1.3693, "step": 94370 }, { "epoch": 0.13, "learning_rate": 4.410125e-05, "loss": 1.4236, "step": 94380 }, { "epoch": 0.13, "learning_rate": 4.4100625000000004e-05, "loss": 1.4178, "step": 94390 }, { "epoch": 0.13, "learning_rate": 4.41e-05, "loss": 1.4309, "step": 94400 }, { "epoch": 0.13, "learning_rate": 4.4099375000000005e-05, "loss": 1.588, "step": 94410 }, { "epoch": 0.13, "learning_rate": 4.409875e-05, "loss": 1.4898, "step": 94420 }, { "epoch": 0.13, "learning_rate": 4.4098125e-05, "loss": 1.6459, "step": 94430 }, { "epoch": 0.13, "learning_rate": 4.40975e-05, "loss": 1.4862, "step": 94440 }, { "epoch": 0.13, "learning_rate": 4.4096875000000006e-05, "loss": 1.48, "step": 94450 }, { "epoch": 0.13, "learning_rate": 4.409625e-05, "loss": 1.5352, "step": 94460 }, { "epoch": 0.13, "learning_rate": 4.4095625e-05, "loss": 1.4232, "step": 94470 }, { "epoch": 0.13, "learning_rate": 4.4095000000000004e-05, "loss": 1.4074, "step": 94480 }, { "epoch": 0.13, "learning_rate": 4.409437500000001e-05, "loss": 1.6424, "step": 94490 }, { "epoch": 0.13, "learning_rate": 4.409375e-05, "loss": 1.4481, "step": 94500 }, { "epoch": 0.13, "learning_rate": 4.4093125e-05, "loss": 1.4186, "step": 94510 }, { "epoch": 0.13, "learning_rate": 4.4092500000000005e-05, "loss": 1.4387, "step": 94520 }, { "epoch": 0.13, "learning_rate": 4.4091875e-05, "loss": 1.2522, "step": 94530 }, { "epoch": 0.13, "learning_rate": 4.409125e-05, "loss": 1.3289, "step": 94540 }, { "epoch": 0.13, "learning_rate": 4.4090625e-05, "loss": 1.1995, "step": 94550 }, { "epoch": 0.13, "learning_rate": 4.4090000000000006e-05, "loss": 1.6025, "step": 94560 }, { "epoch": 0.13, "learning_rate": 4.4089375e-05, "loss": 1.5216, "step": 94570 }, { "epoch": 0.13, "learning_rate": 4.408875e-05, "loss": 1.2445, "step": 94580 }, { "epoch": 0.13, "learning_rate": 4.4088125000000004e-05, "loss": 1.4187, "step": 94590 }, { "epoch": 0.13, "learning_rate": 4.40875e-05, "loss": 1.5593, "step": 94600 }, { "epoch": 0.13, "learning_rate": 4.4086875e-05, "loss": 1.2626, "step": 94610 }, { "epoch": 0.13, "learning_rate": 4.408625e-05, "loss": 1.4327, "step": 94620 }, { "epoch": 0.13, "learning_rate": 4.4085625000000005e-05, "loss": 1.4119, "step": 94630 }, { "epoch": 0.13, "learning_rate": 4.4085e-05, "loss": 1.5863, "step": 94640 }, { "epoch": 0.13, "learning_rate": 4.4084375e-05, "loss": 1.4105, "step": 94650 }, { "epoch": 0.13, "learning_rate": 4.408375e-05, "loss": 1.387, "step": 94660 }, { "epoch": 0.13, "learning_rate": 4.4083125e-05, "loss": 1.4899, "step": 94670 }, { "epoch": 0.13, "learning_rate": 4.4082500000000004e-05, "loss": 1.5019, "step": 94680 }, { "epoch": 0.13, "learning_rate": 4.4081875e-05, "loss": 1.4067, "step": 94690 }, { "epoch": 0.13, "learning_rate": 4.4081250000000004e-05, "loss": 1.3784, "step": 94700 }, { "epoch": 0.13, "learning_rate": 4.4080625e-05, "loss": 1.3809, "step": 94710 }, { "epoch": 0.13, "learning_rate": 4.4080000000000005e-05, "loss": 1.2904, "step": 94720 }, { "epoch": 0.13, "learning_rate": 4.4079375e-05, "loss": 1.3566, "step": 94730 }, { "epoch": 0.13, "learning_rate": 4.407875e-05, "loss": 1.4816, "step": 94740 }, { "epoch": 0.13, "learning_rate": 4.4078125e-05, "loss": 1.2324, "step": 94750 }, { "epoch": 0.13, "learning_rate": 4.40775e-05, "loss": 1.3096, "step": 94760 }, { "epoch": 0.13, "learning_rate": 4.4076875e-05, "loss": 1.3903, "step": 94770 }, { "epoch": 0.13, "learning_rate": 4.407625e-05, "loss": 1.3292, "step": 94780 }, { "epoch": 0.13, "learning_rate": 4.4075625000000004e-05, "loss": 1.4571, "step": 94790 }, { "epoch": 0.13, "learning_rate": 4.4075e-05, "loss": 1.336, "step": 94800 }, { "epoch": 0.13, "learning_rate": 4.4074375000000004e-05, "loss": 1.58, "step": 94810 }, { "epoch": 0.13, "learning_rate": 4.407375e-05, "loss": 1.461, "step": 94820 }, { "epoch": 0.13, "learning_rate": 4.4073125000000005e-05, "loss": 1.3984, "step": 94830 }, { "epoch": 0.13, "learning_rate": 4.40725e-05, "loss": 1.3621, "step": 94840 }, { "epoch": 0.13, "learning_rate": 4.4071875e-05, "loss": 1.328, "step": 94850 }, { "epoch": 0.13, "learning_rate": 4.407125e-05, "loss": 1.2879, "step": 94860 }, { "epoch": 0.13, "learning_rate": 4.4070625000000006e-05, "loss": 1.688, "step": 94870 }, { "epoch": 0.13, "learning_rate": 4.407e-05, "loss": 1.5975, "step": 94880 }, { "epoch": 0.13, "learning_rate": 4.4069375e-05, "loss": 1.3863, "step": 94890 }, { "epoch": 0.13, "learning_rate": 4.4068750000000004e-05, "loss": 1.5073, "step": 94900 }, { "epoch": 0.13, "learning_rate": 4.406812500000001e-05, "loss": 1.6213, "step": 94910 }, { "epoch": 0.13, "learning_rate": 4.40675e-05, "loss": 1.4904, "step": 94920 }, { "epoch": 0.13, "learning_rate": 4.4066875e-05, "loss": 1.2557, "step": 94930 }, { "epoch": 0.13, "learning_rate": 4.4066250000000005e-05, "loss": 1.2852, "step": 94940 }, { "epoch": 0.13, "learning_rate": 4.4065625e-05, "loss": 1.5848, "step": 94950 }, { "epoch": 0.13, "learning_rate": 4.4065e-05, "loss": 1.3188, "step": 94960 }, { "epoch": 0.13, "learning_rate": 4.4064375e-05, "loss": 1.4849, "step": 94970 }, { "epoch": 0.13, "learning_rate": 4.4063750000000007e-05, "loss": 1.2275, "step": 94980 }, { "epoch": 0.13, "learning_rate": 4.4063125e-05, "loss": 1.6144, "step": 94990 }, { "epoch": 0.13, "learning_rate": 4.40625e-05, "loss": 1.3005, "step": 95000 }, { "epoch": 0.13, "eval_loss": 1.37950599193573, "eval_runtime": 1320.8924, "eval_samples_per_second": 43.937, "eval_steps_per_second": 5.492, "step": 95000 }, { "epoch": 0.13, "learning_rate": 4.4061875000000004e-05, "loss": 1.3339, "step": 95010 }, { "epoch": 0.13, "learning_rate": 4.406125e-05, "loss": 1.3371, "step": 95020 }, { "epoch": 0.13, "learning_rate": 4.4060625e-05, "loss": 1.5869, "step": 95030 }, { "epoch": 0.13, "learning_rate": 4.406e-05, "loss": 1.5179, "step": 95040 }, { "epoch": 0.13, "learning_rate": 4.4059375000000005e-05, "loss": 1.5629, "step": 95050 }, { "epoch": 0.13, "learning_rate": 4.405875e-05, "loss": 1.4092, "step": 95060 }, { "epoch": 0.13, "learning_rate": 4.4058125e-05, "loss": 1.5827, "step": 95070 }, { "epoch": 0.13, "learning_rate": 4.40575e-05, "loss": 1.2523, "step": 95080 }, { "epoch": 0.13, "learning_rate": 4.4056875e-05, "loss": 1.4674, "step": 95090 }, { "epoch": 0.13, "learning_rate": 4.4056250000000004e-05, "loss": 1.4654, "step": 95100 }, { "epoch": 0.13, "learning_rate": 4.4055625e-05, "loss": 1.4328, "step": 95110 }, { "epoch": 0.13, "learning_rate": 4.4055000000000004e-05, "loss": 1.4462, "step": 95120 }, { "epoch": 0.13, "learning_rate": 4.4054375e-05, "loss": 1.6577, "step": 95130 }, { "epoch": 0.13, "learning_rate": 4.4053750000000005e-05, "loss": 1.3817, "step": 95140 }, { "epoch": 0.13, "learning_rate": 4.4053125e-05, "loss": 1.278, "step": 95150 }, { "epoch": 0.13, "learning_rate": 4.40525e-05, "loss": 1.436, "step": 95160 }, { "epoch": 0.13, "learning_rate": 4.4051875e-05, "loss": 1.3518, "step": 95170 }, { "epoch": 0.13, "learning_rate": 4.405125e-05, "loss": 1.5514, "step": 95180 }, { "epoch": 0.13, "learning_rate": 4.4050625e-05, "loss": 1.6952, "step": 95190 }, { "epoch": 0.13, "learning_rate": 4.405e-05, "loss": 1.408, "step": 95200 }, { "epoch": 0.13, "learning_rate": 4.4049375000000004e-05, "loss": 1.4465, "step": 95210 }, { "epoch": 0.13, "learning_rate": 4.404875e-05, "loss": 1.3666, "step": 95220 }, { "epoch": 0.13, "learning_rate": 4.4048125000000004e-05, "loss": 1.4847, "step": 95230 }, { "epoch": 0.13, "learning_rate": 4.40475e-05, "loss": 1.3612, "step": 95240 }, { "epoch": 0.13, "learning_rate": 4.4046875000000005e-05, "loss": 1.5337, "step": 95250 }, { "epoch": 0.13, "learning_rate": 4.404625e-05, "loss": 1.2673, "step": 95260 }, { "epoch": 0.13, "learning_rate": 4.4045625e-05, "loss": 1.3687, "step": 95270 }, { "epoch": 0.13, "learning_rate": 4.4045e-05, "loss": 1.3282, "step": 95280 }, { "epoch": 0.13, "learning_rate": 4.4044375000000006e-05, "loss": 1.5108, "step": 95290 }, { "epoch": 0.13, "learning_rate": 4.404375e-05, "loss": 1.3273, "step": 95300 }, { "epoch": 0.13, "learning_rate": 4.4043125e-05, "loss": 1.657, "step": 95310 }, { "epoch": 0.13, "learning_rate": 4.4042500000000004e-05, "loss": 1.2864, "step": 95320 }, { "epoch": 0.13, "learning_rate": 4.404187500000001e-05, "loss": 1.3797, "step": 95330 }, { "epoch": 0.13, "learning_rate": 4.404125e-05, "loss": 1.4421, "step": 95340 }, { "epoch": 0.13, "learning_rate": 4.4040625e-05, "loss": 1.5158, "step": 95350 }, { "epoch": 0.13, "learning_rate": 4.4040000000000005e-05, "loss": 1.389, "step": 95360 }, { "epoch": 0.13, "learning_rate": 4.4039375e-05, "loss": 1.3861, "step": 95370 }, { "epoch": 0.13, "learning_rate": 4.403875e-05, "loss": 1.4524, "step": 95380 }, { "epoch": 0.13, "learning_rate": 4.4038125e-05, "loss": 1.3235, "step": 95390 }, { "epoch": 0.13, "learning_rate": 4.4037500000000007e-05, "loss": 1.1547, "step": 95400 }, { "epoch": 0.13, "learning_rate": 4.4036875e-05, "loss": 1.4211, "step": 95410 }, { "epoch": 0.13, "learning_rate": 4.403625e-05, "loss": 1.3601, "step": 95420 }, { "epoch": 0.13, "learning_rate": 4.4035625000000004e-05, "loss": 1.3873, "step": 95430 }, { "epoch": 0.13, "learning_rate": 4.4035e-05, "loss": 1.5625, "step": 95440 }, { "epoch": 0.13, "learning_rate": 4.4034375e-05, "loss": 1.6381, "step": 95450 }, { "epoch": 0.13, "learning_rate": 4.403375e-05, "loss": 1.3611, "step": 95460 }, { "epoch": 0.13, "learning_rate": 4.4033125000000005e-05, "loss": 1.5928, "step": 95470 }, { "epoch": 0.13, "learning_rate": 4.40325e-05, "loss": 1.2693, "step": 95480 }, { "epoch": 0.13, "learning_rate": 4.4031875e-05, "loss": 1.4336, "step": 95490 }, { "epoch": 0.13, "learning_rate": 4.403125e-05, "loss": 1.5216, "step": 95500 }, { "epoch": 0.13, "learning_rate": 4.4030625e-05, "loss": 1.5549, "step": 95510 }, { "epoch": 0.13, "learning_rate": 4.4030000000000004e-05, "loss": 1.373, "step": 95520 }, { "epoch": 0.13, "learning_rate": 4.4029375e-05, "loss": 1.5094, "step": 95530 }, { "epoch": 0.13, "learning_rate": 4.4028750000000004e-05, "loss": 1.6094, "step": 95540 }, { "epoch": 0.13, "learning_rate": 4.4028125e-05, "loss": 1.7623, "step": 95550 }, { "epoch": 0.13, "learning_rate": 4.4027500000000005e-05, "loss": 1.4062, "step": 95560 }, { "epoch": 0.13, "learning_rate": 4.4026875e-05, "loss": 1.338, "step": 95570 }, { "epoch": 0.13, "learning_rate": 4.4026250000000006e-05, "loss": 1.5322, "step": 95580 }, { "epoch": 0.13, "learning_rate": 4.4025625e-05, "loss": 1.4379, "step": 95590 }, { "epoch": 0.13, "learning_rate": 4.4025e-05, "loss": 1.4336, "step": 95600 }, { "epoch": 0.13, "learning_rate": 4.4024375e-05, "loss": 1.4185, "step": 95610 }, { "epoch": 0.13, "learning_rate": 4.402375e-05, "loss": 1.5758, "step": 95620 }, { "epoch": 0.13, "learning_rate": 4.4023125000000004e-05, "loss": 1.5782, "step": 95630 }, { "epoch": 0.13, "learning_rate": 4.40225e-05, "loss": 1.3617, "step": 95640 }, { "epoch": 0.13, "learning_rate": 4.4021875000000004e-05, "loss": 1.7001, "step": 95650 }, { "epoch": 0.13, "learning_rate": 4.402125e-05, "loss": 1.3457, "step": 95660 }, { "epoch": 0.13, "learning_rate": 4.4020625000000005e-05, "loss": 1.4187, "step": 95670 }, { "epoch": 0.13, "learning_rate": 4.402e-05, "loss": 1.4084, "step": 95680 }, { "epoch": 0.13, "learning_rate": 4.4019375e-05, "loss": 1.3077, "step": 95690 }, { "epoch": 0.13, "learning_rate": 4.401875e-05, "loss": 1.6014, "step": 95700 }, { "epoch": 0.13, "learning_rate": 4.4018125000000006e-05, "loss": 1.5133, "step": 95710 }, { "epoch": 0.13, "learning_rate": 4.40175e-05, "loss": 1.4507, "step": 95720 }, { "epoch": 0.13, "learning_rate": 4.4016875e-05, "loss": 1.4139, "step": 95730 }, { "epoch": 0.13, "learning_rate": 4.4016250000000004e-05, "loss": 1.5589, "step": 95740 }, { "epoch": 0.13, "learning_rate": 4.401562500000001e-05, "loss": 1.4279, "step": 95750 }, { "epoch": 0.13, "learning_rate": 4.4015e-05, "loss": 1.4264, "step": 95760 }, { "epoch": 0.13, "learning_rate": 4.4014375e-05, "loss": 1.4807, "step": 95770 }, { "epoch": 0.13, "learning_rate": 4.4013750000000005e-05, "loss": 1.6669, "step": 95780 }, { "epoch": 0.13, "learning_rate": 4.4013125e-05, "loss": 1.4468, "step": 95790 }, { "epoch": 0.13, "learning_rate": 4.40125e-05, "loss": 1.3805, "step": 95800 }, { "epoch": 0.13, "learning_rate": 4.4011875e-05, "loss": 1.555, "step": 95810 }, { "epoch": 0.13, "learning_rate": 4.4011250000000007e-05, "loss": 1.2329, "step": 95820 }, { "epoch": 0.13, "learning_rate": 4.4010625e-05, "loss": 1.689, "step": 95830 }, { "epoch": 0.13, "learning_rate": 4.401e-05, "loss": 1.4018, "step": 95840 }, { "epoch": 0.13, "learning_rate": 4.4009375000000004e-05, "loss": 1.3187, "step": 95850 }, { "epoch": 0.13, "learning_rate": 4.400875e-05, "loss": 1.3361, "step": 95860 }, { "epoch": 0.13, "learning_rate": 4.4008125e-05, "loss": 1.481, "step": 95870 }, { "epoch": 0.13, "learning_rate": 4.40075e-05, "loss": 1.5259, "step": 95880 }, { "epoch": 0.13, "learning_rate": 4.4006875000000005e-05, "loss": 1.4963, "step": 95890 }, { "epoch": 0.13, "learning_rate": 4.400625e-05, "loss": 1.3987, "step": 95900 }, { "epoch": 0.13, "learning_rate": 4.4005625e-05, "loss": 1.393, "step": 95910 }, { "epoch": 0.13, "learning_rate": 4.4005e-05, "loss": 1.4333, "step": 95920 }, { "epoch": 0.13, "learning_rate": 4.4004375e-05, "loss": 1.6155, "step": 95930 }, { "epoch": 0.13, "learning_rate": 4.4003750000000004e-05, "loss": 1.606, "step": 95940 }, { "epoch": 0.13, "learning_rate": 4.4003125e-05, "loss": 1.5822, "step": 95950 }, { "epoch": 0.13, "learning_rate": 4.4002500000000004e-05, "loss": 1.5346, "step": 95960 }, { "epoch": 0.13, "learning_rate": 4.4001875e-05, "loss": 1.2455, "step": 95970 }, { "epoch": 0.13, "learning_rate": 4.400125e-05, "loss": 1.3218, "step": 95980 }, { "epoch": 0.13, "learning_rate": 4.4000625e-05, "loss": 1.3688, "step": 95990 }, { "epoch": 0.13, "learning_rate": 4.4000000000000006e-05, "loss": 1.5393, "step": 96000 }, { "epoch": 0.13, "eval_loss": 1.3718936443328857, "eval_runtime": 1303.3151, "eval_samples_per_second": 44.53, "eval_steps_per_second": 5.567, "step": 96000 }, { "epoch": 0.13, "learning_rate": 4.3999375e-05, "loss": 1.5231, "step": 96010 }, { "epoch": 0.13, "learning_rate": 4.399875e-05, "loss": 1.3562, "step": 96020 }, { "epoch": 0.13, "learning_rate": 4.3998125e-05, "loss": 1.2402, "step": 96030 }, { "epoch": 0.13, "learning_rate": 4.39975e-05, "loss": 1.5917, "step": 96040 }, { "epoch": 0.13, "learning_rate": 4.3996875000000004e-05, "loss": 1.5433, "step": 96050 }, { "epoch": 0.13, "learning_rate": 4.399625e-05, "loss": 1.4622, "step": 96060 }, { "epoch": 0.13, "learning_rate": 4.3995625000000005e-05, "loss": 1.6371, "step": 96070 }, { "epoch": 0.13, "learning_rate": 4.3995e-05, "loss": 1.2302, "step": 96080 }, { "epoch": 0.13, "learning_rate": 4.3994375000000005e-05, "loss": 1.4961, "step": 96090 }, { "epoch": 0.13, "learning_rate": 4.399375e-05, "loss": 1.3436, "step": 96100 }, { "epoch": 0.13, "learning_rate": 4.3993125e-05, "loss": 1.3962, "step": 96110 }, { "epoch": 0.13, "learning_rate": 4.39925e-05, "loss": 1.1797, "step": 96120 }, { "epoch": 0.13, "learning_rate": 4.3991875000000006e-05, "loss": 1.5016, "step": 96130 }, { "epoch": 0.13, "learning_rate": 4.399125e-05, "loss": 1.3432, "step": 96140 }, { "epoch": 0.13, "learning_rate": 4.3990625e-05, "loss": 1.3733, "step": 96150 }, { "epoch": 0.13, "learning_rate": 4.3990000000000004e-05, "loss": 1.6093, "step": 96160 }, { "epoch": 0.13, "learning_rate": 4.3989375e-05, "loss": 1.5604, "step": 96170 }, { "epoch": 0.13, "learning_rate": 4.398875e-05, "loss": 1.3781, "step": 96180 }, { "epoch": 0.13, "learning_rate": 4.3988125e-05, "loss": 1.4831, "step": 96190 }, { "epoch": 0.13, "learning_rate": 4.3987500000000005e-05, "loss": 1.2771, "step": 96200 }, { "epoch": 0.13, "learning_rate": 4.3986875e-05, "loss": 1.3696, "step": 96210 }, { "epoch": 0.13, "learning_rate": 4.398625e-05, "loss": 1.5831, "step": 96220 }, { "epoch": 0.13, "learning_rate": 4.3985625e-05, "loss": 1.5303, "step": 96230 }, { "epoch": 0.13, "learning_rate": 4.398500000000001e-05, "loss": 1.3109, "step": 96240 }, { "epoch": 0.13, "learning_rate": 4.3984375e-05, "loss": 1.4265, "step": 96250 }, { "epoch": 0.13, "learning_rate": 4.398375e-05, "loss": 1.4608, "step": 96260 }, { "epoch": 0.13, "learning_rate": 4.3983125000000004e-05, "loss": 1.4147, "step": 96270 }, { "epoch": 0.13, "learning_rate": 4.39825e-05, "loss": 1.4755, "step": 96280 }, { "epoch": 0.13, "learning_rate": 4.3981875e-05, "loss": 1.4012, "step": 96290 }, { "epoch": 0.13, "learning_rate": 4.398125e-05, "loss": 1.2153, "step": 96300 }, { "epoch": 0.13, "learning_rate": 4.3980625000000006e-05, "loss": 1.3044, "step": 96310 }, { "epoch": 0.13, "learning_rate": 4.398e-05, "loss": 1.4598, "step": 96320 }, { "epoch": 0.13, "learning_rate": 4.3979375e-05, "loss": 1.6441, "step": 96330 }, { "epoch": 0.13, "learning_rate": 4.397875e-05, "loss": 1.4339, "step": 96340 }, { "epoch": 0.13, "learning_rate": 4.3978125e-05, "loss": 1.2824, "step": 96350 }, { "epoch": 0.13, "learning_rate": 4.3977500000000004e-05, "loss": 1.5076, "step": 96360 }, { "epoch": 0.13, "learning_rate": 4.3976875e-05, "loss": 1.4008, "step": 96370 }, { "epoch": 0.13, "learning_rate": 4.3976250000000004e-05, "loss": 1.5697, "step": 96380 }, { "epoch": 0.13, "learning_rate": 4.3975625e-05, "loss": 1.5203, "step": 96390 }, { "epoch": 0.13, "learning_rate": 4.3975e-05, "loss": 1.4071, "step": 96400 }, { "epoch": 0.13, "learning_rate": 4.3974375e-05, "loss": 1.6181, "step": 96410 }, { "epoch": 0.13, "learning_rate": 4.3973750000000006e-05, "loss": 1.3642, "step": 96420 }, { "epoch": 0.13, "learning_rate": 4.3973125e-05, "loss": 1.4061, "step": 96430 }, { "epoch": 0.13, "learning_rate": 4.39725e-05, "loss": 1.3878, "step": 96440 }, { "epoch": 0.13, "learning_rate": 4.3971875e-05, "loss": 1.217, "step": 96450 }, { "epoch": 0.13, "learning_rate": 4.397125e-05, "loss": 1.4672, "step": 96460 }, { "epoch": 0.13, "learning_rate": 4.3970625000000004e-05, "loss": 1.3473, "step": 96470 }, { "epoch": 0.13, "learning_rate": 4.397e-05, "loss": 1.431, "step": 96480 }, { "epoch": 0.13, "learning_rate": 4.3969375000000005e-05, "loss": 1.3847, "step": 96490 }, { "epoch": 0.13, "learning_rate": 4.396875e-05, "loss": 1.397, "step": 96500 }, { "epoch": 0.13, "learning_rate": 4.3968125000000005e-05, "loss": 1.4784, "step": 96510 }, { "epoch": 0.13, "learning_rate": 4.39675e-05, "loss": 1.731, "step": 96520 }, { "epoch": 0.13, "learning_rate": 4.3966875e-05, "loss": 1.4252, "step": 96530 }, { "epoch": 0.13, "learning_rate": 4.396625e-05, "loss": 1.5477, "step": 96540 }, { "epoch": 0.13, "learning_rate": 4.3965625000000006e-05, "loss": 1.3511, "step": 96550 }, { "epoch": 0.13, "learning_rate": 4.3965000000000003e-05, "loss": 1.5624, "step": 96560 }, { "epoch": 0.13, "learning_rate": 4.3964375e-05, "loss": 1.6047, "step": 96570 }, { "epoch": 0.13, "learning_rate": 4.3963750000000004e-05, "loss": 1.4655, "step": 96580 }, { "epoch": 0.13, "learning_rate": 4.3963125e-05, "loss": 1.5227, "step": 96590 }, { "epoch": 0.13, "learning_rate": 4.39625e-05, "loss": 1.4177, "step": 96600 }, { "epoch": 0.13, "learning_rate": 4.3961875e-05, "loss": 1.5533, "step": 96610 }, { "epoch": 0.13, "learning_rate": 4.3961250000000005e-05, "loss": 1.5185, "step": 96620 }, { "epoch": 0.13, "learning_rate": 4.3960625e-05, "loss": 1.3115, "step": 96630 }, { "epoch": 0.13, "learning_rate": 4.396e-05, "loss": 1.5952, "step": 96640 }, { "epoch": 0.13, "learning_rate": 4.3959375e-05, "loss": 1.421, "step": 96650 }, { "epoch": 0.13, "learning_rate": 4.395875000000001e-05, "loss": 1.5137, "step": 96660 }, { "epoch": 0.13, "learning_rate": 4.3958125e-05, "loss": 1.3669, "step": 96670 }, { "epoch": 0.13, "learning_rate": 4.39575e-05, "loss": 1.3581, "step": 96680 }, { "epoch": 0.13, "learning_rate": 4.3956875000000004e-05, "loss": 1.4235, "step": 96690 }, { "epoch": 0.13, "learning_rate": 4.395625e-05, "loss": 1.5109, "step": 96700 }, { "epoch": 0.13, "learning_rate": 4.3955625e-05, "loss": 1.5665, "step": 96710 }, { "epoch": 0.13, "learning_rate": 4.3955e-05, "loss": 1.4002, "step": 96720 }, { "epoch": 0.13, "learning_rate": 4.3954375000000006e-05, "loss": 1.204, "step": 96730 }, { "epoch": 0.13, "learning_rate": 4.395375e-05, "loss": 1.3894, "step": 96740 }, { "epoch": 0.13, "learning_rate": 4.3953125e-05, "loss": 1.5287, "step": 96750 }, { "epoch": 0.13, "learning_rate": 4.39525e-05, "loss": 1.4548, "step": 96760 }, { "epoch": 0.13, "learning_rate": 4.395187500000001e-05, "loss": 1.3915, "step": 96770 }, { "epoch": 0.13, "learning_rate": 4.395125e-05, "loss": 1.4658, "step": 96780 }, { "epoch": 0.13, "learning_rate": 4.3950625e-05, "loss": 1.4163, "step": 96790 }, { "epoch": 0.13, "learning_rate": 4.3950000000000004e-05, "loss": 1.6936, "step": 96800 }, { "epoch": 0.13, "learning_rate": 4.3949375e-05, "loss": 1.3899, "step": 96810 }, { "epoch": 0.13, "learning_rate": 4.394875e-05, "loss": 1.3326, "step": 96820 }, { "epoch": 0.13, "learning_rate": 4.3948125e-05, "loss": 1.5171, "step": 96830 }, { "epoch": 0.13, "learning_rate": 4.3947500000000006e-05, "loss": 1.2988, "step": 96840 }, { "epoch": 0.13, "learning_rate": 4.3946875e-05, "loss": 1.2827, "step": 96850 }, { "epoch": 0.13, "learning_rate": 4.394625e-05, "loss": 1.3397, "step": 96860 }, { "epoch": 0.13, "learning_rate": 4.3945625e-05, "loss": 1.3143, "step": 96870 }, { "epoch": 0.13, "learning_rate": 4.3945e-05, "loss": 1.3252, "step": 96880 }, { "epoch": 0.13, "learning_rate": 4.3944375000000004e-05, "loss": 1.5797, "step": 96890 }, { "epoch": 0.13, "learning_rate": 4.394375e-05, "loss": 1.2518, "step": 96900 }, { "epoch": 0.13, "learning_rate": 4.3943125000000005e-05, "loss": 1.3062, "step": 96910 }, { "epoch": 0.13, "learning_rate": 4.39425e-05, "loss": 1.4574, "step": 96920 }, { "epoch": 0.13, "learning_rate": 4.3941875000000005e-05, "loss": 1.4343, "step": 96930 }, { "epoch": 0.13, "learning_rate": 4.394125e-05, "loss": 1.4848, "step": 96940 }, { "epoch": 0.13, "learning_rate": 4.3940625e-05, "loss": 1.6212, "step": 96950 }, { "epoch": 0.14, "learning_rate": 4.394e-05, "loss": 1.276, "step": 96960 }, { "epoch": 0.14, "learning_rate": 4.3939375000000007e-05, "loss": 1.327, "step": 96970 }, { "epoch": 0.14, "learning_rate": 4.3938750000000003e-05, "loss": 1.5038, "step": 96980 }, { "epoch": 0.14, "learning_rate": 4.3938125e-05, "loss": 1.6831, "step": 96990 }, { "epoch": 0.14, "learning_rate": 4.3937500000000004e-05, "loss": 1.5303, "step": 97000 }, { "epoch": 0.14, "eval_loss": 1.368253469467163, "eval_runtime": 1300.0688, "eval_samples_per_second": 44.641, "eval_steps_per_second": 5.58, "step": 97000 }, { "epoch": 0.14, "learning_rate": 4.3936875e-05, "loss": 1.3329, "step": 97010 }, { "epoch": 0.14, "learning_rate": 4.393625e-05, "loss": 1.4297, "step": 97020 }, { "epoch": 0.14, "learning_rate": 4.3935625e-05, "loss": 1.4196, "step": 97030 }, { "epoch": 0.14, "learning_rate": 4.3935000000000005e-05, "loss": 1.4657, "step": 97040 }, { "epoch": 0.14, "learning_rate": 4.3934375e-05, "loss": 1.5418, "step": 97050 }, { "epoch": 0.14, "learning_rate": 4.393375e-05, "loss": 1.5312, "step": 97060 }, { "epoch": 0.14, "learning_rate": 4.3933125e-05, "loss": 1.5585, "step": 97070 }, { "epoch": 0.14, "learning_rate": 4.393250000000001e-05, "loss": 1.2271, "step": 97080 }, { "epoch": 0.14, "learning_rate": 4.3931875e-05, "loss": 1.3322, "step": 97090 }, { "epoch": 0.14, "learning_rate": 4.393125e-05, "loss": 1.4709, "step": 97100 }, { "epoch": 0.14, "learning_rate": 4.3930625000000004e-05, "loss": 1.4496, "step": 97110 }, { "epoch": 0.14, "learning_rate": 4.393e-05, "loss": 1.4952, "step": 97120 }, { "epoch": 0.14, "learning_rate": 4.3929375e-05, "loss": 1.5564, "step": 97130 }, { "epoch": 0.14, "learning_rate": 4.392875e-05, "loss": 1.5511, "step": 97140 }, { "epoch": 0.14, "learning_rate": 4.3928125000000006e-05, "loss": 1.4358, "step": 97150 }, { "epoch": 0.14, "learning_rate": 4.39275e-05, "loss": 1.4407, "step": 97160 }, { "epoch": 0.14, "learning_rate": 4.3926875e-05, "loss": 1.6275, "step": 97170 }, { "epoch": 0.14, "learning_rate": 4.392625e-05, "loss": 1.2369, "step": 97180 }, { "epoch": 0.14, "learning_rate": 4.392562500000001e-05, "loss": 1.4836, "step": 97190 }, { "epoch": 0.14, "learning_rate": 4.3925e-05, "loss": 1.4849, "step": 97200 }, { "epoch": 0.14, "learning_rate": 4.3924375e-05, "loss": 1.4292, "step": 97210 }, { "epoch": 0.14, "learning_rate": 4.3923750000000004e-05, "loss": 1.3742, "step": 97220 }, { "epoch": 0.14, "learning_rate": 4.3923125e-05, "loss": 1.462, "step": 97230 }, { "epoch": 0.14, "learning_rate": 4.39225e-05, "loss": 1.3081, "step": 97240 }, { "epoch": 0.14, "learning_rate": 4.3921875e-05, "loss": 1.4913, "step": 97250 }, { "epoch": 0.14, "learning_rate": 4.3921250000000006e-05, "loss": 1.5869, "step": 97260 }, { "epoch": 0.14, "learning_rate": 4.3920625e-05, "loss": 1.4277, "step": 97270 }, { "epoch": 0.14, "learning_rate": 4.392e-05, "loss": 1.353, "step": 97280 }, { "epoch": 0.14, "learning_rate": 4.3919375e-05, "loss": 1.5365, "step": 97290 }, { "epoch": 0.14, "learning_rate": 4.391875e-05, "loss": 1.5113, "step": 97300 }, { "epoch": 0.14, "learning_rate": 4.3918125000000004e-05, "loss": 1.3835, "step": 97310 }, { "epoch": 0.14, "learning_rate": 4.39175e-05, "loss": 1.6042, "step": 97320 }, { "epoch": 0.14, "learning_rate": 4.3916875000000005e-05, "loss": 1.5915, "step": 97330 }, { "epoch": 0.14, "learning_rate": 4.391625e-05, "loss": 1.3977, "step": 97340 }, { "epoch": 0.14, "learning_rate": 4.3915625000000005e-05, "loss": 1.4166, "step": 97350 }, { "epoch": 0.14, "learning_rate": 4.3915e-05, "loss": 1.3005, "step": 97360 }, { "epoch": 0.14, "learning_rate": 4.3914375e-05, "loss": 1.512, "step": 97370 }, { "epoch": 0.14, "learning_rate": 4.391375e-05, "loss": 1.3625, "step": 97380 }, { "epoch": 0.14, "learning_rate": 4.3913125e-05, "loss": 1.5091, "step": 97390 }, { "epoch": 0.14, "learning_rate": 4.3912500000000004e-05, "loss": 1.1919, "step": 97400 }, { "epoch": 0.14, "learning_rate": 4.3911875e-05, "loss": 1.494, "step": 97410 }, { "epoch": 0.14, "learning_rate": 4.3911250000000004e-05, "loss": 1.7644, "step": 97420 }, { "epoch": 0.14, "learning_rate": 4.3910625e-05, "loss": 1.4366, "step": 97430 }, { "epoch": 0.14, "learning_rate": 4.391e-05, "loss": 1.4097, "step": 97440 }, { "epoch": 0.14, "learning_rate": 4.3909375e-05, "loss": 1.5102, "step": 97450 }, { "epoch": 0.14, "learning_rate": 4.3908750000000005e-05, "loss": 1.4891, "step": 97460 }, { "epoch": 0.14, "learning_rate": 4.3908125e-05, "loss": 1.3623, "step": 97470 }, { "epoch": 0.14, "learning_rate": 4.39075e-05, "loss": 1.2787, "step": 97480 }, { "epoch": 0.14, "learning_rate": 4.3906875e-05, "loss": 1.377, "step": 97490 }, { "epoch": 0.14, "learning_rate": 4.390625000000001e-05, "loss": 1.5319, "step": 97500 }, { "epoch": 0.14, "learning_rate": 4.3905625e-05, "loss": 1.5688, "step": 97510 }, { "epoch": 0.14, "learning_rate": 4.3905e-05, "loss": 1.3685, "step": 97520 }, { "epoch": 0.14, "learning_rate": 4.3904375000000004e-05, "loss": 1.4684, "step": 97530 }, { "epoch": 0.14, "learning_rate": 4.390375e-05, "loss": 1.3997, "step": 97540 }, { "epoch": 0.14, "learning_rate": 4.3903125e-05, "loss": 1.3961, "step": 97550 }, { "epoch": 0.14, "learning_rate": 4.39025e-05, "loss": 1.5587, "step": 97560 }, { "epoch": 0.14, "learning_rate": 4.3901875000000006e-05, "loss": 1.236, "step": 97570 }, { "epoch": 0.14, "learning_rate": 4.390125e-05, "loss": 1.4191, "step": 97580 }, { "epoch": 0.14, "learning_rate": 4.3900625e-05, "loss": 1.6038, "step": 97590 }, { "epoch": 0.14, "learning_rate": 4.39e-05, "loss": 1.233, "step": 97600 }, { "epoch": 0.14, "learning_rate": 4.389937500000001e-05, "loss": 1.3017, "step": 97610 }, { "epoch": 0.14, "learning_rate": 4.389875e-05, "loss": 1.4006, "step": 97620 }, { "epoch": 0.14, "learning_rate": 4.3898125e-05, "loss": 1.32, "step": 97630 }, { "epoch": 0.14, "learning_rate": 4.3897500000000004e-05, "loss": 1.2751, "step": 97640 }, { "epoch": 0.14, "learning_rate": 4.3896875e-05, "loss": 1.5087, "step": 97650 }, { "epoch": 0.14, "learning_rate": 4.389625e-05, "loss": 1.488, "step": 97660 }, { "epoch": 0.14, "learning_rate": 4.3895625e-05, "loss": 1.2719, "step": 97670 }, { "epoch": 0.14, "learning_rate": 4.3895000000000006e-05, "loss": 1.5142, "step": 97680 }, { "epoch": 0.14, "learning_rate": 4.3894375e-05, "loss": 1.2836, "step": 97690 }, { "epoch": 0.14, "learning_rate": 4.389375e-05, "loss": 1.6781, "step": 97700 }, { "epoch": 0.14, "learning_rate": 4.3893125e-05, "loss": 1.371, "step": 97710 }, { "epoch": 0.14, "learning_rate": 4.38925e-05, "loss": 1.4092, "step": 97720 }, { "epoch": 0.14, "learning_rate": 4.3891875000000004e-05, "loss": 1.3025, "step": 97730 }, { "epoch": 0.14, "learning_rate": 4.389125e-05, "loss": 1.4083, "step": 97740 }, { "epoch": 0.14, "learning_rate": 4.3890625000000005e-05, "loss": 1.2855, "step": 97750 }, { "epoch": 0.14, "learning_rate": 4.389e-05, "loss": 1.5428, "step": 97760 }, { "epoch": 0.14, "learning_rate": 4.3889375000000005e-05, "loss": 1.3059, "step": 97770 }, { "epoch": 0.14, "learning_rate": 4.388875e-05, "loss": 1.4892, "step": 97780 }, { "epoch": 0.14, "learning_rate": 4.3888125e-05, "loss": 1.2747, "step": 97790 }, { "epoch": 0.14, "learning_rate": 4.38875e-05, "loss": 1.4191, "step": 97800 }, { "epoch": 0.14, "learning_rate": 4.3886875e-05, "loss": 1.4822, "step": 97810 }, { "epoch": 0.14, "learning_rate": 4.3886250000000004e-05, "loss": 1.2137, "step": 97820 }, { "epoch": 0.14, "learning_rate": 4.3885625e-05, "loss": 1.4142, "step": 97830 }, { "epoch": 0.14, "learning_rate": 4.3885000000000004e-05, "loss": 1.501, "step": 97840 }, { "epoch": 0.14, "learning_rate": 4.3884375e-05, "loss": 1.4527, "step": 97850 }, { "epoch": 0.14, "learning_rate": 4.388375e-05, "loss": 1.4963, "step": 97860 }, { "epoch": 0.14, "learning_rate": 4.3883125e-05, "loss": 1.4935, "step": 97870 }, { "epoch": 0.14, "learning_rate": 4.3882500000000005e-05, "loss": 1.6229, "step": 97880 }, { "epoch": 0.14, "learning_rate": 4.3881875e-05, "loss": 1.3544, "step": 97890 }, { "epoch": 0.14, "learning_rate": 4.388125e-05, "loss": 1.6249, "step": 97900 }, { "epoch": 0.14, "learning_rate": 4.3880625e-05, "loss": 1.2368, "step": 97910 }, { "epoch": 0.14, "learning_rate": 4.388000000000001e-05, "loss": 1.5224, "step": 97920 }, { "epoch": 0.14, "learning_rate": 4.3879375000000004e-05, "loss": 1.5016, "step": 97930 }, { "epoch": 0.14, "learning_rate": 4.387875e-05, "loss": 1.4088, "step": 97940 }, { "epoch": 0.14, "learning_rate": 4.3878125000000004e-05, "loss": 1.4054, "step": 97950 }, { "epoch": 0.14, "learning_rate": 4.387750000000001e-05, "loss": 1.2834, "step": 97960 }, { "epoch": 0.14, "learning_rate": 4.3876875e-05, "loss": 1.5453, "step": 97970 }, { "epoch": 0.14, "learning_rate": 4.387625e-05, "loss": 1.3334, "step": 97980 }, { "epoch": 0.14, "learning_rate": 4.3875625000000006e-05, "loss": 1.3617, "step": 97990 }, { "epoch": 0.14, "learning_rate": 4.3875e-05, "loss": 1.3319, "step": 98000 }, { "epoch": 0.14, "eval_loss": 1.3564341068267822, "eval_runtime": 1621.2293, "eval_samples_per_second": 35.798, "eval_steps_per_second": 4.475, "step": 98000 }, { "epoch": 0.14, "learning_rate": 4.3874375e-05, "loss": 1.6211, "step": 98010 }, { "epoch": 0.14, "learning_rate": 4.387375e-05, "loss": 1.3628, "step": 98020 }, { "epoch": 0.14, "learning_rate": 4.387312500000001e-05, "loss": 1.1819, "step": 98030 }, { "epoch": 0.14, "learning_rate": 4.38725e-05, "loss": 1.3509, "step": 98040 }, { "epoch": 0.14, "learning_rate": 4.3871875e-05, "loss": 1.3985, "step": 98050 }, { "epoch": 0.14, "learning_rate": 4.3871250000000005e-05, "loss": 1.4005, "step": 98060 }, { "epoch": 0.14, "learning_rate": 4.3870625e-05, "loss": 1.3921, "step": 98070 }, { "epoch": 0.14, "learning_rate": 4.387e-05, "loss": 1.3322, "step": 98080 }, { "epoch": 0.14, "learning_rate": 4.3869375e-05, "loss": 1.4114, "step": 98090 }, { "epoch": 0.14, "learning_rate": 4.3868750000000006e-05, "loss": 1.4863, "step": 98100 }, { "epoch": 0.14, "learning_rate": 4.3868125e-05, "loss": 1.4904, "step": 98110 }, { "epoch": 0.14, "learning_rate": 4.38675e-05, "loss": 1.2611, "step": 98120 }, { "epoch": 0.14, "learning_rate": 4.3866875000000003e-05, "loss": 1.3707, "step": 98130 }, { "epoch": 0.14, "learning_rate": 4.386625e-05, "loss": 1.6856, "step": 98140 }, { "epoch": 0.14, "learning_rate": 4.3865625000000004e-05, "loss": 1.5951, "step": 98150 }, { "epoch": 0.14, "learning_rate": 4.3865e-05, "loss": 1.2664, "step": 98160 }, { "epoch": 0.14, "learning_rate": 4.3864375000000005e-05, "loss": 1.2109, "step": 98170 }, { "epoch": 0.14, "learning_rate": 4.386375e-05, "loss": 1.3414, "step": 98180 }, { "epoch": 0.14, "learning_rate": 4.3863125e-05, "loss": 1.5331, "step": 98190 }, { "epoch": 0.14, "learning_rate": 4.38625e-05, "loss": 1.4396, "step": 98200 }, { "epoch": 0.14, "learning_rate": 4.3861875e-05, "loss": 1.8079, "step": 98210 }, { "epoch": 0.14, "learning_rate": 4.386125e-05, "loss": 1.2669, "step": 98220 }, { "epoch": 0.14, "learning_rate": 4.3860625e-05, "loss": 1.5213, "step": 98230 }, { "epoch": 0.14, "learning_rate": 4.3860000000000004e-05, "loss": 1.4468, "step": 98240 }, { "epoch": 0.14, "learning_rate": 4.3859375e-05, "loss": 1.6185, "step": 98250 }, { "epoch": 0.14, "learning_rate": 4.3858750000000004e-05, "loss": 1.5542, "step": 98260 }, { "epoch": 0.14, "learning_rate": 4.3858125e-05, "loss": 1.3051, "step": 98270 }, { "epoch": 0.14, "learning_rate": 4.38575e-05, "loss": 1.4367, "step": 98280 }, { "epoch": 0.14, "learning_rate": 4.3856875e-05, "loss": 1.5208, "step": 98290 }, { "epoch": 0.14, "learning_rate": 4.3856250000000006e-05, "loss": 1.5143, "step": 98300 }, { "epoch": 0.14, "learning_rate": 4.3855625e-05, "loss": 1.3462, "step": 98310 }, { "epoch": 0.14, "learning_rate": 4.3855e-05, "loss": 1.3269, "step": 98320 }, { "epoch": 0.14, "learning_rate": 4.3854375e-05, "loss": 1.4949, "step": 98330 }, { "epoch": 0.14, "learning_rate": 4.385375000000001e-05, "loss": 1.3172, "step": 98340 }, { "epoch": 0.14, "learning_rate": 4.3853125000000004e-05, "loss": 1.202, "step": 98350 }, { "epoch": 0.14, "learning_rate": 4.38525e-05, "loss": 1.6322, "step": 98360 }, { "epoch": 0.14, "learning_rate": 4.3851875000000004e-05, "loss": 1.5441, "step": 98370 }, { "epoch": 0.14, "learning_rate": 4.385125000000001e-05, "loss": 1.2568, "step": 98380 }, { "epoch": 0.14, "learning_rate": 4.3850625e-05, "loss": 1.3598, "step": 98390 }, { "epoch": 0.14, "learning_rate": 4.385e-05, "loss": 1.6014, "step": 98400 }, { "epoch": 0.14, "learning_rate": 4.3849375000000006e-05, "loss": 1.6186, "step": 98410 }, { "epoch": 0.14, "learning_rate": 4.384875e-05, "loss": 1.633, "step": 98420 }, { "epoch": 0.14, "learning_rate": 4.3848125e-05, "loss": 1.2508, "step": 98430 }, { "epoch": 0.14, "learning_rate": 4.38475e-05, "loss": 1.3597, "step": 98440 }, { "epoch": 0.14, "learning_rate": 4.384687500000001e-05, "loss": 1.5442, "step": 98450 }, { "epoch": 0.14, "learning_rate": 4.384625e-05, "loss": 1.4992, "step": 98460 }, { "epoch": 0.14, "learning_rate": 4.3845625e-05, "loss": 1.3477, "step": 98470 }, { "epoch": 0.14, "learning_rate": 4.3845000000000005e-05, "loss": 1.4499, "step": 98480 }, { "epoch": 0.14, "learning_rate": 4.3844375e-05, "loss": 1.43, "step": 98490 }, { "epoch": 0.14, "learning_rate": 4.384375e-05, "loss": 1.5428, "step": 98500 }, { "epoch": 0.14, "learning_rate": 4.3843125e-05, "loss": 1.2477, "step": 98510 }, { "epoch": 0.14, "learning_rate": 4.3842500000000006e-05, "loss": 1.4718, "step": 98520 }, { "epoch": 0.14, "learning_rate": 4.3841875e-05, "loss": 1.4011, "step": 98530 }, { "epoch": 0.14, "learning_rate": 4.384125e-05, "loss": 1.5802, "step": 98540 }, { "epoch": 0.14, "learning_rate": 4.3840625000000003e-05, "loss": 1.7305, "step": 98550 }, { "epoch": 0.14, "learning_rate": 4.384e-05, "loss": 1.3119, "step": 98560 }, { "epoch": 0.14, "learning_rate": 4.3839375000000004e-05, "loss": 1.6239, "step": 98570 }, { "epoch": 0.14, "learning_rate": 4.383875e-05, "loss": 1.5134, "step": 98580 }, { "epoch": 0.14, "learning_rate": 4.3838125000000005e-05, "loss": 1.3987, "step": 98590 }, { "epoch": 0.14, "learning_rate": 4.38375e-05, "loss": 1.4476, "step": 98600 }, { "epoch": 0.14, "learning_rate": 4.3836875e-05, "loss": 1.4066, "step": 98610 }, { "epoch": 0.14, "learning_rate": 4.383625e-05, "loss": 1.2906, "step": 98620 }, { "epoch": 0.14, "learning_rate": 4.3835625e-05, "loss": 1.4863, "step": 98630 }, { "epoch": 0.14, "learning_rate": 4.3835e-05, "loss": 1.4435, "step": 98640 }, { "epoch": 0.14, "learning_rate": 4.3834375e-05, "loss": 1.4935, "step": 98650 }, { "epoch": 0.14, "learning_rate": 4.3833750000000004e-05, "loss": 1.6458, "step": 98660 }, { "epoch": 0.14, "learning_rate": 4.3833125e-05, "loss": 1.5979, "step": 98670 }, { "epoch": 0.14, "learning_rate": 4.3832500000000004e-05, "loss": 1.4878, "step": 98680 }, { "epoch": 0.14, "learning_rate": 4.3831875e-05, "loss": 1.3795, "step": 98690 }, { "epoch": 0.14, "learning_rate": 4.383125e-05, "loss": 1.1427, "step": 98700 }, { "epoch": 0.14, "learning_rate": 4.3830625e-05, "loss": 1.2598, "step": 98710 }, { "epoch": 0.14, "learning_rate": 4.3830000000000006e-05, "loss": 1.3041, "step": 98720 }, { "epoch": 0.14, "learning_rate": 4.3829375e-05, "loss": 1.3726, "step": 98730 }, { "epoch": 0.14, "learning_rate": 4.382875e-05, "loss": 1.418, "step": 98740 }, { "epoch": 0.14, "learning_rate": 4.3828125e-05, "loss": 1.4607, "step": 98750 }, { "epoch": 0.14, "learning_rate": 4.382750000000001e-05, "loss": 1.3643, "step": 98760 }, { "epoch": 0.14, "learning_rate": 4.3826875000000004e-05, "loss": 1.3872, "step": 98770 }, { "epoch": 0.14, "learning_rate": 4.382625e-05, "loss": 1.3497, "step": 98780 }, { "epoch": 0.14, "learning_rate": 4.3825625000000004e-05, "loss": 1.4096, "step": 98790 }, { "epoch": 0.14, "learning_rate": 4.3825e-05, "loss": 1.4184, "step": 98800 }, { "epoch": 0.14, "learning_rate": 4.3824375e-05, "loss": 1.5794, "step": 98810 }, { "epoch": 0.14, "learning_rate": 4.382375e-05, "loss": 1.2194, "step": 98820 }, { "epoch": 0.14, "learning_rate": 4.3823125000000006e-05, "loss": 1.3339, "step": 98830 }, { "epoch": 0.14, "learning_rate": 4.38225e-05, "loss": 1.4253, "step": 98840 }, { "epoch": 0.14, "learning_rate": 4.3821875e-05, "loss": 1.5361, "step": 98850 }, { "epoch": 0.14, "learning_rate": 4.382125e-05, "loss": 1.3634, "step": 98860 }, { "epoch": 0.14, "learning_rate": 4.382062500000001e-05, "loss": 1.2555, "step": 98870 }, { "epoch": 0.14, "learning_rate": 4.382e-05, "loss": 1.3652, "step": 98880 }, { "epoch": 0.14, "learning_rate": 4.3819375e-05, "loss": 1.568, "step": 98890 }, { "epoch": 0.14, "learning_rate": 4.3818750000000005e-05, "loss": 1.2885, "step": 98900 }, { "epoch": 0.14, "learning_rate": 4.3818125e-05, "loss": 1.4942, "step": 98910 }, { "epoch": 0.14, "learning_rate": 4.38175e-05, "loss": 1.2961, "step": 98920 }, { "epoch": 0.14, "learning_rate": 4.3816875e-05, "loss": 1.6207, "step": 98930 }, { "epoch": 0.14, "learning_rate": 4.3816250000000006e-05, "loss": 1.5693, "step": 98940 }, { "epoch": 0.14, "learning_rate": 4.3815625e-05, "loss": 1.4153, "step": 98950 }, { "epoch": 0.14, "learning_rate": 4.3815e-05, "loss": 1.4892, "step": 98960 }, { "epoch": 0.14, "learning_rate": 4.3814375000000003e-05, "loss": 1.42, "step": 98970 }, { "epoch": 0.14, "learning_rate": 4.381375e-05, "loss": 1.2888, "step": 98980 }, { "epoch": 0.14, "learning_rate": 4.3813125000000004e-05, "loss": 1.5041, "step": 98990 }, { "epoch": 0.14, "learning_rate": 4.38125e-05, "loss": 1.3808, "step": 99000 }, { "epoch": 0.14, "eval_loss": 1.3637181520462036, "eval_runtime": 1592.1748, "eval_samples_per_second": 36.451, "eval_steps_per_second": 4.557, "step": 99000 }, { "epoch": 0.14, "learning_rate": 4.3811875000000005e-05, "loss": 1.2578, "step": 99010 }, { "epoch": 0.14, "learning_rate": 4.381125e-05, "loss": 1.5427, "step": 99020 }, { "epoch": 0.14, "learning_rate": 4.3810625e-05, "loss": 1.5357, "step": 99030 }, { "epoch": 0.14, "learning_rate": 4.381e-05, "loss": 1.4738, "step": 99040 }, { "epoch": 0.14, "learning_rate": 4.3809375e-05, "loss": 1.4264, "step": 99050 }, { "epoch": 0.14, "learning_rate": 4.380875e-05, "loss": 1.337, "step": 99060 }, { "epoch": 0.14, "learning_rate": 4.3808125e-05, "loss": 1.5118, "step": 99070 }, { "epoch": 0.14, "learning_rate": 4.3807500000000004e-05, "loss": 1.5669, "step": 99080 }, { "epoch": 0.14, "learning_rate": 4.3806875e-05, "loss": 1.5749, "step": 99090 }, { "epoch": 0.14, "learning_rate": 4.3806250000000004e-05, "loss": 1.3627, "step": 99100 }, { "epoch": 0.14, "learning_rate": 4.3805625e-05, "loss": 1.4672, "step": 99110 }, { "epoch": 0.14, "learning_rate": 4.3805000000000005e-05, "loss": 1.581, "step": 99120 }, { "epoch": 0.14, "learning_rate": 4.3804375e-05, "loss": 1.4735, "step": 99130 }, { "epoch": 0.14, "learning_rate": 4.3803750000000006e-05, "loss": 1.2917, "step": 99140 }, { "epoch": 0.14, "learning_rate": 4.3803125e-05, "loss": 1.2608, "step": 99150 }, { "epoch": 0.14, "learning_rate": 4.38025e-05, "loss": 1.6439, "step": 99160 }, { "epoch": 0.14, "learning_rate": 4.3801875e-05, "loss": 1.3784, "step": 99170 }, { "epoch": 0.14, "learning_rate": 4.380125000000001e-05, "loss": 1.58, "step": 99180 }, { "epoch": 0.14, "learning_rate": 4.3800625000000004e-05, "loss": 1.4018, "step": 99190 }, { "epoch": 0.14, "learning_rate": 4.38e-05, "loss": 1.2804, "step": 99200 }, { "epoch": 0.14, "learning_rate": 4.3799375000000004e-05, "loss": 1.469, "step": 99210 }, { "epoch": 0.14, "learning_rate": 4.379875e-05, "loss": 1.5091, "step": 99220 }, { "epoch": 0.14, "learning_rate": 4.3798125e-05, "loss": 1.3578, "step": 99230 }, { "epoch": 0.14, "learning_rate": 4.37975e-05, "loss": 1.481, "step": 99240 }, { "epoch": 0.14, "learning_rate": 4.3796875000000006e-05, "loss": 1.5206, "step": 99250 }, { "epoch": 0.14, "learning_rate": 4.379625e-05, "loss": 1.5543, "step": 99260 }, { "epoch": 0.14, "learning_rate": 4.3795625e-05, "loss": 1.5013, "step": 99270 }, { "epoch": 0.14, "learning_rate": 4.3795e-05, "loss": 1.2939, "step": 99280 }, { "epoch": 0.14, "learning_rate": 4.379437500000001e-05, "loss": 1.3803, "step": 99290 }, { "epoch": 0.14, "learning_rate": 4.379375e-05, "loss": 1.5605, "step": 99300 }, { "epoch": 0.14, "learning_rate": 4.3793125e-05, "loss": 1.4695, "step": 99310 }, { "epoch": 0.14, "learning_rate": 4.3792500000000005e-05, "loss": 1.7419, "step": 99320 }, { "epoch": 0.14, "learning_rate": 4.3791875e-05, "loss": 1.4605, "step": 99330 }, { "epoch": 0.14, "learning_rate": 4.379125e-05, "loss": 1.4443, "step": 99340 }, { "epoch": 0.14, "learning_rate": 4.3790625e-05, "loss": 1.4939, "step": 99350 }, { "epoch": 0.14, "learning_rate": 4.3790000000000006e-05, "loss": 1.4275, "step": 99360 }, { "epoch": 0.14, "learning_rate": 4.3789375e-05, "loss": 1.4459, "step": 99370 }, { "epoch": 0.14, "learning_rate": 4.378875e-05, "loss": 1.523, "step": 99380 }, { "epoch": 0.14, "learning_rate": 4.3788125000000004e-05, "loss": 1.5843, "step": 99390 }, { "epoch": 0.14, "learning_rate": 4.37875e-05, "loss": 1.4307, "step": 99400 }, { "epoch": 0.14, "learning_rate": 4.3786875e-05, "loss": 1.5393, "step": 99410 }, { "epoch": 0.14, "learning_rate": 4.378625e-05, "loss": 1.1878, "step": 99420 }, { "epoch": 0.14, "learning_rate": 4.3785625000000005e-05, "loss": 1.3942, "step": 99430 }, { "epoch": 0.14, "learning_rate": 4.3785e-05, "loss": 1.4439, "step": 99440 }, { "epoch": 0.14, "learning_rate": 4.3784375e-05, "loss": 1.5034, "step": 99450 }, { "epoch": 0.14, "learning_rate": 4.378375e-05, "loss": 1.5641, "step": 99460 }, { "epoch": 0.14, "learning_rate": 4.3783125e-05, "loss": 1.4685, "step": 99470 }, { "epoch": 0.14, "learning_rate": 4.37825e-05, "loss": 1.489, "step": 99480 }, { "epoch": 0.14, "learning_rate": 4.3781875e-05, "loss": 1.244, "step": 99490 }, { "epoch": 0.14, "learning_rate": 4.3781250000000004e-05, "loss": 1.5348, "step": 99500 }, { "epoch": 0.14, "learning_rate": 4.3780625e-05, "loss": 1.5083, "step": 99510 }, { "epoch": 0.14, "learning_rate": 4.3780000000000004e-05, "loss": 1.5247, "step": 99520 }, { "epoch": 0.14, "learning_rate": 4.3779375e-05, "loss": 1.3993, "step": 99530 }, { "epoch": 0.14, "learning_rate": 4.3778750000000005e-05, "loss": 1.4768, "step": 99540 }, { "epoch": 0.14, "learning_rate": 4.3778125e-05, "loss": 1.334, "step": 99550 }, { "epoch": 0.14, "learning_rate": 4.3777500000000006e-05, "loss": 1.3544, "step": 99560 }, { "epoch": 0.14, "learning_rate": 4.3776875e-05, "loss": 1.4939, "step": 99570 }, { "epoch": 0.14, "learning_rate": 4.377625e-05, "loss": 1.4184, "step": 99580 }, { "epoch": 0.14, "learning_rate": 4.3775625e-05, "loss": 1.4975, "step": 99590 }, { "epoch": 0.14, "learning_rate": 4.3775e-05, "loss": 1.4329, "step": 99600 }, { "epoch": 0.14, "learning_rate": 4.3774375000000004e-05, "loss": 1.0794, "step": 99610 }, { "epoch": 0.14, "learning_rate": 4.377375e-05, "loss": 1.5545, "step": 99620 }, { "epoch": 0.14, "learning_rate": 4.3773125000000005e-05, "loss": 1.4619, "step": 99630 }, { "epoch": 0.14, "learning_rate": 4.37725e-05, "loss": 1.2533, "step": 99640 }, { "epoch": 0.14, "learning_rate": 4.3771875e-05, "loss": 1.6456, "step": 99650 }, { "epoch": 0.14, "learning_rate": 4.377125e-05, "loss": 1.4262, "step": 99660 }, { "epoch": 0.14, "learning_rate": 4.3770625000000006e-05, "loss": 1.5021, "step": 99670 }, { "epoch": 0.14, "learning_rate": 4.377e-05, "loss": 1.4341, "step": 99680 }, { "epoch": 0.14, "learning_rate": 4.3769375e-05, "loss": 1.4117, "step": 99690 }, { "epoch": 0.14, "learning_rate": 4.3768750000000003e-05, "loss": 1.4263, "step": 99700 }, { "epoch": 0.14, "learning_rate": 4.376812500000001e-05, "loss": 1.3181, "step": 99710 }, { "epoch": 0.14, "learning_rate": 4.37675e-05, "loss": 1.3074, "step": 99720 }, { "epoch": 0.14, "learning_rate": 4.3766875e-05, "loss": 1.6021, "step": 99730 }, { "epoch": 0.14, "learning_rate": 4.3766250000000005e-05, "loss": 1.3924, "step": 99740 }, { "epoch": 0.14, "learning_rate": 4.3765625e-05, "loss": 1.6198, "step": 99750 }, { "epoch": 0.14, "learning_rate": 4.3765e-05, "loss": 1.4835, "step": 99760 }, { "epoch": 0.14, "learning_rate": 4.3764375e-05, "loss": 1.5179, "step": 99770 }, { "epoch": 0.14, "learning_rate": 4.3763750000000006e-05, "loss": 1.4496, "step": 99780 }, { "epoch": 0.14, "learning_rate": 4.3763125e-05, "loss": 1.4249, "step": 99790 }, { "epoch": 0.14, "learning_rate": 4.37625e-05, "loss": 1.3703, "step": 99800 }, { "epoch": 0.14, "learning_rate": 4.3761875000000004e-05, "loss": 1.5277, "step": 99810 }, { "epoch": 0.14, "learning_rate": 4.376125e-05, "loss": 1.3995, "step": 99820 }, { "epoch": 0.14, "learning_rate": 4.3760625e-05, "loss": 1.2827, "step": 99830 }, { "epoch": 0.14, "learning_rate": 4.376e-05, "loss": 1.7796, "step": 99840 }, { "epoch": 0.14, "learning_rate": 4.3759375000000005e-05, "loss": 1.3804, "step": 99850 }, { "epoch": 0.14, "learning_rate": 4.375875e-05, "loss": 1.5744, "step": 99860 }, { "epoch": 0.14, "learning_rate": 4.3758125e-05, "loss": 1.5205, "step": 99870 }, { "epoch": 0.14, "learning_rate": 4.37575e-05, "loss": 1.781, "step": 99880 }, { "epoch": 0.14, "learning_rate": 4.3756875e-05, "loss": 1.2722, "step": 99890 }, { "epoch": 0.14, "learning_rate": 4.375625e-05, "loss": 1.4972, "step": 99900 }, { "epoch": 0.14, "learning_rate": 4.3755625e-05, "loss": 1.4029, "step": 99910 }, { "epoch": 0.14, "learning_rate": 4.3755000000000004e-05, "loss": 1.3669, "step": 99920 }, { "epoch": 0.14, "learning_rate": 4.3754375e-05, "loss": 1.4457, "step": 99930 }, { "epoch": 0.14, "learning_rate": 4.3753750000000004e-05, "loss": 1.3882, "step": 99940 }, { "epoch": 0.14, "learning_rate": 4.3753125e-05, "loss": 1.6418, "step": 99950 }, { "epoch": 0.14, "learning_rate": 4.3752500000000005e-05, "loss": 1.6337, "step": 99960 }, { "epoch": 0.14, "learning_rate": 4.3751875e-05, "loss": 1.4789, "step": 99970 }, { "epoch": 0.14, "learning_rate": 4.3751250000000006e-05, "loss": 1.4218, "step": 99980 }, { "epoch": 0.14, "learning_rate": 4.3750625e-05, "loss": 1.4138, "step": 99990 }, { "epoch": 0.14, "learning_rate": 4.375e-05, "loss": 1.6022, "step": 100000 }, { "epoch": 0.14, "eval_loss": 1.3524357080459595, "eval_runtime": 1293.3508, "eval_samples_per_second": 44.873, "eval_steps_per_second": 5.609, "step": 100000 } ], "max_steps": 800000, "num_train_epochs": 2, "total_flos": 2.1056385024e+17, "trial_name": null, "trial_params": null }