Release prep: Phase 3 LM (checkpoints/phase3/p3_lm_full), versions/voices/settings, auto-tagger, Gradio app, CC-BY-SA-4.0 weights + Apache-2.0 code, model card
85bc603 verified | {"step": 0, "val": {"np": 3.563712294792896, "p": 3.571646510972262, "anchor_p": 3.2171030044555664, "mean": 3.567679402882579}} | |
| {"step": 25, "ce": 3.7304, "acc": 0.187, "adv": 8.175, "adv_acc": 0.129, "gn": 2.12, "B": 18, "lr": 1.25e-06, "skips": 0, "oom": 0, "min": 0.24} | |
| {"step": 50, "ce": 4.0527, "acc": 0.1759, "adv": 7.922, "adv_acc": 0.0, "gn": 3.5, "B": 10, "lr": 2.499e-06, "skips": 0, "oom": 0, "min": 0.36} | |
| {"step": 75, "ce": 4.1871, "acc": 0.1463, "adv": 8.59, "adv_acc": 0.009, "gn": 2.74, "B": 53, "lr": 3.748e-06, "skips": 0, "oom": 0, "min": 0.49} | |
| {"step": 100, "ce": 3.4694, "acc": 0.2184, "adv": 4.838, "adv_acc": 0.0, "gn": 2.66, "B": 6, "lr": 4.996e-06, "skips": 1, "oom": 0, "min": 0.61} | |
| {"step": 125, "ce": 3.7797, "acc": 0.1892, "adv": 5.106, "adv_acc": 0.029, "gn": 2.78, "B": 29, "lr": 6.241e-06, "skips": 1, "oom": 0, "min": 0.73} | |
| {"step": 150, "ce": 3.6674, "acc": 0.2085, "adv": 4.72, "adv_acc": 0.001, "gn": 2.17, "B": 14, "lr": 7.485e-06, "skips": 2, "oom": 0, "min": 0.85} | |
| {"step": 175, "ce": 3.8224, "acc": 0.178, "adv": 6.07, "adv_acc": 0.007, "gn": 2.08, "B": 34, "lr": 8.726e-06, "skips": 2, "oom": 0, "min": 0.97} | |
| {"step": 200, "ce": 3.4966, "acc": 0.231, "adv": 5.114, "adv_acc": 0.18, "gn": 2.23, "B": 13, "lr": 9.965e-06, "skips": 2, "oom": 0, "min": 1.09} | |
| {"step": 225, "ce": 3.6002, "acc": 0.2083, "adv": 4.485, "adv_acc": 0.172, "gn": 2.06, "B": 28, "lr": 9.956e-06, "skips": 2, "oom": 0, "min": 1.22} | |
| {"step": 250, "ce": 3.831, "acc": 0.175, "adv": 5.128, "adv_acc": 0.004, "gn": 2.1, "B": 13, "lr": 9.945e-06, "skips": 2, "oom": 0, "min": 1.35} | |
| {"step": 275, "ce": 3.5115, "acc": 0.2121, "adv": 5.416, "adv_acc": 0.0, "gn": 2.1, "B": 17, "lr": 9.933e-06, "skips": 2, "oom": 0, "min": 1.49} | |
| {"step": 300, "ce": 3.6931, "acc": 0.1905, "adv": 5.637, "adv_acc": 0.011, "gn": 1.87, "B": 14, "lr": 9.921e-06, "skips": 2, "oom": 0, "min": 1.61} | |
| {"step": 325, "ce": 3.933, "acc": 0.1594, "adv": 6.282, "adv_acc": 0.022, "gn": 2.38, "B": 41, "lr": 9.907e-06, "skips": 2, "oom": 0, "min": 1.73} | |
| {"step": 350, "ce": 3.3591, "acc": 0.2514, "adv": 4.969, "adv_acc": 0.001, "gn": 2.06, "B": 10, "lr": 9.892e-06, "skips": 2, "oom": 0, "min": 1.85} | |
| {"step": 375, "ce": 3.6679, "acc": 0.1897, "adv": 5.176, "adv_acc": 0.12, "gn": 1.9, "B": 15, "lr": 9.876e-06, "skips": 2, "oom": 0, "min": 1.97} | |
| {"step": 400, "ce": 3.8537, "acc": 0.1769, "adv": 5.109, "adv_acc": 0.024, "gn": 1.81, "B": 23, "lr": 9.859e-06, "skips": 2, "oom": 0, "min": 2.07} | |
| {"step": 425, "ce": 3.6942, "acc": 0.1925, "adv": 4.691, "adv_acc": 0.125, "gn": 2.17, "B": 26, "lr": 9.841e-06, "skips": 2, "oom": 0, "min": 2.2} | |
| {"step": 450, "ce": 3.9306, "acc": 0.1615, "adv": 6.244, "adv_acc": 0.022, "gn": 2.23, "B": 44, "lr": 9.822e-06, "skips": 2, "oom": 0, "min": 2.33} | |
| {"step": 475, "ce": 3.343, "acc": 0.2304, "adv": 4.348, "adv_acc": 0.115, "gn": 2.17, "B": 16, "lr": 9.802e-06, "skips": 2, "oom": 0, "min": 2.45} | |
| {"step": 500, "ce": 3.6741, "acc": 0.1936, "adv": 4.489, "adv_acc": 0.025, "gn": 1.83, "B": 21, "lr": 9.781e-06, "skips": 3, "oom": 0, "min": 2.57} | |
| {"step": 500, "val": {"np": 3.507689257786406, "p": 3.512645748616818, "anchor_p": 3.1165452003479004, "mean": 3.510167503201612}, "best": true} | |
| {"step": 525, "ce": 3.7693, "acc": 0.1915, "adv": 8.908, "adv_acc": 0.0, "gn": 2.37, "B": 59, "lr": 9.758e-06, "skips": 3, "oom": 0, "min": 2.77} | |
| {"step": 550, "ce": 3.5181, "acc": 0.2223, "adv": 4.695, "adv_acc": 0.103, "gn": 2.16, "B": 14, "lr": 9.735e-06, "skips": 4, "oom": 0, "min": 2.89} | |
| {"step": 575, "ce": 3.5077, "acc": 0.2158, "adv": 4.792, "adv_acc": 0.0, "gn": 3.13, "B": 7, "lr": 9.71e-06, "skips": 4, "oom": 0, "min": 3.01} | |
| {"step": 600, "ce": 3.4617, "acc": 0.2337, "adv": 4.699, "adv_acc": 0.032, "gn": 2.14, "B": 10, "lr": 9.685e-06, "skips": 4, "oom": 0, "min": 3.14} | |
| {"step": 625, "ce": 3.5601, "acc": 0.1991, "adv": 4.77, "adv_acc": 0.021, "gn": 2.13, "B": 24, "lr": 9.659e-06, "skips": 5, "oom": 0, "min": 3.25} | |
| {"step": 650, "ce": 3.3011, "acc": 0.2631, "adv": 4.122, "adv_acc": 0.011, "gn": 1.85, "B": 14, "lr": 9.631e-06, "skips": 5, "oom": 0, "min": 3.38} | |
| {"step": 675, "ce": 3.5935, "acc": 0.2031, "adv": 4.653, "adv_acc": 0.033, "gn": 1.88, "B": 19, "lr": 9.602e-06, "skips": 5, "oom": 0, "min": 3.5} | |
| {"step": 700, "ce": 3.2415, "acc": 0.2494, "adv": 4.693, "adv_acc": 0.003, "gn": 2.14, "B": 7, "lr": 9.573e-06, "skips": 5, "oom": 0, "min": 3.63} | |
| {"step": 725, "ce": 3.2985, "acc": 0.2415, "adv": 4.372, "adv_acc": 0.129, "gn": 1.99, "B": 11, "lr": 9.542e-06, "skips": 5, "oom": 0, "min": 3.76} | |
| {"step": 750, "ce": 3.5693, "acc": 0.1964, "adv": 4.211, "adv_acc": 0.047, "gn": 2.05, "B": 27, "lr": 9.511e-06, "skips": 5, "oom": 0, "min": 3.88} | |
| {"step": 775, "ce": 3.26, "acc": 0.2431, "adv": 4.574, "adv_acc": 0.007, "gn": 2.29, "B": 8, "lr": 9.478e-06, "skips": 5, "oom": 0, "min": 4.01} | |
| {"step": 800, "ce": 3.3265, "acc": 0.2363, "adv": 4.495, "adv_acc": 0.047, "gn": 2.32, "B": 9, "lr": 9.445e-06, "skips": 5, "oom": 0, "min": 4.13} | |
| {"step": 825, "ce": 3.7204, "acc": 0.1812, "adv": 7.24, "adv_acc": 0.002, "gn": 2.3, "B": 29, "lr": 9.41e-06, "skips": 5, "oom": 0, "min": 4.27} | |
| {"step": 850, "ce": 3.5532, "acc": 0.2079, "adv": 15.118, "adv_acc": 0.006, "gn": 3.38, "B": 11, "lr": 9.375e-06, "skips": 5, "oom": 0, "min": 4.38} | |
| {"step": 875, "ce": 3.9259, "acc": 0.1713, "adv": 40.286, "adv_acc": 0.017, "gn": 4.35, "B": 41, "lr": 9.338e-06, "skips": 5, "oom": 0, "min": 4.51} | |
| {"step": 900, "ce": 3.9721, "acc": 0.1658, "adv": 43.79, "adv_acc": 0.02, "gn": 6.45, "B": 51, "lr": 9.301e-06, "skips": 5, "oom": 0, "min": 4.62} | |
| {"step": 925, "ce": 3.4603, "acc": 0.2349, "adv": 10.369, "adv_acc": 0.0, "gn": 2.68, "B": 14, "lr": 9.263e-06, "skips": 5, "oom": 0, "min": 4.74} | |
| {"step": 950, "ce": 3.5721, "acc": 0.2092, "adv": 9.951, "adv_acc": 0.032, "gn": 2.45, "B": 20, "lr": 9.223e-06, "skips": 5, "oom": 0, "min": 4.85} | |
| {"step": 975, "ce": 3.7512, "acc": 0.1921, "adv": 38.525, "adv_acc": 0.0, "gn": 5.27, "B": 65, "lr": 9.183e-06, "skips": 5, "oom": 0, "min": 4.97} | |
| {"step": 1000, "ce": 3.4875, "acc": 0.2139, "adv": 9.408, "adv_acc": 0.104, "gn": 2.57, "B": 19, "lr": 9.142e-06, "skips": 5, "oom": 0, "min": 5.1} | |
| {"step": 1000, "val": {"np": 3.4977280627632763, "p": 3.5005001353906886, "anchor_p": 3.1229519844055176, "mean": 3.4991140990769827}, "best": true} | |
| {"step": 1025, "ce": 3.4877, "acc": 0.2195, "adv": 15.951, "adv_acc": 0.0, "gn": 3.29, "B": 16, "lr": 9.1e-06, "skips": 6, "oom": 0, "min": 5.31} | |
| {"step": 1050, "ce": 3.4113, "acc": 0.2332, "adv": 19.778, "adv_acc": 0.08, "gn": 3.95, "B": 9, "lr": 9.057e-06, "skips": 6, "oom": 0, "min": 5.43} | |
| {"step": 1075, "ce": 3.1604, "acc": 0.2612, "adv": 10.481, "adv_acc": 0.001, "gn": 3.24, "B": 7, "lr": 9.014e-06, "skips": 6, "oom": 0, "min": 5.54} | |
| {"step": 1100, "ce": 3.9686, "acc": 0.1677, "adv": 33.668, "adv_acc": 0.0, "gn": 4.65, "B": 58, "lr": 8.969e-06, "skips": 6, "oom": 0, "min": 5.66} | |
| {"step": 1125, "ce": 3.8179, "acc": 0.1783, "adv": 16.777, "adv_acc": 0.002, "gn": 3.08, "B": 36, "lr": 8.924e-06, "skips": 6, "oom": 0, "min": 5.78} | |
| {"step": 1150, "ce": 3.6925, "acc": 0.1956, "adv": 10.705, "adv_acc": 0.012, "gn": 2.96, "B": 14, "lr": 8.877e-06, "skips": 6, "oom": 0, "min": 5.89} | |
| {"step": 1175, "ce": 3.746, "acc": 0.1831, "adv": 11.042, "adv_acc": 0.001, "gn": 2.63, "B": 18, "lr": 8.83e-06, "skips": 6, "oom": 0, "min": 6.01} | |
| {"step": 1200, "ce": 3.6621, "acc": 0.195, "adv": 8.882, "adv_acc": 0.0, "gn": 2.38, "B": 10, "lr": 8.782e-06, "skips": 6, "oom": 0, "min": 6.14} | |
| {"step": 1225, "ce": 3.4611, "acc": 0.2234, "adv": 12.372, "adv_acc": 0.047, "gn": 2.56, "B": 18, "lr": 8.734e-06, "skips": 6, "oom": 0, "min": 6.26} | |
| {"step": 1250, "ce": 3.7915, "acc": 0.1761, "adv": 9.404, "adv_acc": 0.045, "gn": 2.54, "B": 13, "lr": 8.684e-06, "skips": 6, "oom": 0, "min": 6.37} | |
| {"step": 1275, "ce": 3.766, "acc": 0.1962, "adv": 22.222, "adv_acc": 0.0, "gn": 3.98, "B": 57, "lr": 8.634e-06, "skips": 6, "oom": 0, "min": 6.49} | |
| {"step": 1300, "ce": 3.772, "acc": 0.1855, "adv": 9.375, "adv_acc": 0.001, "gn": 2.33, "B": 22, "lr": 8.583e-06, "skips": 6, "oom": 0, "min": 6.61} | |
| {"step": 1325, "ce": 3.2919, "acc": 0.2332, "adv": 4.917, "adv_acc": 0.053, "gn": 2.88, "B": 7, "lr": 8.531e-06, "skips": 6, "oom": 0, "min": 6.73} | |
| {"step": 1350, "ce": 3.3606, "acc": 0.2269, "adv": 8.513, "adv_acc": 0.001, "gn": 2.57, "B": 13, "lr": 8.478e-06, "skips": 6, "oom": 0, "min": 6.85} | |
| {"step": 1375, "ce": 3.4161, "acc": 0.2352, "adv": 6.922, "adv_acc": 0.013, "gn": 2.26, "B": 14, "lr": 8.425e-06, "skips": 6, "oom": 0, "min": 6.96} | |
| {"step": 1400, "ce": 3.325, "acc": 0.2363, "adv": 8.211, "adv_acc": 0.007, "gn": 2.91, "B": 9, "lr": 8.371e-06, "skips": 6, "oom": 0, "min": 7.08} | |
| {"step": 1425, "ce": 3.7772, "acc": 0.1893, "adv": 22.674, "adv_acc": 0.0, "gn": 4.1, "B": 61, "lr": 8.316e-06, "skips": 6, "oom": 0, "min": 7.2} | |
| {"step": 1450, "ce": 3.5065, "acc": 0.2128, "adv": 7.335, "adv_acc": 0.032, "gn": 2.11, "B": 29, "lr": 8.26e-06, "skips": 6, "oom": 0, "min": 7.33} | |
| {"step": 1475, "ce": 3.5474, "acc": 0.2084, "adv": 6.431, "adv_acc": 0.009, "gn": 2.18, "B": 26, "lr": 8.204e-06, "skips": 6, "oom": 0, "min": 7.44} | |
| {"step": 1500, "ce": 3.7764, "acc": 0.1759, "adv": 11.695, "adv_acc": 0.041, "gn": 2.27, "B": 43, "lr": 8.147e-06, "skips": 7, "oom": 0, "min": 7.56} | |
| {"step": 1500, "val": {"np": 3.483489041219705, "p": 3.4874038346815577, "anchor_p": 3.090611457824707, "mean": 3.4854464379506314}, "best": true} | |
| {"step": 1525, "ce": 3.5611, "acc": 0.2128, "adv": 6.366, "adv_acc": 0.017, "gn": 2.16, "B": 12, "lr": 8.09e-06, "skips": 7, "oom": 0, "min": 7.76} | |
| {"step": 1550, "ce": 3.571, "acc": 0.2182, "adv": 11.321, "adv_acc": 0.004, "gn": 3.66, "B": 76, "lr": 8.032e-06, "skips": 7, "oom": 0, "min": 7.89} | |
| {"step": 1575, "ce": 3.4323, "acc": 0.2329, "adv": 6.34, "adv_acc": 0.015, "gn": 2.26, "B": 10, "lr": 7.973e-06, "skips": 7, "oom": 0, "min": 7.99} | |
| {"step": 1600, "ce": 3.7183, "acc": 0.1903, "adv": 11.932, "adv_acc": 0.005, "gn": 3.07, "B": 51, "lr": 7.914e-06, "skips": 7, "oom": 0, "min": 8.1} | |
| {"step": 1625, "ce": 3.4348, "acc": 0.2274, "adv": 6.191, "adv_acc": 0.005, "gn": 2.24, "B": 12, "lr": 7.854e-06, "skips": 7, "oom": 0, "min": 8.23} | |
| {"step": 1650, "ce": 3.5439, "acc": 0.1908, "adv": 7.033, "adv_acc": 0.0, "gn": 2.77, "B": 6, "lr": 7.793e-06, "skips": 7, "oom": 0, "min": 8.35} | |
| {"step": 1675, "ce": 3.7204, "acc": 0.1895, "adv": 6.672, "adv_acc": 0.041, "gn": 1.98, "B": 27, "lr": 7.732e-06, "skips": 7, "oom": 0, "min": 8.47} | |
| {"step": 1700, "ce": 3.6902, "acc": 0.1898, "adv": 6.503, "adv_acc": 0.031, "gn": 2.17, "B": 29, "lr": 7.67e-06, "skips": 8, "oom": 0, "min": 8.6} | |
| {"step": 1725, "ce": 3.5412, "acc": 0.2078, "adv": 6.138, "adv_acc": 0.022, "gn": 2.09, "B": 21, "lr": 7.608e-06, "skips": 8, "oom": 0, "min": 8.73} | |
| {"step": 1750, "ce": 3.5848, "acc": 0.208, "adv": 6.601, "adv_acc": 0.056, "gn": 2.2, "B": 10, "lr": 7.545e-06, "skips": 8, "oom": 0, "min": 8.84} | |
| {"step": 1775, "ce": 3.4334, "acc": 0.2558, "adv": 18.173, "adv_acc": 0.0, "gn": 3.74, "B": 69, "lr": 7.482e-06, "skips": 8, "oom": 0, "min": 8.95} | |
| {"step": 1800, "ce": 3.3952, "acc": 0.2212, "adv": 4.707, "adv_acc": 0.064, "gn": 2.3, "B": 9, "lr": 7.419e-06, "skips": 8, "oom": 0, "min": 9.07} | |
| {"step": 1825, "ce": 3.6546, "acc": 0.2024, "adv": 6.154, "adv_acc": 0.044, "gn": 2.32, "B": 33, "lr": 7.354e-06, "skips": 8, "oom": 0, "min": 9.19} | |
| {"step": 1850, "ce": 3.5126, "acc": 0.2026, "adv": 6.185, "adv_acc": 0.01, "gn": 2.16, "B": 26, "lr": 7.29e-06, "skips": 8, "oom": 0, "min": 9.3} | |
| {"step": 1875, "ce": 3.0871, "acc": 0.2778, "adv": 4.731, "adv_acc": 0.098, "gn": 2.17, "B": 10, "lr": 7.225e-06, "skips": 8, "oom": 0, "min": 9.41} | |
| {"step": 1900, "ce": 3.1692, "acc": 0.2457, "adv": 5.498, "adv_acc": 0.015, "gn": 2.32, "B": 6, "lr": 7.159e-06, "skips": 9, "oom": 0, "min": 9.53} | |
| {"step": 1925, "ce": 3.6654, "acc": 0.1816, "adv": 11.98, "adv_acc": 0.001, "gn": 2.36, "B": 41, "lr": 7.093e-06, "skips": 9, "oom": 0, "min": 9.65} | |
| {"step": 1950, "ce": 3.691, "acc": 0.1926, "adv": 15.862, "adv_acc": 0.0, "gn": 3.8, "B": 67, "lr": 7.027e-06, "skips": 9, "oom": 0, "min": 9.76} | |
| {"step": 1975, "ce": 3.569, "acc": 0.2165, "adv": 5.811, "adv_acc": 0.034, "gn": 2.0, "B": 16, "lr": 6.96e-06, "skips": 9, "oom": 0, "min": 9.89} | |
| {"step": 2000, "ce": 3.6569, "acc": 0.2025, "adv": 5.672, "adv_acc": 0.011, "gn": 2.62, "B": 24, "lr": 6.893e-06, "skips": 9, "oom": 0, "min": 10.01} | |
| {"step": 2000, "val": {"np": 3.4673069890236623, "p": 3.473822727265498, "anchor_p": 3.0952401161193848, "mean": 3.47056485814458}, "best": true} | |
| {"step": 2025, "ce": 3.5417, "acc": 0.2055, "adv": 6.049, "adv_acc": 0.031, "gn": 2.45, "B": 10, "lr": 6.826e-06, "skips": 9, "oom": 0, "min": 10.2} | |
| {"step": 2050, "ce": 3.3296, "acc": 0.2364, "adv": 5.457, "adv_acc": 0.018, "gn": 2.29, "B": 8, "lr": 6.758e-06, "skips": 9, "oom": 0, "min": 10.32} | |
| {"step": 2075, "ce": 3.39, "acc": 0.2286, "adv": 5.628, "adv_acc": 0.001, "gn": 2.08, "B": 13, "lr": 6.69e-06, "skips": 9, "oom": 0, "min": 10.43} | |
| {"step": 2100, "ce": 3.7423, "acc": 0.1898, "adv": 10.36, "adv_acc": 0.003, "gn": 2.33, "B": 43, "lr": 6.622e-06, "skips": 9, "oom": 0, "min": 10.56} | |
| {"step": 2125, "ce": 3.6225, "acc": 0.2003, "adv": 6.223, "adv_acc": 0.005, "gn": 1.99, "B": 13, "lr": 6.553e-06, "skips": 9, "oom": 0, "min": 10.68} | |
| {"step": 2150, "ce": 3.4632, "acc": 0.2213, "adv": 5.662, "adv_acc": 0.0, "gn": 1.88, "B": 16, "lr": 6.484e-06, "skips": 9, "oom": 0, "min": 10.8} | |
| {"step": 2175, "ce": 3.2873, "acc": 0.2562, "adv": 5.891, "adv_acc": 0.004, "gn": 2.14, "B": 10, "lr": 6.415e-06, "skips": 9, "oom": 0, "min": 10.92} | |
| {"step": 2200, "ce": 3.9016, "acc": 0.1671, "adv": 8.882, "adv_acc": 0.005, "gn": 2.36, "B": 41, "lr": 6.346e-06, "skips": 9, "oom": 0, "min": 11.03} | |
| {"step": 2225, "ce": 2.9077, "acc": 0.3194, "adv": 5.976, "adv_acc": 0.013, "gn": 2.83, "B": 7, "lr": 6.276e-06, "skips": 9, "oom": 0, "min": 11.15} | |
| {"step": 2250, "ce": 3.6433, "acc": 0.2093, "adv": 13.079, "adv_acc": 0.001, "gn": 3.92, "B": 54, "lr": 6.207e-06, "skips": 9, "oom": 0, "min": 11.26} | |
| {"step": 2275, "ce": 3.4893, "acc": 0.2168, "adv": 5.574, "adv_acc": 0.06, "gn": 1.87, "B": 14, "lr": 6.137e-06, "skips": 10, "oom": 0, "min": 11.38} | |
| {"step": 2300, "ce": 3.704, "acc": 0.1807, "adv": 10.287, "adv_acc": 0.0, "gn": 2.77, "B": 50, "lr": 6.067e-06, "skips": 10, "oom": 0, "min": 11.5} | |
| {"step": 2325, "ce": 3.8156, "acc": 0.1893, "adv": 13.466, "adv_acc": 0.0, "gn": 3.07, "B": 55, "lr": 5.997e-06, "skips": 10, "oom": 0, "min": 11.61} | |
| {"step": 2350, "ce": 3.5805, "acc": 0.2117, "adv": 14.332, "adv_acc": 0.005, "gn": 3.27, "B": 68, "lr": 5.926e-06, "skips": 10, "oom": 0, "min": 11.74} | |
| {"step": 2375, "ce": 3.3738, "acc": 0.2044, "adv": 6.482, "adv_acc": 0.0, "gn": 2.93, "B": 6, "lr": 5.856e-06, "skips": 10, "oom": 0, "min": 11.87} | |
| {"step": 2400, "ce": 3.6756, "acc": 0.1828, "adv": 6.471, "adv_acc": 0.005, "gn": 1.93, "B": 12, "lr": 5.785e-06, "skips": 10, "oom": 0, "min": 11.99} | |
| {"step": 2425, "ce": 3.3128, "acc": 0.2365, "adv": 6.476, "adv_acc": 0.013, "gn": 1.89, "B": 10, "lr": 5.715e-06, "skips": 10, "oom": 0, "min": 12.1} | |
| {"step": 2450, "ce": 3.4673, "acc": 0.2181, "adv": 6.946, "adv_acc": 0.015, "gn": 2.02, "B": 33, "lr": 5.644e-06, "skips": 10, "oom": 0, "min": 12.22} | |
| {"step": 2475, "ce": 3.5298, "acc": 0.2094, "adv": 6.09, "adv_acc": 0.039, "gn": 1.71, "B": 19, "lr": 5.574e-06, "skips": 10, "oom": 0, "min": 12.32} | |
| {"step": 2500, "ce": 3.2715, "acc": 0.2515, "adv": 6.632, "adv_acc": 0.121, "gn": 1.87, "B": 17, "lr": 5.503e-06, "skips": 10, "oom": 0, "min": 12.43} | |
| {"step": 2500, "val": {"np": 3.455619011329129, "p": 3.4601697774197455, "anchor_p": 3.0822691917419434, "mean": 3.4578943943744376}, "best": true} | |
| {"step": 2525, "ce": 3.7506, "acc": 0.183, "adv": 15.727, "adv_acc": 0.002, "gn": 2.97, "B": 55, "lr": 5.432e-06, "skips": 10, "oom": 0, "min": 12.63} | |
| {"step": 2550, "ce": 3.2503, "acc": 0.2422, "adv": 6.394, "adv_acc": 0.012, "gn": 2.1, "B": 10, "lr": 5.361e-06, "skips": 11, "oom": 0, "min": 12.75} | |
| {"step": 2575, "ce": 3.3457, "acc": 0.2401, "adv": 6.05, "adv_acc": 0.119, "gn": 1.93, "B": 17, "lr": 5.291e-06, "skips": 11, "oom": 0, "min": 12.86} | |
| {"step": 2600, "ce": 3.6437, "acc": 0.1846, "adv": 6.671, "adv_acc": 0.061, "gn": 1.87, "B": 29, "lr": 5.22e-06, "skips": 11, "oom": 0, "min": 12.97} | |
| {"step": 2625, "ce": 3.5463, "acc": 0.2367, "adv": 14.848, "adv_acc": 0.003, "gn": 3.72, "B": 74, "lr": 5.15e-06, "skips": 11, "oom": 0, "min": 13.09} | |
| {"step": 2650, "ce": 3.3847, "acc": 0.2314, "adv": 6.816, "adv_acc": 0.016, "gn": 2.39, "B": 9, "lr": 5.079e-06, "skips": 11, "oom": 0, "min": 13.2} | |
| {"step": 2675, "ce": 3.5453, "acc": 0.2041, "adv": 6.815, "adv_acc": 0.003, "gn": 2.07, "B": 23, "lr": 5.009e-06, "skips": 11, "oom": 0, "min": 13.31} | |
| {"step": 2700, "ce": 3.3028, "acc": 0.2437, "adv": 7.003, "adv_acc": 0.128, "gn": 1.87, "B": 15, "lr": 4.939e-06, "skips": 11, "oom": 0, "min": 13.43} | |
| {"step": 2725, "ce": 3.8053, "acc": 0.1819, "adv": 9.202, "adv_acc": 0.03, "gn": 2.73, "B": 42, "lr": 4.869e-06, "skips": 11, "oom": 0, "min": 13.54} | |
| {"step": 2750, "ce": 3.586, "acc": 0.2259, "adv": 10.705, "adv_acc": 0.038, "gn": 3.32, "B": 76, "lr": 4.799e-06, "skips": 12, "oom": 0, "min": 13.66} | |
| {"step": 2775, "ce": 3.242, "acc": 0.2543, "adv": 7.569, "adv_acc": 0.147, "gn": 1.83, "B": 14, "lr": 4.729e-06, "skips": 12, "oom": 0, "min": 13.77} | |
| {"step": 2800, "ce": 3.5247, "acc": 0.2002, "adv": 7.495, "adv_acc": 0.0, "gn": 1.98, "B": 12, "lr": 4.66e-06, "skips": 12, "oom": 0, "min": 13.88} | |
| {"step": 2825, "ce": 3.5414, "acc": 0.2004, "adv": 7.73, "adv_acc": 0.06, "gn": 2.6, "B": 40, "lr": 4.59e-06, "skips": 12, "oom": 0, "min": 13.99} | |
| {"step": 2850, "ce": 3.2731, "acc": 0.2393, "adv": 7.595, "adv_acc": 0.051, "gn": 1.9, "B": 21, "lr": 4.521e-06, "skips": 12, "oom": 0, "min": 14.1} | |
| {"step": 2875, "ce": 3.427, "acc": 0.2152, "adv": 7.72, "adv_acc": 0.0, "gn": 1.96, "B": 10, "lr": 4.452e-06, "skips": 13, "oom": 0, "min": 14.23} | |
| {"step": 2900, "ce": 3.4472, "acc": 0.2124, "adv": 7.726, "adv_acc": 0.0, "gn": 1.9, "B": 14, "lr": 4.384e-06, "skips": 13, "oom": 0, "min": 14.34} | |
| {"step": 2925, "ce": 3.3561, "acc": 0.228, "adv": 7.77, "adv_acc": 0.139, "gn": 1.95, "B": 17, "lr": 4.315e-06, "skips": 13, "oom": 0, "min": 14.45} | |
| {"step": 2950, "ce": 3.6581, "acc": 0.2103, "adv": 8.242, "adv_acc": 0.001, "gn": 2.15, "B": 53, "lr": 4.247e-06, "skips": 13, "oom": 0, "min": 14.58} | |
| {"step": 2975, "ce": 3.5687, "acc": 0.2064, "adv": 7.748, "adv_acc": 0.0, "gn": 2.09, "B": 11, "lr": 4.18e-06, "skips": 13, "oom": 0, "min": 14.69} | |
| {"step": 3000, "ce": 3.8188, "acc": 0.1662, "adv": 8.003, "adv_acc": 0.01, "gn": 11.11, "B": 52, "lr": 4.112e-06, "skips": 13, "oom": 0, "min": 14.81} | |
| {"step": 3000, "val": {"np": 3.458494051271619, "p": 3.4631192987050605, "anchor_p": 3.0834555625915527, "mean": 3.4608066749883397}, "best": false} | |
| {"step": 3025, "ce": 3.8608, "acc": 0.1688, "adv": 7.726, "adv_acc": 0.001, "gn": 2.45, "B": 47, "lr": 4.045e-06, "skips": 14, "oom": 0, "min": 14.96} | |
| {"step": 3050, "ce": 3.5531, "acc": 0.2074, "adv": 7.65, "adv_acc": 0.193, "gn": 2.03, "B": 13, "lr": 3.978e-06, "skips": 14, "oom": 0, "min": 15.07} | |
| {"step": 3075, "ce": 3.6485, "acc": 0.2358, "adv": 8.172, "adv_acc": 0.0, "gn": 2.18, "B": 54, "lr": 3.912e-06, "skips": 14, "oom": 0, "min": 15.19} | |
| {"step": 3100, "ce": 3.6049, "acc": 0.1975, "adv": 7.879, "adv_acc": 0.015, "gn": 1.97, "B": 43, "lr": 3.846e-06, "skips": 14, "oom": 0, "min": 15.3} | |
| {"step": 3125, "ce": 3.4728, "acc": 0.2281, "adv": 7.707, "adv_acc": 0.002, "gn": 1.97, "B": 15, "lr": 3.781e-06, "skips": 14, "oom": 0, "min": 15.4} | |
| {"step": 3150, "ce": 3.1627, "acc": 0.248, "adv": 7.661, "adv_acc": 0.001, "gn": 2.15, "B": 7, "lr": 3.715e-06, "skips": 14, "oom": 0, "min": 15.51} | |
| {"step": 3175, "ce": 3.632, "acc": 0.1925, "adv": 8.086, "adv_acc": 0.0, "gn": 2.32, "B": 61, "lr": 3.651e-06, "skips": 15, "oom": 0, "min": 15.64} | |
| {"step": 3200, "ce": 3.1466, "acc": 0.2697, "adv": 7.568, "adv_acc": 0.001, "gn": 2.21, "B": 8, "lr": 3.587e-06, "skips": 15, "oom": 0, "min": 15.76} | |
| {"step": 3225, "ce": 3.4873, "acc": 0.2071, "adv": 7.532, "adv_acc": 0.0, "gn": 2.03, "B": 15, "lr": 3.523e-06, "skips": 15, "oom": 0, "min": 15.87} | |
| {"step": 3250, "ce": 3.4786, "acc": 0.2021, "adv": 7.501, "adv_acc": 0.078, "gn": 1.92, "B": 24, "lr": 3.46e-06, "skips": 15, "oom": 0, "min": 15.97} | |
| {"step": 3275, "ce": 3.8398, "acc": 0.1564, "adv": 7.779, "adv_acc": 0.0, "gn": 2.0, "B": 45, "lr": 3.397e-06, "skips": 15, "oom": 0, "min": 16.09} | |
| {"step": 3300, "ce": 3.4721, "acc": 0.1948, "adv": 7.401, "adv_acc": 0.0, "gn": 2.6, "B": 6, "lr": 3.335e-06, "skips": 15, "oom": 0, "min": 16.2} | |
| {"step": 3325, "ce": 3.4505, "acc": 0.2151, "adv": 7.363, "adv_acc": 0.037, "gn": 2.08, "B": 19, "lr": 3.273e-06, "skips": 17, "oom": 0, "min": 16.32} | |
| {"step": 3350, "ce": 3.6933, "acc": 0.1805, "adv": 8.033, "adv_acc": 0.0, "gn": 2.17, "B": 54, "lr": 3.212e-06, "skips": 17, "oom": 0, "min": 16.43} | |
| {"step": 3375, "ce": 3.6515, "acc": 0.1935, "adv": 7.364, "adv_acc": 0.001, "gn": 1.94, "B": 26, "lr": 3.151e-06, "skips": 17, "oom": 0, "min": 16.54} | |
| {"step": 3400, "ce": 3.3099, "acc": 0.2384, "adv": 7.2, "adv_acc": 0.241, "gn": 1.99, "B": 18, "lr": 3.091e-06, "skips": 18, "oom": 0, "min": 16.66} | |
| {"step": 3425, "ce": 3.4313, "acc": 0.2122, "adv": 7.262, "adv_acc": 0.0, "gn": 2.29, "B": 11, "lr": 3.032e-06, "skips": 18, "oom": 0, "min": 16.77} | |
| {"step": 3450, "ce": 3.1878, "acc": 0.2464, "adv": 7.173, "adv_acc": 0.109, "gn": 1.89, "B": 18, "lr": 2.973e-06, "skips": 18, "oom": 0, "min": 16.88} | |
| {"step": 3475, "ce": 3.1116, "acc": 0.2702, "adv": 7.052, "adv_acc": 0.115, "gn": 2.0, "B": 12, "lr": 2.915e-06, "skips": 18, "oom": 0, "min": 16.98} | |
| {"step": 3500, "ce": 3.69, "acc": 0.183, "adv": 7.387, "adv_acc": 0.001, "gn": 2.04, "B": 44, "lr": 2.857e-06, "skips": 18, "oom": 0, "min": 17.1} | |
| {"step": 3500, "val": {"np": 3.450007374201225, "p": 3.453779433371578, "anchor_p": 3.084826946258545, "mean": 3.4518934037864017}, "best": true} | |
| {"step": 3525, "ce": 3.8386, "acc": 0.1708, "adv": 7.872, "adv_acc": 0.0, "gn": 2.22, "B": 59, "lr": 2.8e-06, "skips": 18, "oom": 0, "min": 17.29} | |
| {"step": 3550, "ce": 3.3008, "acc": 0.2418, "adv": 6.87, "adv_acc": 0.256, "gn": 2.36, "B": 18, "lr": 2.744e-06, "skips": 18, "oom": 0, "min": 17.39} | |
| {"step": 3575, "ce": 3.2114, "acc": 0.2499, "adv": 6.952, "adv_acc": 0.143, "gn": 2.01, "B": 25, "lr": 2.689e-06, "skips": 18, "oom": 0, "min": 17.49} | |
| {"step": 3600, "ce": 3.631, "acc": 0.2007, "adv": 7.465, "adv_acc": 0.0, "gn": 2.28, "B": 48, "lr": 2.634e-06, "skips": 18, "oom": 0, "min": 17.59} | |
| {"step": 3625, "ce": 3.0395, "acc": 0.2746, "adv": 6.853, "adv_acc": 0.365, "gn": 2.04, "B": 14, "lr": 2.58e-06, "skips": 19, "oom": 0, "min": 17.71} | |
| {"step": 3650, "ce": 3.2907, "acc": 0.2407, "adv": 6.957, "adv_acc": 0.059, "gn": 2.04, "B": 19, "lr": 2.526e-06, "skips": 19, "oom": 0, "min": 17.82} | |
| {"step": 3675, "ce": 3.4421, "acc": 0.219, "adv": 6.903, "adv_acc": 0.078, "gn": 2.07, "B": 18, "lr": 2.474e-06, "skips": 20, "oom": 0, "min": 17.94} | |
| {"step": 3700, "ce": 3.3729, "acc": 0.2236, "adv": 6.87, "adv_acc": 0.087, "gn": 2.08, "B": 11, "lr": 2.422e-06, "skips": 21, "oom": 0, "min": 18.05} | |
| {"step": 3725, "ce": 3.8656, "acc": 0.1712, "adv": 7.638, "adv_acc": 0.001, "gn": 2.12, "B": 48, "lr": 2.37e-06, "skips": 21, "oom": 0, "min": 18.16} | |
| {"step": 3750, "ce": 3.1599, "acc": 0.251, "adv": 6.81, "adv_acc": 0.002, "gn": 2.31, "B": 9, "lr": 2.32e-06, "skips": 21, "oom": 0, "min": 18.26} | |
| {"step": 3775, "ce": 3.597, "acc": 0.2018, "adv": 7.29, "adv_acc": 0.0, "gn": 2.19, "B": 45, "lr": 2.27e-06, "skips": 21, "oom": 0, "min": 18.35} | |
| {"step": 3800, "ce": 3.4677, "acc": 0.2096, "adv": 7.013, "adv_acc": 0.021, "gn": 2.12, "B": 38, "lr": 2.222e-06, "skips": 21, "oom": 0, "min": 18.46} | |
| {"step": 3825, "ce": 3.8072, "acc": 0.162, "adv": 7.709, "adv_acc": 0.0, "gn": 2.54, "B": 56, "lr": 2.174e-06, "skips": 21, "oom": 0, "min": 18.57} | |
| {"step": 3850, "ce": 3.3768, "acc": 0.2393, "adv": 6.691, "adv_acc": 0.07, "gn": 2.16, "B": 15, "lr": 2.126e-06, "skips": 22, "oom": 0, "min": 18.68} | |
| {"step": 3875, "ce": 3.4887, "acc": 0.2255, "adv": 8.748, "adv_acc": 0.001, "gn": 2.49, "B": 57, "lr": 2.08e-06, "skips": 22, "oom": 0, "min": 18.78} | |
| {"step": 3900, "ce": 3.3054, "acc": 0.2282, "adv": 6.594, "adv_acc": 0.169, "gn": 1.96, "B": 19, "lr": 2.034e-06, "skips": 23, "oom": 0, "min": 18.89} | |
| {"step": 3925, "ce": 3.366, "acc": 0.2343, "adv": 6.587, "adv_acc": 0.001, "gn": 1.94, "B": 25, "lr": 1.99e-06, "skips": 24, "oom": 0, "min": 19.01} | |
| {"step": 3950, "ce": 3.3814, "acc": 0.2161, "adv": 6.627, "adv_acc": 0.0, "gn": 2.39, "B": 8, "lr": 1.946e-06, "skips": 24, "oom": 0, "min": 19.13} | |
| {"step": 3975, "ce": 3.8303, "acc": 0.1731, "adv": 7.142, "adv_acc": 0.001, "gn": 2.05, "B": 40, "lr": 1.903e-06, "skips": 24, "oom": 0, "min": 19.25} | |
| {"step": 4000, "ce": 3.3377, "acc": 0.2298, "adv": 6.495, "adv_acc": 0.094, "gn": 2.11, "B": 14, "lr": 1.861e-06, "skips": 24, "oom": 0, "min": 19.34} | |
| {"step": 4000, "val": {"np": 3.4431558952269414, "p": 3.446689893057758, "anchor_p": 3.0781209468841553, "mean": 3.4449228941423495}, "best": true} | |
| {"step": 4025, "ce": 3.6726, "acc": 0.2001, "adv": 7.533, "adv_acc": 0.0, "gn": 2.45, "B": 48, "lr": 1.82e-06, "skips": 25, "oom": 0, "min": 19.52} | |
| {"step": 4050, "ce": 3.3085, "acc": 0.241, "adv": 6.482, "adv_acc": 0.087, "gn": 1.9, "B": 19, "lr": 1.78e-06, "skips": 26, "oom": 0, "min": 19.63} | |
| {"step": 4075, "ce": 3.3112, "acc": 0.2284, "adv": 6.608, "adv_acc": 0.0, "gn": 2.34, "B": 7, "lr": 1.74e-06, "skips": 26, "oom": 0, "min": 19.73} | |
| {"step": 4100, "ce": 3.2196, "acc": 0.2551, "adv": 6.462, "adv_acc": 0.062, "gn": 2.0, "B": 18, "lr": 1.702e-06, "skips": 27, "oom": 0, "min": 19.83} | |
| {"step": 4125, "ce": 3.3942, "acc": 0.2193, "adv": 6.398, "adv_acc": 0.072, "gn": 2.13, "B": 24, "lr": 1.665e-06, "skips": 27, "oom": 0, "min": 19.93} | |
| {"step": 4150, "ce": 3.5767, "acc": 0.1964, "adv": 6.64, "adv_acc": 0.015, "gn": 2.21, "B": 37, "lr": 1.628e-06, "skips": 27, "oom": 0, "min": 20.05} | |
| {"step": 4175, "ce": 3.1631, "acc": 0.2804, "adv": 8.278, "adv_acc": 0.0, "gn": 2.84, "B": 70, "lr": 1.593e-06, "skips": 27, "oom": 0, "min": 20.15} | |
| {"step": 4200, "ce": 3.5769, "acc": 0.2021, "adv": 6.396, "adv_acc": 0.045, "gn": 3.32, "B": 29, "lr": 1.558e-06, "skips": 27, "oom": 0, "min": 20.25} | |
| {"step": 4225, "ce": 3.4182, "acc": 0.228, "adv": 6.34, "adv_acc": 0.026, "gn": 1.91, "B": 23, "lr": 1.524e-06, "skips": 27, "oom": 0, "min": 20.34} | |
| {"step": 4250, "ce": 3.4323, "acc": 0.1862, "adv": 6.416, "adv_acc": 0.0, "gn": 4.24, "B": 4, "lr": 1.492e-06, "skips": 27, "oom": 0, "min": 20.46} | |
| {"step": 4275, "ce": 3.2959, "acc": 0.2454, "adv": 6.336, "adv_acc": 0.074, "gn": 2.14, "B": 10, "lr": 1.46e-06, "skips": 27, "oom": 0, "min": 20.56} | |
| {"step": 4300, "ce": 3.3964, "acc": 0.2241, "adv": 6.433, "adv_acc": 0.0, "gn": 2.44, "B": 9, "lr": 1.429e-06, "skips": 27, "oom": 0, "min": 20.67} | |
| {"step": 4325, "ce": 3.1359, "acc": 0.2704, "adv": 6.4, "adv_acc": 0.0, "gn": 2.44, "B": 8, "lr": 1.4e-06, "skips": 27, "oom": 0, "min": 20.78} | |
| {"step": 4350, "ce": 3.3125, "acc": 0.2263, "adv": 6.296, "adv_acc": 0.024, "gn": 2.0, "B": 21, "lr": 1.371e-06, "skips": 27, "oom": 0, "min": 20.87} | |
| {"step": 4375, "ce": 3.7765, "acc": 0.1722, "adv": 6.958, "adv_acc": 0.003, "gn": 2.27, "B": 44, "lr": 1.344e-06, "skips": 28, "oom": 0, "min": 20.98} | |
| {"step": 4400, "ce": 3.802, "acc": 0.1807, "adv": 6.978, "adv_acc": 0.048, "gn": 2.29, "B": 44, "lr": 1.317e-06, "skips": 28, "oom": 0, "min": 21.1} | |
| {"step": 4425, "ce": 3.9622, "acc": 0.1576, "adv": 7.175, "adv_acc": 0.001, "gn": 2.23, "B": 53, "lr": 1.292e-06, "skips": 28, "oom": 0, "min": 21.22} | |
| {"step": 4450, "ce": 3.5589, "acc": 0.1894, "adv": 6.487, "adv_acc": 0.121, "gn": 2.24, "B": 38, "lr": 1.267e-06, "skips": 28, "oom": 0, "min": 21.33} | |
| {"step": 4475, "ce": 3.5869, "acc": 0.1958, "adv": 6.216, "adv_acc": 0.091, "gn": 2.13, "B": 21, "lr": 1.244e-06, "skips": 28, "oom": 0, "min": 21.44} | |
| {"step": 4500, "ce": 3.5748, "acc": 0.2104, "adv": 8.296, "adv_acc": 0.001, "gn": 2.57, "B": 62, "lr": 1.221e-06, "skips": 28, "oom": 0, "min": 21.54} | |
| {"step": 4500, "val": {"np": 3.4411249983970817, "p": 3.4445827232509947, "anchor_p": 3.079256296157837, "mean": 3.442853860824038}, "best": true} | |
| {"step": 4525, "ce": 3.6671, "acc": 0.1935, "adv": 7.579, "adv_acc": 0.0, "gn": 2.41, "B": 61, "lr": 1.2e-06, "skips": 28, "oom": 0, "min": 21.73} | |
| {"step": 4550, "ce": 3.5396, "acc": 0.2035, "adv": 6.112, "adv_acc": 0.033, "gn": 2.28, "B": 28, "lr": 1.179e-06, "skips": 28, "oom": 0, "min": 21.84} | |
| {"step": 4575, "ce": 3.0832, "acc": 0.2803, "adv": 6.234, "adv_acc": 0.0, "gn": 2.13, "B": 8, "lr": 1.16e-06, "skips": 29, "oom": 0, "min": 21.94} | |
| {"step": 4600, "ce": 3.5924, "acc": 0.1956, "adv": 6.252, "adv_acc": 0.001, "gn": 2.17, "B": 29, "lr": 1.142e-06, "skips": 29, "oom": 0, "min": 22.05} | |
| {"step": 4625, "ce": 3.4295, "acc": 0.2224, "adv": 6.193, "adv_acc": 0.043, "gn": 2.09, "B": 20, "lr": 1.125e-06, "skips": 30, "oom": 0, "min": 22.17} | |
| {"step": 4650, "ce": 3.3532, "acc": 0.2302, "adv": 6.142, "adv_acc": 0.181, "gn": 2.09, "B": 22, "lr": 1.109e-06, "skips": 30, "oom": 0, "min": 22.29} | |
| {"step": 4675, "ce": 3.6147, "acc": 0.2032, "adv": 6.351, "adv_acc": 0.031, "gn": 2.01, "B": 35, "lr": 1.094e-06, "skips": 30, "oom": 0, "min": 22.39} | |
| {"step": 4700, "ce": 3.5387, "acc": 0.2016, "adv": 6.802, "adv_acc": 0.0, "gn": 2.13, "B": 43, "lr": 1.08e-06, "skips": 30, "oom": 0, "min": 22.51} | |
| {"step": 4725, "ce": 3.5325, "acc": 0.2115, "adv": 6.492, "adv_acc": 0.001, "gn": 2.37, "B": 41, "lr": 1.067e-06, "skips": 30, "oom": 0, "min": 22.62} | |
| {"step": 4750, "ce": 3.3788, "acc": 0.2414, "adv": 6.013, "adv_acc": 0.256, "gn": 1.93, "B": 16, "lr": 1.056e-06, "skips": 30, "oom": 0, "min": 22.73} | |
| {"step": 4775, "ce": 3.4268, "acc": 0.228, "adv": 6.181, "adv_acc": 0.0, "gn": 2.18, "B": 13, "lr": 1.045e-06, "skips": 30, "oom": 0, "min": 22.83} | |
| {"step": 4800, "ce": 3.4523, "acc": 0.2268, "adv": 6.0, "adv_acc": 0.077, "gn": 2.02, "B": 17, "lr": 1.036e-06, "skips": 30, "oom": 0, "min": 22.94} | |
| {"step": 4825, "ce": 3.5242, "acc": 0.2296, "adv": 7.628, "adv_acc": 0.044, "gn": 2.09, "B": 51, "lr": 1.027e-06, "skips": 30, "oom": 0, "min": 23.05} | |
| {"step": 4850, "ce": 3.1302, "acc": 0.2662, "adv": 5.944, "adv_acc": 0.324, "gn": 2.04, "B": 12, "lr": 1.02e-06, "skips": 30, "oom": 0, "min": 23.14} | |
| {"step": 4875, "ce": 3.4428, "acc": 0.2185, "adv": 6.03, "adv_acc": 0.165, "gn": 2.1, "B": 15, "lr": 1.014e-06, "skips": 30, "oom": 0, "min": 23.25} | |
| {"step": 4900, "ce": 3.1318, "acc": 0.2708, "adv": 6.019, "adv_acc": 0.0, "gn": 2.6, "B": 6, "lr": 1.009e-06, "skips": 30, "oom": 0, "min": 23.35} | |
| {"step": 4925, "ce": 3.5066, "acc": 0.2196, "adv": 6.043, "adv_acc": 0.039, "gn": 1.91, "B": 24, "lr": 1.005e-06, "skips": 30, "oom": 0, "min": 23.44} | |
| {"step": 4950, "ce": 3.4736, "acc": 0.2202, "adv": 7.73, "adv_acc": 0.0, "gn": 3.54, "B": 69, "lr": 1.002e-06, "skips": 30, "oom": 0, "min": 23.54} | |
| {"step": 4975, "ce": 3.0291, "acc": 0.2834, "adv": 5.771, "adv_acc": 0.001, "gn": 2.29, "B": 9, "lr": 1.001e-06, "skips": 30, "oom": 0, "min": 23.64} | |
| {"step": 5000, "ce": 3.4741, "acc": 0.2143, "adv": 5.99, "adv_acc": 0.0, "gn": 1.92, "B": 15, "lr": 1e-06, "skips": 30, "oom": 0, "min": 23.74} | |
| {"step": 5000, "val": {"np": 3.4417521021653465, "p": 3.445160358658831, "anchor_p": 3.078298807144165, "mean": 3.443456230412089}, "best": false} | |