{ "best_metric": null, "best_model_checkpoint": null, "epoch": 6.085192697768763, "global_step": 9000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "acc": 0.0, "epoch": 0.0, "learning_rate": 2e-05, "loss": 0.2016, "rl_loss": -0.0068, "step": 1 }, { "acc": 0.0, "epoch": 0.01, "learning_rate": 2e-05, "loss": 0.1913, "rl_loss": -0.0196, "step": 10 }, { "acc": 0.0, "epoch": 0.01, "learning_rate": 2e-05, "loss": 0.204, "rl_loss": -0.069, "step": 20 }, { "acc": 0.0, "epoch": 0.02, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -0.0423, "step": 30 }, { "acc": 0.0, "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.1903, "rl_loss": -0.1927, "step": 40 }, { "acc": 0.0, "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.2237, "rl_loss": -0.4978, "step": 50 }, { "acc": 0.0, "epoch": 0.04, "learning_rate": 2e-05, "loss": 0.247, "rl_loss": -0.627, "step": 60 }, { "acc": 0.0, "epoch": 0.05, "learning_rate": 2e-05, "loss": 0.309, "rl_loss": -0.4734, "step": 70 }, { "acc": 0.0, "epoch": 0.05, "learning_rate": 2e-05, "loss": 0.2298, "rl_loss": -0.6488, "step": 80 }, { "acc": 0.0, "epoch": 0.06, "learning_rate": 2e-05, "loss": 0.218, "rl_loss": -0.8969, "step": 90 }, { "acc": 0.0, "epoch": 0.07, "learning_rate": 2e-05, "loss": 0.2409, "rl_loss": -0.9568, "step": 100 }, { "acc": 0.0, "epoch": 0.07, "learning_rate": 2e-05, "loss": 0.2302, "rl_loss": -0.0831, "step": 110 }, { "acc": 0.0, "epoch": 0.08, "learning_rate": 2e-05, "loss": 0.2382, "rl_loss": -0.7247, "step": 120 }, { "acc": 0.0, "epoch": 0.09, "learning_rate": 2e-05, "loss": 0.2293, "rl_loss": -1.1691, "step": 130 }, { "acc": 0.0, "epoch": 0.09, "learning_rate": 2e-05, "loss": 0.2124, "rl_loss": -1.1364, "step": 140 }, { "acc": 0.0, "epoch": 0.1, "learning_rate": 2e-05, "loss": 0.221, "rl_loss": -0.6203, "step": 150 }, { "acc": 0.0, "epoch": 0.11, "learning_rate": 2e-05, "loss": 0.2165, "rl_loss": -0.7486, "step": 160 }, { "acc": 0.0, "epoch": 0.11, "learning_rate": 2e-05, "loss": 0.2134, "rl_loss": -1.13, "step": 170 }, { "acc": 0.0, "epoch": 0.12, "learning_rate": 2e-05, "loss": 0.214, "rl_loss": -0.5846, "step": 180 }, { "acc": 0.0, "epoch": 0.13, "learning_rate": 2e-05, "loss": 0.2095, "rl_loss": -0.9399, "step": 190 }, { "acc": 0.0, "epoch": 0.14, "learning_rate": 2e-05, "loss": 0.2143, "rl_loss": -0.4508, "step": 200 }, { "acc": 0.0, "epoch": 0.14, "learning_rate": 2e-05, "loss": 0.2287, "rl_loss": -0.9757, "step": 210 }, { "acc": 0.0, "epoch": 0.15, "learning_rate": 2e-05, "loss": 0.2165, "rl_loss": -1.6187, "step": 220 }, { "acc": 0.0, "epoch": 0.16, "learning_rate": 2e-05, "loss": 0.2084, "rl_loss": -0.5423, "step": 230 }, { "acc": 0.0, "epoch": 0.16, "learning_rate": 2e-05, "loss": 0.2025, "rl_loss": -0.4129, "step": 240 }, { "acc": 0.0, "epoch": 0.17, "learning_rate": 2e-05, "loss": 0.1911, "rl_loss": -1.5743, "step": 250 }, { "acc": 0.0, "epoch": 0.18, "learning_rate": 2e-05, "loss": 0.2866, "rl_loss": -0.8082, "step": 260 }, { "acc": 0.0, "epoch": 0.18, "learning_rate": 2e-05, "loss": 0.2133, "rl_loss": -1.1171, "step": 270 }, { "acc": 0.0, "epoch": 0.19, "learning_rate": 2e-05, "loss": 0.2113, "rl_loss": -0.597, "step": 280 }, { "acc": 0.0, "epoch": 0.2, "learning_rate": 2e-05, "loss": 0.2001, "rl_loss": -0.679, "step": 290 }, { "acc": 0.0, "epoch": 0.2, "learning_rate": 2e-05, "loss": 0.2386, "rl_loss": -1.5896, "step": 300 }, { "acc": 0.0, "epoch": 0.21, "learning_rate": 2e-05, "loss": 0.2029, "rl_loss": -0.5881, "step": 310 }, { "acc": 0.0, "epoch": 0.22, "learning_rate": 2e-05, "loss": 0.2047, "rl_loss": -0.6694, "step": 320 }, { "acc": 0.0, "epoch": 0.22, "learning_rate": 2e-05, "loss": 0.1989, "rl_loss": -0.9078, "step": 330 }, { "acc": 0.0, "epoch": 0.23, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -1.1072, "step": 340 }, { "acc": 0.0, "epoch": 0.24, "learning_rate": 2e-05, "loss": 0.3267, "rl_loss": -0.9213, "step": 350 }, { "acc": 0.0, "epoch": 0.24, "learning_rate": 2e-05, "loss": 0.212, "rl_loss": -0.414, "step": 360 }, { "acc": 0.0, "epoch": 0.25, "learning_rate": 2e-05, "loss": 0.2015, "rl_loss": -0.7487, "step": 370 }, { "acc": 0.0, "epoch": 0.26, "learning_rate": 2e-05, "loss": 0.2086, "rl_loss": -0.823, "step": 380 }, { "acc": 0.0, "epoch": 0.26, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.1136, "step": 390 }, { "acc": 0.0, "epoch": 0.27, "learning_rate": 2e-05, "loss": 0.2032, "rl_loss": -0.4595, "step": 400 }, { "acc": 0.0, "epoch": 0.28, "learning_rate": 2e-05, "loss": 0.2216, "rl_loss": -1.2084, "step": 410 }, { "acc": 0.0, "epoch": 0.28, "learning_rate": 2e-05, "loss": 0.2121, "rl_loss": -0.7541, "step": 420 }, { "acc": 0.0, "epoch": 0.29, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -1.5264, "step": 430 }, { "acc": 0.0, "epoch": 0.3, "learning_rate": 2e-05, "loss": 0.2136, "rl_loss": -0.8737, "step": 440 }, { "acc": 0.0, "epoch": 0.3, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -1.0759, "step": 450 }, { "acc": 0.0, "epoch": 0.31, "learning_rate": 2e-05, "loss": 0.1945, "rl_loss": -0.7129, "step": 460 }, { "acc": 0.0, "epoch": 0.32, "learning_rate": 2e-05, "loss": 0.2045, "rl_loss": -0.6577, "step": 470 }, { "acc": 0.0, "epoch": 0.32, "learning_rate": 2e-05, "loss": 0.1961, "rl_loss": -0.9649, "step": 480 }, { "acc": 0.0, "epoch": 0.33, "learning_rate": 2e-05, "loss": 0.2154, "rl_loss": -0.7029, "step": 490 }, { "acc": 0.0, "epoch": 0.34, "learning_rate": 2e-05, "loss": 0.226, "rl_loss": -0.7143, "step": 500 }, { "acc": 0.0, "epoch": 0.34, "learning_rate": 2e-05, "loss": 0.2093, "rl_loss": -1.2367, "step": 510 }, { "acc": 0.0, "epoch": 0.35, "learning_rate": 2e-05, "loss": 0.2162, "rl_loss": -1.242, "step": 520 }, { "acc": 0.0, "epoch": 0.36, "learning_rate": 2e-05, "loss": 0.2025, "rl_loss": -0.4002, "step": 530 }, { "acc": 0.0, "epoch": 0.37, "learning_rate": 2e-05, "loss": 0.2131, "rl_loss": -0.832, "step": 540 }, { "acc": 0.0, "epoch": 0.37, "learning_rate": 2e-05, "loss": 0.2111, "rl_loss": -1.4323, "step": 550 }, { "acc": 0.0, "epoch": 0.38, "learning_rate": 2e-05, "loss": 0.217, "rl_loss": -1.5549, "step": 560 }, { "acc": 0.0, "epoch": 0.39, "learning_rate": 2e-05, "loss": 0.2055, "rl_loss": -0.8696, "step": 570 }, { "acc": 0.0, "epoch": 0.39, "learning_rate": 2e-05, "loss": 0.2006, "rl_loss": -1.4912, "step": 580 }, { "acc": 0.0, "epoch": 0.4, "learning_rate": 2e-05, "loss": 0.2088, "rl_loss": -1.8852, "step": 590 }, { "acc": 0.0, "epoch": 0.41, "learning_rate": 2e-05, "loss": 0.212, "rl_loss": -0.9952, "step": 600 }, { "acc": 0.0, "epoch": 0.41, "learning_rate": 2e-05, "loss": 0.1979, "rl_loss": -1.0308, "step": 610 }, { "acc": 0.0, "epoch": 0.42, "learning_rate": 2e-05, "loss": 0.2105, "rl_loss": -0.7833, "step": 620 }, { "acc": 0.0, "epoch": 0.43, "learning_rate": 2e-05, "loss": 0.2021, "rl_loss": -0.7707, "step": 630 }, { "acc": 0.0, "epoch": 0.43, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -1.6439, "step": 640 }, { "acc": 0.0, "epoch": 0.44, "learning_rate": 2e-05, "loss": 0.2098, "rl_loss": -1.0089, "step": 650 }, { "acc": 0.0, "epoch": 0.45, "learning_rate": 2e-05, "loss": 0.2132, "rl_loss": -1.2961, "step": 660 }, { "acc": 0.0, "epoch": 0.45, "learning_rate": 2e-05, "loss": 0.2144, "rl_loss": -0.983, "step": 670 }, { "acc": 0.0, "epoch": 0.46, "learning_rate": 2e-05, "loss": 0.2015, "rl_loss": -1.054, "step": 680 }, { "acc": 0.0, "epoch": 0.47, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -0.7861, "step": 690 }, { "acc": 0.0, "epoch": 0.47, "learning_rate": 2e-05, "loss": 0.2016, "rl_loss": -1.3266, "step": 700 }, { "acc": 0.0, "epoch": 0.48, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -0.8763, "step": 710 }, { "acc": 0.0, "epoch": 0.49, "learning_rate": 2e-05, "loss": 0.2004, "rl_loss": -1.4412, "step": 720 }, { "acc": 0.0, "epoch": 0.49, "learning_rate": 2e-05, "loss": 0.2163, "rl_loss": -1.4161, "step": 730 }, { "acc": 0.0, "epoch": 0.5, "learning_rate": 2e-05, "loss": 0.2046, "rl_loss": -0.9692, "step": 740 }, { "acc": 0.0, "epoch": 0.51, "learning_rate": 2e-05, "loss": 0.2167, "rl_loss": -1.608, "step": 750 }, { "acc": 0.0, "epoch": 0.51, "learning_rate": 2e-05, "loss": 0.2022, "rl_loss": -0.9561, "step": 760 }, { "acc": 0.0, "epoch": 0.52, "learning_rate": 2e-05, "loss": 0.2185, "rl_loss": -1.1544, "step": 770 }, { "acc": 0.0, "epoch": 0.53, "learning_rate": 2e-05, "loss": 0.2519, "rl_loss": -1.724, "step": 780 }, { "acc": 0.0, "epoch": 0.53, "learning_rate": 2e-05, "loss": 0.216, "rl_loss": -1.1172, "step": 790 }, { "acc": 0.0, "epoch": 0.54, "learning_rate": 2e-05, "loss": 0.2134, "rl_loss": -1.5313, "step": 800 }, { "acc": 0.0, "epoch": 0.55, "learning_rate": 2e-05, "loss": 0.2097, "rl_loss": -1.2275, "step": 810 }, { "acc": 0.0, "epoch": 0.55, "learning_rate": 2e-05, "loss": 0.2014, "rl_loss": -0.8528, "step": 820 }, { "acc": 0.0, "epoch": 0.56, "learning_rate": 2e-05, "loss": 0.213, "rl_loss": -1.3257, "step": 830 }, { "acc": 0.0, "epoch": 0.57, "learning_rate": 2e-05, "loss": 0.2056, "rl_loss": -0.4148, "step": 840 }, { "acc": 0.0, "epoch": 0.57, "learning_rate": 2e-05, "loss": 0.2088, "rl_loss": -0.9069, "step": 850 }, { "acc": 0.0, "epoch": 0.58, "learning_rate": 2e-05, "loss": 0.2067, "rl_loss": -0.4816, "step": 860 }, { "acc": 0.0, "epoch": 0.59, "learning_rate": 2e-05, "loss": 0.2047, "rl_loss": -0.7858, "step": 870 }, { "acc": 0.0, "epoch": 0.59, "learning_rate": 2e-05, "loss": 0.2021, "rl_loss": -1.4338, "step": 880 }, { "acc": 0.0, "epoch": 0.6, "learning_rate": 2e-05, "loss": 0.1971, "rl_loss": -1.5757, "step": 890 }, { "acc": 0.0, "epoch": 0.61, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -0.7151, "step": 900 }, { "acc": 0.0, "epoch": 0.62, "learning_rate": 2e-05, "loss": 0.1993, "rl_loss": -1.4565, "step": 910 }, { "acc": 0.0, "epoch": 0.62, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -1.7945, "step": 920 }, { "acc": 0.0, "epoch": 0.63, "learning_rate": 2e-05, "loss": 0.1928, "rl_loss": -1.2473, "step": 930 }, { "acc": 0.0, "epoch": 0.64, "learning_rate": 2e-05, "loss": 0.2092, "rl_loss": -1.2106, "step": 940 }, { "acc": 0.0, "epoch": 0.64, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -1.1657, "step": 950 }, { "acc": 0.0, "epoch": 0.65, "learning_rate": 2e-05, "loss": 0.2104, "rl_loss": -1.7675, "step": 960 }, { "acc": 0.0, "epoch": 0.66, "learning_rate": 2e-05, "loss": 0.2104, "rl_loss": -1.2402, "step": 970 }, { "acc": 0.0, "epoch": 0.66, "learning_rate": 2e-05, "loss": 0.2059, "rl_loss": -1.5079, "step": 980 }, { "acc": 0.0, "epoch": 0.67, "learning_rate": 2e-05, "loss": 0.2071, "rl_loss": -1.8573, "step": 990 }, { "acc": 0.0, "epoch": 0.68, "learning_rate": 2e-05, "loss": 0.216, "rl_loss": -1.1563, "step": 1000 }, { "acc": 0.0, "epoch": 0.68, "learning_rate": 2e-05, "loss": 0.2178, "rl_loss": -1.1678, "step": 1010 }, { "acc": 0.0, "epoch": 0.69, "learning_rate": 2e-05, "loss": 0.2276, "rl_loss": -1.4955, "step": 1020 }, { "acc": 0.0, "epoch": 0.7, "learning_rate": 2e-05, "loss": 0.216, "rl_loss": -0.8749, "step": 1030 }, { "acc": 0.0, "epoch": 0.7, "learning_rate": 2e-05, "loss": 0.2179, "rl_loss": -0.617, "step": 1040 }, { "acc": 0.0, "epoch": 0.71, "learning_rate": 2e-05, "loss": 0.2078, "rl_loss": -0.9727, "step": 1050 }, { "acc": 0.0, "epoch": 0.72, "learning_rate": 2e-05, "loss": 0.2102, "rl_loss": -0.9591, "step": 1060 }, { "acc": 0.0, "epoch": 0.72, "learning_rate": 2e-05, "loss": 0.2069, "rl_loss": -0.8018, "step": 1070 }, { "acc": 0.0, "epoch": 0.73, "learning_rate": 2e-05, "loss": 0.2051, "rl_loss": -1.371, "step": 1080 }, { "acc": 0.0, "epoch": 0.74, "learning_rate": 2e-05, "loss": 0.193, "rl_loss": -1.5617, "step": 1090 }, { "acc": 0.0, "epoch": 0.74, "learning_rate": 2e-05, "loss": 0.2108, "rl_loss": -1.4672, "step": 1100 }, { "acc": 0.0, "epoch": 0.75, "learning_rate": 2e-05, "loss": 0.2096, "rl_loss": -1.8175, "step": 1110 }, { "acc": 0.0, "epoch": 0.76, "learning_rate": 2e-05, "loss": 0.2121, "rl_loss": -1.5995, "step": 1120 }, { "acc": 0.0, "epoch": 0.76, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -1.0439, "step": 1130 }, { "acc": 0.0, "epoch": 0.77, "learning_rate": 2e-05, "loss": 0.2188, "rl_loss": -1.602, "step": 1140 }, { "acc": 0.0, "epoch": 0.78, "learning_rate": 2e-05, "loss": 0.2064, "rl_loss": -2.0899, "step": 1150 }, { "acc": 0.0, "epoch": 0.78, "learning_rate": 2e-05, "loss": 0.2073, "rl_loss": -1.0716, "step": 1160 }, { "acc": 0.0, "epoch": 0.79, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -0.5594, "step": 1170 }, { "acc": 0.0, "epoch": 0.8, "learning_rate": 2e-05, "loss": 0.2111, "rl_loss": -1.6328, "step": 1180 }, { "acc": 0.0, "epoch": 0.8, "learning_rate": 2e-05, "loss": 0.2099, "rl_loss": -1.2663, "step": 1190 }, { "acc": 0.0, "epoch": 0.81, "learning_rate": 2e-05, "loss": 0.22, "rl_loss": -1.9216, "step": 1200 }, { "acc": 0.0, "epoch": 0.82, "learning_rate": 2e-05, "loss": 0.2, "rl_loss": -1.7567, "step": 1210 }, { "acc": 0.0, "epoch": 0.82, "learning_rate": 2e-05, "loss": 0.2091, "rl_loss": -1.7256, "step": 1220 }, { "acc": 0.0, "epoch": 0.83, "learning_rate": 2e-05, "loss": 0.218, "rl_loss": -1.4322, "step": 1230 }, { "acc": 0.0, "epoch": 0.84, "learning_rate": 2e-05, "loss": 0.2102, "rl_loss": -1.4049, "step": 1240 }, { "acc": 0.0, "epoch": 0.85, "learning_rate": 2e-05, "loss": 0.2215, "rl_loss": -1.453, "step": 1250 }, { "acc": 0.0, "epoch": 0.85, "learning_rate": 2e-05, "loss": 0.203, "rl_loss": -1.5857, "step": 1260 }, { "acc": 0.0, "epoch": 0.86, "learning_rate": 2e-05, "loss": 0.2165, "rl_loss": -1.3901, "step": 1270 }, { "acc": 0.0, "epoch": 0.87, "learning_rate": 2e-05, "loss": 0.2034, "rl_loss": -1.2783, "step": 1280 }, { "acc": 0.0, "epoch": 0.87, "learning_rate": 2e-05, "loss": 0.2022, "rl_loss": -1.0224, "step": 1290 }, { "acc": 0.0, "epoch": 0.88, "learning_rate": 2e-05, "loss": 0.2173, "rl_loss": -2.0539, "step": 1300 }, { "acc": 0.0, "epoch": 0.89, "learning_rate": 2e-05, "loss": 0.202, "rl_loss": -1.5235, "step": 1310 }, { "acc": 0.0, "epoch": 0.89, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -1.5547, "step": 1320 }, { "acc": 0.0, "epoch": 0.9, "learning_rate": 2e-05, "loss": 0.2125, "rl_loss": -2.0365, "step": 1330 }, { "acc": 0.0, "epoch": 0.91, "learning_rate": 2e-05, "loss": 0.2056, "rl_loss": -1.3627, "step": 1340 }, { "acc": 0.0, "epoch": 0.91, "learning_rate": 2e-05, "loss": 0.2028, "rl_loss": -1.4812, "step": 1350 }, { "acc": 0.0, "epoch": 0.92, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -1.6724, "step": 1360 }, { "acc": 0.0, "epoch": 0.93, "learning_rate": 2e-05, "loss": 0.2005, "rl_loss": -0.4073, "step": 1370 }, { "acc": 0.0, "epoch": 0.93, "learning_rate": 2e-05, "loss": 0.21, "rl_loss": -0.9003, "step": 1380 }, { "acc": 0.0, "epoch": 0.94, "learning_rate": 2e-05, "loss": 0.2146, "rl_loss": -1.3524, "step": 1390 }, { "acc": 0.0, "epoch": 0.95, "learning_rate": 2e-05, "loss": 0.2027, "rl_loss": -0.911, "step": 1400 }, { "acc": 0.0, "epoch": 0.95, "learning_rate": 2e-05, "loss": 0.2044, "rl_loss": -0.7583, "step": 1410 }, { "acc": 0.0, "epoch": 0.96, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -1.582, "step": 1420 }, { "acc": 0.0, "epoch": 0.97, "learning_rate": 2e-05, "loss": 0.2213, "rl_loss": -1.073, "step": 1430 }, { "acc": 0.0, "epoch": 0.97, "learning_rate": 2e-05, "loss": 0.2054, "rl_loss": -1.4395, "step": 1440 }, { "acc": 0.0, "epoch": 0.98, "learning_rate": 2e-05, "loss": 0.1974, "rl_loss": -1.7853, "step": 1450 }, { "acc": 0.0, "epoch": 0.99, "learning_rate": 2e-05, "loss": 0.2065, "rl_loss": -1.2973, "step": 1460 }, { "acc": 0.0, "epoch": 0.99, "learning_rate": 2e-05, "loss": 0.2075, "rl_loss": -1.7771, "step": 1470 }, { "acc": 0.0, "epoch": 1.0, "learning_rate": 2e-05, "loss": 0.2045, "rl_loss": -1.1747, "step": 1480 }, { "acc": 0.0, "epoch": 1.01, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -1.189, "step": 1490 }, { "acc": 0.0, "epoch": 1.01, "learning_rate": 2e-05, "loss": 0.2177, "rl_loss": -1.6834, "step": 1500 }, { "acc": 0.0, "epoch": 1.02, "learning_rate": 2e-05, "loss": 0.1941, "rl_loss": -1.4439, "step": 1510 }, { "acc": 0.0, "epoch": 1.03, "learning_rate": 2e-05, "loss": 0.2113, "rl_loss": -1.2772, "step": 1520 }, { "acc": 0.0, "epoch": 1.03, "learning_rate": 2e-05, "loss": 0.2093, "rl_loss": -1.7273, "step": 1530 }, { "acc": 0.0, "epoch": 1.04, "learning_rate": 2e-05, "loss": 0.2078, "rl_loss": -2.2682, "step": 1540 }, { "acc": 0.0, "epoch": 1.05, "learning_rate": 2e-05, "loss": 0.2144, "rl_loss": -1.7264, "step": 1550 }, { "acc": 0.0, "epoch": 1.05, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -2.5611, "step": 1560 }, { "acc": 0.0, "epoch": 1.06, "learning_rate": 2e-05, "loss": 0.2032, "rl_loss": -1.9682, "step": 1570 }, { "acc": 0.0, "epoch": 1.07, "learning_rate": 2e-05, "loss": 0.2045, "rl_loss": -1.0156, "step": 1580 }, { "acc": 0.0, "epoch": 1.08, "learning_rate": 2e-05, "loss": 0.1924, "rl_loss": -1.1269, "step": 1590 }, { "acc": 0.0, "epoch": 1.08, "learning_rate": 2e-05, "loss": 0.2026, "rl_loss": -1.0348, "step": 1600 }, { "acc": 0.0, "epoch": 1.09, "learning_rate": 2e-05, "loss": 0.2109, "rl_loss": -1.4111, "step": 1610 }, { "acc": 0.0, "epoch": 1.1, "learning_rate": 2e-05, "loss": 0.2006, "rl_loss": -1.1935, "step": 1620 }, { "acc": 0.0, "epoch": 1.1, "learning_rate": 2e-05, "loss": 0.2018, "rl_loss": -1.2109, "step": 1630 }, { "acc": 0.0, "epoch": 1.11, "learning_rate": 2e-05, "loss": 0.2002, "rl_loss": -1.9524, "step": 1640 }, { "acc": 0.0, "epoch": 1.12, "learning_rate": 2e-05, "loss": 0.204, "rl_loss": -1.4621, "step": 1650 }, { "acc": 0.0, "epoch": 1.12, "learning_rate": 2e-05, "loss": 0.1969, "rl_loss": -1.2072, "step": 1660 }, { "acc": 0.0, "epoch": 1.13, "learning_rate": 2e-05, "loss": 0.1973, "rl_loss": -1.8943, "step": 1670 }, { "acc": 0.0, "epoch": 1.14, "learning_rate": 2e-05, "loss": 0.203, "rl_loss": -2.2548, "step": 1680 }, { "acc": 0.0, "epoch": 1.14, "learning_rate": 2e-05, "loss": 0.2103, "rl_loss": -1.1583, "step": 1690 }, { "acc": 0.0, "epoch": 1.15, "learning_rate": 2e-05, "loss": 0.2041, "rl_loss": -1.7588, "step": 1700 }, { "acc": 0.0, "epoch": 1.16, "learning_rate": 2e-05, "loss": 0.2077, "rl_loss": -1.8442, "step": 1710 }, { "acc": 0.0, "epoch": 1.16, "learning_rate": 2e-05, "loss": 0.2036, "rl_loss": -1.8368, "step": 1720 }, { "acc": 0.0, "epoch": 1.17, "learning_rate": 2e-05, "loss": 0.1921, "rl_loss": -2.1755, "step": 1730 }, { "acc": 0.0, "epoch": 1.18, "learning_rate": 2e-05, "loss": 0.2115, "rl_loss": -0.9972, "step": 1740 }, { "acc": 0.0, "epoch": 1.18, "learning_rate": 2e-05, "loss": 0.2051, "rl_loss": -1.5256, "step": 1750 }, { "acc": 0.0, "epoch": 1.19, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -1.7492, "step": 1760 }, { "acc": 0.0, "epoch": 1.2, "learning_rate": 2e-05, "loss": 0.2028, "rl_loss": -1.8514, "step": 1770 }, { "acc": 0.0, "epoch": 1.2, "learning_rate": 2e-05, "loss": 0.1973, "rl_loss": -1.6839, "step": 1780 }, { "acc": 0.0, "epoch": 1.21, "learning_rate": 2e-05, "loss": 0.1972, "rl_loss": -1.9182, "step": 1790 }, { "acc": 0.0, "epoch": 1.22, "learning_rate": 2e-05, "loss": 0.1917, "rl_loss": -2.0456, "step": 1800 }, { "acc": 0.0, "epoch": 1.22, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -1.6735, "step": 1810 }, { "acc": 0.0, "epoch": 1.23, "learning_rate": 2e-05, "loss": 0.213, "rl_loss": -1.2171, "step": 1820 }, { "acc": 0.0, "epoch": 1.24, "learning_rate": 2e-05, "loss": 0.2064, "rl_loss": -1.8816, "step": 1830 }, { "acc": 0.0, "epoch": 1.24, "learning_rate": 2e-05, "loss": 0.2118, "rl_loss": -1.6733, "step": 1840 }, { "acc": 0.0, "epoch": 1.25, "learning_rate": 2e-05, "loss": 0.2006, "rl_loss": -3.33, "step": 1850 }, { "acc": 0.0, "epoch": 1.26, "learning_rate": 2e-05, "loss": 0.2107, "rl_loss": -2.4915, "step": 1860 }, { "acc": 0.0, "epoch": 1.26, "learning_rate": 2e-05, "loss": 0.1992, "rl_loss": -1.5826, "step": 1870 }, { "acc": 0.0, "epoch": 1.27, "learning_rate": 2e-05, "loss": 0.2069, "rl_loss": -2.3705, "step": 1880 }, { "acc": 0.0, "epoch": 1.28, "learning_rate": 2e-05, "loss": 0.2136, "rl_loss": -1.7253, "step": 1890 }, { "acc": 0.0, "epoch": 1.28, "learning_rate": 2e-05, "loss": 0.1993, "rl_loss": -1.3736, "step": 1900 }, { "acc": 0.0, "epoch": 1.29, "learning_rate": 2e-05, "loss": 0.1928, "rl_loss": -1.7456, "step": 1910 }, { "acc": 0.0, "epoch": 1.3, "learning_rate": 2e-05, "loss": 0.2089, "rl_loss": -2.2405, "step": 1920 }, { "acc": 0.0, "epoch": 1.3, "learning_rate": 2e-05, "loss": 0.2081, "rl_loss": -1.2207, "step": 1930 }, { "acc": 0.0, "epoch": 1.31, "learning_rate": 2e-05, "loss": 0.2083, "rl_loss": -1.7201, "step": 1940 }, { "acc": 0.0, "epoch": 1.32, "learning_rate": 2e-05, "loss": 0.2093, "rl_loss": -1.2544, "step": 1950 }, { "acc": 0.0, "epoch": 1.33, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -1.4828, "step": 1960 }, { "acc": 0.0, "epoch": 1.33, "learning_rate": 2e-05, "loss": 0.2027, "rl_loss": -2.005, "step": 1970 }, { "acc": 0.0, "epoch": 1.34, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -1.5474, "step": 1980 }, { "acc": 0.0, "epoch": 1.35, "learning_rate": 2e-05, "loss": 0.2006, "rl_loss": -2.2947, "step": 1990 }, { "acc": 0.0, "epoch": 1.35, "learning_rate": 2e-05, "loss": 0.2041, "rl_loss": -1.4322, "step": 2000 }, { "acc": 0.0, "epoch": 1.36, "learning_rate": 2e-05, "loss": 0.2029, "rl_loss": -1.4151, "step": 2010 }, { "acc": 0.0, "epoch": 1.37, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -1.9008, "step": 2020 }, { "acc": 0.0, "epoch": 1.37, "learning_rate": 2e-05, "loss": 0.1981, "rl_loss": -1.8018, "step": 2030 }, { "acc": 0.0, "epoch": 1.38, "learning_rate": 2e-05, "loss": 0.2118, "rl_loss": -1.2058, "step": 2040 }, { "acc": 0.0, "epoch": 1.39, "learning_rate": 2e-05, "loss": 0.1971, "rl_loss": -2.4733, "step": 2050 }, { "acc": 0.0, "epoch": 1.39, "learning_rate": 2e-05, "loss": 0.1938, "rl_loss": -1.3028, "step": 2060 }, { "acc": 0.0, "epoch": 1.4, "learning_rate": 2e-05, "loss": 0.1976, "rl_loss": -1.8747, "step": 2070 }, { "acc": 0.0, "epoch": 1.41, "learning_rate": 2e-05, "loss": 0.2067, "rl_loss": -1.6249, "step": 2080 }, { "acc": 0.0, "epoch": 1.41, "learning_rate": 2e-05, "loss": 0.2036, "rl_loss": -2.5536, "step": 2090 }, { "acc": 0.0, "epoch": 1.42, "learning_rate": 2e-05, "loss": 0.2005, "rl_loss": -1.3202, "step": 2100 }, { "acc": 0.0, "epoch": 1.43, "learning_rate": 2e-05, "loss": 0.2088, "rl_loss": -1.5091, "step": 2110 }, { "acc": 0.0, "epoch": 1.43, "learning_rate": 2e-05, "loss": 0.1951, "rl_loss": -1.4308, "step": 2120 }, { "acc": 0.0, "epoch": 1.44, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -1.5221, "step": 2130 }, { "acc": 0.0, "epoch": 1.45, "learning_rate": 2e-05, "loss": 0.1852, "rl_loss": -1.4614, "step": 2140 }, { "acc": 0.0, "epoch": 1.45, "learning_rate": 2e-05, "loss": 0.1936, "rl_loss": -2.3238, "step": 2150 }, { "acc": 0.0, "epoch": 1.46, "learning_rate": 2e-05, "loss": 0.2083, "rl_loss": -1.9778, "step": 2160 }, { "acc": 0.0, "epoch": 1.47, "learning_rate": 2e-05, "loss": 0.1921, "rl_loss": -1.9141, "step": 2170 }, { "acc": 0.0, "epoch": 1.47, "learning_rate": 2e-05, "loss": 0.1998, "rl_loss": -1.1636, "step": 2180 }, { "acc": 0.0, "epoch": 1.48, "learning_rate": 2e-05, "loss": 0.2025, "rl_loss": -1.938, "step": 2190 }, { "acc": 0.0, "epoch": 1.49, "learning_rate": 2e-05, "loss": 0.2074, "rl_loss": -1.7501, "step": 2200 }, { "acc": 0.0, "epoch": 1.49, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -1.8256, "step": 2210 }, { "acc": 0.0, "epoch": 1.5, "learning_rate": 2e-05, "loss": 0.2075, "rl_loss": -2.5085, "step": 2220 }, { "acc": 0.0, "epoch": 1.51, "learning_rate": 2e-05, "loss": 0.1981, "rl_loss": -1.7291, "step": 2230 }, { "acc": 0.0, "epoch": 1.51, "learning_rate": 2e-05, "loss": 0.2093, "rl_loss": -1.3699, "step": 2240 }, { "acc": 0.0, "epoch": 1.52, "learning_rate": 2e-05, "loss": 0.1903, "rl_loss": -2.0905, "step": 2250 }, { "acc": 0.0, "epoch": 1.53, "learning_rate": 2e-05, "loss": 0.2085, "rl_loss": -2.188, "step": 2260 }, { "acc": 0.0, "epoch": 1.53, "learning_rate": 2e-05, "loss": 0.202, "rl_loss": -1.4118, "step": 2270 }, { "acc": 0.0, "epoch": 1.54, "learning_rate": 2e-05, "loss": 0.2117, "rl_loss": -1.8315, "step": 2280 }, { "acc": 0.0, "epoch": 1.55, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.5906, "step": 2290 }, { "acc": 0.0, "epoch": 1.56, "learning_rate": 2e-05, "loss": 0.2007, "rl_loss": -2.1578, "step": 2300 }, { "acc": 0.0, "epoch": 1.56, "learning_rate": 2e-05, "loss": 0.2098, "rl_loss": -1.6494, "step": 2310 }, { "acc": 0.0, "epoch": 1.57, "learning_rate": 2e-05, "loss": 0.2058, "rl_loss": -2.5481, "step": 2320 }, { "acc": 0.0, "epoch": 1.58, "learning_rate": 2e-05, "loss": 0.2149, "rl_loss": -2.0375, "step": 2330 }, { "acc": 0.0, "epoch": 1.58, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.1528, "step": 2340 }, { "acc": 0.0, "epoch": 1.59, "learning_rate": 2e-05, "loss": 0.2122, "rl_loss": -2.1616, "step": 2350 }, { "acc": 0.0, "epoch": 1.6, "learning_rate": 2e-05, "loss": 0.2069, "rl_loss": -1.7392, "step": 2360 }, { "acc": 0.0, "epoch": 1.6, "learning_rate": 2e-05, "loss": 0.2055, "rl_loss": -1.774, "step": 2370 }, { "acc": 0.0, "epoch": 1.61, "learning_rate": 2e-05, "loss": 0.2078, "rl_loss": -1.8271, "step": 2380 }, { "acc": 0.0, "epoch": 1.62, "learning_rate": 2e-05, "loss": 0.1955, "rl_loss": -1.9991, "step": 2390 }, { "acc": 0.0, "epoch": 1.62, "learning_rate": 2e-05, "loss": 0.2111, "rl_loss": -1.4246, "step": 2400 }, { "acc": 0.0, "epoch": 1.63, "learning_rate": 2e-05, "loss": 0.2151, "rl_loss": -1.4902, "step": 2410 }, { "acc": 0.0, "epoch": 1.64, "learning_rate": 2e-05, "loss": 0.1931, "rl_loss": -1.464, "step": 2420 }, { "acc": 0.0, "epoch": 1.64, "learning_rate": 2e-05, "loss": 0.2009, "rl_loss": -1.7579, "step": 2430 }, { "acc": 0.0, "epoch": 1.65, "learning_rate": 2e-05, "loss": 0.2026, "rl_loss": -1.7974, "step": 2440 }, { "acc": 0.0, "epoch": 1.66, "learning_rate": 2e-05, "loss": 0.1958, "rl_loss": -2.4374, "step": 2450 }, { "acc": 0.0, "epoch": 1.66, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -1.5141, "step": 2460 }, { "acc": 0.0, "epoch": 1.67, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -2.6986, "step": 2470 }, { "acc": 0.0, "epoch": 1.68, "learning_rate": 2e-05, "loss": 0.2092, "rl_loss": -2.2837, "step": 2480 }, { "acc": 0.0, "epoch": 1.68, "learning_rate": 2e-05, "loss": 0.2093, "rl_loss": -1.8279, "step": 2490 }, { "acc": 0.0, "epoch": 1.69, "learning_rate": 2e-05, "loss": 0.2151, "rl_loss": -1.1879, "step": 2500 }, { "acc": 0.0, "epoch": 1.7, "learning_rate": 2e-05, "loss": 0.1897, "rl_loss": -1.9185, "step": 2510 }, { "acc": 0.0, "epoch": 1.7, "learning_rate": 2e-05, "loss": 0.2137, "rl_loss": -2.0637, "step": 2520 }, { "acc": 0.0, "epoch": 1.71, "learning_rate": 2e-05, "loss": 0.1908, "rl_loss": -1.5085, "step": 2530 }, { "acc": 0.0, "epoch": 1.72, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -0.7273, "step": 2540 }, { "acc": 0.0, "epoch": 1.72, "learning_rate": 2e-05, "loss": 0.23, "rl_loss": -2.5821, "step": 2550 }, { "acc": 0.0, "epoch": 1.73, "learning_rate": 2e-05, "loss": 0.2124, "rl_loss": -1.4476, "step": 2560 }, { "acc": 0.0, "epoch": 1.74, "learning_rate": 2e-05, "loss": 0.1991, "rl_loss": -1.7364, "step": 2570 }, { "acc": 0.0, "epoch": 1.74, "learning_rate": 2e-05, "loss": 0.2125, "rl_loss": -1.4094, "step": 2580 }, { "acc": 0.0, "epoch": 1.75, "learning_rate": 2e-05, "loss": 0.2057, "rl_loss": -2.2976, "step": 2590 }, { "acc": 0.0, "epoch": 1.76, "learning_rate": 2e-05, "loss": 0.1983, "rl_loss": -2.9695, "step": 2600 }, { "acc": 0.0, "epoch": 1.76, "learning_rate": 2e-05, "loss": 0.2029, "rl_loss": -2.38, "step": 2610 }, { "acc": 0.0, "epoch": 1.77, "learning_rate": 2e-05, "loss": 0.2132, "rl_loss": -1.2959, "step": 2620 }, { "acc": 0.0, "epoch": 1.78, "learning_rate": 2e-05, "loss": 0.1939, "rl_loss": -1.0997, "step": 2630 }, { "acc": 0.0, "epoch": 1.78, "learning_rate": 2e-05, "loss": 0.2126, "rl_loss": -2.3755, "step": 2640 }, { "acc": 0.0, "epoch": 1.79, "learning_rate": 2e-05, "loss": 0.1884, "rl_loss": -2.5333, "step": 2650 }, { "acc": 0.0, "epoch": 1.8, "learning_rate": 2e-05, "loss": 0.2359, "rl_loss": -0.9339, "step": 2660 }, { "acc": 0.0, "epoch": 1.81, "learning_rate": 2e-05, "loss": 0.2073, "rl_loss": -1.9454, "step": 2670 }, { "acc": 0.0, "epoch": 1.81, "learning_rate": 2e-05, "loss": 0.1917, "rl_loss": -2.2108, "step": 2680 }, { "acc": 0.0, "epoch": 1.82, "learning_rate": 2e-05, "loss": 0.2194, "rl_loss": -2.1191, "step": 2690 }, { "acc": 0.0, "epoch": 1.83, "learning_rate": 2e-05, "loss": 0.2036, "rl_loss": -1.8452, "step": 2700 }, { "acc": 0.0, "epoch": 1.83, "learning_rate": 2e-05, "loss": 0.2073, "rl_loss": -1.5664, "step": 2710 }, { "acc": 0.0, "epoch": 1.84, "learning_rate": 2e-05, "loss": 0.2132, "rl_loss": -0.9847, "step": 2720 }, { "acc": 0.0, "epoch": 1.85, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -1.9664, "step": 2730 }, { "acc": 0.0, "epoch": 1.85, "learning_rate": 2e-05, "loss": 0.1909, "rl_loss": -2.4553, "step": 2740 }, { "acc": 0.0, "epoch": 1.86, "learning_rate": 2e-05, "loss": 0.1953, "rl_loss": -2.4401, "step": 2750 }, { "acc": 0.0, "epoch": 1.87, "learning_rate": 2e-05, "loss": 0.2102, "rl_loss": -2.1955, "step": 2760 }, { "acc": 0.0, "epoch": 1.87, "learning_rate": 2e-05, "loss": 0.2021, "rl_loss": -2.6245, "step": 2770 }, { "acc": 0.0, "epoch": 1.88, "learning_rate": 2e-05, "loss": 0.2054, "rl_loss": -2.2406, "step": 2780 }, { "acc": 0.0, "epoch": 1.89, "learning_rate": 2e-05, "loss": 0.2026, "rl_loss": -2.0128, "step": 2790 }, { "acc": 0.0, "epoch": 1.89, "learning_rate": 2e-05, "loss": 0.2017, "rl_loss": -2.3864, "step": 2800 }, { "acc": 0.0, "epoch": 1.9, "learning_rate": 2e-05, "loss": 0.2098, "rl_loss": -1.575, "step": 2810 }, { "acc": 0.0, "epoch": 1.91, "learning_rate": 2e-05, "loss": 0.1974, "rl_loss": -2.1752, "step": 2820 }, { "acc": 0.0, "epoch": 1.91, "learning_rate": 2e-05, "loss": 0.2072, "rl_loss": -1.9743, "step": 2830 }, { "acc": 0.0, "epoch": 1.92, "learning_rate": 2e-05, "loss": 0.2081, "rl_loss": -1.5731, "step": 2840 }, { "acc": 0.0, "epoch": 1.93, "learning_rate": 2e-05, "loss": 0.2095, "rl_loss": -2.2315, "step": 2850 }, { "acc": 0.0, "epoch": 1.93, "learning_rate": 2e-05, "loss": 0.1961, "rl_loss": -1.9078, "step": 2860 }, { "acc": 0.0, "epoch": 1.94, "learning_rate": 2e-05, "loss": 0.2119, "rl_loss": -2.1196, "step": 2870 }, { "acc": 0.0, "epoch": 1.95, "learning_rate": 2e-05, "loss": 0.2177, "rl_loss": -1.7101, "step": 2880 }, { "acc": 0.0, "epoch": 1.95, "learning_rate": 2e-05, "loss": 0.2063, "rl_loss": -2.7389, "step": 2890 }, { "acc": 0.0, "epoch": 1.96, "learning_rate": 2e-05, "loss": 0.1838, "rl_loss": -1.7766, "step": 2900 }, { "acc": 0.0, "epoch": 1.97, "learning_rate": 2e-05, "loss": 0.2009, "rl_loss": -1.9517, "step": 2910 }, { "acc": 0.0, "epoch": 1.97, "learning_rate": 2e-05, "loss": 0.2053, "rl_loss": -1.8588, "step": 2920 }, { "acc": 0.0, "epoch": 1.98, "learning_rate": 2e-05, "loss": 0.202, "rl_loss": -1.2567, "step": 2930 }, { "acc": 0.0, "epoch": 1.99, "learning_rate": 2e-05, "loss": 0.2106, "rl_loss": -1.6925, "step": 2940 }, { "acc": 0.0, "epoch": 1.99, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -1.3853, "step": 2950 }, { "acc": 0.0, "epoch": 2.0, "learning_rate": 2e-05, "loss": 0.2053, "rl_loss": -2.0613, "step": 2960 }, { "acc": 0.0, "epoch": 2.01, "learning_rate": 2e-05, "loss": 0.2055, "rl_loss": -1.7382, "step": 2970 }, { "acc": 0.0, "epoch": 2.01, "learning_rate": 2e-05, "loss": 0.2133, "rl_loss": -2.3319, "step": 2980 }, { "acc": 0.0, "epoch": 2.02, "learning_rate": 2e-05, "loss": 0.2006, "rl_loss": -2.7122, "step": 2990 }, { "acc": 0.0, "epoch": 2.03, "learning_rate": 2e-05, "loss": 0.2211, "rl_loss": -1.998, "step": 3000 }, { "acc": 0.0, "epoch": 2.04, "learning_rate": 2e-05, "loss": 0.1998, "rl_loss": -1.9436, "step": 3010 }, { "acc": 0.0, "epoch": 2.04, "learning_rate": 2e-05, "loss": 0.2216, "rl_loss": -1.3867, "step": 3020 }, { "acc": 0.0, "epoch": 2.05, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.5493, "step": 3030 }, { "acc": 0.0, "epoch": 2.06, "learning_rate": 2e-05, "loss": 0.2019, "rl_loss": -2.5932, "step": 3040 }, { "acc": 0.0, "epoch": 2.06, "learning_rate": 2e-05, "loss": 0.202, "rl_loss": -1.9979, "step": 3050 }, { "acc": 0.0, "epoch": 2.07, "learning_rate": 2e-05, "loss": 0.2083, "rl_loss": -2.15, "step": 3060 }, { "acc": 0.0, "epoch": 2.08, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -2.0852, "step": 3070 }, { "acc": 0.0, "epoch": 2.08, "learning_rate": 2e-05, "loss": 0.2024, "rl_loss": -1.9339, "step": 3080 }, { "acc": 0.0, "epoch": 2.09, "learning_rate": 2e-05, "loss": 0.2092, "rl_loss": -1.7885, "step": 3090 }, { "acc": 0.0, "epoch": 2.1, "learning_rate": 2e-05, "loss": 0.2021, "rl_loss": -1.8393, "step": 3100 }, { "acc": 0.0, "epoch": 2.1, "learning_rate": 2e-05, "loss": 0.1984, "rl_loss": -1.9165, "step": 3110 }, { "acc": 0.0, "epoch": 2.11, "learning_rate": 2e-05, "loss": 0.2067, "rl_loss": -1.8556, "step": 3120 }, { "acc": 0.0, "epoch": 2.12, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.7716, "step": 3130 }, { "acc": 0.0, "epoch": 2.12, "learning_rate": 2e-05, "loss": 0.2004, "rl_loss": -2.4081, "step": 3140 }, { "acc": 0.0, "epoch": 2.13, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -2.8044, "step": 3150 }, { "acc": 0.0, "epoch": 2.14, "learning_rate": 2e-05, "loss": 0.2009, "rl_loss": -1.3568, "step": 3160 }, { "acc": 0.0, "epoch": 2.14, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -2.2277, "step": 3170 }, { "acc": 0.0, "epoch": 2.15, "learning_rate": 2e-05, "loss": 0.2034, "rl_loss": -2.1072, "step": 3180 }, { "acc": 0.0, "epoch": 2.16, "learning_rate": 2e-05, "loss": 0.2107, "rl_loss": -1.8421, "step": 3190 }, { "acc": 0.0, "epoch": 2.16, "learning_rate": 2e-05, "loss": 0.1981, "rl_loss": -1.8021, "step": 3200 }, { "acc": 0.0, "epoch": 2.17, "learning_rate": 2e-05, "loss": 0.2024, "rl_loss": -1.8409, "step": 3210 }, { "acc": 0.0, "epoch": 2.18, "learning_rate": 2e-05, "loss": 0.2078, "rl_loss": -2.4022, "step": 3220 }, { "acc": 0.0, "epoch": 2.18, "learning_rate": 2e-05, "loss": 0.1977, "rl_loss": -2.2737, "step": 3230 }, { "acc": 0.0, "epoch": 2.19, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -2.4403, "step": 3240 }, { "acc": 0.0, "epoch": 2.2, "learning_rate": 2e-05, "loss": 0.227, "rl_loss": -1.9865, "step": 3250 }, { "acc": 0.0, "epoch": 2.2, "learning_rate": 2e-05, "loss": 0.2074, "rl_loss": -1.817, "step": 3260 }, { "acc": 0.0, "epoch": 2.21, "learning_rate": 2e-05, "loss": 0.1881, "rl_loss": -1.8318, "step": 3270 }, { "acc": 0.0, "epoch": 2.22, "learning_rate": 2e-05, "loss": 0.2101, "rl_loss": -1.5954, "step": 3280 }, { "acc": 0.0, "epoch": 2.22, "learning_rate": 2e-05, "loss": 0.2022, "rl_loss": -1.576, "step": 3290 }, { "acc": 0.0, "epoch": 2.23, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.8416, "step": 3300 }, { "acc": 0.0, "epoch": 2.24, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -1.8762, "step": 3310 }, { "acc": 0.0, "epoch": 2.24, "learning_rate": 2e-05, "loss": 0.2051, "rl_loss": -1.9596, "step": 3320 }, { "acc": 0.0, "epoch": 2.25, "learning_rate": 2e-05, "loss": 0.2029, "rl_loss": -1.7614, "step": 3330 }, { "acc": 0.0, "epoch": 2.26, "learning_rate": 2e-05, "loss": 0.2035, "rl_loss": -2.8926, "step": 3340 }, { "acc": 0.0, "epoch": 2.27, "learning_rate": 2e-05, "loss": 0.208, "rl_loss": -2.1869, "step": 3350 }, { "acc": 0.0, "epoch": 2.27, "learning_rate": 2e-05, "loss": 0.1989, "rl_loss": -1.9421, "step": 3360 }, { "acc": 0.0, "epoch": 2.28, "learning_rate": 2e-05, "loss": 0.198, "rl_loss": -1.6054, "step": 3370 }, { "acc": 0.0, "epoch": 2.29, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -1.8191, "step": 3380 }, { "acc": 0.0, "epoch": 2.29, "learning_rate": 2e-05, "loss": 0.2013, "rl_loss": -1.6489, "step": 3390 }, { "acc": 0.0, "epoch": 2.3, "learning_rate": 2e-05, "loss": 0.1952, "rl_loss": -2.7371, "step": 3400 }, { "acc": 0.0, "epoch": 2.31, "learning_rate": 2e-05, "loss": 0.2059, "rl_loss": -2.1396, "step": 3410 }, { "acc": 0.0, "epoch": 2.31, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -1.7093, "step": 3420 }, { "acc": 0.0, "epoch": 2.32, "learning_rate": 2e-05, "loss": 0.2075, "rl_loss": -2.3022, "step": 3430 }, { "acc": 0.0, "epoch": 2.33, "learning_rate": 2e-05, "loss": 0.2064, "rl_loss": -1.1935, "step": 3440 }, { "acc": 0.0, "epoch": 2.33, "learning_rate": 2e-05, "loss": 0.1925, "rl_loss": -1.9742, "step": 3450 }, { "acc": 0.0, "epoch": 2.34, "learning_rate": 2e-05, "loss": 0.1993, "rl_loss": -2.668, "step": 3460 }, { "acc": 0.0, "epoch": 2.35, "learning_rate": 2e-05, "loss": 0.2001, "rl_loss": -2.6062, "step": 3470 }, { "acc": 0.0, "epoch": 2.35, "learning_rate": 2e-05, "loss": 0.2053, "rl_loss": -2.2569, "step": 3480 }, { "acc": 0.0, "epoch": 2.36, "learning_rate": 2e-05, "loss": 0.198, "rl_loss": -2.1492, "step": 3490 }, { "acc": 0.0, "epoch": 2.37, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -1.5569, "step": 3500 }, { "acc": 0.0, "epoch": 2.37, "learning_rate": 2e-05, "loss": 0.208, "rl_loss": -1.8926, "step": 3510 }, { "acc": 0.0, "epoch": 2.38, "learning_rate": 2e-05, "loss": 0.1992, "rl_loss": -2.1013, "step": 3520 }, { "acc": 0.0, "epoch": 2.39, "learning_rate": 2e-05, "loss": 0.1944, "rl_loss": -2.5324, "step": 3530 }, { "acc": 0.0, "epoch": 2.39, "learning_rate": 2e-05, "loss": 0.197, "rl_loss": -1.5396, "step": 3540 }, { "acc": 0.0, "epoch": 2.4, "learning_rate": 2e-05, "loss": 0.2125, "rl_loss": -2.4613, "step": 3550 }, { "acc": 0.0, "epoch": 2.41, "learning_rate": 2e-05, "loss": 0.191, "rl_loss": -1.7846, "step": 3560 }, { "acc": 0.0, "epoch": 2.41, "learning_rate": 2e-05, "loss": 0.2127, "rl_loss": -2.0609, "step": 3570 }, { "acc": 0.0, "epoch": 2.42, "learning_rate": 2e-05, "loss": 0.2128, "rl_loss": -2.7034, "step": 3580 }, { "acc": 0.0, "epoch": 2.43, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -1.9699, "step": 3590 }, { "acc": 0.0, "epoch": 2.43, "learning_rate": 2e-05, "loss": 0.1883, "rl_loss": -3.29, "step": 3600 }, { "acc": 0.0, "epoch": 2.44, "learning_rate": 2e-05, "loss": 0.1962, "rl_loss": -2.5387, "step": 3610 }, { "acc": 0.0, "epoch": 2.45, "learning_rate": 2e-05, "loss": 0.1835, "rl_loss": -2.3881, "step": 3620 }, { "acc": 0.0, "epoch": 2.45, "learning_rate": 2e-05, "loss": 0.2099, "rl_loss": -1.8734, "step": 3630 }, { "acc": 0.0, "epoch": 2.46, "learning_rate": 2e-05, "loss": 0.1859, "rl_loss": -1.7912, "step": 3640 }, { "acc": 0.0, "epoch": 2.47, "learning_rate": 2e-05, "loss": 0.2066, "rl_loss": -2.1104, "step": 3650 }, { "acc": 0.0, "epoch": 2.47, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -1.8194, "step": 3660 }, { "acc": 0.0, "epoch": 2.48, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -1.8905, "step": 3670 }, { "acc": 0.0, "epoch": 2.49, "learning_rate": 2e-05, "loss": 0.2027, "rl_loss": -1.6544, "step": 3680 }, { "acc": 0.0, "epoch": 2.49, "learning_rate": 2e-05, "loss": 0.2007, "rl_loss": -1.959, "step": 3690 }, { "acc": 0.0, "epoch": 2.5, "learning_rate": 2e-05, "loss": 0.209, "rl_loss": -2.0811, "step": 3700 }, { "acc": 0.0, "epoch": 2.51, "learning_rate": 2e-05, "loss": 0.1894, "rl_loss": -2.9403, "step": 3710 }, { "acc": 0.0, "epoch": 2.52, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -2.178, "step": 3720 }, { "acc": 0.0, "epoch": 2.52, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -1.8048, "step": 3730 }, { "acc": 0.0, "epoch": 2.53, "learning_rate": 2e-05, "loss": 0.1964, "rl_loss": -1.5694, "step": 3740 }, { "acc": 0.0, "epoch": 2.54, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -1.6905, "step": 3750 }, { "acc": 0.0, "epoch": 2.54, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -1.5757, "step": 3760 }, { "acc": 0.0, "epoch": 2.55, "learning_rate": 2e-05, "loss": 0.1993, "rl_loss": -1.7275, "step": 3770 }, { "acc": 0.0, "epoch": 2.56, "learning_rate": 2e-05, "loss": 0.2066, "rl_loss": -1.975, "step": 3780 }, { "acc": 0.0, "epoch": 2.56, "learning_rate": 2e-05, "loss": 0.1983, "rl_loss": -1.7712, "step": 3790 }, { "acc": 0.0, "epoch": 2.57, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -2.3119, "step": 3800 }, { "acc": 0.0, "epoch": 2.58, "learning_rate": 2e-05, "loss": 0.2031, "rl_loss": -2.1734, "step": 3810 }, { "acc": 0.0, "epoch": 2.58, "learning_rate": 2e-05, "loss": 0.1924, "rl_loss": -1.2348, "step": 3820 }, { "acc": 0.0, "epoch": 2.59, "learning_rate": 2e-05, "loss": 0.1948, "rl_loss": -2.5681, "step": 3830 }, { "acc": 0.0, "epoch": 2.6, "learning_rate": 2e-05, "loss": 0.2122, "rl_loss": -2.1672, "step": 3840 }, { "acc": 0.0, "epoch": 2.6, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -1.8657, "step": 3850 }, { "acc": 0.0, "epoch": 2.61, "learning_rate": 2e-05, "loss": 0.1936, "rl_loss": -2.3622, "step": 3860 }, { "acc": 0.0, "epoch": 2.62, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -1.8407, "step": 3870 }, { "acc": 0.0, "epoch": 2.62, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -1.9087, "step": 3880 }, { "acc": 0.0, "epoch": 2.63, "learning_rate": 2e-05, "loss": 0.1936, "rl_loss": -2.891, "step": 3890 }, { "acc": 0.0, "epoch": 2.64, "learning_rate": 2e-05, "loss": 0.2082, "rl_loss": -2.6062, "step": 3900 }, { "acc": 0.0, "epoch": 2.64, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -2.2284, "step": 3910 }, { "acc": 0.0, "epoch": 2.65, "learning_rate": 2e-05, "loss": 0.2109, "rl_loss": -2.3717, "step": 3920 }, { "acc": 0.0, "epoch": 2.66, "learning_rate": 2e-05, "loss": 0.2131, "rl_loss": -2.5808, "step": 3930 }, { "acc": 0.0, "epoch": 2.66, "learning_rate": 2e-05, "loss": 0.1939, "rl_loss": -1.9021, "step": 3940 }, { "acc": 0.0, "epoch": 2.67, "learning_rate": 2e-05, "loss": 0.2229, "rl_loss": -1.9056, "step": 3950 }, { "acc": 0.0, "epoch": 2.68, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -2.1623, "step": 3960 }, { "acc": 0.0, "epoch": 2.68, "learning_rate": 2e-05, "loss": 0.2074, "rl_loss": -2.3669, "step": 3970 }, { "acc": 0.0, "epoch": 2.69, "learning_rate": 2e-05, "loss": 0.2043, "rl_loss": -2.2813, "step": 3980 }, { "acc": 0.0, "epoch": 2.7, "learning_rate": 2e-05, "loss": 0.1899, "rl_loss": -1.4695, "step": 3990 }, { "acc": 0.0, "epoch": 2.7, "learning_rate": 2e-05, "loss": 0.2026, "rl_loss": -2.2627, "step": 4000 }, { "acc": 0.0, "epoch": 2.71, "learning_rate": 2e-05, "loss": 0.2088, "rl_loss": -2.0507, "step": 4010 }, { "acc": 0.0, "epoch": 2.72, "learning_rate": 2e-05, "loss": 0.2087, "rl_loss": -1.9129, "step": 4020 }, { "acc": 0.0, "epoch": 2.72, "learning_rate": 2e-05, "loss": 0.2018, "rl_loss": -2.0182, "step": 4030 }, { "acc": 0.0, "epoch": 2.73, "learning_rate": 2e-05, "loss": 0.1958, "rl_loss": -2.1713, "step": 4040 }, { "acc": 0.0, "epoch": 2.74, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -1.4516, "step": 4050 }, { "acc": 0.0, "epoch": 2.75, "learning_rate": 2e-05, "loss": 0.1882, "rl_loss": -1.8775, "step": 4060 }, { "acc": 0.0, "epoch": 2.75, "learning_rate": 2e-05, "loss": 0.1863, "rl_loss": -1.5827, "step": 4070 }, { "acc": 0.0, "epoch": 2.76, "learning_rate": 2e-05, "loss": 0.2037, "rl_loss": -2.2443, "step": 4080 }, { "acc": 0.0, "epoch": 2.77, "learning_rate": 2e-05, "loss": 0.1878, "rl_loss": -1.7416, "step": 4090 }, { "acc": 0.0, "epoch": 2.77, "learning_rate": 2e-05, "loss": 0.2031, "rl_loss": -2.7883, "step": 4100 }, { "acc": 0.0, "epoch": 2.78, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -2.31, "step": 4110 }, { "acc": 0.0, "epoch": 2.79, "learning_rate": 2e-05, "loss": 0.2176, "rl_loss": -1.5883, "step": 4120 }, { "acc": 0.0, "epoch": 2.79, "learning_rate": 2e-05, "loss": 0.2036, "rl_loss": -2.4506, "step": 4130 }, { "acc": 0.0, "epoch": 2.8, "learning_rate": 2e-05, "loss": 0.1972, "rl_loss": -1.9433, "step": 4140 }, { "acc": 0.0, "epoch": 2.81, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -1.8733, "step": 4150 }, { "acc": 0.0, "epoch": 2.81, "learning_rate": 2e-05, "loss": 0.2023, "rl_loss": -2.2307, "step": 4160 }, { "acc": 0.0, "epoch": 2.82, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -2.1388, "step": 4170 }, { "acc": 0.0, "epoch": 2.83, "learning_rate": 2e-05, "loss": 0.1941, "rl_loss": -3.0965, "step": 4180 }, { "acc": 0.0, "epoch": 2.83, "learning_rate": 2e-05, "loss": 0.198, "rl_loss": -2.1522, "step": 4190 }, { "acc": 0.0, "epoch": 2.84, "learning_rate": 2e-05, "loss": 0.1936, "rl_loss": -2.4873, "step": 4200 }, { "acc": 0.0, "epoch": 2.85, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -2.1431, "step": 4210 }, { "acc": 0.0, "epoch": 2.85, "learning_rate": 2e-05, "loss": 0.1983, "rl_loss": -2.2953, "step": 4220 }, { "acc": 0.0, "epoch": 2.86, "learning_rate": 2e-05, "loss": 0.1993, "rl_loss": -1.5354, "step": 4230 }, { "acc": 0.0, "epoch": 2.87, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.7655, "step": 4240 }, { "acc": 0.0, "epoch": 2.87, "learning_rate": 2e-05, "loss": 0.2001, "rl_loss": -2.3103, "step": 4250 }, { "acc": 0.0, "epoch": 2.88, "learning_rate": 2e-05, "loss": 0.1917, "rl_loss": -1.8755, "step": 4260 }, { "acc": 0.0, "epoch": 2.89, "learning_rate": 2e-05, "loss": 0.1924, "rl_loss": -2.3443, "step": 4270 }, { "acc": 0.0, "epoch": 2.89, "learning_rate": 2e-05, "loss": 0.2158, "rl_loss": -2.5232, "step": 4280 }, { "acc": 0.0, "epoch": 2.9, "learning_rate": 2e-05, "loss": 0.1988, "rl_loss": -1.7692, "step": 4290 }, { "acc": 0.0, "epoch": 2.91, "learning_rate": 2e-05, "loss": 0.3165, "rl_loss": -1.9035, "step": 4300 }, { "acc": 0.0, "epoch": 2.91, "learning_rate": 2e-05, "loss": 0.2011, "rl_loss": -2.7394, "step": 4310 }, { "acc": 0.0, "epoch": 2.92, "learning_rate": 2e-05, "loss": 0.1981, "rl_loss": -1.9746, "step": 4320 }, { "acc": 0.0, "epoch": 2.93, "learning_rate": 2e-05, "loss": 0.1959, "rl_loss": -2.2058, "step": 4330 }, { "acc": 0.0, "epoch": 2.93, "learning_rate": 2e-05, "loss": 0.1932, "rl_loss": -1.9834, "step": 4340 }, { "acc": 0.0, "epoch": 2.94, "learning_rate": 2e-05, "loss": 0.1991, "rl_loss": -2.3905, "step": 4350 }, { "acc": 0.0, "epoch": 2.95, "learning_rate": 2e-05, "loss": 0.2034, "rl_loss": -2.4909, "step": 4360 }, { "acc": 0.0, "epoch": 2.95, "learning_rate": 2e-05, "loss": 0.1958, "rl_loss": -2.0517, "step": 4370 }, { "acc": 0.0, "epoch": 2.96, "learning_rate": 2e-05, "loss": 0.2007, "rl_loss": -2.0089, "step": 4380 }, { "acc": 0.0, "epoch": 2.97, "learning_rate": 2e-05, "loss": 0.1929, "rl_loss": -1.9276, "step": 4390 }, { "acc": 0.0, "epoch": 2.97, "learning_rate": 2e-05, "loss": 0.1894, "rl_loss": -3.0913, "step": 4400 }, { "acc": 0.0, "epoch": 2.98, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -1.6282, "step": 4410 }, { "acc": 0.0, "epoch": 2.99, "learning_rate": 2e-05, "loss": 0.2063, "rl_loss": -1.9378, "step": 4420 }, { "acc": 0.0, "epoch": 3.0, "learning_rate": 2e-05, "loss": 0.2003, "rl_loss": -2.3594, "step": 4430 }, { "acc": 0.0, "epoch": 3.0, "learning_rate": 2e-05, "loss": 0.2064, "rl_loss": -2.0177, "step": 4440 }, { "acc": 0.0, "epoch": 3.01, "learning_rate": 2e-05, "loss": 0.199, "rl_loss": -2.4657, "step": 4450 }, { "acc": 0.0, "epoch": 3.02, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -1.314, "step": 4460 }, { "acc": 0.0, "epoch": 3.02, "learning_rate": 2e-05, "loss": 0.1979, "rl_loss": -2.4745, "step": 4470 }, { "acc": 0.0, "epoch": 3.03, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -3.4262, "step": 4480 }, { "acc": 0.0, "epoch": 3.04, "learning_rate": 2e-05, "loss": 0.2085, "rl_loss": -1.7261, "step": 4490 }, { "acc": 0.0, "epoch": 3.04, "learning_rate": 2e-05, "loss": 0.1807, "rl_loss": -2.2682, "step": 4500 }, { "acc": 0.0, "epoch": 3.05, "learning_rate": 2e-05, "loss": 0.2019, "rl_loss": -1.8529, "step": 4510 }, { "acc": 0.0, "epoch": 3.06, "learning_rate": 2e-05, "loss": 0.2116, "rl_loss": -2.4143, "step": 4520 }, { "acc": 0.0, "epoch": 3.06, "learning_rate": 2e-05, "loss": 0.1982, "rl_loss": -2.91, "step": 4530 }, { "acc": 0.0, "epoch": 3.07, "learning_rate": 2e-05, "loss": 0.2033, "rl_loss": -2.7137, "step": 4540 }, { "acc": 0.0, "epoch": 3.08, "learning_rate": 2e-05, "loss": 0.1953, "rl_loss": -2.8363, "step": 4550 }, { "acc": 0.0, "epoch": 3.08, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -1.6648, "step": 4560 }, { "acc": 0.0, "epoch": 3.09, "learning_rate": 2e-05, "loss": 0.2003, "rl_loss": -2.9286, "step": 4570 }, { "acc": 0.0, "epoch": 3.1, "learning_rate": 2e-05, "loss": 0.1844, "rl_loss": -1.7505, "step": 4580 }, { "acc": 0.0, "epoch": 3.1, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -2.6774, "step": 4590 }, { "acc": 0.0, "epoch": 3.11, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -2.055, "step": 4600 }, { "acc": 0.0, "epoch": 3.12, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -2.5754, "step": 4610 }, { "acc": 0.0, "epoch": 3.12, "learning_rate": 2e-05, "loss": 0.1938, "rl_loss": -1.89, "step": 4620 }, { "acc": 0.0, "epoch": 3.13, "learning_rate": 2e-05, "loss": 0.2016, "rl_loss": -1.8629, "step": 4630 }, { "acc": 0.0, "epoch": 3.14, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -3.426, "step": 4640 }, { "acc": 0.0, "epoch": 3.14, "learning_rate": 2e-05, "loss": 0.1959, "rl_loss": -2.0629, "step": 4650 }, { "acc": 0.0, "epoch": 3.15, "learning_rate": 2e-05, "loss": 0.1881, "rl_loss": -2.1224, "step": 4660 }, { "acc": 0.0, "epoch": 3.16, "learning_rate": 2e-05, "loss": 0.1932, "rl_loss": -2.3354, "step": 4670 }, { "acc": 0.0, "epoch": 3.16, "learning_rate": 2e-05, "loss": 0.1964, "rl_loss": -2.3875, "step": 4680 }, { "acc": 0.0, "epoch": 3.17, "learning_rate": 2e-05, "loss": 0.2122, "rl_loss": -1.7637, "step": 4690 }, { "acc": 0.0, "epoch": 3.18, "learning_rate": 2e-05, "loss": 0.2009, "rl_loss": -2.158, "step": 4700 }, { "acc": 0.0, "epoch": 3.18, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -2.6828, "step": 4710 }, { "acc": 0.0, "epoch": 3.19, "learning_rate": 2e-05, "loss": 0.1873, "rl_loss": -2.1975, "step": 4720 }, { "acc": 0.0, "epoch": 3.2, "learning_rate": 2e-05, "loss": 0.2, "rl_loss": -2.7453, "step": 4730 }, { "acc": 0.0, "epoch": 3.2, "learning_rate": 2e-05, "loss": 0.1824, "rl_loss": -2.0108, "step": 4740 }, { "acc": 0.0, "epoch": 3.21, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -2.7777, "step": 4750 }, { "acc": 0.0, "epoch": 3.22, "learning_rate": 2e-05, "loss": 0.1881, "rl_loss": -2.0824, "step": 4760 }, { "acc": 0.0, "epoch": 3.23, "learning_rate": 2e-05, "loss": 0.1935, "rl_loss": -1.9981, "step": 4770 }, { "acc": 0.0, "epoch": 3.23, "learning_rate": 2e-05, "loss": 0.2005, "rl_loss": -2.8281, "step": 4780 }, { "acc": 0.0, "epoch": 3.24, "learning_rate": 2e-05, "loss": 0.2031, "rl_loss": -2.7581, "step": 4790 }, { "acc": 0.0, "epoch": 3.25, "learning_rate": 2e-05, "loss": 0.1962, "rl_loss": -1.7212, "step": 4800 }, { "acc": 0.0, "epoch": 3.25, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -1.7026, "step": 4810 }, { "acc": 0.0, "epoch": 3.26, "learning_rate": 2e-05, "loss": 0.2015, "rl_loss": -3.2029, "step": 4820 }, { "acc": 0.0, "epoch": 3.27, "learning_rate": 2e-05, "loss": 0.2018, "rl_loss": -2.1036, "step": 4830 }, { "acc": 0.0, "epoch": 3.27, "learning_rate": 2e-05, "loss": 0.2107, "rl_loss": -2.7463, "step": 4840 }, { "acc": 0.0, "epoch": 3.28, "learning_rate": 2e-05, "loss": 0.1863, "rl_loss": -3.1595, "step": 4850 }, { "acc": 0.0, "epoch": 3.29, "learning_rate": 2e-05, "loss": 0.1955, "rl_loss": -2.1052, "step": 4860 }, { "acc": 0.0, "epoch": 3.29, "learning_rate": 2e-05, "loss": 0.1898, "rl_loss": -2.8484, "step": 4870 }, { "acc": 0.0, "epoch": 3.3, "learning_rate": 2e-05, "loss": 0.198, "rl_loss": -2.4653, "step": 4880 }, { "acc": 0.0, "epoch": 3.31, "learning_rate": 2e-05, "loss": 0.1896, "rl_loss": -1.8699, "step": 4890 }, { "acc": 0.0, "epoch": 3.31, "learning_rate": 2e-05, "loss": 0.1998, "rl_loss": -2.1023, "step": 4900 }, { "acc": 0.0, "epoch": 3.32, "learning_rate": 2e-05, "loss": 0.1944, "rl_loss": -2.4873, "step": 4910 }, { "acc": 0.0, "epoch": 3.33, "learning_rate": 2e-05, "loss": 0.3689, "rl_loss": -1.8524, "step": 4920 }, { "acc": 0.0, "epoch": 3.33, "learning_rate": 2e-05, "loss": 0.2047, "rl_loss": -1.9567, "step": 4930 }, { "acc": 0.0, "epoch": 3.34, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -2.6545, "step": 4940 }, { "acc": 0.0, "epoch": 3.35, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -2.6681, "step": 4950 }, { "acc": 0.0, "epoch": 3.35, "learning_rate": 2e-05, "loss": 0.203, "rl_loss": -2.9652, "step": 4960 }, { "acc": 0.0, "epoch": 3.36, "learning_rate": 2e-05, "loss": 0.2177, "rl_loss": -1.4795, "step": 4970 }, { "acc": 0.0, "epoch": 3.37, "learning_rate": 2e-05, "loss": 0.1979, "rl_loss": -1.8465, "step": 4980 }, { "acc": 0.0, "epoch": 3.37, "learning_rate": 2e-05, "loss": 0.1864, "rl_loss": -1.8854, "step": 4990 }, { "acc": 0.0, "epoch": 3.38, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -2.5208, "step": 5000 }, { "acc": 0.0, "epoch": 3.39, "learning_rate": 2e-05, "loss": 0.1923, "rl_loss": -1.9017, "step": 5010 }, { "acc": 0.0, "epoch": 3.39, "learning_rate": 2e-05, "loss": 0.191, "rl_loss": -2.6415, "step": 5020 }, { "acc": 0.0, "epoch": 3.4, "learning_rate": 2e-05, "loss": 0.2011, "rl_loss": -3.6356, "step": 5030 }, { "acc": 0.0, "epoch": 3.41, "learning_rate": 2e-05, "loss": 0.1982, "rl_loss": -2.2966, "step": 5040 }, { "acc": 0.0, "epoch": 3.41, "learning_rate": 2e-05, "loss": 0.2051, "rl_loss": -1.2086, "step": 5050 }, { "acc": 0.0, "epoch": 3.42, "learning_rate": 2e-05, "loss": 0.2024, "rl_loss": -2.5252, "step": 5060 }, { "acc": 0.0, "epoch": 3.43, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -2.3811, "step": 5070 }, { "acc": 0.0, "epoch": 3.43, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -2.3807, "step": 5080 }, { "acc": 0.0, "epoch": 3.44, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.1399, "step": 5090 }, { "acc": 0.0, "epoch": 3.45, "learning_rate": 2e-05, "loss": 0.1945, "rl_loss": -1.7478, "step": 5100 }, { "acc": 0.0, "epoch": 3.46, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -2.4084, "step": 5110 }, { "acc": 0.0, "epoch": 3.46, "learning_rate": 2e-05, "loss": 0.1938, "rl_loss": -2.5162, "step": 5120 }, { "acc": 0.0, "epoch": 3.47, "learning_rate": 2e-05, "loss": 0.2075, "rl_loss": -2.4883, "step": 5130 }, { "acc": 0.0, "epoch": 3.48, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -1.5761, "step": 5140 }, { "acc": 0.0, "epoch": 3.48, "learning_rate": 2e-05, "loss": 0.1972, "rl_loss": -1.7353, "step": 5150 }, { "acc": 0.0, "epoch": 3.49, "learning_rate": 2e-05, "loss": 0.2057, "rl_loss": -2.8966, "step": 5160 }, { "acc": 0.0, "epoch": 3.5, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -3.1249, "step": 5170 }, { "acc": 0.0, "epoch": 3.5, "learning_rate": 2e-05, "loss": 0.1969, "rl_loss": -2.3385, "step": 5180 }, { "acc": 0.0, "epoch": 3.51, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -1.7078, "step": 5190 }, { "acc": 0.0, "epoch": 3.52, "learning_rate": 2e-05, "loss": 0.195, "rl_loss": -1.6956, "step": 5200 }, { "acc": 0.0, "epoch": 3.52, "learning_rate": 2e-05, "loss": 0.2005, "rl_loss": -2.2946, "step": 5210 }, { "acc": 0.0, "epoch": 3.53, "learning_rate": 2e-05, "loss": 0.1931, "rl_loss": -2.2494, "step": 5220 }, { "acc": 0.0, "epoch": 3.54, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -2.1455, "step": 5230 }, { "acc": 0.0, "epoch": 3.54, "learning_rate": 2e-05, "loss": 0.2165, "rl_loss": -2.3963, "step": 5240 }, { "acc": 0.0, "epoch": 3.55, "learning_rate": 2e-05, "loss": 0.1964, "rl_loss": -2.5725, "step": 5250 }, { "acc": 0.0, "epoch": 3.56, "learning_rate": 2e-05, "loss": 0.1889, "rl_loss": -2.22, "step": 5260 }, { "acc": 0.0, "epoch": 3.56, "learning_rate": 2e-05, "loss": 0.1997, "rl_loss": -2.3766, "step": 5270 }, { "acc": 0.0, "epoch": 3.57, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -2.6519, "step": 5280 }, { "acc": 0.0, "epoch": 3.58, "learning_rate": 2e-05, "loss": 0.2079, "rl_loss": -1.5913, "step": 5290 }, { "acc": 0.0, "epoch": 3.58, "learning_rate": 2e-05, "loss": 0.181, "rl_loss": -2.7583, "step": 5300 }, { "acc": 0.0, "epoch": 3.59, "learning_rate": 2e-05, "loss": 0.1887, "rl_loss": -2.4896, "step": 5310 }, { "acc": 0.0, "epoch": 3.6, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -2.263, "step": 5320 }, { "acc": 0.0, "epoch": 3.6, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -2.1162, "step": 5330 }, { "acc": 0.0, "epoch": 3.61, "learning_rate": 2e-05, "loss": 0.1981, "rl_loss": -1.9419, "step": 5340 }, { "acc": 0.0, "epoch": 3.62, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -2.951, "step": 5350 }, { "acc": 0.0, "epoch": 3.62, "learning_rate": 2e-05, "loss": 0.1941, "rl_loss": -2.3458, "step": 5360 }, { "acc": 0.0, "epoch": 3.63, "learning_rate": 2e-05, "loss": 0.1951, "rl_loss": -2.4543, "step": 5370 }, { "acc": 0.0, "epoch": 3.64, "learning_rate": 2e-05, "loss": 0.2071, "rl_loss": -2.5265, "step": 5380 }, { "acc": 0.0, "epoch": 3.64, "learning_rate": 2e-05, "loss": 0.1961, "rl_loss": -2.0825, "step": 5390 }, { "acc": 0.0, "epoch": 3.65, "learning_rate": 2e-05, "loss": 0.2055, "rl_loss": -2.7325, "step": 5400 }, { "acc": 0.0, "epoch": 3.66, "learning_rate": 2e-05, "loss": 0.1881, "rl_loss": -2.2273, "step": 5410 }, { "acc": 0.0, "epoch": 3.66, "learning_rate": 2e-05, "loss": 0.1896, "rl_loss": -2.3256, "step": 5420 }, { "acc": 0.0, "epoch": 3.67, "learning_rate": 2e-05, "loss": 0.2989, "rl_loss": -3.3343, "step": 5430 }, { "acc": 0.0, "epoch": 3.68, "learning_rate": 2e-05, "loss": 0.2031, "rl_loss": -3.2835, "step": 5440 }, { "acc": 0.0, "epoch": 3.68, "learning_rate": 2e-05, "loss": 0.1972, "rl_loss": -2.5404, "step": 5450 }, { "acc": 0.0, "epoch": 3.69, "learning_rate": 2e-05, "loss": 0.193, "rl_loss": -2.2415, "step": 5460 }, { "acc": 0.0, "epoch": 3.7, "learning_rate": 2e-05, "loss": 0.1942, "rl_loss": -2.6968, "step": 5470 }, { "acc": 0.0, "epoch": 3.71, "learning_rate": 2e-05, "loss": 0.2045, "rl_loss": -1.0487, "step": 5480 }, { "acc": 0.0, "epoch": 3.71, "learning_rate": 2e-05, "loss": 0.1899, "rl_loss": -2.1441, "step": 5490 }, { "acc": 0.0, "epoch": 3.72, "learning_rate": 2e-05, "loss": 0.1842, "rl_loss": -2.9598, "step": 5500 }, { "acc": 0.0, "epoch": 3.73, "learning_rate": 2e-05, "loss": 0.1886, "rl_loss": -2.0872, "step": 5510 }, { "acc": 0.0, "epoch": 3.73, "learning_rate": 2e-05, "loss": 0.192, "rl_loss": -3.5614, "step": 5520 }, { "acc": 0.0, "epoch": 3.74, "learning_rate": 2e-05, "loss": 0.1933, "rl_loss": -2.3861, "step": 5530 }, { "acc": 0.0, "epoch": 3.75, "learning_rate": 2e-05, "loss": 0.1988, "rl_loss": -2.3681, "step": 5540 }, { "acc": 0.0, "epoch": 3.75, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.7298, "step": 5550 }, { "acc": 0.0, "epoch": 3.76, "learning_rate": 2e-05, "loss": 0.1938, "rl_loss": -2.3424, "step": 5560 }, { "acc": 0.0, "epoch": 3.77, "learning_rate": 2e-05, "loss": 0.2195, "rl_loss": -2.02, "step": 5570 }, { "acc": 0.0, "epoch": 3.77, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -2.7591, "step": 5580 }, { "acc": 0.0, "epoch": 3.78, "learning_rate": 2e-05, "loss": 0.1928, "rl_loss": -3.3643, "step": 5590 }, { "acc": 0.0, "epoch": 3.79, "learning_rate": 2e-05, "loss": 0.2044, "rl_loss": -2.0494, "step": 5600 }, { "acc": 0.0, "epoch": 3.79, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -2.4601, "step": 5610 }, { "acc": 0.0, "epoch": 3.8, "learning_rate": 2e-05, "loss": 0.2, "rl_loss": -2.7308, "step": 5620 }, { "acc": 0.0, "epoch": 3.81, "learning_rate": 2e-05, "loss": 0.1918, "rl_loss": -2.8355, "step": 5630 }, { "acc": 0.0, "epoch": 3.81, "learning_rate": 2e-05, "loss": 0.1974, "rl_loss": -1.7876, "step": 5640 }, { "acc": 0.0, "epoch": 3.82, "learning_rate": 2e-05, "loss": 0.1939, "rl_loss": -1.9107, "step": 5650 }, { "acc": 0.0, "epoch": 3.83, "learning_rate": 2e-05, "loss": 0.2016, "rl_loss": -2.0189, "step": 5660 }, { "acc": 0.0, "epoch": 3.83, "learning_rate": 2e-05, "loss": 0.1913, "rl_loss": -2.3644, "step": 5670 }, { "acc": 0.0, "epoch": 3.84, "learning_rate": 2e-05, "loss": 0.2013, "rl_loss": -2.0528, "step": 5680 }, { "acc": 0.0, "epoch": 3.85, "learning_rate": 2e-05, "loss": 0.1962, "rl_loss": -3.0767, "step": 5690 }, { "acc": 0.0, "epoch": 3.85, "learning_rate": 2e-05, "loss": 0.1942, "rl_loss": -1.6827, "step": 5700 }, { "acc": 0.0, "epoch": 3.86, "learning_rate": 2e-05, "loss": 0.2016, "rl_loss": -2.0426, "step": 5710 }, { "acc": 0.0, "epoch": 3.87, "learning_rate": 2e-05, "loss": 0.206, "rl_loss": -2.6912, "step": 5720 }, { "acc": 0.0, "epoch": 3.87, "learning_rate": 2e-05, "loss": 0.2117, "rl_loss": -2.7239, "step": 5730 }, { "acc": 0.0, "epoch": 3.88, "learning_rate": 2e-05, "loss": 0.1854, "rl_loss": -2.6038, "step": 5740 }, { "acc": 0.0, "epoch": 3.89, "learning_rate": 2e-05, "loss": 0.1988, "rl_loss": -1.3471, "step": 5750 }, { "acc": 0.0, "epoch": 3.89, "learning_rate": 2e-05, "loss": 0.196, "rl_loss": -1.9074, "step": 5760 }, { "acc": 0.0, "epoch": 3.9, "learning_rate": 2e-05, "loss": 0.2898, "rl_loss": -2.8468, "step": 5770 }, { "acc": 0.0, "epoch": 3.91, "learning_rate": 2e-05, "loss": 0.1948, "rl_loss": -2.8875, "step": 5780 }, { "acc": 0.0, "epoch": 3.91, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -3.0429, "step": 5790 }, { "acc": 0.0, "epoch": 3.92, "learning_rate": 2e-05, "loss": 0.202, "rl_loss": -2.9436, "step": 5800 }, { "acc": 0.0, "epoch": 3.93, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -2.601, "step": 5810 }, { "acc": 0.0, "epoch": 3.94, "learning_rate": 2e-05, "loss": 0.1875, "rl_loss": -2.748, "step": 5820 }, { "acc": 0.0, "epoch": 3.94, "learning_rate": 2e-05, "loss": 0.195, "rl_loss": -2.2702, "step": 5830 }, { "acc": 0.0, "epoch": 3.95, "learning_rate": 2e-05, "loss": 0.1925, "rl_loss": -3.0517, "step": 5840 }, { "acc": 0.0, "epoch": 3.96, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -2.5624, "step": 5850 }, { "acc": 0.0, "epoch": 3.96, "learning_rate": 2e-05, "loss": 0.1884, "rl_loss": -2.791, "step": 5860 }, { "acc": 0.0, "epoch": 3.97, "learning_rate": 2e-05, "loss": 0.1908, "rl_loss": -2.5692, "step": 5870 }, { "acc": 0.0, "epoch": 3.98, "learning_rate": 2e-05, "loss": 0.1939, "rl_loss": -2.8593, "step": 5880 }, { "acc": 0.0, "epoch": 3.98, "learning_rate": 2e-05, "loss": 0.2063, "rl_loss": -2.3636, "step": 5890 }, { "acc": 0.0, "epoch": 3.99, "learning_rate": 2e-05, "loss": 0.1848, "rl_loss": -3.1319, "step": 5900 }, { "acc": 0.0, "epoch": 4.0, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -2.2009, "step": 5910 }, { "acc": 0.0, "epoch": 4.0, "learning_rate": 2e-05, "loss": 0.2017, "rl_loss": -2.9681, "step": 5920 }, { "acc": 0.0, "epoch": 4.01, "learning_rate": 2e-05, "loss": 0.2053, "rl_loss": -1.858, "step": 5930 }, { "acc": 0.0, "epoch": 4.02, "learning_rate": 2e-05, "loss": 0.1835, "rl_loss": -2.5722, "step": 5940 }, { "acc": 0.0, "epoch": 4.02, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -3.0652, "step": 5950 }, { "acc": 0.0, "epoch": 4.03, "learning_rate": 2e-05, "loss": 0.184, "rl_loss": -2.8385, "step": 5960 }, { "acc": 0.0, "epoch": 4.04, "learning_rate": 2e-05, "loss": 0.2058, "rl_loss": -2.4548, "step": 5970 }, { "acc": 0.0, "epoch": 4.04, "learning_rate": 2e-05, "loss": 0.198, "rl_loss": -3.1167, "step": 5980 }, { "acc": 0.0, "epoch": 4.05, "learning_rate": 2e-05, "loss": 0.2022, "rl_loss": -2.1764, "step": 5990 }, { "acc": 0.0, "epoch": 4.06, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -2.7141, "step": 6000 }, { "acc": 0.0, "epoch": 4.06, "learning_rate": 2e-05, "loss": 0.1876, "rl_loss": -1.6952, "step": 6010 }, { "acc": 0.0, "epoch": 4.07, "learning_rate": 2e-05, "loss": 0.1885, "rl_loss": -1.969, "step": 6020 }, { "acc": 0.0, "epoch": 4.08, "learning_rate": 2e-05, "loss": 0.1885, "rl_loss": -2.2169, "step": 6030 }, { "acc": 0.0, "epoch": 4.08, "learning_rate": 2e-05, "loss": 0.189, "rl_loss": -2.0601, "step": 6040 }, { "acc": 0.0, "epoch": 4.09, "learning_rate": 2e-05, "loss": 0.1962, "rl_loss": -3.3515, "step": 6050 }, { "acc": 0.0, "epoch": 4.1, "learning_rate": 2e-05, "loss": 0.1888, "rl_loss": -2.0714, "step": 6060 }, { "acc": 0.0, "epoch": 4.1, "learning_rate": 2e-05, "loss": 0.1983, "rl_loss": -2.2821, "step": 6070 }, { "acc": 0.0, "epoch": 4.11, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -3.4232, "step": 6080 }, { "acc": 0.0, "epoch": 4.12, "learning_rate": 2e-05, "loss": 0.1885, "rl_loss": -2.4265, "step": 6090 }, { "acc": 0.0, "epoch": 4.12, "learning_rate": 2e-05, "loss": 0.1857, "rl_loss": -2.2742, "step": 6100 }, { "acc": 0.0, "epoch": 4.13, "learning_rate": 2e-05, "loss": 0.1799, "rl_loss": -2.324, "step": 6110 }, { "acc": 0.0, "epoch": 4.14, "learning_rate": 2e-05, "loss": 0.2019, "rl_loss": -1.7835, "step": 6120 }, { "acc": 0.0, "epoch": 4.14, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -2.7385, "step": 6130 }, { "acc": 0.0, "epoch": 4.15, "learning_rate": 2e-05, "loss": 0.2012, "rl_loss": -3.2146, "step": 6140 }, { "acc": 0.0, "epoch": 4.16, "learning_rate": 2e-05, "loss": 0.1962, "rl_loss": -3.5467, "step": 6150 }, { "acc": 0.0, "epoch": 4.16, "learning_rate": 2e-05, "loss": 0.1948, "rl_loss": -2.8846, "step": 6160 }, { "acc": 0.0, "epoch": 4.17, "learning_rate": 2e-05, "loss": 0.1961, "rl_loss": -2.9109, "step": 6170 }, { "acc": 0.0, "epoch": 4.18, "learning_rate": 2e-05, "loss": 0.1931, "rl_loss": -2.758, "step": 6180 }, { "acc": 0.0, "epoch": 4.19, "learning_rate": 2e-05, "loss": 0.1871, "rl_loss": -2.3659, "step": 6190 }, { "acc": 0.0, "epoch": 4.19, "learning_rate": 2e-05, "loss": 0.1827, "rl_loss": -3.368, "step": 6200 }, { "acc": 0.0, "epoch": 4.2, "learning_rate": 2e-05, "loss": 0.1923, "rl_loss": -3.397, "step": 6210 }, { "acc": 0.0, "epoch": 4.21, "learning_rate": 2e-05, "loss": 0.1871, "rl_loss": -2.2373, "step": 6220 }, { "acc": 0.0, "epoch": 4.21, "learning_rate": 2e-05, "loss": 0.1911, "rl_loss": -2.8355, "step": 6230 }, { "acc": 0.0, "epoch": 4.22, "learning_rate": 2e-05, "loss": 0.1919, "rl_loss": -2.4808, "step": 6240 }, { "acc": 0.0, "epoch": 4.23, "learning_rate": 2e-05, "loss": 0.1942, "rl_loss": -2.2186, "step": 6250 }, { "acc": 0.0, "epoch": 4.23, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -3.2056, "step": 6260 }, { "acc": 0.0, "epoch": 4.24, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -2.0922, "step": 6270 }, { "acc": 0.0, "epoch": 4.25, "learning_rate": 2e-05, "loss": 0.1885, "rl_loss": -3.3753, "step": 6280 }, { "acc": 0.0, "epoch": 4.25, "learning_rate": 2e-05, "loss": 0.1987, "rl_loss": -2.4124, "step": 6290 }, { "acc": 0.0, "epoch": 4.26, "learning_rate": 2e-05, "loss": 0.2012, "rl_loss": -2.7648, "step": 6300 }, { "acc": 0.0, "epoch": 4.27, "learning_rate": 2e-05, "loss": 0.204, "rl_loss": -2.7979, "step": 6310 }, { "acc": 0.0, "epoch": 4.27, "learning_rate": 2e-05, "loss": 0.1825, "rl_loss": -2.9201, "step": 6320 }, { "acc": 0.0, "epoch": 4.28, "learning_rate": 2e-05, "loss": 0.1989, "rl_loss": -1.9986, "step": 6330 }, { "acc": 0.0, "epoch": 4.29, "learning_rate": 2e-05, "loss": 0.2018, "rl_loss": -2.5842, "step": 6340 }, { "acc": 0.0, "epoch": 4.29, "learning_rate": 2e-05, "loss": 0.2012, "rl_loss": -2.9912, "step": 6350 }, { "acc": 0.0, "epoch": 4.3, "learning_rate": 2e-05, "loss": 0.2062, "rl_loss": -2.8943, "step": 6360 }, { "acc": 0.0, "epoch": 4.31, "learning_rate": 2e-05, "loss": 0.192, "rl_loss": -2.7264, "step": 6370 }, { "acc": 0.0, "epoch": 4.31, "learning_rate": 2e-05, "loss": 0.1971, "rl_loss": -2.6842, "step": 6380 }, { "acc": 0.0, "epoch": 4.32, "learning_rate": 2e-05, "loss": 0.188, "rl_loss": -3.3447, "step": 6390 }, { "acc": 0.0, "epoch": 4.33, "learning_rate": 2e-05, "loss": 0.1953, "rl_loss": -3.1123, "step": 6400 }, { "acc": 0.0, "epoch": 4.33, "learning_rate": 2e-05, "loss": 0.1946, "rl_loss": -2.6241, "step": 6410 }, { "acc": 0.0, "epoch": 4.34, "learning_rate": 2e-05, "loss": 0.1798, "rl_loss": -2.4209, "step": 6420 }, { "acc": 0.0, "epoch": 4.35, "learning_rate": 2e-05, "loss": 0.1874, "rl_loss": -2.791, "step": 6430 }, { "acc": 0.0, "epoch": 4.35, "learning_rate": 2e-05, "loss": 0.1911, "rl_loss": -2.9924, "step": 6440 }, { "acc": 0.0, "epoch": 4.36, "learning_rate": 2e-05, "loss": 0.1879, "rl_loss": -2.8797, "step": 6450 }, { "acc": 0.0, "epoch": 4.37, "learning_rate": 2e-05, "loss": 0.183, "rl_loss": -2.486, "step": 6460 }, { "acc": 0.0, "epoch": 4.37, "learning_rate": 2e-05, "loss": 0.1828, "rl_loss": -2.0302, "step": 6470 }, { "acc": 0.0, "epoch": 4.38, "learning_rate": 2e-05, "loss": 0.1985, "rl_loss": -3.1149, "step": 6480 }, { "acc": 0.0, "epoch": 4.39, "learning_rate": 2e-05, "loss": 0.1964, "rl_loss": -2.6101, "step": 6490 }, { "acc": 0.0, "epoch": 4.39, "learning_rate": 2e-05, "loss": 0.1889, "rl_loss": -3.5678, "step": 6500 }, { "acc": 0.0, "epoch": 4.4, "learning_rate": 2e-05, "loss": 0.2112, "rl_loss": -2.3722, "step": 6510 }, { "acc": 0.0, "epoch": 4.41, "learning_rate": 2e-05, "loss": 0.1832, "rl_loss": -2.641, "step": 6520 }, { "acc": 0.0, "epoch": 4.42, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -3.0656, "step": 6530 }, { "acc": 0.0, "epoch": 4.42, "learning_rate": 2e-05, "loss": 0.1931, "rl_loss": -3.8367, "step": 6540 }, { "acc": 0.0, "epoch": 4.43, "learning_rate": 2e-05, "loss": 0.1911, "rl_loss": -2.1095, "step": 6550 }, { "acc": 0.0, "epoch": 4.44, "learning_rate": 2e-05, "loss": 0.1946, "rl_loss": -2.7825, "step": 6560 }, { "acc": 0.0, "epoch": 4.44, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.7297, "step": 6570 }, { "acc": 0.0, "epoch": 4.45, "learning_rate": 2e-05, "loss": 0.1902, "rl_loss": -3.9339, "step": 6580 }, { "acc": 0.0, "epoch": 4.46, "learning_rate": 2e-05, "loss": 0.1978, "rl_loss": -2.2615, "step": 6590 }, { "acc": 0.0, "epoch": 4.46, "learning_rate": 2e-05, "loss": 0.1893, "rl_loss": -2.7209, "step": 6600 }, { "acc": 0.0, "epoch": 4.47, "learning_rate": 2e-05, "loss": 0.192, "rl_loss": -2.0833, "step": 6610 }, { "acc": 0.0, "epoch": 4.48, "learning_rate": 2e-05, "loss": 0.1884, "rl_loss": -1.9636, "step": 6620 }, { "acc": 0.0, "epoch": 4.48, "learning_rate": 2e-05, "loss": 0.1995, "rl_loss": -3.1072, "step": 6630 }, { "acc": 0.0, "epoch": 4.49, "learning_rate": 2e-05, "loss": 0.1868, "rl_loss": -2.4482, "step": 6640 }, { "acc": 0.0, "epoch": 4.5, "learning_rate": 2e-05, "loss": 0.1923, "rl_loss": -2.1615, "step": 6650 }, { "acc": 0.0, "epoch": 4.5, "learning_rate": 2e-05, "loss": 0.1968, "rl_loss": -3.1933, "step": 6660 }, { "acc": 0.0, "epoch": 4.51, "learning_rate": 2e-05, "loss": 0.2019, "rl_loss": -2.1685, "step": 6670 }, { "acc": 0.0, "epoch": 4.52, "learning_rate": 2e-05, "loss": 0.1976, "rl_loss": -2.65, "step": 6680 }, { "acc": 0.0, "epoch": 4.52, "learning_rate": 2e-05, "loss": 0.1903, "rl_loss": -2.9864, "step": 6690 }, { "acc": 0.0, "epoch": 4.53, "learning_rate": 2e-05, "loss": 0.2339, "rl_loss": -2.327, "step": 6700 }, { "acc": 0.0, "epoch": 4.54, "learning_rate": 2e-05, "loss": 0.1954, "rl_loss": -4.2312, "step": 6710 }, { "acc": 0.0, "epoch": 4.54, "learning_rate": 2e-05, "loss": 0.1846, "rl_loss": -2.6488, "step": 6720 }, { "acc": 0.0, "epoch": 4.55, "learning_rate": 2e-05, "loss": 0.1889, "rl_loss": -2.6731, "step": 6730 }, { "acc": 0.0, "epoch": 4.56, "learning_rate": 2e-05, "loss": 0.187, "rl_loss": -2.8007, "step": 6740 }, { "acc": 0.0, "epoch": 4.56, "learning_rate": 2e-05, "loss": 0.1886, "rl_loss": -3.1589, "step": 6750 }, { "acc": 0.0, "epoch": 4.57, "learning_rate": 2e-05, "loss": 0.186, "rl_loss": -2.9446, "step": 6760 }, { "acc": 0.0, "epoch": 4.58, "learning_rate": 2e-05, "loss": 0.2044, "rl_loss": -1.8589, "step": 6770 }, { "acc": 0.0, "epoch": 4.58, "learning_rate": 2e-05, "loss": 0.1967, "rl_loss": -2.5799, "step": 6780 }, { "acc": 0.0, "epoch": 4.59, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -3.484, "step": 6790 }, { "acc": 0.0, "epoch": 4.6, "learning_rate": 2e-05, "loss": 0.2082, "rl_loss": -3.1572, "step": 6800 }, { "acc": 0.0, "epoch": 4.6, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -3.3015, "step": 6810 }, { "acc": 0.0, "epoch": 4.61, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -1.997, "step": 6820 }, { "acc": 0.0, "epoch": 4.62, "learning_rate": 2e-05, "loss": 0.1938, "rl_loss": -2.5738, "step": 6830 }, { "acc": 0.0, "epoch": 4.62, "learning_rate": 2e-05, "loss": 0.18, "rl_loss": -1.9034, "step": 6840 }, { "acc": 0.0, "epoch": 4.63, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -3.464, "step": 6850 }, { "acc": 0.0, "epoch": 4.64, "learning_rate": 2e-05, "loss": 0.1994, "rl_loss": -2.0304, "step": 6860 }, { "acc": 0.0, "epoch": 4.65, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -3.0455, "step": 6870 }, { "acc": 0.0, "epoch": 4.65, "learning_rate": 2e-05, "loss": 0.1945, "rl_loss": -2.6431, "step": 6880 }, { "acc": 0.0, "epoch": 4.66, "learning_rate": 2e-05, "loss": 0.189, "rl_loss": -2.8907, "step": 6890 }, { "acc": 0.0, "epoch": 4.67, "learning_rate": 2e-05, "loss": 0.1909, "rl_loss": -2.0786, "step": 6900 }, { "acc": 0.0, "epoch": 4.67, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -2.0547, "step": 6910 }, { "acc": 0.0, "epoch": 4.68, "learning_rate": 2e-05, "loss": 0.1959, "rl_loss": -2.3965, "step": 6920 }, { "acc": 0.0, "epoch": 4.69, "learning_rate": 2e-05, "loss": 0.1805, "rl_loss": -2.9684, "step": 6930 }, { "acc": 0.0, "epoch": 4.69, "learning_rate": 2e-05, "loss": 0.189, "rl_loss": -3.0414, "step": 6940 }, { "acc": 0.0, "epoch": 4.7, "learning_rate": 2e-05, "loss": 0.2018, "rl_loss": -3.0954, "step": 6950 }, { "acc": 0.0, "epoch": 4.71, "learning_rate": 2e-05, "loss": 0.1931, "rl_loss": -2.4459, "step": 6960 }, { "acc": 0.0, "epoch": 4.71, "learning_rate": 2e-05, "loss": 0.1988, "rl_loss": -2.1799, "step": 6970 }, { "acc": 0.0, "epoch": 4.72, "learning_rate": 2e-05, "loss": 0.1857, "rl_loss": -2.2469, "step": 6980 }, { "acc": 0.0, "epoch": 4.73, "learning_rate": 2e-05, "loss": 0.1854, "rl_loss": -1.6519, "step": 6990 }, { "acc": 0.0, "epoch": 4.73, "learning_rate": 2e-05, "loss": 0.1892, "rl_loss": -2.2516, "step": 7000 }, { "acc": 0.0, "epoch": 4.74, "learning_rate": 2e-05, "loss": 0.2088, "rl_loss": -2.3415, "step": 7010 }, { "acc": 0.0, "epoch": 4.75, "learning_rate": 2e-05, "loss": 0.217, "rl_loss": -2.4478, "step": 7020 }, { "acc": 0.0, "epoch": 4.75, "learning_rate": 2e-05, "loss": 0.197, "rl_loss": -1.5287, "step": 7030 }, { "acc": 0.0, "epoch": 4.76, "learning_rate": 2e-05, "loss": 0.1848, "rl_loss": -3.2543, "step": 7040 }, { "acc": 0.0, "epoch": 4.77, "learning_rate": 2e-05, "loss": 0.1934, "rl_loss": -1.9223, "step": 7050 }, { "acc": 0.0, "epoch": 4.77, "learning_rate": 2e-05, "loss": 0.1896, "rl_loss": -2.7441, "step": 7060 }, { "acc": 0.0, "epoch": 4.78, "learning_rate": 2e-05, "loss": 0.1941, "rl_loss": -3.1355, "step": 7070 }, { "acc": 0.0, "epoch": 4.79, "learning_rate": 2e-05, "loss": 0.1901, "rl_loss": -2.551, "step": 7080 }, { "acc": 0.0, "epoch": 4.79, "learning_rate": 2e-05, "loss": 0.1947, "rl_loss": -2.7906, "step": 7090 }, { "acc": 0.0, "epoch": 4.8, "learning_rate": 2e-05, "loss": 0.1772, "rl_loss": -1.386, "step": 7100 }, { "acc": 0.0, "epoch": 4.81, "learning_rate": 2e-05, "loss": 0.2013, "rl_loss": -3.4789, "step": 7110 }, { "acc": 0.0, "epoch": 4.81, "learning_rate": 2e-05, "loss": 0.1903, "rl_loss": -3.727, "step": 7120 }, { "acc": 0.0, "epoch": 4.82, "learning_rate": 2e-05, "loss": 0.1843, "rl_loss": -2.2961, "step": 7130 }, { "acc": 0.0, "epoch": 4.83, "learning_rate": 2e-05, "loss": 0.179, "rl_loss": -3.1504, "step": 7140 }, { "acc": 0.0, "epoch": 4.83, "learning_rate": 2e-05, "loss": 0.1742, "rl_loss": -3.0242, "step": 7150 }, { "acc": 0.0, "epoch": 4.84, "learning_rate": 2e-05, "loss": 0.1952, "rl_loss": -4.6317, "step": 7160 }, { "acc": 0.0, "epoch": 4.85, "learning_rate": 2e-05, "loss": 0.1852, "rl_loss": -2.747, "step": 7170 }, { "acc": 0.0, "epoch": 4.85, "learning_rate": 2e-05, "loss": 0.1867, "rl_loss": -2.6207, "step": 7180 }, { "acc": 0.0, "epoch": 4.86, "learning_rate": 2e-05, "loss": 0.1925, "rl_loss": -3.3217, "step": 7190 }, { "acc": 0.0, "epoch": 4.87, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -2.8516, "step": 7200 }, { "acc": 0.0, "epoch": 4.87, "learning_rate": 2e-05, "loss": 0.1874, "rl_loss": -3.4229, "step": 7210 }, { "acc": 0.0, "epoch": 4.88, "learning_rate": 2e-05, "loss": 0.1917, "rl_loss": -3.0031, "step": 7220 }, { "acc": 0.0, "epoch": 4.89, "learning_rate": 2e-05, "loss": 0.1887, "rl_loss": -3.6536, "step": 7230 }, { "acc": 0.0, "epoch": 4.9, "learning_rate": 2e-05, "loss": 0.1847, "rl_loss": -1.9608, "step": 7240 }, { "acc": 0.0, "epoch": 4.9, "learning_rate": 2e-05, "loss": 0.1886, "rl_loss": -3.1278, "step": 7250 }, { "acc": 0.0, "epoch": 4.91, "learning_rate": 2e-05, "loss": 0.1916, "rl_loss": -2.8768, "step": 7260 }, { "acc": 0.0, "epoch": 4.92, "learning_rate": 2e-05, "loss": 0.1915, "rl_loss": -2.3052, "step": 7270 }, { "acc": 0.0, "epoch": 4.92, "learning_rate": 2e-05, "loss": 0.19, "rl_loss": -3.1146, "step": 7280 }, { "acc": 0.0, "epoch": 4.93, "learning_rate": 2e-05, "loss": 0.1861, "rl_loss": -2.9576, "step": 7290 }, { "acc": 0.0, "epoch": 4.94, "learning_rate": 2e-05, "loss": 0.1856, "rl_loss": -2.1348, "step": 7300 }, { "acc": 0.0, "epoch": 4.94, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -2.7981, "step": 7310 }, { "acc": 0.0, "epoch": 4.95, "learning_rate": 2e-05, "loss": 0.1856, "rl_loss": -2.0344, "step": 7320 }, { "acc": 0.0, "epoch": 4.96, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -3.8394, "step": 7330 }, { "acc": 0.0, "epoch": 4.96, "learning_rate": 2e-05, "loss": 0.196, "rl_loss": -2.5963, "step": 7340 }, { "acc": 0.0, "epoch": 4.97, "learning_rate": 2e-05, "loss": 0.1862, "rl_loss": -2.7614, "step": 7350 }, { "acc": 0.0, "epoch": 4.98, "learning_rate": 2e-05, "loss": 0.2075, "rl_loss": -2.6204, "step": 7360 }, { "acc": 0.0, "epoch": 4.98, "learning_rate": 2e-05, "loss": 0.1944, "rl_loss": -3.5221, "step": 7370 }, { "acc": 0.0, "epoch": 4.99, "learning_rate": 2e-05, "loss": 0.1871, "rl_loss": -3.2469, "step": 7380 }, { "acc": 0.0, "epoch": 5.0, "learning_rate": 2e-05, "loss": 0.1866, "rl_loss": -3.4487, "step": 7390 }, { "acc": 0.0, "epoch": 5.0, "learning_rate": 2e-05, "loss": 0.179, "rl_loss": -3.0974, "step": 7400 }, { "acc": 0.0, "epoch": 5.01, "learning_rate": 2e-05, "loss": 0.1866, "rl_loss": -3.307, "step": 7410 }, { "acc": 0.0, "epoch": 5.02, "learning_rate": 2e-05, "loss": 0.2019, "rl_loss": -1.9082, "step": 7420 }, { "acc": 0.0, "epoch": 5.02, "learning_rate": 2e-05, "loss": 0.1792, "rl_loss": -2.6152, "step": 7430 }, { "acc": 0.0, "epoch": 5.03, "learning_rate": 2e-05, "loss": 0.1825, "rl_loss": -2.6146, "step": 7440 }, { "acc": 0.0, "epoch": 5.04, "learning_rate": 2e-05, "loss": 0.2038, "rl_loss": -2.9781, "step": 7450 }, { "acc": 0.0, "epoch": 5.04, "learning_rate": 2e-05, "loss": 0.2084, "rl_loss": -2.9026, "step": 7460 }, { "acc": 0.0, "epoch": 5.05, "learning_rate": 2e-05, "loss": 0.2008, "rl_loss": -2.6283, "step": 7470 }, { "acc": 0.0, "epoch": 5.06, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -2.3691, "step": 7480 }, { "acc": 0.0, "epoch": 5.06, "learning_rate": 2e-05, "loss": 0.1889, "rl_loss": -3.0735, "step": 7490 }, { "acc": 0.0, "epoch": 5.07, "learning_rate": 2e-05, "loss": 0.187, "rl_loss": -3.6196, "step": 7500 }, { "acc": 0.0, "epoch": 5.08, "learning_rate": 2e-05, "loss": 0.1863, "rl_loss": -3.1925, "step": 7510 }, { "acc": 0.0, "epoch": 5.08, "learning_rate": 2e-05, "loss": 0.2152, "rl_loss": -3.3089, "step": 7520 }, { "acc": 0.0, "epoch": 5.09, "learning_rate": 2e-05, "loss": 0.1877, "rl_loss": -2.0276, "step": 7530 }, { "acc": 0.0, "epoch": 5.1, "learning_rate": 2e-05, "loss": 0.1927, "rl_loss": -2.8, "step": 7540 }, { "acc": 0.0, "epoch": 5.1, "learning_rate": 2e-05, "loss": 0.1955, "rl_loss": -2.1008, "step": 7550 }, { "acc": 0.0, "epoch": 5.11, "learning_rate": 2e-05, "loss": 0.206, "rl_loss": -3.1376, "step": 7560 }, { "acc": 0.0, "epoch": 5.12, "learning_rate": 2e-05, "loss": 0.188, "rl_loss": -2.7592, "step": 7570 }, { "acc": 0.0, "epoch": 5.13, "learning_rate": 2e-05, "loss": 0.1831, "rl_loss": -3.8855, "step": 7580 }, { "acc": 0.0, "epoch": 5.13, "learning_rate": 2e-05, "loss": 0.1857, "rl_loss": -2.8628, "step": 7590 }, { "acc": 0.0, "epoch": 5.14, "learning_rate": 2e-05, "loss": 0.2021, "rl_loss": -2.9993, "step": 7600 }, { "acc": 0.0, "epoch": 5.15, "learning_rate": 2e-05, "loss": 0.2077, "rl_loss": -2.313, "step": 7610 }, { "acc": 0.0, "epoch": 5.15, "learning_rate": 2e-05, "loss": 0.199, "rl_loss": -2.6262, "step": 7620 }, { "acc": 0.0, "epoch": 5.16, "learning_rate": 2e-05, "loss": 0.1944, "rl_loss": -2.8302, "step": 7630 }, { "acc": 0.0, "epoch": 5.17, "learning_rate": 2e-05, "loss": 0.1925, "rl_loss": -2.7592, "step": 7640 }, { "acc": 0.0, "epoch": 5.17, "learning_rate": 2e-05, "loss": 0.1917, "rl_loss": -1.8957, "step": 7650 }, { "acc": 0.0, "epoch": 5.18, "learning_rate": 2e-05, "loss": 0.1918, "rl_loss": -2.2624, "step": 7660 }, { "acc": 0.0, "epoch": 5.19, "learning_rate": 2e-05, "loss": 0.1857, "rl_loss": -2.2221, "step": 7670 }, { "acc": 0.0, "epoch": 5.19, "learning_rate": 2e-05, "loss": 0.2995, "rl_loss": -2.2804, "step": 7680 }, { "acc": 0.0, "epoch": 5.2, "learning_rate": 2e-05, "loss": 0.1755, "rl_loss": -2.1464, "step": 7690 }, { "acc": 0.0, "epoch": 5.21, "learning_rate": 2e-05, "loss": 0.2031, "rl_loss": -2.777, "step": 7700 }, { "acc": 0.0, "epoch": 5.21, "learning_rate": 2e-05, "loss": 0.1898, "rl_loss": -4.2044, "step": 7710 }, { "acc": 0.0, "epoch": 5.22, "learning_rate": 2e-05, "loss": 0.1966, "rl_loss": -3.1803, "step": 7720 }, { "acc": 0.0, "epoch": 5.23, "learning_rate": 2e-05, "loss": 0.1999, "rl_loss": -3.1848, "step": 7730 }, { "acc": 0.0, "epoch": 5.23, "learning_rate": 2e-05, "loss": 0.1861, "rl_loss": -2.442, "step": 7740 }, { "acc": 0.0, "epoch": 5.24, "learning_rate": 2e-05, "loss": 0.1975, "rl_loss": -2.3335, "step": 7750 }, { "acc": 0.0, "epoch": 5.25, "learning_rate": 2e-05, "loss": 0.1835, "rl_loss": -2.6741, "step": 7760 }, { "acc": 0.0, "epoch": 5.25, "learning_rate": 2e-05, "loss": 0.189, "rl_loss": -2.6599, "step": 7770 }, { "acc": 0.0, "epoch": 5.26, "learning_rate": 2e-05, "loss": 0.1837, "rl_loss": -3.1247, "step": 7780 }, { "acc": 0.0, "epoch": 5.27, "learning_rate": 2e-05, "loss": 0.1779, "rl_loss": -2.9899, "step": 7790 }, { "acc": 0.0, "epoch": 5.27, "learning_rate": 2e-05, "loss": 0.1943, "rl_loss": -1.8916, "step": 7800 }, { "acc": 0.0, "epoch": 5.28, "learning_rate": 2e-05, "loss": 0.1949, "rl_loss": -3.501, "step": 7810 }, { "acc": 0.0, "epoch": 5.29, "learning_rate": 2e-05, "loss": 0.1943, "rl_loss": -2.3796, "step": 7820 }, { "acc": 0.0, "epoch": 5.29, "learning_rate": 2e-05, "loss": 0.1943, "rl_loss": -3.0778, "step": 7830 }, { "acc": 0.0, "epoch": 5.3, "learning_rate": 2e-05, "loss": 0.1864, "rl_loss": -3.5929, "step": 7840 }, { "acc": 0.0, "epoch": 5.31, "learning_rate": 2e-05, "loss": 0.1767, "rl_loss": -3.171, "step": 7850 }, { "acc": 0.0, "epoch": 5.31, "learning_rate": 2e-05, "loss": 0.1877, "rl_loss": -2.9619, "step": 7860 }, { "acc": 0.0, "epoch": 5.32, "learning_rate": 2e-05, "loss": 0.1842, "rl_loss": -2.5713, "step": 7870 }, { "acc": 0.0, "epoch": 5.33, "learning_rate": 2e-05, "loss": 0.1897, "rl_loss": -2.2853, "step": 7880 }, { "acc": 0.0, "epoch": 5.33, "learning_rate": 2e-05, "loss": 0.2, "rl_loss": -2.4228, "step": 7890 }, { "acc": 0.0, "epoch": 5.34, "learning_rate": 2e-05, "loss": 0.1977, "rl_loss": -3.3886, "step": 7900 }, { "acc": 0.0, "epoch": 5.35, "learning_rate": 2e-05, "loss": 0.1984, "rl_loss": -3.5538, "step": 7910 }, { "acc": 0.0, "epoch": 5.35, "learning_rate": 2e-05, "loss": 0.1937, "rl_loss": -3.549, "step": 7920 }, { "acc": 0.0, "epoch": 5.36, "learning_rate": 2e-05, "loss": 0.1907, "rl_loss": -2.5696, "step": 7930 }, { "acc": 0.0, "epoch": 5.37, "learning_rate": 2e-05, "loss": 0.1926, "rl_loss": -2.5746, "step": 7940 }, { "acc": 0.0, "epoch": 5.38, "learning_rate": 2e-05, "loss": 0.1777, "rl_loss": -2.2632, "step": 7950 }, { "acc": 0.0, "epoch": 5.38, "learning_rate": 2e-05, "loss": 0.1897, "rl_loss": -2.8944, "step": 7960 }, { "acc": 0.0, "epoch": 5.39, "learning_rate": 2e-05, "loss": 0.182, "rl_loss": -2.9216, "step": 7970 }, { "acc": 0.0, "epoch": 5.4, "learning_rate": 2e-05, "loss": 0.1915, "rl_loss": -2.5142, "step": 7980 }, { "acc": 0.0, "epoch": 5.4, "learning_rate": 2e-05, "loss": 0.1877, "rl_loss": -2.9806, "step": 7990 }, { "acc": 0.0, "epoch": 5.41, "learning_rate": 2e-05, "loss": 0.1774, "rl_loss": -2.7414, "step": 8000 }, { "acc": 0.0, "epoch": 5.42, "learning_rate": 2e-05, "loss": 0.179, "rl_loss": -2.5406, "step": 8010 }, { "acc": 0.0, "epoch": 5.42, "learning_rate": 2e-05, "loss": 0.2011, "rl_loss": -2.3767, "step": 8020 }, { "acc": 0.0, "epoch": 5.43, "learning_rate": 2e-05, "loss": 0.1853, "rl_loss": -2.4299, "step": 8030 }, { "acc": 0.0, "epoch": 5.44, "learning_rate": 2e-05, "loss": 0.1821, "rl_loss": -3.6629, "step": 8040 }, { "acc": 0.0, "epoch": 5.44, "learning_rate": 2e-05, "loss": 0.1888, "rl_loss": -1.9189, "step": 8050 }, { "acc": 0.0, "epoch": 5.45, "learning_rate": 2e-05, "loss": 0.1808, "rl_loss": -3.5717, "step": 8060 }, { "acc": 0.0, "epoch": 5.46, "learning_rate": 2e-05, "loss": 0.1929, "rl_loss": -2.5349, "step": 8070 }, { "acc": 0.0, "epoch": 5.46, "learning_rate": 2e-05, "loss": 0.1859, "rl_loss": -2.2485, "step": 8080 }, { "acc": 0.0, "epoch": 5.47, "learning_rate": 2e-05, "loss": 0.1842, "rl_loss": -2.1744, "step": 8090 }, { "acc": 0.0, "epoch": 5.48, "learning_rate": 2e-05, "loss": 0.1963, "rl_loss": -2.9456, "step": 8100 }, { "acc": 0.0, "epoch": 5.48, "learning_rate": 2e-05, "loss": 0.1968, "rl_loss": -3.5672, "step": 8110 }, { "acc": 0.0, "epoch": 5.49, "learning_rate": 2e-05, "loss": 0.1863, "rl_loss": -2.4744, "step": 8120 }, { "acc": 0.0, "epoch": 5.5, "learning_rate": 2e-05, "loss": 0.1804, "rl_loss": -2.7228, "step": 8130 }, { "acc": 0.0, "epoch": 5.5, "learning_rate": 2e-05, "loss": 0.1877, "rl_loss": -2.6649, "step": 8140 }, { "acc": 0.0, "epoch": 5.51, "learning_rate": 2e-05, "loss": 0.1844, "rl_loss": -2.7186, "step": 8150 }, { "acc": 0.0, "epoch": 5.52, "learning_rate": 2e-05, "loss": 0.1884, "rl_loss": -2.895, "step": 8160 }, { "acc": 0.0, "epoch": 5.52, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -2.3746, "step": 8170 }, { "acc": 0.0, "epoch": 5.53, "learning_rate": 2e-05, "loss": 0.1859, "rl_loss": -2.1799, "step": 8180 }, { "acc": 0.0, "epoch": 5.54, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -3.0748, "step": 8190 }, { "acc": 0.0, "epoch": 5.54, "learning_rate": 2e-05, "loss": 0.1892, "rl_loss": -2.7951, "step": 8200 }, { "acc": 0.0, "epoch": 5.55, "learning_rate": 2e-05, "loss": 0.1875, "rl_loss": -2.0033, "step": 8210 }, { "acc": 0.0, "epoch": 5.56, "learning_rate": 2e-05, "loss": 0.1809, "rl_loss": -2.9939, "step": 8220 }, { "acc": 0.0, "epoch": 5.56, "learning_rate": 2e-05, "loss": 0.1898, "rl_loss": -2.7836, "step": 8230 }, { "acc": 0.0, "epoch": 5.57, "learning_rate": 2e-05, "loss": 0.182, "rl_loss": -3.3567, "step": 8240 }, { "acc": 0.0, "epoch": 5.58, "learning_rate": 2e-05, "loss": 0.1946, "rl_loss": -3.4702, "step": 8250 }, { "acc": 0.0, "epoch": 5.58, "learning_rate": 2e-05, "loss": 0.1849, "rl_loss": -2.9167, "step": 8260 }, { "acc": 0.0, "epoch": 5.59, "learning_rate": 2e-05, "loss": 0.1905, "rl_loss": -1.9328, "step": 8270 }, { "acc": 0.0, "epoch": 5.6, "learning_rate": 2e-05, "loss": 0.2001, "rl_loss": -2.7231, "step": 8280 }, { "acc": 0.0, "epoch": 5.61, "learning_rate": 2e-05, "loss": 0.1814, "rl_loss": -2.8048, "step": 8290 }, { "acc": 0.0, "epoch": 5.61, "learning_rate": 2e-05, "loss": 0.1942, "rl_loss": -2.9153, "step": 8300 }, { "acc": 0.0, "epoch": 5.62, "learning_rate": 2e-05, "loss": 0.1956, "rl_loss": -2.3119, "step": 8310 }, { "acc": 0.0, "epoch": 5.63, "learning_rate": 2e-05, "loss": 0.1893, "rl_loss": -2.149, "step": 8320 }, { "acc": 0.0, "epoch": 5.63, "learning_rate": 2e-05, "loss": 0.1867, "rl_loss": -3.0405, "step": 8330 }, { "acc": 0.0, "epoch": 5.64, "learning_rate": 2e-05, "loss": 0.1932, "rl_loss": -1.9545, "step": 8340 }, { "acc": 0.0, "epoch": 5.65, "learning_rate": 2e-05, "loss": 0.1757, "rl_loss": -2.4746, "step": 8350 }, { "acc": 0.0, "epoch": 5.65, "learning_rate": 2e-05, "loss": 0.178, "rl_loss": -2.7607, "step": 8360 }, { "acc": 0.0, "epoch": 5.66, "learning_rate": 2e-05, "loss": 0.1687, "rl_loss": -3.9055, "step": 8370 }, { "acc": 0.0, "epoch": 5.67, "learning_rate": 2e-05, "loss": 0.1869, "rl_loss": -2.5216, "step": 8380 }, { "acc": 0.0, "epoch": 5.67, "learning_rate": 2e-05, "loss": 0.1827, "rl_loss": -3.3682, "step": 8390 }, { "acc": 0.0, "epoch": 5.68, "learning_rate": 2e-05, "loss": 0.1932, "rl_loss": -4.0057, "step": 8400 }, { "acc": 0.0, "epoch": 5.69, "learning_rate": 2e-05, "loss": 0.2032, "rl_loss": -3.6295, "step": 8410 }, { "acc": 0.0, "epoch": 5.69, "learning_rate": 2e-05, "loss": 0.1955, "rl_loss": -3.3139, "step": 8420 }, { "acc": 0.0, "epoch": 5.7, "learning_rate": 2e-05, "loss": 0.1884, "rl_loss": -3.9557, "step": 8430 }, { "acc": 0.0, "epoch": 5.71, "learning_rate": 2e-05, "loss": 0.193, "rl_loss": -3.0584, "step": 8440 }, { "acc": 0.0, "epoch": 5.71, "learning_rate": 2e-05, "loss": 0.1912, "rl_loss": -3.6201, "step": 8450 }, { "acc": 0.0, "epoch": 5.72, "learning_rate": 2e-05, "loss": 0.2938, "rl_loss": -2.5195, "step": 8460 }, { "acc": 0.0, "epoch": 5.73, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -2.4443, "step": 8470 }, { "acc": 0.0, "epoch": 5.73, "learning_rate": 2e-05, "loss": 0.1789, "rl_loss": -2.4472, "step": 8480 }, { "acc": 0.0, "epoch": 5.74, "learning_rate": 2e-05, "loss": 0.1809, "rl_loss": -2.73, "step": 8490 }, { "acc": 0.0, "epoch": 5.75, "learning_rate": 2e-05, "loss": 0.1881, "rl_loss": -2.4838, "step": 8500 }, { "acc": 0.0, "epoch": 5.75, "learning_rate": 2e-05, "loss": 0.1893, "rl_loss": -2.83, "step": 8510 }, { "acc": 0.0, "epoch": 5.76, "learning_rate": 2e-05, "loss": 0.189, "rl_loss": -4.0273, "step": 8520 }, { "acc": 0.0, "epoch": 5.77, "learning_rate": 2e-05, "loss": 0.1806, "rl_loss": -3.859, "step": 8530 }, { "acc": 0.0, "epoch": 5.77, "learning_rate": 2e-05, "loss": 0.2739, "rl_loss": -3.5589, "step": 8540 }, { "acc": 0.0, "epoch": 5.78, "learning_rate": 2e-05, "loss": 0.1856, "rl_loss": -3.5805, "step": 8550 }, { "acc": 0.0, "epoch": 5.79, "learning_rate": 2e-05, "loss": 0.192, "rl_loss": -3.3353, "step": 8560 }, { "acc": 0.0, "epoch": 5.79, "learning_rate": 2e-05, "loss": 0.1822, "rl_loss": -2.3245, "step": 8570 }, { "acc": 0.0, "epoch": 5.8, "learning_rate": 2e-05, "loss": 0.1872, "rl_loss": -1.9409, "step": 8580 }, { "acc": 0.0, "epoch": 5.81, "learning_rate": 2e-05, "loss": 0.1957, "rl_loss": -2.0536, "step": 8590 }, { "acc": 0.0, "epoch": 5.81, "learning_rate": 2e-05, "loss": 0.1846, "rl_loss": -2.0565, "step": 8600 }, { "acc": 0.0, "epoch": 5.82, "learning_rate": 2e-05, "loss": 0.201, "rl_loss": -2.8909, "step": 8610 }, { "acc": 0.0, "epoch": 5.83, "learning_rate": 2e-05, "loss": 0.1928, "rl_loss": -2.1485, "step": 8620 }, { "acc": 0.0, "epoch": 5.84, "learning_rate": 2e-05, "loss": 0.1991, "rl_loss": -2.2741, "step": 8630 }, { "acc": 0.0, "epoch": 5.84, "learning_rate": 2e-05, "loss": 0.1831, "rl_loss": -2.6712, "step": 8640 }, { "acc": 0.0, "epoch": 5.85, "learning_rate": 2e-05, "loss": 0.1925, "rl_loss": -3.8293, "step": 8650 }, { "acc": 0.0, "epoch": 5.86, "learning_rate": 2e-05, "loss": 0.1935, "rl_loss": -2.8727, "step": 8660 }, { "acc": 0.0, "epoch": 5.86, "learning_rate": 2e-05, "loss": 0.1889, "rl_loss": -3.497, "step": 8670 }, { "acc": 0.0, "epoch": 5.87, "learning_rate": 2e-05, "loss": 0.2116, "rl_loss": -2.9592, "step": 8680 }, { "acc": 0.0, "epoch": 5.88, "learning_rate": 2e-05, "loss": 0.1886, "rl_loss": -2.7006, "step": 8690 }, { "acc": 0.0, "epoch": 5.88, "learning_rate": 2e-05, "loss": 0.1926, "rl_loss": -2.5719, "step": 8700 }, { "acc": 0.0, "epoch": 5.89, "learning_rate": 2e-05, "loss": 0.1828, "rl_loss": -3.8962, "step": 8710 }, { "acc": 0.0, "epoch": 5.9, "learning_rate": 2e-05, "loss": 0.1788, "rl_loss": -3.0283, "step": 8720 }, { "acc": 0.0, "epoch": 5.9, "learning_rate": 2e-05, "loss": 0.1839, "rl_loss": -2.7229, "step": 8730 }, { "acc": 0.0, "epoch": 5.91, "learning_rate": 2e-05, "loss": 0.1747, "rl_loss": -3.4372, "step": 8740 }, { "acc": 0.0, "epoch": 5.92, "learning_rate": 2e-05, "loss": 0.1845, "rl_loss": -2.3397, "step": 8750 }, { "acc": 0.0, "epoch": 5.92, "learning_rate": 2e-05, "loss": 0.1953, "rl_loss": -3.8894, "step": 8760 }, { "acc": 0.0, "epoch": 5.93, "learning_rate": 2e-05, "loss": 0.185, "rl_loss": -3.4764, "step": 8770 }, { "acc": 0.0, "epoch": 5.94, "learning_rate": 2e-05, "loss": 0.1819, "rl_loss": -3.6318, "step": 8780 }, { "acc": 0.0, "epoch": 5.94, "learning_rate": 2e-05, "loss": 0.1849, "rl_loss": -4.4896, "step": 8790 }, { "acc": 0.0, "epoch": 5.95, "learning_rate": 2e-05, "loss": 0.1944, "rl_loss": -3.7055, "step": 8800 }, { "acc": 0.0, "epoch": 5.96, "learning_rate": 2e-05, "loss": 0.1843, "rl_loss": -2.4664, "step": 8810 }, { "acc": 0.0, "epoch": 5.96, "learning_rate": 2e-05, "loss": 0.1789, "rl_loss": -2.7583, "step": 8820 }, { "acc": 0.0, "epoch": 5.97, "learning_rate": 2e-05, "loss": 0.1842, "rl_loss": -3.0288, "step": 8830 }, { "acc": 0.0, "epoch": 5.98, "learning_rate": 2e-05, "loss": 0.1844, "rl_loss": -2.8853, "step": 8840 }, { "acc": 0.0, "epoch": 5.98, "learning_rate": 2e-05, "loss": 0.1874, "rl_loss": -3.2796, "step": 8850 }, { "acc": 0.0, "epoch": 5.99, "learning_rate": 2e-05, "loss": 0.1958, "rl_loss": -2.8333, "step": 8860 }, { "acc": 0.0, "epoch": 6.0, "learning_rate": 2e-05, "loss": 0.1898, "rl_loss": -2.2042, "step": 8870 }, { "acc": 0.0, "epoch": 6.0, "learning_rate": 2e-05, "loss": 0.1906, "rl_loss": -3.2543, "step": 8880 }, { "acc": 0.0, "epoch": 6.01, "learning_rate": 2e-05, "loss": 0.1941, "rl_loss": -3.0033, "step": 8890 }, { "acc": 0.0, "epoch": 6.02, "learning_rate": 2e-05, "loss": 0.1779, "rl_loss": -3.2559, "step": 8900 }, { "acc": 0.0, "epoch": 6.02, "learning_rate": 2e-05, "loss": 0.1918, "rl_loss": -4.0929, "step": 8910 }, { "acc": 0.0, "epoch": 6.03, "learning_rate": 2e-05, "loss": 0.2028, "rl_loss": -2.9524, "step": 8920 }, { "acc": 0.0, "epoch": 6.04, "learning_rate": 2e-05, "loss": 0.1886, "rl_loss": -3.6603, "step": 8930 }, { "acc": 0.0, "epoch": 6.04, "learning_rate": 2e-05, "loss": 0.1972, "rl_loss": -3.4814, "step": 8940 }, { "acc": 0.0, "epoch": 6.05, "learning_rate": 2e-05, "loss": 0.186, "rl_loss": -2.7249, "step": 8950 }, { "acc": 0.0, "epoch": 6.06, "learning_rate": 2e-05, "loss": 0.1876, "rl_loss": -2.9313, "step": 8960 }, { "acc": 0.0, "epoch": 6.06, "learning_rate": 2e-05, "loss": 0.2073, "rl_loss": -3.348, "step": 8970 }, { "acc": 0.0, "epoch": 6.07, "learning_rate": 2e-05, "loss": 0.1778, "rl_loss": -2.7439, "step": 8980 }, { "acc": 0.0, "epoch": 6.08, "learning_rate": 2e-05, "loss": 0.1922, "rl_loss": -2.8209, "step": 8990 }, { "acc": 0.0, "epoch": 6.09, "learning_rate": 2e-05, "loss": 0.1754, "rl_loss": -3.3048, "step": 9000 } ], "max_steps": 14790, "num_train_epochs": 10, "total_flos": 1.4614069248e+18, "trial_name": null, "trial_params": null }