{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9999180932099271, "global_step": 6104, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.8e-07, "loss": 1.4588, "step": 10 }, { "epoch": 0.0, "learning_rate": 3.8e-07, "loss": 1.5163, "step": 20 }, { "epoch": 0.0, "learning_rate": 5.800000000000001e-07, "loss": 1.5383, "step": 30 }, { "epoch": 0.01, "learning_rate": 7.8e-07, "loss": 1.4794, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.800000000000001e-07, "loss": 1.4675, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.1800000000000001e-06, "loss": 1.392, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.3800000000000001e-06, "loss": 1.4392, "step": 70 }, { "epoch": 0.01, "learning_rate": 1.5800000000000001e-06, "loss": 1.3785, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.7800000000000001e-06, "loss": 1.3863, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.98e-06, "loss": 1.4214, "step": 100 }, { "epoch": 0.02, "learning_rate": 2.1800000000000003e-06, "loss": 1.4741, "step": 110 }, { "epoch": 0.02, "learning_rate": 2.38e-06, "loss": 1.3113, "step": 120 }, { "epoch": 0.02, "learning_rate": 2.5800000000000003e-06, "loss": 1.3297, "step": 130 }, { "epoch": 0.02, "learning_rate": 2.7800000000000005e-06, "loss": 1.3179, "step": 140 }, { "epoch": 0.02, "learning_rate": 2.9800000000000003e-06, "loss": 1.3273, "step": 150 }, { "epoch": 0.03, "learning_rate": 3.1800000000000005e-06, "loss": 1.3637, "step": 160 }, { "epoch": 0.03, "learning_rate": 3.3800000000000007e-06, "loss": 1.4357, "step": 170 }, { "epoch": 0.03, "learning_rate": 3.58e-06, "loss": 1.3675, "step": 180 }, { "epoch": 0.03, "learning_rate": 3.7800000000000002e-06, "loss": 1.3398, "step": 190 }, { "epoch": 0.03, "learning_rate": 3.980000000000001e-06, "loss": 1.3694, "step": 200 }, { "epoch": 0.03, "learning_rate": 4.18e-06, "loss": 1.3456, "step": 210 }, { "epoch": 0.04, "learning_rate": 4.38e-06, "loss": 1.3203, "step": 220 }, { "epoch": 0.04, "learning_rate": 4.58e-06, "loss": 1.3678, "step": 230 }, { "epoch": 0.04, "learning_rate": 4.78e-06, "loss": 1.3973, "step": 240 }, { "epoch": 0.04, "learning_rate": 4.980000000000001e-06, "loss": 1.3017, "step": 250 }, { "epoch": 0.04, "learning_rate": 5.18e-06, "loss": 1.2614, "step": 260 }, { "epoch": 0.04, "learning_rate": 5.380000000000001e-06, "loss": 1.3756, "step": 270 }, { "epoch": 0.05, "learning_rate": 5.580000000000001e-06, "loss": 1.4691, "step": 280 }, { "epoch": 0.05, "learning_rate": 5.78e-06, "loss": 1.2958, "step": 290 }, { "epoch": 0.05, "learning_rate": 5.98e-06, "loss": 1.3148, "step": 300 }, { "epoch": 0.05, "learning_rate": 6.18e-06, "loss": 1.418, "step": 310 }, { "epoch": 0.05, "learning_rate": 6.380000000000001e-06, "loss": 1.3144, "step": 320 }, { "epoch": 0.05, "learning_rate": 6.5800000000000005e-06, "loss": 1.3724, "step": 330 }, { "epoch": 0.06, "learning_rate": 6.780000000000001e-06, "loss": 1.363, "step": 340 }, { "epoch": 0.06, "learning_rate": 6.98e-06, "loss": 1.2373, "step": 350 }, { "epoch": 0.06, "learning_rate": 7.180000000000001e-06, "loss": 1.3572, "step": 360 }, { "epoch": 0.06, "learning_rate": 7.3800000000000005e-06, "loss": 1.3494, "step": 370 }, { "epoch": 0.06, "learning_rate": 7.58e-06, "loss": 1.3055, "step": 380 }, { "epoch": 0.06, "learning_rate": 7.78e-06, "loss": 1.3626, "step": 390 }, { "epoch": 0.07, "learning_rate": 7.980000000000002e-06, "loss": 1.3851, "step": 400 }, { "epoch": 0.07, "learning_rate": 8.18e-06, "loss": 1.2608, "step": 410 }, { "epoch": 0.07, "learning_rate": 8.380000000000001e-06, "loss": 1.3282, "step": 420 }, { "epoch": 0.07, "learning_rate": 8.580000000000001e-06, "loss": 1.3424, "step": 430 }, { "epoch": 0.07, "learning_rate": 8.78e-06, "loss": 1.3204, "step": 440 }, { "epoch": 0.07, "learning_rate": 8.98e-06, "loss": 1.3753, "step": 450 }, { "epoch": 0.08, "learning_rate": 9.180000000000002e-06, "loss": 1.2741, "step": 460 }, { "epoch": 0.08, "learning_rate": 9.38e-06, "loss": 1.4024, "step": 470 }, { "epoch": 0.08, "learning_rate": 9.58e-06, "loss": 1.2724, "step": 480 }, { "epoch": 0.08, "learning_rate": 9.780000000000001e-06, "loss": 1.4165, "step": 490 }, { "epoch": 0.08, "learning_rate": 9.980000000000001e-06, "loss": 1.3768, "step": 500 }, { "epoch": 0.08, "learning_rate": 1e-05, "loss": 1.3327, "step": 510 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.3627, "step": 520 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.3674, "step": 530 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.4293, "step": 540 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.2829, "step": 550 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.4258, "step": 560 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 1.3552, "step": 570 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.4267, "step": 580 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.3482, "step": 590 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.3768, "step": 600 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.4167, "step": 610 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.3562, "step": 620 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.3819, "step": 630 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 1.3362, "step": 640 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.3477, "step": 650 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.3238, "step": 660 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.3467, "step": 670 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.3679, "step": 680 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.325, "step": 690 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 1.2288, "step": 700 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.2394, "step": 710 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.2945, "step": 720 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.2176, "step": 730 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.3438, "step": 740 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.3514, "step": 750 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 1.3976, "step": 760 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.3542, "step": 770 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.3002, "step": 780 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.2712, "step": 790 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.427, "step": 800 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.329, "step": 810 }, { "epoch": 0.13, "learning_rate": 1e-05, "loss": 1.3053, "step": 820 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.3724, "step": 830 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.2535, "step": 840 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.3213, "step": 850 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.23, "step": 860 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.2782, "step": 870 }, { "epoch": 0.14, "learning_rate": 1e-05, "loss": 1.3225, "step": 880 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.2838, "step": 890 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.3434, "step": 900 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.2737, "step": 910 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.3334, "step": 920 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.2737, "step": 930 }, { "epoch": 0.15, "learning_rate": 1e-05, "loss": 1.3841, "step": 940 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.3208, "step": 950 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.2985, "step": 960 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.3755, "step": 970 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.2616, "step": 980 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.3529, "step": 990 }, { "epoch": 0.16, "learning_rate": 1e-05, "loss": 1.3163, "step": 1000 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.2914, "step": 1010 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.4547, "step": 1020 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.2758, "step": 1030 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.3591, "step": 1040 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.3162, "step": 1050 }, { "epoch": 0.17, "learning_rate": 1e-05, "loss": 1.2589, "step": 1060 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.3123, "step": 1070 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.3089, "step": 1080 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.2332, "step": 1090 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.3301, "step": 1100 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.3039, "step": 1110 }, { "epoch": 0.18, "learning_rate": 1e-05, "loss": 1.3132, "step": 1120 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.3389, "step": 1130 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.2938, "step": 1140 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.2528, "step": 1150 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.3062, "step": 1160 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.2864, "step": 1170 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.2356, "step": 1180 }, { "epoch": 0.19, "learning_rate": 1e-05, "loss": 1.2604, "step": 1190 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.2845, "step": 1200 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.3801, "step": 1210 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.3173, "step": 1220 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.272, "step": 1230 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.3123, "step": 1240 }, { "epoch": 0.2, "learning_rate": 1e-05, "loss": 1.3632, "step": 1250 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.3333, "step": 1260 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.3122, "step": 1270 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.317, "step": 1280 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.2972, "step": 1290 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.2496, "step": 1300 }, { "epoch": 0.21, "learning_rate": 1e-05, "loss": 1.2843, "step": 1310 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.3373, "step": 1320 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.2508, "step": 1330 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.4132, "step": 1340 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.2656, "step": 1350 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.3105, "step": 1360 }, { "epoch": 0.22, "learning_rate": 1e-05, "loss": 1.3734, "step": 1370 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.3538, "step": 1380 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.2767, "step": 1390 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.4437, "step": 1400 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.2782, "step": 1410 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.2588, "step": 1420 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 1.3658, "step": 1430 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.3161, "step": 1440 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.3481, "step": 1450 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.3329, "step": 1460 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.2643, "step": 1470 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.3022, "step": 1480 }, { "epoch": 0.24, "learning_rate": 1e-05, "loss": 1.2482, "step": 1490 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.3112, "step": 1500 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.335, "step": 1510 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.2738, "step": 1520 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.3219, "step": 1530 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.3446, "step": 1540 }, { "epoch": 0.25, "learning_rate": 1e-05, "loss": 1.2233, "step": 1550 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.2723, "step": 1560 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.3748, "step": 1570 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.3433, "step": 1580 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.2581, "step": 1590 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.2938, "step": 1600 }, { "epoch": 0.26, "learning_rate": 1e-05, "loss": 1.3463, "step": 1610 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.3424, "step": 1620 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.2639, "step": 1630 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.3822, "step": 1640 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.3105, "step": 1650 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.3273, "step": 1660 }, { "epoch": 0.27, "learning_rate": 1e-05, "loss": 1.4105, "step": 1670 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.3346, "step": 1680 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.3361, "step": 1690 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.3085, "step": 1700 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.3822, "step": 1710 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.2784, "step": 1720 }, { "epoch": 0.28, "learning_rate": 1e-05, "loss": 1.3267, "step": 1730 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.2102, "step": 1740 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.2318, "step": 1750 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.2865, "step": 1760 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.4117, "step": 1770 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.2947, "step": 1780 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.3536, "step": 1790 }, { "epoch": 0.29, "learning_rate": 1e-05, "loss": 1.3179, "step": 1800 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.3403, "step": 1810 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.3984, "step": 1820 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.2817, "step": 1830 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.3239, "step": 1840 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.3182, "step": 1850 }, { "epoch": 0.3, "learning_rate": 1e-05, "loss": 1.3002, "step": 1860 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.2652, "step": 1870 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.2937, "step": 1880 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.3046, "step": 1890 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.3152, "step": 1900 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.3017, "step": 1910 }, { "epoch": 0.31, "learning_rate": 1e-05, "loss": 1.2774, "step": 1920 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.2409, "step": 1930 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.2773, "step": 1940 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.3688, "step": 1950 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.3041, "step": 1960 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.3089, "step": 1970 }, { "epoch": 0.32, "learning_rate": 1e-05, "loss": 1.2131, "step": 1980 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.3107, "step": 1990 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.2902, "step": 2000 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.2635, "step": 2010 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.3283, "step": 2020 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.37, "step": 2030 }, { "epoch": 0.33, "learning_rate": 1e-05, "loss": 1.3471, "step": 2040 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.2582, "step": 2050 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.2883, "step": 2060 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.3454, "step": 2070 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.2927, "step": 2080 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.337, "step": 2090 }, { "epoch": 0.34, "learning_rate": 1e-05, "loss": 1.2725, "step": 2100 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.3667, "step": 2110 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.2807, "step": 2120 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.283, "step": 2130 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.2922, "step": 2140 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.2685, "step": 2150 }, { "epoch": 0.35, "learning_rate": 1e-05, "loss": 1.2355, "step": 2160 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.2418, "step": 2170 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.2759, "step": 2180 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.2873, "step": 2190 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.2961, "step": 2200 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.3602, "step": 2210 }, { "epoch": 0.36, "learning_rate": 1e-05, "loss": 1.463, "step": 2220 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.2493, "step": 2230 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.2928, "step": 2240 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.2797, "step": 2250 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.2727, "step": 2260 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.3153, "step": 2270 }, { "epoch": 0.37, "learning_rate": 1e-05, "loss": 1.3328, "step": 2280 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.312, "step": 2290 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.3117, "step": 2300 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.3142, "step": 2310 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.3553, "step": 2320 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.3259, "step": 2330 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.2908, "step": 2340 }, { "epoch": 0.38, "learning_rate": 1e-05, "loss": 1.3421, "step": 2350 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.2947, "step": 2360 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.2784, "step": 2370 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.2878, "step": 2380 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.2465, "step": 2390 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.2567, "step": 2400 }, { "epoch": 0.39, "learning_rate": 1e-05, "loss": 1.3259, "step": 2410 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.3074, "step": 2420 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.2679, "step": 2430 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.2891, "step": 2440 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.2942, "step": 2450 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.309, "step": 2460 }, { "epoch": 0.4, "learning_rate": 1e-05, "loss": 1.2833, "step": 2470 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.2876, "step": 2480 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.3055, "step": 2490 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.2866, "step": 2500 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.2896, "step": 2510 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.2554, "step": 2520 }, { "epoch": 0.41, "learning_rate": 1e-05, "loss": 1.3605, "step": 2530 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.2808, "step": 2540 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.3809, "step": 2550 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.2405, "step": 2560 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.3486, "step": 2570 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.2717, "step": 2580 }, { "epoch": 0.42, "learning_rate": 1e-05, "loss": 1.2951, "step": 2590 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.3206, "step": 2600 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.2649, "step": 2610 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.3111, "step": 2620 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.3412, "step": 2630 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.3971, "step": 2640 }, { "epoch": 0.43, "learning_rate": 1e-05, "loss": 1.2913, "step": 2650 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.3284, "step": 2660 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.233, "step": 2670 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.2013, "step": 2680 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.3606, "step": 2690 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.3042, "step": 2700 }, { "epoch": 0.44, "learning_rate": 1e-05, "loss": 1.3331, "step": 2710 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.3148, "step": 2720 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.2421, "step": 2730 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.3807, "step": 2740 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.3048, "step": 2750 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.3048, "step": 2760 }, { "epoch": 0.45, "learning_rate": 1e-05, "loss": 1.3163, "step": 2770 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.2634, "step": 2780 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.36, "step": 2790 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.2809, "step": 2800 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.283, "step": 2810 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.3317, "step": 2820 }, { "epoch": 0.46, "learning_rate": 1e-05, "loss": 1.3165, "step": 2830 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.3155, "step": 2840 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.1975, "step": 2850 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.1896, "step": 2860 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.208, "step": 2870 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.2704, "step": 2880 }, { "epoch": 0.47, "learning_rate": 1e-05, "loss": 1.2951, "step": 2890 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.1931, "step": 2900 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.2505, "step": 2910 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.2654, "step": 2920 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.378, "step": 2930 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.3106, "step": 2940 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.2436, "step": 2950 }, { "epoch": 0.48, "learning_rate": 1e-05, "loss": 1.3263, "step": 2960 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.315, "step": 2970 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.3096, "step": 2980 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.3328, "step": 2990 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.273, "step": 3000 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.2463, "step": 3010 }, { "epoch": 0.49, "learning_rate": 1e-05, "loss": 1.313, "step": 3020 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.2886, "step": 3030 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.2971, "step": 3040 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.3871, "step": 3050 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.2677, "step": 3060 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.3094, "step": 3070 }, { "epoch": 0.5, "learning_rate": 1e-05, "loss": 1.2652, "step": 3080 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.2308, "step": 3090 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.3286, "step": 3100 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.2933, "step": 3110 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.2355, "step": 3120 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.3138, "step": 3130 }, { "epoch": 0.51, "learning_rate": 1e-05, "loss": 1.2825, "step": 3140 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.258, "step": 3150 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.2926, "step": 3160 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.3267, "step": 3170 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.334, "step": 3180 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.2268, "step": 3190 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.3248, "step": 3200 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.2997, "step": 3210 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.259, "step": 3220 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.3282, "step": 3230 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.2567, "step": 3240 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.3052, "step": 3250 }, { "epoch": 0.53, "learning_rate": 1e-05, "loss": 1.2964, "step": 3260 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.2399, "step": 3270 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.2876, "step": 3280 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.2853, "step": 3290 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.3239, "step": 3300 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.2694, "step": 3310 }, { "epoch": 0.54, "learning_rate": 1e-05, "loss": 1.3172, "step": 3320 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.2988, "step": 3330 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.246, "step": 3340 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.302, "step": 3350 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.2601, "step": 3360 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.3388, "step": 3370 }, { "epoch": 0.55, "learning_rate": 1e-05, "loss": 1.1941, "step": 3380 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.1759, "step": 3390 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.2403, "step": 3400 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.2024, "step": 3410 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.2752, "step": 3420 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.2273, "step": 3430 }, { "epoch": 0.56, "learning_rate": 1e-05, "loss": 1.2261, "step": 3440 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.2652, "step": 3450 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.3601, "step": 3460 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.3543, "step": 3470 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.3679, "step": 3480 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.2009, "step": 3490 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.2414, "step": 3500 }, { "epoch": 0.57, "learning_rate": 1e-05, "loss": 1.1929, "step": 3510 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.2947, "step": 3520 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.2843, "step": 3530 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.2223, "step": 3540 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.2569, "step": 3550 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.1927, "step": 3560 }, { "epoch": 0.58, "learning_rate": 1e-05, "loss": 1.3239, "step": 3570 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.3273, "step": 3580 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.2522, "step": 3590 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.3084, "step": 3600 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.2368, "step": 3610 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.2644, "step": 3620 }, { "epoch": 0.59, "learning_rate": 1e-05, "loss": 1.2636, "step": 3630 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.2676, "step": 3640 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.2844, "step": 3650 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.3151, "step": 3660 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.2261, "step": 3670 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.2124, "step": 3680 }, { "epoch": 0.6, "learning_rate": 1e-05, "loss": 1.2555, "step": 3690 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.2169, "step": 3700 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.3497, "step": 3710 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.2615, "step": 3720 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.28, "step": 3730 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.2943, "step": 3740 }, { "epoch": 0.61, "learning_rate": 1e-05, "loss": 1.2596, "step": 3750 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.2416, "step": 3760 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.2909, "step": 3770 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.2281, "step": 3780 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.2071, "step": 3790 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.279, "step": 3800 }, { "epoch": 0.62, "learning_rate": 1e-05, "loss": 1.2921, "step": 3810 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.2398, "step": 3820 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.2424, "step": 3830 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.2615, "step": 3840 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.3171, "step": 3850 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.3122, "step": 3860 }, { "epoch": 0.63, "learning_rate": 1e-05, "loss": 1.2333, "step": 3870 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.3382, "step": 3880 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.2945, "step": 3890 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.2718, "step": 3900 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.3351, "step": 3910 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.2759, "step": 3920 }, { "epoch": 0.64, "learning_rate": 1e-05, "loss": 1.3349, "step": 3930 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.2619, "step": 3940 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.2832, "step": 3950 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.2646, "step": 3960 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.3354, "step": 3970 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.3257, "step": 3980 }, { "epoch": 0.65, "learning_rate": 1e-05, "loss": 1.2637, "step": 3990 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.3233, "step": 4000 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.2305, "step": 4010 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.3121, "step": 4020 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.2502, "step": 4030 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.3405, "step": 4040 }, { "epoch": 0.66, "learning_rate": 1e-05, "loss": 1.265, "step": 4050 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.3124, "step": 4060 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.2664, "step": 4070 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.3427, "step": 4080 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.2858, "step": 4090 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.2414, "step": 4100 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.1726, "step": 4110 }, { "epoch": 0.67, "learning_rate": 1e-05, "loss": 1.2786, "step": 4120 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.2386, "step": 4130 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.2194, "step": 4140 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.3317, "step": 4150 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.3069, "step": 4160 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.182, "step": 4170 }, { "epoch": 0.68, "learning_rate": 1e-05, "loss": 1.2819, "step": 4180 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.2253, "step": 4190 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.2657, "step": 4200 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.2418, "step": 4210 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.2285, "step": 4220 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.3214, "step": 4230 }, { "epoch": 0.69, "learning_rate": 1e-05, "loss": 1.2696, "step": 4240 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.3138, "step": 4250 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.2085, "step": 4260 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.2584, "step": 4270 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.2333, "step": 4280 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.3303, "step": 4290 }, { "epoch": 0.7, "learning_rate": 1e-05, "loss": 1.2612, "step": 4300 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.1736, "step": 4310 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.2627, "step": 4320 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.3033, "step": 4330 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.2415, "step": 4340 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.2739, "step": 4350 }, { "epoch": 0.71, "learning_rate": 1e-05, "loss": 1.2591, "step": 4360 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.2699, "step": 4370 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.3777, "step": 4380 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.2144, "step": 4390 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.3081, "step": 4400 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.2445, "step": 4410 }, { "epoch": 0.72, "learning_rate": 1e-05, "loss": 1.1593, "step": 4420 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.2987, "step": 4430 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.2467, "step": 4440 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.2139, "step": 4450 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.2999, "step": 4460 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.2681, "step": 4470 }, { "epoch": 0.73, "learning_rate": 1e-05, "loss": 1.3139, "step": 4480 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.2685, "step": 4490 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.3077, "step": 4500 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.2559, "step": 4510 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.2181, "step": 4520 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.3011, "step": 4530 }, { "epoch": 0.74, "learning_rate": 1e-05, "loss": 1.3051, "step": 4540 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.2037, "step": 4550 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.2504, "step": 4560 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.212, "step": 4570 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.2554, "step": 4580 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.2767, "step": 4590 }, { "epoch": 0.75, "learning_rate": 1e-05, "loss": 1.2845, "step": 4600 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.2142, "step": 4610 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.2353, "step": 4620 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.2294, "step": 4630 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.2984, "step": 4640 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.279, "step": 4650 }, { "epoch": 0.76, "learning_rate": 1e-05, "loss": 1.2119, "step": 4660 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.2493, "step": 4670 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.1979, "step": 4680 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.2422, "step": 4690 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.2443, "step": 4700 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.3392, "step": 4710 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.315, "step": 4720 }, { "epoch": 0.77, "learning_rate": 1e-05, "loss": 1.2761, "step": 4730 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2308, "step": 4740 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2628, "step": 4750 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2887, "step": 4760 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2942, "step": 4770 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2921, "step": 4780 }, { "epoch": 0.78, "learning_rate": 1e-05, "loss": 1.2017, "step": 4790 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.2328, "step": 4800 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.2817, "step": 4810 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.2046, "step": 4820 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.2962, "step": 4830 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.2223, "step": 4840 }, { "epoch": 0.79, "learning_rate": 1e-05, "loss": 1.3326, "step": 4850 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.2584, "step": 4860 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.1835, "step": 4870 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.2967, "step": 4880 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.2358, "step": 4890 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.2331, "step": 4900 }, { "epoch": 0.8, "learning_rate": 1e-05, "loss": 1.2126, "step": 4910 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.3323, "step": 4920 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.2067, "step": 4930 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.2419, "step": 4940 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.2713, "step": 4950 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.1894, "step": 4960 }, { "epoch": 0.81, "learning_rate": 1e-05, "loss": 1.2695, "step": 4970 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.2709, "step": 4980 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.2718, "step": 4990 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.31, "step": 5000 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.2691, "step": 5010 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.2704, "step": 5020 }, { "epoch": 0.82, "learning_rate": 1e-05, "loss": 1.2374, "step": 5030 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.3116, "step": 5040 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.3039, "step": 5050 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.223, "step": 5060 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.2047, "step": 5070 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.2721, "step": 5080 }, { "epoch": 0.83, "learning_rate": 1e-05, "loss": 1.292, "step": 5090 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2201, "step": 5100 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2874, "step": 5110 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2849, "step": 5120 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2056, "step": 5130 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2918, "step": 5140 }, { "epoch": 0.84, "learning_rate": 1e-05, "loss": 1.2618, "step": 5150 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.2592, "step": 5160 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.221, "step": 5170 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.2401, "step": 5180 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.2268, "step": 5190 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.2221, "step": 5200 }, { "epoch": 0.85, "learning_rate": 1e-05, "loss": 1.1223, "step": 5210 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.1579, "step": 5220 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.2795, "step": 5230 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.2462, "step": 5240 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.2158, "step": 5250 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.1765, "step": 5260 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.2117, "step": 5270 }, { "epoch": 0.86, "learning_rate": 1e-05, "loss": 1.2117, "step": 5280 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.204, "step": 5290 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.2754, "step": 5300 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.2453, "step": 5310 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.2434, "step": 5320 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.2598, "step": 5330 }, { "epoch": 0.87, "learning_rate": 1e-05, "loss": 1.2409, "step": 5340 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.2302, "step": 5350 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.2618, "step": 5360 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.3057, "step": 5370 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.2942, "step": 5380 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.1696, "step": 5390 }, { "epoch": 0.88, "learning_rate": 1e-05, "loss": 1.2329, "step": 5400 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.2805, "step": 5410 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.2741, "step": 5420 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.2481, "step": 5430 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.2315, "step": 5440 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.3099, "step": 5450 }, { "epoch": 0.89, "learning_rate": 1e-05, "loss": 1.2454, "step": 5460 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.2796, "step": 5470 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.2591, "step": 5480 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.2764, "step": 5490 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.2662, "step": 5500 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.2407, "step": 5510 }, { "epoch": 0.9, "learning_rate": 1e-05, "loss": 1.219, "step": 5520 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.2577, "step": 5530 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.2551, "step": 5540 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.1574, "step": 5550 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.2744, "step": 5560 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.2122, "step": 5570 }, { "epoch": 0.91, "learning_rate": 1e-05, "loss": 1.2706, "step": 5580 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.2703, "step": 5590 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.2772, "step": 5600 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.2145, "step": 5610 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.3134, "step": 5620 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.234, "step": 5630 }, { "epoch": 0.92, "learning_rate": 1e-05, "loss": 1.2484, "step": 5640 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.1774, "step": 5650 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.3045, "step": 5660 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.2948, "step": 5670 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.25, "step": 5680 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.2265, "step": 5690 }, { "epoch": 0.93, "learning_rate": 1e-05, "loss": 1.3164, "step": 5700 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2688, "step": 5710 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2615, "step": 5720 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2785, "step": 5730 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2813, "step": 5740 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2287, "step": 5750 }, { "epoch": 0.94, "learning_rate": 1e-05, "loss": 1.2914, "step": 5760 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.2133, "step": 5770 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.2405, "step": 5780 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.3153, "step": 5790 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.2094, "step": 5800 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.2593, "step": 5810 }, { "epoch": 0.95, "learning_rate": 1e-05, "loss": 1.3192, "step": 5820 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.2264, "step": 5830 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.267, "step": 5840 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.1878, "step": 5850 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.2058, "step": 5860 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.1589, "step": 5870 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.1993, "step": 5880 }, { "epoch": 0.96, "learning_rate": 1e-05, "loss": 1.2312, "step": 5890 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.2511, "step": 5900 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.3163, "step": 5910 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.2523, "step": 5920 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.3137, "step": 5930 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.2225, "step": 5940 }, { "epoch": 0.97, "learning_rate": 1e-05, "loss": 1.2642, "step": 5950 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.2146, "step": 5960 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.3105, "step": 5970 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.2762, "step": 5980 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.2242, "step": 5990 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.2385, "step": 6000 }, { "epoch": 0.98, "learning_rate": 1e-05, "loss": 1.2467, "step": 6010 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.2547, "step": 6020 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.2799, "step": 6030 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.2916, "step": 6040 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.2668, "step": 6050 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.1854, "step": 6060 }, { "epoch": 0.99, "learning_rate": 1e-05, "loss": 1.2529, "step": 6070 }, { "epoch": 1.0, "learning_rate": 1e-05, "loss": 1.2653, "step": 6080 }, { "epoch": 1.0, "learning_rate": 1e-05, "loss": 1.2418, "step": 6090 }, { "epoch": 1.0, "learning_rate": 1e-05, "loss": 1.246, "step": 6100 }, { "epoch": 1.0, "eval_oasst_export_accuracy": 0.6878104734550841, "eval_oasst_export_loss": 1.349609375, "eval_oasst_export_runtime": 19.3636, "eval_oasst_export_samples_per_second": 15.493, "eval_oasst_export_steps_per_second": 1.962, "step": 6104 }, { "epoch": 1.0, "eval_dolly15k_accuracy": 0.6627859972511925, "eval_dolly15k_loss": 1.2138671875, "eval_dolly15k_runtime": 15.5602, "eval_dolly15k_samples_per_second": 19.28, "eval_dolly15k_steps_per_second": 2.442, "step": 6104 }, { "epoch": 1.0, "eval__ruozhiba_accuracy": 0.5154994259471871, "eval__ruozhiba_loss": 1.7412109375, "eval__ruozhiba_runtime": 0.6619, "eval__ruozhiba_samples_per_second": 10.576, "eval__ruozhiba_steps_per_second": 1.511, "step": 6104 }, { "epoch": 1.0, "eval__sharegpt_format_accuracy": 0.8137771622392778, "eval__sharegpt_format_loss": 1.0361328125, "eval__sharegpt_format_runtime": 32.5298, "eval__sharegpt_format_samples_per_second": 15.371, "eval__sharegpt_format_steps_per_second": 1.937, "step": 6104 }, { "epoch": 1.0, "eval__similar_question_accuracy": 0.577728776185226, "eval__similar_question_loss": 1.6435546875, "eval__similar_question_runtime": 0.7886, "eval__similar_question_samples_per_second": 8.877, "eval__similar_question_steps_per_second": 1.268, "step": 6104 }, { "epoch": 1.0, "eval__open_domain_subject_accuracy": 0.6082298514730947, "eval__open_domain_subject_loss": 1.451171875, "eval__open_domain_subject_runtime": 2.7365, "eval__open_domain_subject_samples_per_second": 16.81, "eval__open_domain_subject_steps_per_second": 2.193, "step": 6104 }, { "epoch": 1.0, "eval__empdia_ly_07_17_accuracy": 0.6261699064074874, "eval__empdia_ly_07_17_loss": 1.33984375, "eval__empdia_ly_07_17_runtime": 5.1144, "eval__empdia_ly_07_17_samples_per_second": 17.988, "eval__empdia_ly_07_17_steps_per_second": 2.346, "step": 6104 }, { "epoch": 1.0, "eval__zuowen_accuracy": 0.9144507380228243, "eval__zuowen_loss": 0.299560546875, "eval__zuowen_runtime": 6.9808, "eval__zuowen_samples_per_second": 14.325, "eval__zuowen_steps_per_second": 1.862, "step": 6104 }, { "epoch": 1.0, "eval__psychat_6_29_and_11_15_mix_accuracy": 0.5055779183438758, "eval__psychat_6_29_and_11_15_mix_loss": 1.9296875, "eval__psychat_6_29_and_11_15_mix_runtime": 6.0911, "eval__psychat_6_29_and_11_15_mix_samples_per_second": 16.417, "eval__psychat_6_29_and_11_15_mix_steps_per_second": 2.134, "step": 6104 }, { "epoch": 1.0, "eval__psy_qa_1208_accuracy": 0.49772520473157417, "eval__psy_qa_1208_loss": 2.26171875, "eval__psy_qa_1208_runtime": 2.7612, "eval__psy_qa_1208_samples_per_second": 16.297, "eval__psy_qa_1208_steps_per_second": 2.173, "step": 6104 }, { "epoch": 1.0, "eval__socrates_teaching_accuracy": 0.6412340032115226, "eval__socrates_teaching_loss": 1.3828125, "eval__socrates_teaching_runtime": 5.1769, "eval__socrates_teaching_samples_per_second": 16.612, "eval__socrates_teaching_steps_per_second": 2.125, "step": 6104 }, { "epoch": 1.0, "eval__socrates_teaching_math1213_accuracy": 0.9996854356715948, "eval__socrates_teaching_math1213_loss": 0.0015439987182617188, "eval__socrates_teaching_math1213_runtime": 1.6723, "eval__socrates_teaching_math1213_samples_per_second": 14.949, "eval__socrates_teaching_math1213_steps_per_second": 2.392, "step": 6104 }, { "epoch": 1.0, "eval__search_accuracy": 0.6996122053270742, "eval__search_loss": 1.099609375, "eval__search_runtime": 5.9225, "eval__search_samples_per_second": 16.885, "eval__search_steps_per_second": 2.195, "step": 6104 } ], "max_steps": 12208, "num_train_epochs": 2, "total_flos": 5281029455085568.0, "trial_name": null, "trial_params": null }