diff --git "a/train_loop2_full.json" "b/train_loop2_full.json" new file mode 100644--- /dev/null +++ "b/train_loop2_full.json" @@ -0,0 +1,56862 @@ +{ + "tag": "loop2_full", + "arch": "loop2", + "params_M": 23.85, + "dim": 384, + "in": 3, + "core": 4, + "out": 3, + "T": 3, + "eff_depth": 18, + "seq": 256, + "batch": 16, + "seen_tok": 150003712, + "wall_s": 6654.5, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain_final": 0.5099, + "role_scale_final": 0.7688, + "vasana_final": [ + -0.1064, + -0.6145, + -0.2595, + 0.023, + -0.2102, + -0.0031, + -0.0943, + -0.3938, + -0.0241, + -0.0532, + -0.0539, + -0.016, + -0.0003, + -0.0185, + -0.0267, + -0.0645 + ], + "final": { + "step": 36620, + "tok": 149995520, + "ce": 2.24, + "ppl": 9.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5099, + "loop_stats": [ + { + "tau": 0.1306, + "v": 0.8945, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.048, + "v": 0.8408, + "H_a": 0.054, + "d_bind": 0.03744 + }, + { + "tau": 0.039, + "v": 0.8223, + "H_a": 0.046, + "d_bind": 0.01905 + } + ] + }, + "curve": [ + { + "step": 1, + "tok": 4096, + "ce": 9.7336, + "ppl": 16875.0, + "pref_transfer_pron": 0.34, + "pref_train_sv": 0.548, + "tok_s": 2783, + "peak_MiB": 5047, + "v_gain": -0.0006, + "loop_stats": [ + { + "tau": 0.4871, + "v": 0.4854, + "H_a": 2.772, + "d_bind": 0.0 + }, + { + "tau": 0.4839, + "v": 0.481, + "H_a": 2.771, + "d_bind": 0.00158 + }, + { + "tau": 0.4813, + "v": 0.4779, + "H_a": 2.77, + "d_bind": 0.00162 + } + ] + }, + { + "step": 20, + "tok": 81920, + "ce": 6.4122, + "ppl": 609.2, + "pref_transfer_pron": 0.667, + "pref_train_sv": 0.705, + "tok_s": 18873, + "peak_MiB": 5297, + "v_gain": 0.0005, + "loop_stats": [ + { + "tau": 0.3138, + "v": 0.5841, + "H_a": 2.339, + "d_bind": 0.0 + }, + { + "tau": 0.1228, + "v": 0.6549, + "H_a": 1.573, + "d_bind": 0.04079 + }, + { + "tau": 0.0656, + "v": 0.7156, + "H_a": 1.212, + "d_bind": 0.01392 + } + ] + }, + { + "step": 40, + "tok": 163840, + "ce": 5.9805, + "ppl": 395.6, + "pref_transfer_pron": 0.782, + "pref_train_sv": 0.72, + "tok_s": 20927, + "peak_MiB": 5297, + "v_gain": -0.0023, + "loop_stats": [ + { + "tau": 0.1525, + "v": 0.6445, + "H_a": 1.723, + "d_bind": 0.0 + }, + { + "tau": 0.0123, + "v": 0.6834, + "H_a": 0.922, + "d_bind": 0.05207 + }, + { + "tau": 0.0021, + "v": 0.7086, + "H_a": 0.572, + "d_bind": 0.02277 + } + ] + }, + { + "step": 60, + "tok": 245760, + "ce": 5.7769, + "ppl": 322.7, + "pref_transfer_pron": 0.775, + "pref_train_sv": 0.715, + "tok_s": 21732, + "peak_MiB": 5297, + "v_gain": -0.0045, + "loop_stats": [ + { + "tau": 0.1374, + "v": 0.8815, + "H_a": 0.61, + "d_bind": 0.0 + }, + { + "tau": 0.0182, + "v": 0.9185, + "H_a": 0.267, + "d_bind": 0.01413 + }, + { + "tau": 0.0059, + "v": 0.9342, + "H_a": 0.209, + "d_bind": 0.00516 + } + ] + }, + { + "step": 80, + "tok": 327680, + "ce": 5.7933, + "ppl": 328.1, + "pref_transfer_pron": 0.712, + "pref_train_sv": 0.717, + "tok_s": 22782, + "peak_MiB": 5298, + "v_gain": -0.0028, + "loop_stats": [ + { + "tau": 0.0817, + "v": 0.8589, + "H_a": 0.567, + "d_bind": 0.0 + }, + { + "tau": 0.0167, + "v": 0.886, + "H_a": 0.297, + "d_bind": 0.01669 + }, + { + "tau": 0.0066, + "v": 0.8974, + "H_a": 0.23, + "d_bind": 0.0069 + } + ] + }, + { + "step": 100, + "tok": 409600, + "ce": 5.0894, + "ppl": 162.3, + "pref_transfer_pron": 0.752, + "pref_train_sv": 0.727, + "tok_s": 22516, + "peak_MiB": 5298, + "v_gain": -0.0019, + "loop_stats": [ + { + "tau": 0.0547, + "v": 0.7706, + "H_a": 0.502, + "d_bind": 0.0 + }, + { + "tau": 0.0088, + "v": 0.8063, + "H_a": 0.345, + "d_bind": 0.01329 + }, + { + "tau": 0.0028, + "v": 0.8249, + "H_a": 0.306, + "d_bind": 0.00781 + } + ] + }, + { + "step": 120, + "tok": 491520, + "ce": 5.3205, + "ppl": 204.5, + "pref_transfer_pron": 0.765, + "pref_train_sv": 0.757, + "tok_s": 21859, + "peak_MiB": 5298, + "v_gain": -0.0016, + "loop_stats": [ + { + "tau": 0.0761, + "v": 0.7406, + "H_a": 0.359, + "d_bind": 0.0 + }, + { + "tau": 0.0251, + "v": 0.7755, + "H_a": 0.234, + "d_bind": 0.00814 + }, + { + "tau": 0.0141, + "v": 0.7936, + "H_a": 0.211, + "d_bind": 0.00434 + } + ] + }, + { + "step": 140, + "tok": 573440, + "ce": 4.7126, + "ppl": 111.3, + "pref_transfer_pron": 0.722, + "pref_train_sv": 0.775, + "tok_s": 22062, + "peak_MiB": 5298, + "v_gain": -0.001, + "loop_stats": [ + { + "tau": 0.0848, + "v": 0.7148, + "H_a": 0.343, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.745, + "H_a": 0.237, + "d_bind": 0.00692 + }, + { + "tau": 0.0174, + "v": 0.7634, + "H_a": 0.221, + "d_bind": 0.0047 + } + ] + }, + { + "step": 160, + "tok": 655360, + "ce": 4.4023, + "ppl": 81.6, + "pref_transfer_pron": 0.697, + "pref_train_sv": 0.792, + "tok_s": 22219, + "peak_MiB": 5298, + "v_gain": 0.001, + "loop_stats": [ + { + "tau": 0.061, + "v": 0.6366, + "H_a": 0.285, + "d_bind": 0.0 + }, + { + "tau": 0.01, + "v": 0.6532, + "H_a": 0.182, + "d_bind": 0.00617 + }, + { + "tau": 0.0036, + "v": 0.6647, + "H_a": 0.159, + "d_bind": 0.00342 + } + ] + }, + { + "step": 180, + "tok": 737280, + "ce": 4.6156, + "ppl": 101.1, + "pref_transfer_pron": 0.77, + "pref_train_sv": 0.782, + "tok_s": 22341, + "peak_MiB": 5298, + "v_gain": 0.0021, + "loop_stats": [ + { + "tau": 0.0801, + "v": 0.6404, + "H_a": 0.281, + "d_bind": 0.0 + }, + { + "tau": 0.0151, + "v": 0.6631, + "H_a": 0.175, + "d_bind": 0.00553 + }, + { + "tau": 0.0064, + "v": 0.6813, + "H_a": 0.156, + "d_bind": 0.00276 + } + ] + }, + { + "step": 200, + "tok": 819200, + "ce": 4.9193, + "ppl": 136.9, + "pref_transfer_pron": 0.722, + "pref_train_sv": 0.785, + "tok_s": 22694, + "peak_MiB": 5298, + "v_gain": 0.004, + "loop_stats": [ + { + "tau": 0.0591, + "v": 0.634, + "H_a": 0.213, + "d_bind": 0.0 + }, + { + "tau": 0.0166, + "v": 0.6497, + "H_a": 0.163, + "d_bind": 0.00462 + }, + { + "tau": 0.0084, + "v": 0.66, + "H_a": 0.154, + "d_bind": 0.00335 + } + ] + }, + { + "step": 220, + "tok": 901120, + "ce": 4.8195, + "ppl": 123.9, + "pref_transfer_pron": 0.75, + "pref_train_sv": 0.803, + "tok_s": 22577, + "peak_MiB": 5298, + "v_gain": 0.006, + "loop_stats": [ + { + "tau": 0.0613, + "v": 0.6088, + "H_a": 0.187, + "d_bind": 0.0 + }, + { + "tau": 0.0141, + "v": 0.6402, + "H_a": 0.118, + "d_bind": 0.00431 + }, + { + "tau": 0.0076, + "v": 0.6608, + "H_a": 0.099, + "d_bind": 0.00201 + } + ] + }, + { + "step": 240, + "tok": 983040, + "ce": 4.1379, + "ppl": 62.7, + "pref_transfer_pron": 0.785, + "pref_train_sv": 0.797, + "tok_s": 22224, + "peak_MiB": 5299, + "v_gain": 0.0056, + "loop_stats": [ + { + "tau": 0.1007, + "v": 0.5046, + "H_a": 0.354, + "d_bind": 0.0 + }, + { + "tau": 0.0245, + "v": 0.5219, + "H_a": 0.214, + "d_bind": 0.00879 + }, + { + "tau": 0.0124, + "v": 0.535, + "H_a": 0.18, + "d_bind": 0.00413 + } + ] + }, + { + "step": 260, + "tok": 1064960, + "ce": 4.3383, + "ppl": 76.6, + "pref_transfer_pron": 0.78, + "pref_train_sv": 0.8, + "tok_s": 22301, + "peak_MiB": 5299, + "v_gain": 0.0055, + "loop_stats": [ + { + "tau": 0.0613, + "v": 0.5885, + "H_a": 0.238, + "d_bind": 0.0 + }, + { + "tau": 0.0146, + "v": 0.6015, + "H_a": 0.168, + "d_bind": 0.00537 + }, + { + "tau": 0.0074, + "v": 0.611, + "H_a": 0.158, + "d_bind": 0.00331 + } + ] + }, + { + "step": 280, + "tok": 1146880, + "ce": 4.9439, + "ppl": 140.3, + "pref_transfer_pron": 0.772, + "pref_train_sv": 0.785, + "tok_s": 22364, + "peak_MiB": 5299, + "v_gain": 0.0049, + "loop_stats": [ + { + "tau": 0.0941, + "v": 0.5857, + "H_a": 0.224, + "d_bind": 0.0 + }, + { + "tau": 0.0318, + "v": 0.6104, + "H_a": 0.192, + "d_bind": 0.00517 + }, + { + "tau": 0.0206, + "v": 0.6285, + "H_a": 0.191, + "d_bind": 0.00444 + } + ] + }, + { + "step": 300, + "tok": 1228800, + "ce": 4.6457, + "ppl": 104.1, + "pref_transfer_pron": 0.83, + "pref_train_sv": 0.785, + "tok_s": 22419, + "peak_MiB": 5299, + "v_gain": 0.0053, + "loop_stats": [ + { + "tau": 0.0771, + "v": 0.5498, + "H_a": 0.155, + "d_bind": 0.0 + }, + { + "tau": 0.0135, + "v": 0.5654, + "H_a": 0.14, + "d_bind": 0.00509 + }, + { + "tau": 0.0054, + "v": 0.5764, + "H_a": 0.135, + "d_bind": 0.00369 + } + ] + }, + { + "step": 320, + "tok": 1310720, + "ce": 4.8357, + "ppl": 125.9, + "pref_transfer_pron": 0.79, + "pref_train_sv": 0.785, + "tok_s": 22627, + "peak_MiB": 5299, + "v_gain": 0.0064, + "loop_stats": [ + { + "tau": 0.0612, + "v": 0.5976, + "H_a": 0.207, + "d_bind": 0.0 + }, + { + "tau": 0.0116, + "v": 0.6171, + "H_a": 0.182, + "d_bind": 0.00579 + }, + { + "tau": 0.0054, + "v": 0.6308, + "H_a": 0.174, + "d_bind": 0.00436 + } + ] + }, + { + "step": 340, + "tok": 1392640, + "ce": 4.4359, + "ppl": 84.4, + "pref_transfer_pron": 0.812, + "pref_train_sv": 0.812, + "tok_s": 22541, + "peak_MiB": 5299, + "v_gain": 0.0068, + "loop_stats": [ + { + "tau": 0.0472, + "v": 0.4789, + "H_a": 0.208, + "d_bind": 0.0 + }, + { + "tau": 0.013, + "v": 0.4881, + "H_a": 0.169, + "d_bind": 0.00617 + }, + { + "tau": 0.0108, + "v": 0.4945, + "H_a": 0.163, + "d_bind": 0.00396 + } + ] + }, + { + "step": 360, + "tok": 1474560, + "ce": 4.7602, + "ppl": 116.8, + "pref_transfer_pron": 0.785, + "pref_train_sv": 0.8, + "tok_s": 22303, + "peak_MiB": 5299, + "v_gain": 0.0068, + "loop_stats": [ + { + "tau": 0.0645, + "v": 0.4795, + "H_a": 0.249, + "d_bind": 0.0 + }, + { + "tau": 0.017, + "v": 0.4857, + "H_a": 0.21, + "d_bind": 0.00624 + }, + { + "tau": 0.0107, + "v": 0.4892, + "H_a": 0.209, + "d_bind": 0.00493 + } + ] + }, + { + "step": 380, + "tok": 1556480, + "ce": 4.4399, + "ppl": 84.8, + "pref_transfer_pron": 0.815, + "pref_train_sv": 0.82, + "tok_s": 22354, + "peak_MiB": 5299, + "v_gain": 0.0041, + "loop_stats": [ + { + "tau": 0.0543, + "v": 0.4765, + "H_a": 0.24, + "d_bind": 0.0 + }, + { + "tau": 0.0054, + "v": 0.4869, + "H_a": 0.208, + "d_bind": 0.00794 + }, + { + "tau": 0.0024, + "v": 0.4958, + "H_a": 0.197, + "d_bind": 0.00502 + } + ] + }, + { + "step": 400, + "tok": 1638400, + "ce": 4.1162, + "ppl": 61.3, + "pref_transfer_pron": 0.787, + "pref_train_sv": 0.83, + "tok_s": 22380, + "peak_MiB": 5299, + "v_gain": 0.0053, + "loop_stats": [ + { + "tau": 0.0674, + "v": 0.5156, + "H_a": 0.234, + "d_bind": 0.0 + }, + { + "tau": 0.0089, + "v": 0.5309, + "H_a": 0.214, + "d_bind": 0.00852 + }, + { + "tau": 0.0053, + "v": 0.5444, + "H_a": 0.205, + "d_bind": 0.00638 + } + ] + }, + { + "step": 420, + "tok": 1720320, + "ce": 4.4925, + "ppl": 89.3, + "pref_transfer_pron": 0.827, + "pref_train_sv": 0.837, + "tok_s": 22425, + "peak_MiB": 5299, + "v_gain": 0.0044, + "loop_stats": [ + { + "tau": 0.0954, + "v": 0.4623, + "H_a": 0.217, + "d_bind": 0.0 + }, + { + "tau": 0.0188, + "v": 0.4741, + "H_a": 0.158, + "d_bind": 0.00781 + }, + { + "tau": 0.0065, + "v": 0.4842, + "H_a": 0.147, + "d_bind": 0.00459 + } + ] + }, + { + "step": 440, + "tok": 1802240, + "ce": 4.6471, + "ppl": 104.3, + "pref_transfer_pron": 0.815, + "pref_train_sv": 0.8, + "tok_s": 22581, + "peak_MiB": 5299, + "v_gain": 0.0044, + "loop_stats": [ + { + "tau": 0.0733, + "v": 0.5058, + "H_a": 0.254, + "d_bind": 0.0 + }, + { + "tau": 0.0126, + "v": 0.5254, + "H_a": 0.216, + "d_bind": 0.00846 + }, + { + "tau": 0.0056, + "v": 0.5441, + "H_a": 0.199, + "d_bind": 0.00585 + } + ] + }, + { + "step": 460, + "tok": 1884160, + "ce": 4.4215, + "ppl": 83.2, + "pref_transfer_pron": 0.832, + "pref_train_sv": 0.842, + "tok_s": 22531, + "peak_MiB": 5299, + "v_gain": 0.0041, + "loop_stats": [ + { + "tau": 0.0872, + "v": 0.4973, + "H_a": 0.269, + "d_bind": 0.0 + }, + { + "tau": 0.02, + "v": 0.5002, + "H_a": 0.224, + "d_bind": 0.01085 + }, + { + "tau": 0.0144, + "v": 0.5052, + "H_a": 0.206, + "d_bind": 0.00685 + } + ] + }, + { + "step": 480, + "tok": 1966080, + "ce": 4.8649, + "ppl": 129.7, + "pref_transfer_pron": 0.812, + "pref_train_sv": 0.817, + "tok_s": 22356, + "peak_MiB": 5299, + "v_gain": 0.0011, + "loop_stats": [ + { + "tau": 0.0708, + "v": 0.4885, + "H_a": 0.333, + "d_bind": 0.0 + }, + { + "tau": 0.0224, + "v": 0.5003, + "H_a": 0.251, + "d_bind": 0.01115 + }, + { + "tau": 0.017, + "v": 0.5097, + "H_a": 0.227, + "d_bind": 0.0065 + } + ] + }, + { + "step": 500, + "tok": 2048000, + "ce": 4.3891, + "ppl": 80.6, + "pref_transfer_pron": 0.847, + "pref_train_sv": 0.795, + "tok_s": 22393, + "peak_MiB": 5299, + "v_gain": 0.0024, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.5101, + "H_a": 0.33, + "d_bind": 0.0 + }, + { + "tau": 0.026, + "v": 0.5318, + "H_a": 0.248, + "d_bind": 0.01366 + }, + { + "tau": 0.0137, + "v": 0.5501, + "H_a": 0.222, + "d_bind": 0.00793 + } + ] + }, + { + "step": 520, + "tok": 2129920, + "ce": 3.9695, + "ppl": 53.0, + "pref_transfer_pron": 0.827, + "pref_train_sv": 0.837, + "tok_s": 22424, + "peak_MiB": 5299, + "v_gain": 0.0038, + "loop_stats": [ + { + "tau": 0.0646, + "v": 0.5813, + "H_a": 0.291, + "d_bind": 0.0 + }, + { + "tau": 0.0122, + "v": 0.6082, + "H_a": 0.239, + "d_bind": 0.01234 + }, + { + "tau": 0.0082, + "v": 0.6278, + "H_a": 0.222, + "d_bind": 0.00918 + } + ] + }, + { + "step": 540, + "tok": 2211840, + "ce": 4.5336, + "ppl": 93.1, + "pref_transfer_pron": 0.862, + "pref_train_sv": 0.835, + "tok_s": 22459, + "peak_MiB": 5299, + "v_gain": 0.0024, + "loop_stats": [ + { + "tau": 0.0846, + "v": 0.5995, + "H_a": 0.292, + "d_bind": 0.0 + }, + { + "tau": 0.0105, + "v": 0.6376, + "H_a": 0.237, + "d_bind": 0.01147 + }, + { + "tau": 0.004, + "v": 0.667, + "H_a": 0.221, + "d_bind": 0.00654 + } + ] + }, + { + "step": 560, + "tok": 2293760, + "ce": 4.1812, + "ppl": 65.4, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.822, + "tok_s": 22582, + "peak_MiB": 5299, + "v_gain": 0.0022, + "loop_stats": [ + { + "tau": 0.0695, + "v": 0.6311, + "H_a": 0.297, + "d_bind": 0.0 + }, + { + "tau": 0.0126, + "v": 0.6664, + "H_a": 0.239, + "d_bind": 0.01313 + }, + { + "tau": 0.0069, + "v": 0.6904, + "H_a": 0.22, + "d_bind": 0.00828 + } + ] + }, + { + "step": 580, + "tok": 2375680, + "ce": 4.3648, + "ppl": 78.6, + "pref_transfer_pron": 0.837, + "pref_train_sv": 0.827, + "tok_s": 22542, + "peak_MiB": 5299, + "v_gain": 0.0002, + "loop_stats": [ + { + "tau": 0.0367, + "v": 0.6152, + "H_a": 0.264, + "d_bind": 0.0 + }, + { + "tau": 0.0053, + "v": 0.6378, + "H_a": 0.221, + "d_bind": 0.0115 + }, + { + "tau": 0.0022, + "v": 0.6516, + "H_a": 0.215, + "d_bind": 0.00788 + } + ] + }, + { + "step": 600, + "tok": 2457600, + "ce": 4.3114, + "ppl": 74.5, + "pref_transfer_pron": 0.862, + "pref_train_sv": 0.85, + "tok_s": 22401, + "peak_MiB": 5299, + "v_gain": -0.0027, + "loop_stats": [ + { + "tau": 0.1066, + "v": 0.6715, + "H_a": 0.292, + "d_bind": 0.0 + }, + { + "tau": 0.0233, + "v": 0.7066, + "H_a": 0.242, + "d_bind": 0.01154 + }, + { + "tau": 0.0115, + "v": 0.7259, + "H_a": 0.23, + "d_bind": 0.00779 + } + ] + }, + { + "step": 620, + "tok": 2539520, + "ce": 4.352, + "ppl": 77.6, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.837, + "tok_s": 22428, + "peak_MiB": 5299, + "v_gain": -0.0002, + "loop_stats": [ + { + "tau": 0.0909, + "v": 0.6028, + "H_a": 0.264, + "d_bind": 0.0 + }, + { + "tau": 0.013, + "v": 0.6276, + "H_a": 0.225, + "d_bind": 0.01309 + }, + { + "tau": 0.0068, + "v": 0.6444, + "H_a": 0.212, + "d_bind": 0.00872 + } + ] + }, + { + "step": 640, + "tok": 2621440, + "ce": 4.2219, + "ppl": 68.2, + "pref_transfer_pron": 0.87, + "pref_train_sv": 0.827, + "tok_s": 22457, + "peak_MiB": 5299, + "v_gain": 0.0025, + "loop_stats": [ + { + "tau": 0.0743, + "v": 0.686, + "H_a": 0.261, + "d_bind": 0.0 + }, + { + "tau": 0.0074, + "v": 0.7107, + "H_a": 0.206, + "d_bind": 0.01251 + }, + { + "tau": 0.0032, + "v": 0.7245, + "H_a": 0.18, + "d_bind": 0.00788 + } + ] + }, + { + "step": 660, + "tok": 2703360, + "ce": 4.0812, + "ppl": 59.2, + "pref_transfer_pron": 0.835, + "pref_train_sv": 0.847, + "tok_s": 22484, + "peak_MiB": 5299, + "v_gain": 0.0001, + "loop_stats": [ + { + "tau": 0.0795, + "v": 0.6619, + "H_a": 0.25, + "d_bind": 0.0 + }, + { + "tau": 0.011, + "v": 0.6909, + "H_a": 0.209, + "d_bind": 0.01334 + }, + { + "tau": 0.0048, + "v": 0.7115, + "H_a": 0.201, + "d_bind": 0.0081 + } + ] + }, + { + "step": 680, + "tok": 2785280, + "ce": 4.2451, + "ppl": 69.8, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.837, + "tok_s": 22585, + "peak_MiB": 5299, + "v_gain": 0.0023, + "loop_stats": [ + { + "tau": 0.0959, + "v": 0.6583, + "H_a": 0.265, + "d_bind": 0.0 + }, + { + "tau": 0.0156, + "v": 0.6802, + "H_a": 0.236, + "d_bind": 0.01203 + }, + { + "tau": 0.0064, + "v": 0.6952, + "H_a": 0.231, + "d_bind": 0.00893 + } + ] + }, + { + "step": 700, + "tok": 2867200, + "ce": 3.8362, + "ppl": 46.3, + "pref_transfer_pron": 0.855, + "pref_train_sv": 0.835, + "tok_s": 22551, + "peak_MiB": 5300, + "v_gain": 0.0034, + "loop_stats": [ + { + "tau": 0.0696, + "v": 0.6296, + "H_a": 0.247, + "d_bind": 0.0 + }, + { + "tau": 0.0102, + "v": 0.6548, + "H_a": 0.212, + "d_bind": 0.01269 + }, + { + "tau": 0.0054, + "v": 0.6711, + "H_a": 0.201, + "d_bind": 0.00821 + } + ] + }, + { + "step": 720, + "tok": 2949120, + "ce": 3.8954, + "ppl": 49.2, + "pref_transfer_pron": 0.847, + "pref_train_sv": 0.865, + "tok_s": 22434, + "peak_MiB": 5300, + "v_gain": 0.0025, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.629, + "H_a": 0.254, + "d_bind": 0.0 + }, + { + "tau": 0.0152, + "v": 0.6621, + "H_a": 0.214, + "d_bind": 0.01377 + }, + { + "tau": 0.0073, + "v": 0.6828, + "H_a": 0.2, + "d_bind": 0.00865 + } + ] + }, + { + "step": 740, + "tok": 3031040, + "ce": 4.6819, + "ppl": 108.0, + "pref_transfer_pron": 0.862, + "pref_train_sv": 0.847, + "tok_s": 22457, + "peak_MiB": 5300, + "v_gain": 0.0061, + "loop_stats": [ + { + "tau": 0.0948, + "v": 0.7328, + "H_a": 0.276, + "d_bind": 0.0 + }, + { + "tau": 0.0131, + "v": 0.7726, + "H_a": 0.24, + "d_bind": 0.01282 + }, + { + "tau": 0.0065, + "v": 0.797, + "H_a": 0.237, + "d_bind": 0.00868 + } + ] + }, + { + "step": 760, + "tok": 3112960, + "ce": 4.0884, + "ppl": 59.6, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.847, + "tok_s": 22479, + "peak_MiB": 5300, + "v_gain": 0.0065, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.6968, + "H_a": 0.283, + "d_bind": 0.0 + }, + { + "tau": 0.0178, + "v": 0.7236, + "H_a": 0.226, + "d_bind": 0.01459 + }, + { + "tau": 0.0092, + "v": 0.7385, + "H_a": 0.211, + "d_bind": 0.0087 + } + ] + }, + { + "step": 780, + "tok": 3194880, + "ce": 4.3305, + "ppl": 76.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.845, + "tok_s": 22502, + "peak_MiB": 5300, + "v_gain": 0.006, + "loop_stats": [ + { + "tau": 0.1075, + "v": 0.7607, + "H_a": 0.306, + "d_bind": 0.0 + }, + { + "tau": 0.019, + "v": 0.7903, + "H_a": 0.269, + "d_bind": 0.01404 + }, + { + "tau": 0.0074, + "v": 0.8022, + "H_a": 0.246, + "d_bind": 0.00911 + } + ] + }, + { + "step": 800, + "tok": 3276800, + "ce": 3.8869, + "ppl": 48.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.842, + "tok_s": 22576, + "peak_MiB": 5300, + "v_gain": 0.0049, + "loop_stats": [ + { + "tau": 0.0767, + "v": 0.6946, + "H_a": 0.26, + "d_bind": 0.0 + }, + { + "tau": 0.0129, + "v": 0.7308, + "H_a": 0.238, + "d_bind": 0.01452 + }, + { + "tau": 0.0055, + "v": 0.7502, + "H_a": 0.227, + "d_bind": 0.01013 + } + ] + }, + { + "step": 820, + "tok": 3358720, + "ce": 3.5197, + "ppl": 33.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.857, + "tok_s": 22549, + "peak_MiB": 5300, + "v_gain": 0.0049, + "loop_stats": [ + { + "tau": 0.0849, + "v": 0.6808, + "H_a": 0.268, + "d_bind": 0.0 + }, + { + "tau": 0.0158, + "v": 0.7107, + "H_a": 0.239, + "d_bind": 0.01652 + }, + { + "tau": 0.007, + "v": 0.7289, + "H_a": 0.212, + "d_bind": 0.01027 + } + ] + }, + { + "step": 840, + "tok": 3440640, + "ce": 4.3906, + "ppl": 80.7, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.877, + "tok_s": 22448, + "peak_MiB": 5300, + "v_gain": 0.008, + "loop_stats": [ + { + "tau": 0.0856, + "v": 0.6555, + "H_a": 0.261, + "d_bind": 0.0 + }, + { + "tau": 0.0172, + "v": 0.683, + "H_a": 0.228, + "d_bind": 0.01463 + }, + { + "tau": 0.0089, + "v": 0.6997, + "H_a": 0.212, + "d_bind": 0.00966 + } + ] + }, + { + "step": 860, + "tok": 3522560, + "ce": 3.5552, + "ppl": 35.0, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.85, + "tok_s": 22461, + "peak_MiB": 5300, + "v_gain": 0.0079, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.6683, + "H_a": 0.232, + "d_bind": 0.0 + }, + { + "tau": 0.0061, + "v": 0.6943, + "H_a": 0.211, + "d_bind": 0.01451 + }, + { + "tau": 0.0025, + "v": 0.7108, + "H_a": 0.201, + "d_bind": 0.00917 + } + ] + }, + { + "step": 880, + "tok": 3604480, + "ce": 3.7696, + "ppl": 43.4, + "pref_transfer_pron": 0.873, + "pref_train_sv": 0.837, + "tok_s": 22477, + "peak_MiB": 5300, + "v_gain": 0.0054, + "loop_stats": [ + { + "tau": 0.1386, + "v": 0.7147, + "H_a": 0.293, + "d_bind": 0.0 + }, + { + "tau": 0.0247, + "v": 0.7381, + "H_a": 0.262, + "d_bind": 0.01533 + }, + { + "tau": 0.0138, + "v": 0.7524, + "H_a": 0.246, + "d_bind": 0.01011 + } + ] + }, + { + "step": 900, + "tok": 3686400, + "ce": 4.0677, + "ppl": 58.4, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.868, + "tok_s": 22496, + "peak_MiB": 5300, + "v_gain": 0.0036, + "loop_stats": [ + { + "tau": 0.1116, + "v": 0.7161, + "H_a": 0.303, + "d_bind": 0.0 + }, + { + "tau": 0.0417, + "v": 0.7455, + "H_a": 0.291, + "d_bind": 0.01495 + }, + { + "tau": 0.0322, + "v": 0.7639, + "H_a": 0.28, + "d_bind": 0.0114 + } + ] + }, + { + "step": 920, + "tok": 3768320, + "ce": 4.1651, + "ppl": 64.4, + "pref_transfer_pron": 0.87, + "pref_train_sv": 0.87, + "tok_s": 22568, + "peak_MiB": 5300, + "v_gain": 0.0071, + "loop_stats": [ + { + "tau": 0.0911, + "v": 0.7293, + "H_a": 0.284, + "d_bind": 0.0 + }, + { + "tau": 0.0135, + "v": 0.7661, + "H_a": 0.238, + "d_bind": 0.01545 + }, + { + "tau": 0.0067, + "v": 0.7907, + "H_a": 0.222, + "d_bind": 0.01073 + } + ] + }, + { + "step": 940, + "tok": 3850240, + "ce": 4.1558, + "ppl": 63.8, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.868, + "tok_s": 22542, + "peak_MiB": 5300, + "v_gain": 0.0103, + "loop_stats": [ + { + "tau": 0.0834, + "v": 0.7081, + "H_a": 0.294, + "d_bind": 0.0 + }, + { + "tau": 0.0143, + "v": 0.7355, + "H_a": 0.266, + "d_bind": 0.01573 + }, + { + "tau": 0.0081, + "v": 0.7541, + "H_a": 0.253, + "d_bind": 0.01116 + } + ] + }, + { + "step": 960, + "tok": 3932160, + "ce": 4.2811, + "ppl": 72.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.852, + "tok_s": 22454, + "peak_MiB": 5300, + "v_gain": 0.011, + "loop_stats": [ + { + "tau": 0.0894, + "v": 0.7229, + "H_a": 0.294, + "d_bind": 0.0 + }, + { + "tau": 0.0106, + "v": 0.7433, + "H_a": 0.272, + "d_bind": 0.01597 + }, + { + "tau": 0.0049, + "v": 0.7561, + "H_a": 0.243, + "d_bind": 0.01054 + } + ] + }, + { + "step": 980, + "tok": 4014080, + "ce": 3.986, + "ppl": 53.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.882, + "tok_s": 22470, + "peak_MiB": 5300, + "v_gain": 0.0099, + "loop_stats": [ + { + "tau": 0.0991, + "v": 0.7493, + "H_a": 0.254, + "d_bind": 0.0 + }, + { + "tau": 0.0233, + "v": 0.7619, + "H_a": 0.231, + "d_bind": 0.01452 + }, + { + "tau": 0.0115, + "v": 0.7642, + "H_a": 0.214, + "d_bind": 0.01012 + } + ] + }, + { + "step": 1000, + "tok": 4096000, + "ce": 4.1419, + "ppl": 62.9, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.865, + "tok_s": 22486, + "peak_MiB": 5300, + "v_gain": 0.0107, + "loop_stats": [ + { + "tau": 0.0697, + "v": 0.7462, + "H_a": 0.317, + "d_bind": 0.0 + }, + { + "tau": 0.0143, + "v": 0.7651, + "H_a": 0.302, + "d_bind": 0.01692 + }, + { + "tau": 0.0111, + "v": 0.7741, + "H_a": 0.279, + "d_bind": 0.01204 + } + ] + }, + { + "step": 1020, + "tok": 4177920, + "ce": 4.7328, + "ppl": 113.6, + "pref_transfer_pron": 0.862, + "pref_train_sv": 0.855, + "tok_s": 22503, + "peak_MiB": 5300, + "v_gain": 0.0127, + "loop_stats": [ + { + "tau": 0.0737, + "v": 0.8362, + "H_a": 0.335, + "d_bind": 0.0 + }, + { + "tau": 0.0091, + "v": 0.8546, + "H_a": 0.305, + "d_bind": 0.01674 + }, + { + "tau": 0.0059, + "v": 0.8599, + "H_a": 0.286, + "d_bind": 0.01244 + } + ] + }, + { + "step": 1040, + "tok": 4259840, + "ce": 4.3015, + "ppl": 73.8, + "pref_transfer_pron": 0.857, + "pref_train_sv": 0.86, + "tok_s": 22567, + "peak_MiB": 5300, + "v_gain": 0.0139, + "loop_stats": [ + { + "tau": 0.0707, + "v": 0.7847, + "H_a": 0.294, + "d_bind": 0.0 + }, + { + "tau": 0.0106, + "v": 0.8031, + "H_a": 0.286, + "d_bind": 0.01766 + }, + { + "tau": 0.0062, + "v": 0.8127, + "H_a": 0.264, + "d_bind": 0.0129 + } + ] + }, + { + "step": 1060, + "tok": 4341760, + "ce": 4.1817, + "ppl": 65.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.865, + "tok_s": 22544, + "peak_MiB": 5300, + "v_gain": 0.0147, + "loop_stats": [ + { + "tau": 0.1162, + "v": 0.7909, + "H_a": 0.266, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.7995, + "H_a": 0.225, + "d_bind": 0.01642 + }, + { + "tau": 0.0198, + "v": 0.8041, + "H_a": 0.194, + "d_bind": 0.00982 + } + ] + }, + { + "step": 1080, + "tok": 4423680, + "ce": 4.3918, + "ppl": 80.8, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.88, + "tok_s": 22462, + "peak_MiB": 5300, + "v_gain": 0.0192, + "loop_stats": [ + { + "tau": 0.0952, + "v": 0.7959, + "H_a": 0.262, + "d_bind": 0.0 + }, + { + "tau": 0.0223, + "v": 0.8142, + "H_a": 0.249, + "d_bind": 0.01678 + }, + { + "tau": 0.0161, + "v": 0.8237, + "H_a": 0.237, + "d_bind": 0.01169 + } + ] + }, + { + "step": 1100, + "tok": 4505600, + "ce": 4.2221, + "ppl": 68.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.877, + "tok_s": 22473, + "peak_MiB": 5300, + "v_gain": 0.0218, + "loop_stats": [ + { + "tau": 0.066, + "v": 0.8044, + "H_a": 0.28, + "d_bind": 0.0 + }, + { + "tau": 0.013, + "v": 0.8148, + "H_a": 0.268, + "d_bind": 0.01812 + }, + { + "tau": 0.0101, + "v": 0.8212, + "H_a": 0.255, + "d_bind": 0.01269 + } + ] + }, + { + "step": 1120, + "tok": 4587520, + "ce": 3.9146, + "ppl": 50.1, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.868, + "tok_s": 22487, + "peak_MiB": 5300, + "v_gain": 0.0247, + "loop_stats": [ + { + "tau": 0.1561, + "v": 0.7649, + "H_a": 0.256, + "d_bind": 0.0 + }, + { + "tau": 0.0202, + "v": 0.7641, + "H_a": 0.264, + "d_bind": 0.01635 + }, + { + "tau": 0.0142, + "v": 0.7633, + "H_a": 0.247, + "d_bind": 0.01284 + } + ] + }, + { + "step": 1140, + "tok": 4669440, + "ce": 3.8407, + "ppl": 46.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.907, + "tok_s": 22500, + "peak_MiB": 5300, + "v_gain": 0.026, + "loop_stats": [ + { + "tau": 0.0931, + "v": 0.7208, + "H_a": 0.263, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.7186, + "H_a": 0.248, + "d_bind": 0.01702 + }, + { + "tau": 0.0313, + "v": 0.7194, + "H_a": 0.241, + "d_bind": 0.01231 + } + ] + }, + { + "step": 1160, + "tok": 4751360, + "ce": 4.3134, + "ppl": 74.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.905, + "tok_s": 22550, + "peak_MiB": 5300, + "v_gain": 0.0267, + "loop_stats": [ + { + "tau": 0.0693, + "v": 0.7929, + "H_a": 0.276, + "d_bind": 0.0 + }, + { + "tau": 0.0199, + "v": 0.7949, + "H_a": 0.251, + "d_bind": 0.01668 + }, + { + "tau": 0.0174, + "v": 0.7932, + "H_a": 0.233, + "d_bind": 0.01211 + } + ] + }, + { + "step": 1180, + "tok": 4833280, + "ce": 4.096, + "ppl": 60.1, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.905, + "tok_s": 22527, + "peak_MiB": 5300, + "v_gain": 0.0275, + "loop_stats": [ + { + "tau": 0.0818, + "v": 0.7873, + "H_a": 0.266, + "d_bind": 0.0 + }, + { + "tau": 0.0256, + "v": 0.7793, + "H_a": 0.248, + "d_bind": 0.01856 + }, + { + "tau": 0.0269, + "v": 0.7779, + "H_a": 0.224, + "d_bind": 0.01176 + } + ] + }, + { + "step": 1200, + "tok": 4915200, + "ce": 4.0396, + "ppl": 56.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.89, + "tok_s": 22446, + "peak_MiB": 5300, + "v_gain": 0.0288, + "loop_stats": [ + { + "tau": 0.0839, + "v": 0.7826, + "H_a": 0.239, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.7843, + "H_a": 0.237, + "d_bind": 0.01767 + }, + { + "tau": 0.0281, + "v": 0.7847, + "H_a": 0.216, + "d_bind": 0.01213 + } + ] + }, + { + "step": 1220, + "tok": 4997120, + "ce": 3.7197, + "ppl": 41.3, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.9, + "tok_s": 22457, + "peak_MiB": 5300, + "v_gain": 0.032, + "loop_stats": [ + { + "tau": 0.1278, + "v": 0.8063, + "H_a": 0.232, + "d_bind": 0.0 + }, + { + "tau": 0.0516, + "v": 0.8089, + "H_a": 0.221, + "d_bind": 0.01741 + }, + { + "tau": 0.0454, + "v": 0.8059, + "H_a": 0.204, + "d_bind": 0.01139 + } + ] + }, + { + "step": 1240, + "tok": 5079040, + "ce": 3.4284, + "ppl": 30.8, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.9, + "tok_s": 22468, + "peak_MiB": 5300, + "v_gain": 0.0335, + "loop_stats": [ + { + "tau": 0.1434, + "v": 0.7846, + "H_a": 0.221, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.7767, + "H_a": 0.214, + "d_bind": 0.0166 + }, + { + "tau": 0.0317, + "v": 0.7647, + "H_a": 0.199, + "d_bind": 0.01163 + } + ] + }, + { + "step": 1260, + "tok": 5160960, + "ce": 4.6186, + "ppl": 101.4, + "pref_transfer_pron": 0.865, + "pref_train_sv": 0.9, + "tok_s": 22479, + "peak_MiB": 5300, + "v_gain": 0.0361, + "loop_stats": [ + { + "tau": 0.0667, + "v": 0.8659, + "H_a": 0.249, + "d_bind": 0.0 + }, + { + "tau": 0.0105, + "v": 0.8669, + "H_a": 0.22, + "d_bind": 0.01625 + }, + { + "tau": 0.0074, + "v": 0.866, + "H_a": 0.206, + "d_bind": 0.01116 + } + ] + }, + { + "step": 1280, + "tok": 5242880, + "ce": 4.3163, + "ppl": 74.9, + "pref_transfer_pron": 0.875, + "pref_train_sv": 0.887, + "tok_s": 22530, + "peak_MiB": 5300, + "v_gain": 0.0385, + "loop_stats": [ + { + "tau": 0.0875, + "v": 0.789, + "H_a": 0.246, + "d_bind": 0.0 + }, + { + "tau": 0.0278, + "v": 0.7836, + "H_a": 0.237, + "d_bind": 0.0175 + }, + { + "tau": 0.0238, + "v": 0.7769, + "H_a": 0.223, + "d_bind": 0.01226 + } + ] + }, + { + "step": 1300, + "tok": 5324800, + "ce": 3.7504, + "ppl": 42.5, + "pref_transfer_pron": 0.87, + "pref_train_sv": 0.902, + "tok_s": 22510, + "peak_MiB": 5300, + "v_gain": 0.041, + "loop_stats": [ + { + "tau": 0.0797, + "v": 0.7434, + "H_a": 0.222, + "d_bind": 0.0 + }, + { + "tau": 0.0219, + "v": 0.7205, + "H_a": 0.223, + "d_bind": 0.01615 + }, + { + "tau": 0.0186, + "v": 0.6986, + "H_a": 0.214, + "d_bind": 0.01208 + } + ] + }, + { + "step": 1320, + "tok": 5406720, + "ce": 3.867, + "ppl": 47.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.907, + "tok_s": 22444, + "peak_MiB": 5300, + "v_gain": 0.0409, + "loop_stats": [ + { + "tau": 0.1402, + "v": 0.8586, + "H_a": 0.219, + "d_bind": 0.0 + }, + { + "tau": 0.0229, + "v": 0.8604, + "H_a": 0.243, + "d_bind": 0.01812 + }, + { + "tau": 0.0179, + "v": 0.8587, + "H_a": 0.24, + "d_bind": 0.01461 + } + ] + }, + { + "step": 1340, + "tok": 5488640, + "ce": 3.6346, + "ppl": 37.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.905, + "tok_s": 22454, + "peak_MiB": 5300, + "v_gain": 0.0452, + "loop_stats": [ + { + "tau": 0.122, + "v": 0.8304, + "H_a": 0.209, + "d_bind": 0.0 + }, + { + "tau": 0.0234, + "v": 0.8194, + "H_a": 0.225, + "d_bind": 0.01827 + }, + { + "tau": 0.0188, + "v": 0.8099, + "H_a": 0.209, + "d_bind": 0.01181 + } + ] + }, + { + "step": 1360, + "tok": 5570560, + "ce": 4.1513, + "ppl": 63.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.907, + "tok_s": 22465, + "peak_MiB": 5300, + "v_gain": 0.043, + "loop_stats": [ + { + "tau": 0.0697, + "v": 0.7685, + "H_a": 0.241, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.7572, + "H_a": 0.228, + "d_bind": 0.01718 + }, + { + "tau": 0.0247, + "v": 0.7504, + "H_a": 0.21, + "d_bind": 0.01219 + } + ] + }, + { + "step": 1380, + "tok": 5652480, + "ce": 4.0631, + "ppl": 58.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22477, + "peak_MiB": 5300, + "v_gain": 0.0437, + "loop_stats": [ + { + "tau": 0.0719, + "v": 0.8393, + "H_a": 0.282, + "d_bind": 0.0 + }, + { + "tau": 0.0151, + "v": 0.8355, + "H_a": 0.27, + "d_bind": 0.02024 + }, + { + "tau": 0.0143, + "v": 0.831, + "H_a": 0.242, + "d_bind": 0.01319 + } + ] + }, + { + "step": 1400, + "tok": 5734400, + "ce": 4.0808, + "ppl": 59.2, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.91, + "tok_s": 22525, + "peak_MiB": 5300, + "v_gain": 0.0433, + "loop_stats": [ + { + "tau": 0.1163, + "v": 0.8779, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.8776, + "H_a": 0.214, + "d_bind": 0.01895 + }, + { + "tau": 0.0374, + "v": 0.8765, + "H_a": 0.205, + "d_bind": 0.012 + } + ] + }, + { + "step": 1420, + "tok": 5816320, + "ce": 4.4936, + "ppl": 89.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.9, + "tok_s": 22508, + "peak_MiB": 5300, + "v_gain": 0.0431, + "loop_stats": [ + { + "tau": 0.0866, + "v": 0.8999, + "H_a": 0.239, + "d_bind": 0.0 + }, + { + "tau": 0.0224, + "v": 0.9115, + "H_a": 0.215, + "d_bind": 0.0167 + }, + { + "tau": 0.0142, + "v": 0.9157, + "H_a": 0.198, + "d_bind": 0.01096 + } + ] + }, + { + "step": 1440, + "tok": 5898240, + "ce": 3.8868, + "ppl": 48.8, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.902, + "tok_s": 22450, + "peak_MiB": 5300, + "v_gain": 0.0445, + "loop_stats": [ + { + "tau": 0.1003, + "v": 0.8553, + "H_a": 0.215, + "d_bind": 0.0 + }, + { + "tau": 0.0263, + "v": 0.8555, + "H_a": 0.198, + "d_bind": 0.01774 + }, + { + "tau": 0.0226, + "v": 0.8529, + "H_a": 0.18, + "d_bind": 0.01118 + } + ] + }, + { + "step": 1460, + "tok": 5980160, + "ce": 3.6828, + "ppl": 39.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.91, + "tok_s": 22461, + "peak_MiB": 5300, + "v_gain": 0.0465, + "loop_stats": [ + { + "tau": 0.1246, + "v": 0.8791, + "H_a": 0.213, + "d_bind": 0.0 + }, + { + "tau": 0.0327, + "v": 0.877, + "H_a": 0.201, + "d_bind": 0.0174 + }, + { + "tau": 0.0296, + "v": 0.8767, + "H_a": 0.188, + "d_bind": 0.01096 + } + ] + }, + { + "step": 1480, + "tok": 6062080, + "ce": 3.6088, + "ppl": 36.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22472, + "peak_MiB": 5300, + "v_gain": 0.05, + "loop_stats": [ + { + "tau": 0.127, + "v": 0.8422, + "H_a": 0.223, + "d_bind": 0.0 + }, + { + "tau": 0.0297, + "v": 0.8343, + "H_a": 0.208, + "d_bind": 0.01759 + }, + { + "tau": 0.0287, + "v": 0.8292, + "H_a": 0.195, + "d_bind": 0.01181 + } + ] + }, + { + "step": 1500, + "tok": 6144000, + "ce": 3.9334, + "ppl": 51.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.925, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.0496, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.8546, + "H_a": 0.226, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.8443, + "H_a": 0.205, + "d_bind": 0.01811 + }, + { + "tau": 0.0376, + "v": 0.8346, + "H_a": 0.185, + "d_bind": 0.01227 + } + ] + }, + { + "step": 1520, + "tok": 6225920, + "ce": 4.1166, + "ppl": 61.4, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.895, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.0509, + "loop_stats": [ + { + "tau": 0.0944, + "v": 0.8667, + "H_a": 0.213, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.8466, + "H_a": 0.2, + "d_bind": 0.01704 + }, + { + "tau": 0.0318, + "v": 0.8304, + "H_a": 0.196, + "d_bind": 0.01248 + } + ] + }, + { + "step": 1540, + "tok": 6307840, + "ce": 4.0358, + "ppl": 56.6, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.907, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.0528, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.8516, + "H_a": 0.205, + "d_bind": 0.0 + }, + { + "tau": 0.0303, + "v": 0.8435, + "H_a": 0.21, + "d_bind": 0.01958 + }, + { + "tau": 0.0292, + "v": 0.8346, + "H_a": 0.204, + "d_bind": 0.01388 + } + ] + }, + { + "step": 1560, + "tok": 6389760, + "ce": 3.6009, + "ppl": 36.6, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.915, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.0529, + "loop_stats": [ + { + "tau": 0.1287, + "v": 0.8065, + "H_a": 0.23, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.7949, + "H_a": 0.209, + "d_bind": 0.01987 + }, + { + "tau": 0.0278, + "v": 0.7873, + "H_a": 0.194, + "d_bind": 0.01205 + } + ] + }, + { + "step": 1580, + "tok": 6471680, + "ce": 3.4505, + "ppl": 31.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.922, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.0508, + "loop_stats": [ + { + "tau": 0.1159, + "v": 0.8358, + "H_a": 0.223, + "d_bind": 0.0 + }, + { + "tau": 0.0525, + "v": 0.8249, + "H_a": 0.182, + "d_bind": 0.01928 + }, + { + "tau": 0.0486, + "v": 0.8147, + "H_a": 0.164, + "d_bind": 0.01035 + } + ] + }, + { + "step": 1600, + "tok": 6553600, + "ce": 3.7519, + "ppl": 42.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.92, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.0526, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8545, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.0294, + "v": 0.8435, + "H_a": 0.194, + "d_bind": 0.02009 + }, + { + "tau": 0.0264, + "v": 0.8311, + "H_a": 0.173, + "d_bind": 0.01143 + } + ] + }, + { + "step": 1620, + "tok": 6635520, + "ce": 4.0349, + "ppl": 56.5, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.897, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.0536, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.8303, + "H_a": 0.225, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.8118, + "H_a": 0.22, + "d_bind": 0.01981 + }, + { + "tau": 0.0184, + "v": 0.805, + "H_a": 0.209, + "d_bind": 0.01522 + } + ] + }, + { + "step": 1640, + "tok": 6717440, + "ce": 4.1149, + "ppl": 61.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.885, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.0525, + "loop_stats": [ + { + "tau": 0.0963, + "v": 0.8061, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.0221, + "v": 0.7886, + "H_a": 0.204, + "d_bind": 0.01951 + }, + { + "tau": 0.0219, + "v": 0.7776, + "H_a": 0.201, + "d_bind": 0.01343 + } + ] + }, + { + "step": 1660, + "tok": 6799360, + "ce": 3.9121, + "ppl": 50.0, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.915, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.0507, + "loop_stats": [ + { + "tau": 0.0762, + "v": 0.8028, + "H_a": 0.24, + "d_bind": 0.0 + }, + { + "tau": 0.0307, + "v": 0.7914, + "H_a": 0.209, + "d_bind": 0.01964 + }, + { + "tau": 0.0306, + "v": 0.7833, + "H_a": 0.196, + "d_bind": 0.01235 + } + ] + }, + { + "step": 1680, + "tok": 6881280, + "ce": 3.4341, + "ppl": 31.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.905, + "tok_s": 22457, + "peak_MiB": 5301, + "v_gain": 0.051, + "loop_stats": [ + { + "tau": 0.1222, + "v": 0.7732, + "H_a": 0.227, + "d_bind": 0.0 + }, + { + "tau": 0.0519, + "v": 0.7377, + "H_a": 0.204, + "d_bind": 0.02013 + }, + { + "tau": 0.0479, + "v": 0.7219, + "H_a": 0.18, + "d_bind": 0.01124 + } + ] + }, + { + "step": 1700, + "tok": 6963200, + "ce": 4.0762, + "ppl": 58.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.92, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.0495, + "loop_stats": [ + { + "tau": 0.0846, + "v": 0.8235, + "H_a": 0.233, + "d_bind": 0.0 + }, + { + "tau": 0.0216, + "v": 0.7995, + "H_a": 0.203, + "d_bind": 0.02066 + }, + { + "tau": 0.0199, + "v": 0.7868, + "H_a": 0.187, + "d_bind": 0.01253 + } + ] + }, + { + "step": 1720, + "tok": 7045120, + "ce": 4.1031, + "ppl": 60.5, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.93, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.0488, + "loop_stats": [ + { + "tau": 0.0657, + "v": 0.8974, + "H_a": 0.239, + "d_bind": 0.0 + }, + { + "tau": 0.0239, + "v": 0.8933, + "H_a": 0.218, + "d_bind": 0.02132 + }, + { + "tau": 0.0233, + "v": 0.8873, + "H_a": 0.201, + "d_bind": 0.0137 + } + ] + }, + { + "step": 1740, + "tok": 7127040, + "ce": 4.0573, + "ppl": 57.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.905, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.0491, + "loop_stats": [ + { + "tau": 0.0717, + "v": 0.8536, + "H_a": 0.237, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8421, + "H_a": 0.217, + "d_bind": 0.02147 + }, + { + "tau": 0.0292, + "v": 0.8375, + "H_a": 0.202, + "d_bind": 0.01351 + } + ] + }, + { + "step": 1760, + "tok": 7208960, + "ce": 3.7784, + "ppl": 43.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.902, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.0503, + "loop_stats": [ + { + "tau": 0.0895, + "v": 0.8501, + "H_a": 0.213, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.8467, + "H_a": 0.192, + "d_bind": 0.01808 + }, + { + "tau": 0.0311, + "v": 0.8439, + "H_a": 0.178, + "d_bind": 0.0119 + } + ] + }, + { + "step": 1780, + "tok": 7290880, + "ce": 3.797, + "ppl": 44.6, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.925, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.0528, + "loop_stats": [ + { + "tau": 0.0707, + "v": 0.871, + "H_a": 0.22, + "d_bind": 0.0 + }, + { + "tau": 0.0265, + "v": 0.8663, + "H_a": 0.197, + "d_bind": 0.02019 + }, + { + "tau": 0.0275, + "v": 0.859, + "H_a": 0.182, + "d_bind": 0.01241 + } + ] + }, + { + "step": 1800, + "tok": 7372800, + "ce": 4.2965, + "ppl": 73.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.91, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.0541, + "loop_stats": [ + { + "tau": 0.0993, + "v": 0.8549, + "H_a": 0.205, + "d_bind": 0.0 + }, + { + "tau": 0.0329, + "v": 0.8297, + "H_a": 0.193, + "d_bind": 0.01715 + }, + { + "tau": 0.0241, + "v": 0.8162, + "H_a": 0.188, + "d_bind": 0.01253 + } + ] + }, + { + "step": 1820, + "tok": 7454720, + "ce": 3.7787, + "ppl": 43.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.917, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.0574, + "loop_stats": [ + { + "tau": 0.08, + "v": 0.8705, + "H_a": 0.23, + "d_bind": 0.0 + }, + { + "tau": 0.0222, + "v": 0.8564, + "H_a": 0.21, + "d_bind": 0.01929 + }, + { + "tau": 0.0215, + "v": 0.8461, + "H_a": 0.2, + "d_bind": 0.0124 + } + ] + }, + { + "step": 1840, + "tok": 7536640, + "ce": 3.9545, + "ppl": 52.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.92, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.0567, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8902, + "H_a": 0.236, + "d_bind": 0.0 + }, + { + "tau": 0.0494, + "v": 0.8675, + "H_a": 0.224, + "d_bind": 0.0222 + }, + { + "tau": 0.0463, + "v": 0.854, + "H_a": 0.208, + "d_bind": 0.01422 + } + ] + }, + { + "step": 1860, + "tok": 7618560, + "ce": 4.1927, + "ppl": 66.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.897, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.0587, + "loop_stats": [ + { + "tau": 0.0915, + "v": 0.8946, + "H_a": 0.218, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.8812, + "H_a": 0.203, + "d_bind": 0.0204 + }, + { + "tau": 0.0322, + "v": 0.8716, + "H_a": 0.193, + "d_bind": 0.01249 + } + ] + }, + { + "step": 1880, + "tok": 7700480, + "ce": 3.5087, + "ppl": 33.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.0623, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.8716, + "H_a": 0.217, + "d_bind": 0.0 + }, + { + "tau": 0.0573, + "v": 0.8446, + "H_a": 0.194, + "d_bind": 0.02089 + }, + { + "tau": 0.055, + "v": 0.8321, + "H_a": 0.187, + "d_bind": 0.0131 + } + ] + }, + { + "step": 1900, + "tok": 7782400, + "ce": 3.7482, + "ppl": 42.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.917, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.064, + "loop_stats": [ + { + "tau": 0.0802, + "v": 0.8649, + "H_a": 0.237, + "d_bind": 0.0 + }, + { + "tau": 0.0389, + "v": 0.8357, + "H_a": 0.196, + "d_bind": 0.02023 + }, + { + "tau": 0.0407, + "v": 0.8196, + "H_a": 0.171, + "d_bind": 0.01134 + } + ] + }, + { + "step": 1920, + "tok": 7864320, + "ce": 3.7682, + "ppl": 43.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.92, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.0638, + "loop_stats": [ + { + "tau": 0.0927, + "v": 0.8965, + "H_a": 0.224, + "d_bind": 0.0 + }, + { + "tau": 0.0271, + "v": 0.8863, + "H_a": 0.19, + "d_bind": 0.01977 + }, + { + "tau": 0.0252, + "v": 0.874, + "H_a": 0.176, + "d_bind": 0.01181 + } + ] + }, + { + "step": 1940, + "tok": 7946240, + "ce": 3.777, + "ppl": 43.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.0644, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.9188, + "H_a": 0.217, + "d_bind": 0.0 + }, + { + "tau": 0.036, + "v": 0.9011, + "H_a": 0.203, + "d_bind": 0.02189 + }, + { + "tau": 0.0293, + "v": 0.8837, + "H_a": 0.188, + "d_bind": 0.01432 + } + ] + }, + { + "step": 1960, + "tok": 8028160, + "ce": 3.7749, + "ppl": 43.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.915, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.0653, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.8707, + "H_a": 0.23, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.8402, + "H_a": 0.2, + "d_bind": 0.01909 + }, + { + "tau": 0.0311, + "v": 0.8266, + "H_a": 0.18, + "d_bind": 0.0115 + } + ] + }, + { + "step": 1980, + "tok": 8110080, + "ce": 3.6414, + "ppl": 38.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.917, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.0683, + "loop_stats": [ + { + "tau": 0.0926, + "v": 0.8943, + "H_a": 0.225, + "d_bind": 0.0 + }, + { + "tau": 0.0325, + "v": 0.8699, + "H_a": 0.205, + "d_bind": 0.02203 + }, + { + "tau": 0.0302, + "v": 0.8605, + "H_a": 0.188, + "d_bind": 0.01334 + } + ] + }, + { + "step": 2000, + "tok": 8192000, + "ce": 3.9321, + "ppl": 51.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.915, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.0704, + "loop_stats": [ + { + "tau": 0.0618, + "v": 0.8623, + "H_a": 0.216, + "d_bind": 0.0 + }, + { + "tau": 0.0259, + "v": 0.8301, + "H_a": 0.194, + "d_bind": 0.01825 + }, + { + "tau": 0.0262, + "v": 0.8116, + "H_a": 0.181, + "d_bind": 0.01197 + } + ] + }, + { + "step": 2020, + "tok": 8273920, + "ce": 4.0928, + "ppl": 59.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.0693, + "loop_stats": [ + { + "tau": 0.0789, + "v": 0.9086, + "H_a": 0.239, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8955, + "H_a": 0.221, + "d_bind": 0.02002 + }, + { + "tau": 0.0252, + "v": 0.8831, + "H_a": 0.211, + "d_bind": 0.0134 + } + ] + }, + { + "step": 2040, + "tok": 8355840, + "ce": 4.201, + "ppl": 66.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.93, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.0721, + "loop_stats": [ + { + "tau": 0.0823, + "v": 0.9083, + "H_a": 0.235, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8915, + "H_a": 0.214, + "d_bind": 0.02047 + }, + { + "tau": 0.0316, + "v": 0.8793, + "H_a": 0.2, + "d_bind": 0.01346 + } + ] + }, + { + "step": 2060, + "tok": 8437760, + "ce": 4.1124, + "ppl": 61.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.915, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.0703, + "loop_stats": [ + { + "tau": 0.0952, + "v": 0.8972, + "H_a": 0.246, + "d_bind": 0.0 + }, + { + "tau": 0.0453, + "v": 0.8811, + "H_a": 0.225, + "d_bind": 0.02243 + }, + { + "tau": 0.0414, + "v": 0.8734, + "H_a": 0.21, + "d_bind": 0.01458 + } + ] + }, + { + "step": 2080, + "tok": 8519680, + "ce": 4.3206, + "ppl": 75.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.933, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.0722, + "loop_stats": [ + { + "tau": 0.0759, + "v": 0.9428, + "H_a": 0.248, + "d_bind": 0.0 + }, + { + "tau": 0.0209, + "v": 0.9349, + "H_a": 0.233, + "d_bind": 0.02142 + }, + { + "tau": 0.0193, + "v": 0.9289, + "H_a": 0.224, + "d_bind": 0.01406 + } + ] + }, + { + "step": 2100, + "tok": 8601600, + "ce": 3.2715, + "ppl": 26.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.922, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.0735, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.8866, + "H_a": 0.217, + "d_bind": 0.0 + }, + { + "tau": 0.0317, + "v": 0.8778, + "H_a": 0.191, + "d_bind": 0.02177 + }, + { + "tau": 0.0361, + "v": 0.874, + "H_a": 0.177, + "d_bind": 0.0128 + } + ] + }, + { + "step": 2120, + "tok": 8683520, + "ce": 3.396, + "ppl": 29.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.922, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.0735, + "loop_stats": [ + { + "tau": 0.097, + "v": 0.898, + "H_a": 0.227, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.8819, + "H_a": 0.182, + "d_bind": 0.02179 + }, + { + "tau": 0.0379, + "v": 0.8691, + "H_a": 0.167, + "d_bind": 0.01234 + } + ] + }, + { + "step": 2140, + "tok": 8765440, + "ce": 4.0157, + "ppl": 55.5, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.922, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.0767, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.9037, + "H_a": 0.228, + "d_bind": 0.0 + }, + { + "tau": 0.0295, + "v": 0.8897, + "H_a": 0.215, + "d_bind": 0.02122 + }, + { + "tau": 0.0248, + "v": 0.876, + "H_a": 0.196, + "d_bind": 0.01421 + } + ] + }, + { + "step": 2160, + "tok": 8847360, + "ce": 4.1318, + "ppl": 62.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.0779, + "loop_stats": [ + { + "tau": 0.0999, + "v": 0.9226, + "H_a": 0.246, + "d_bind": 0.0 + }, + { + "tau": 0.0271, + "v": 0.9044, + "H_a": 0.218, + "d_bind": 0.02348 + }, + { + "tau": 0.0238, + "v": 0.8961, + "H_a": 0.198, + "d_bind": 0.01445 + } + ] + }, + { + "step": 2180, + "tok": 8929280, + "ce": 3.5554, + "ppl": 35.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.922, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.0768, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.8792, + "H_a": 0.214, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8679, + "H_a": 0.184, + "d_bind": 0.01955 + }, + { + "tau": 0.0486, + "v": 0.8622, + "H_a": 0.172, + "d_bind": 0.0118 + } + ] + }, + { + "step": 2200, + "tok": 9011200, + "ce": 3.5838, + "ppl": 36.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.91, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.0751, + "loop_stats": [ + { + "tau": 0.123, + "v": 0.9034, + "H_a": 0.232, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.8913, + "H_a": 0.203, + "d_bind": 0.02199 + }, + { + "tau": 0.0361, + "v": 0.8851, + "H_a": 0.181, + "d_bind": 0.0136 + } + ] + }, + { + "step": 2220, + "tok": 9093120, + "ce": 3.6144, + "ppl": 37.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.922, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.0771, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.8903, + "H_a": 0.222, + "d_bind": 0.0 + }, + { + "tau": 0.026, + "v": 0.8717, + "H_a": 0.208, + "d_bind": 0.02265 + }, + { + "tau": 0.0226, + "v": 0.8627, + "H_a": 0.185, + "d_bind": 0.01395 + } + ] + }, + { + "step": 2240, + "tok": 9175040, + "ce": 3.7968, + "ppl": 44.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.933, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.0774, + "loop_stats": [ + { + "tau": 0.0855, + "v": 0.9288, + "H_a": 0.21, + "d_bind": 0.0 + }, + { + "tau": 0.0232, + "v": 0.9123, + "H_a": 0.19, + "d_bind": 0.02062 + }, + { + "tau": 0.0218, + "v": 0.9013, + "H_a": 0.174, + "d_bind": 0.0141 + } + ] + }, + { + "step": 2260, + "tok": 9256960, + "ce": 3.6769, + "ppl": 39.5, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.917, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.0774, + "loop_stats": [ + { + "tau": 0.0931, + "v": 0.9018, + "H_a": 0.197, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.8846, + "H_a": 0.179, + "d_bind": 0.02132 + }, + { + "tau": 0.0298, + "v": 0.876, + "H_a": 0.169, + "d_bind": 0.01325 + } + ] + }, + { + "step": 2280, + "tok": 9338880, + "ce": 3.7, + "ppl": 40.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.922, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.0793, + "loop_stats": [ + { + "tau": 0.0876, + "v": 0.9127, + "H_a": 0.232, + "d_bind": 0.0 + }, + { + "tau": 0.0295, + "v": 0.8995, + "H_a": 0.195, + "d_bind": 0.02218 + }, + { + "tau": 0.0285, + "v": 0.8904, + "H_a": 0.177, + "d_bind": 0.01508 + } + ] + }, + { + "step": 2300, + "tok": 9420800, + "ce": 3.2802, + "ppl": 26.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.927, + "tok_s": 22453, + "peak_MiB": 5301, + "v_gain": 0.0802, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.8938, + "H_a": 0.188, + "d_bind": 0.0 + }, + { + "tau": 0.0533, + "v": 0.8756, + "H_a": 0.158, + "d_bind": 0.01972 + }, + { + "tau": 0.0518, + "v": 0.8627, + "H_a": 0.147, + "d_bind": 0.01192 + } + ] + }, + { + "step": 2320, + "tok": 9502720, + "ce": 3.7232, + "ppl": 41.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.912, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.0814, + "loop_stats": [ + { + "tau": 0.1313, + "v": 0.9076, + "H_a": 0.203, + "d_bind": 0.0 + }, + { + "tau": 0.0533, + "v": 0.8815, + "H_a": 0.192, + "d_bind": 0.02258 + }, + { + "tau": 0.0469, + "v": 0.8636, + "H_a": 0.185, + "d_bind": 0.01544 + } + ] + }, + { + "step": 2340, + "tok": 9584640, + "ce": 3.6485, + "ppl": 38.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.93, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.0823, + "loop_stats": [ + { + "tau": 0.09, + "v": 0.9072, + "H_a": 0.214, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8801, + "H_a": 0.204, + "d_bind": 0.0249 + }, + { + "tau": 0.0357, + "v": 0.8592, + "H_a": 0.183, + "d_bind": 0.01529 + } + ] + }, + { + "step": 2360, + "tok": 9666560, + "ce": 4.2133, + "ppl": 67.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.92, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.086, + "loop_stats": [ + { + "tau": 0.0728, + "v": 0.9208, + "H_a": 0.24, + "d_bind": 0.0 + }, + { + "tau": 0.0266, + "v": 0.9002, + "H_a": 0.2, + "d_bind": 0.0248 + }, + { + "tau": 0.0249, + "v": 0.8846, + "H_a": 0.181, + "d_bind": 0.01437 + } + ] + }, + { + "step": 2380, + "tok": 9748480, + "ce": 3.8296, + "ppl": 46.0, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.917, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.0865, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.9172, + "H_a": 0.217, + "d_bind": 0.0 + }, + { + "tau": 0.0181, + "v": 0.8894, + "H_a": 0.19, + "d_bind": 0.02305 + }, + { + "tau": 0.0158, + "v": 0.8678, + "H_a": 0.175, + "d_bind": 0.01468 + } + ] + }, + { + "step": 2400, + "tok": 9830400, + "ce": 3.5015, + "ppl": 33.2, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.922, + "tok_s": 22436, + "peak_MiB": 5301, + "v_gain": 0.0873, + "loop_stats": [ + { + "tau": 0.1067, + "v": 0.8747, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.019, + "v": 0.8377, + "H_a": 0.183, + "d_bind": 0.02113 + }, + { + "tau": 0.0188, + "v": 0.8139, + "H_a": 0.172, + "d_bind": 0.01396 + } + ] + }, + { + "step": 2420, + "tok": 9912320, + "ce": 3.8096, + "ppl": 45.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.935, + "tok_s": 22441, + "peak_MiB": 5301, + "v_gain": 0.0858, + "loop_stats": [ + { + "tau": 0.098, + "v": 0.9313, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.9001, + "H_a": 0.207, + "d_bind": 0.02537 + }, + { + "tau": 0.0246, + "v": 0.8815, + "H_a": 0.18, + "d_bind": 0.01419 + } + ] + }, + { + "step": 2440, + "tok": 9994240, + "ce": 3.6946, + "ppl": 40.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22445, + "peak_MiB": 5301, + "v_gain": 0.0873, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.9236, + "H_a": 0.238, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8936, + "H_a": 0.201, + "d_bind": 0.02663 + }, + { + "tau": 0.044, + "v": 0.8712, + "H_a": 0.179, + "d_bind": 0.01394 + } + ] + }, + { + "step": 2460, + "tok": 10076160, + "ce": 3.8421, + "ppl": 46.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.927, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.0889, + "loop_stats": [ + { + "tau": 0.0826, + "v": 0.918, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8941, + "H_a": 0.213, + "d_bind": 0.02513 + }, + { + "tau": 0.0343, + "v": 0.8792, + "H_a": 0.197, + "d_bind": 0.01511 + } + ] + }, + { + "step": 2480, + "tok": 10158080, + "ce": 3.7044, + "ppl": 40.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.91, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.0893, + "loop_stats": [ + { + "tau": 0.102, + "v": 0.867, + "H_a": 0.202, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8402, + "H_a": 0.192, + "d_bind": 0.02075 + }, + { + "tau": 0.0438, + "v": 0.8207, + "H_a": 0.174, + "d_bind": 0.01321 + } + ] + }, + { + "step": 2500, + "tok": 10240000, + "ce": 4.0821, + "ppl": 59.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.0882, + "loop_stats": [ + { + "tau": 0.0772, + "v": 0.9079, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.0527, + "v": 0.8775, + "H_a": 0.198, + "d_bind": 0.02373 + }, + { + "tau": 0.0525, + "v": 0.8603, + "H_a": 0.179, + "d_bind": 0.01522 + } + ] + }, + { + "step": 2520, + "tok": 10321920, + "ce": 3.5957, + "ppl": 36.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.925, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.09, + "loop_stats": [ + { + "tau": 0.1336, + "v": 0.8931, + "H_a": 0.197, + "d_bind": 0.0 + }, + { + "tau": 0.0486, + "v": 0.855, + "H_a": 0.19, + "d_bind": 0.0217 + }, + { + "tau": 0.0493, + "v": 0.8296, + "H_a": 0.178, + "d_bind": 0.01497 + } + ] + }, + { + "step": 2540, + "tok": 10403840, + "ce": 4.0223, + "ppl": 55.8, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.922, + "tok_s": 22434, + "peak_MiB": 5301, + "v_gain": 0.09, + "loop_stats": [ + { + "tau": 0.0819, + "v": 0.918, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8969, + "H_a": 0.193, + "d_bind": 0.02416 + }, + { + "tau": 0.0375, + "v": 0.8828, + "H_a": 0.181, + "d_bind": 0.01505 + } + ] + }, + { + "step": 2560, + "tok": 10485760, + "ce": 4.1255, + "ppl": 61.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.925, + "tok_s": 22439, + "peak_MiB": 5301, + "v_gain": 0.0915, + "loop_stats": [ + { + "tau": 0.1019, + "v": 0.9492, + "H_a": 0.236, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.9326, + "H_a": 0.213, + "d_bind": 0.02409 + }, + { + "tau": 0.0357, + "v": 0.9203, + "H_a": 0.195, + "d_bind": 0.01515 + } + ] + }, + { + "step": 2580, + "tok": 10567680, + "ce": 3.5755, + "ppl": 35.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.922, + "tok_s": 22443, + "peak_MiB": 5301, + "v_gain": 0.0955, + "loop_stats": [ + { + "tau": 0.115, + "v": 0.9246, + "H_a": 0.211, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.8929, + "H_a": 0.183, + "d_bind": 0.02504 + }, + { + "tau": 0.0493, + "v": 0.8696, + "H_a": 0.176, + "d_bind": 0.01585 + } + ] + }, + { + "step": 2600, + "tok": 10649600, + "ce": 3.4271, + "ppl": 30.8, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.927, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.0977, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.887, + "H_a": 0.207, + "d_bind": 0.0 + }, + { + "tau": 0.054, + "v": 0.8503, + "H_a": 0.195, + "d_bind": 0.02314 + }, + { + "tau": 0.0549, + "v": 0.8181, + "H_a": 0.178, + "d_bind": 0.01374 + } + ] + }, + { + "step": 2620, + "tok": 10731520, + "ce": 3.7668, + "ppl": 43.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.92, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.1002, + "loop_stats": [ + { + "tau": 0.0987, + "v": 0.8888, + "H_a": 0.191, + "d_bind": 0.0 + }, + { + "tau": 0.0321, + "v": 0.8576, + "H_a": 0.17, + "d_bind": 0.01975 + }, + { + "tau": 0.035, + "v": 0.827, + "H_a": 0.16, + "d_bind": 0.01278 + } + ] + }, + { + "step": 2640, + "tok": 10813440, + "ce": 3.7404, + "ppl": 42.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.922, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.1014, + "loop_stats": [ + { + "tau": 0.0743, + "v": 0.9357, + "H_a": 0.231, + "d_bind": 0.0 + }, + { + "tau": 0.0235, + "v": 0.9106, + "H_a": 0.206, + "d_bind": 0.02579 + }, + { + "tau": 0.0253, + "v": 0.8874, + "H_a": 0.188, + "d_bind": 0.01555 + } + ] + }, + { + "step": 2660, + "tok": 10895360, + "ce": 3.575, + "ppl": 35.7, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.9, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.1017, + "loop_stats": [ + { + "tau": 0.0883, + "v": 0.8958, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8563, + "H_a": 0.177, + "d_bind": 0.02216 + }, + { + "tau": 0.0316, + "v": 0.8246, + "H_a": 0.157, + "d_bind": 0.014 + } + ] + }, + { + "step": 2680, + "tok": 10977280, + "ce": 3.4913, + "ppl": 32.8, + "pref_transfer_pron": 0.862, + "pref_train_sv": 0.915, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.1053, + "loop_stats": [ + { + "tau": 0.1135, + "v": 0.8856, + "H_a": 0.193, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8364, + "H_a": 0.177, + "d_bind": 0.01996 + }, + { + "tau": 0.0437, + "v": 0.8035, + "H_a": 0.162, + "d_bind": 0.01342 + } + ] + }, + { + "step": 2700, + "tok": 11059200, + "ce": 3.8543, + "ppl": 47.2, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.917, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.104, + "loop_stats": [ + { + "tau": 0.0999, + "v": 0.9079, + "H_a": 0.197, + "d_bind": 0.0 + }, + { + "tau": 0.0384, + "v": 0.8683, + "H_a": 0.171, + "d_bind": 0.02076 + }, + { + "tau": 0.0371, + "v": 0.8332, + "H_a": 0.158, + "d_bind": 0.01316 + } + ] + }, + { + "step": 2720, + "tok": 11141120, + "ce": 3.5856, + "ppl": 36.1, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.933, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.1039, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.9301, + "H_a": 0.198, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.908, + "H_a": 0.175, + "d_bind": 0.02058 + }, + { + "tau": 0.0274, + "v": 0.8883, + "H_a": 0.162, + "d_bind": 0.01421 + } + ] + }, + { + "step": 2740, + "tok": 11223040, + "ce": 3.8225, + "ppl": 45.7, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.94, + "tok_s": 22445, + "peak_MiB": 5301, + "v_gain": 0.1078, + "loop_stats": [ + { + "tau": 0.0885, + "v": 0.9215, + "H_a": 0.243, + "d_bind": 0.0 + }, + { + "tau": 0.0203, + "v": 0.8911, + "H_a": 0.191, + "d_bind": 0.02379 + }, + { + "tau": 0.0213, + "v": 0.868, + "H_a": 0.175, + "d_bind": 0.01459 + } + ] + }, + { + "step": 2760, + "tok": 11304960, + "ce": 3.6295, + "ppl": 37.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22414, + "peak_MiB": 5301, + "v_gain": 0.1092, + "loop_stats": [ + { + "tau": 0.1086, + "v": 0.8846, + "H_a": 0.203, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8417, + "H_a": 0.171, + "d_bind": 0.02055 + }, + { + "tau": 0.0467, + "v": 0.802, + "H_a": 0.16, + "d_bind": 0.01277 + } + ] + }, + { + "step": 2780, + "tok": 11386880, + "ce": 4.0929, + "ppl": 59.9, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.94, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.1127, + "loop_stats": [ + { + "tau": 0.0709, + "v": 0.9398, + "H_a": 0.229, + "d_bind": 0.0 + }, + { + "tau": 0.014, + "v": 0.9076, + "H_a": 0.187, + "d_bind": 0.0238 + }, + { + "tau": 0.0141, + "v": 0.8859, + "H_a": 0.165, + "d_bind": 0.01398 + } + ] + }, + { + "step": 2800, + "tok": 11468800, + "ce": 3.7499, + "ppl": 42.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.933, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.1106, + "loop_stats": [ + { + "tau": 0.1009, + "v": 0.9302, + "H_a": 0.204, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.9064, + "H_a": 0.17, + "d_bind": 0.02286 + }, + { + "tau": 0.035, + "v": 0.8819, + "H_a": 0.164, + "d_bind": 0.0152 + } + ] + }, + { + "step": 2820, + "tok": 11550720, + "ce": 4.0153, + "ppl": 55.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.935, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.1118, + "loop_stats": [ + { + "tau": 0.0763, + "v": 0.9304, + "H_a": 0.228, + "d_bind": 0.0 + }, + { + "tau": 0.0295, + "v": 0.9025, + "H_a": 0.185, + "d_bind": 0.02399 + }, + { + "tau": 0.0276, + "v": 0.8842, + "H_a": 0.17, + "d_bind": 0.01527 + } + ] + }, + { + "step": 2840, + "tok": 11632640, + "ce": 3.5268, + "ppl": 34.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22446, + "peak_MiB": 5301, + "v_gain": 0.1117, + "loop_stats": [ + { + "tau": 0.0938, + "v": 0.9013, + "H_a": 0.205, + "d_bind": 0.0 + }, + { + "tau": 0.0327, + "v": 0.8766, + "H_a": 0.186, + "d_bind": 0.02437 + }, + { + "tau": 0.0322, + "v": 0.8594, + "H_a": 0.169, + "d_bind": 0.01544 + } + ] + }, + { + "step": 2860, + "tok": 11714560, + "ce": 3.0505, + "ppl": 21.1, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.933, + "tok_s": 22437, + "peak_MiB": 5301, + "v_gain": 0.113, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.8685, + "H_a": 0.194, + "d_bind": 0.0 + }, + { + "tau": 0.0565, + "v": 0.8275, + "H_a": 0.16, + "d_bind": 0.02343 + }, + { + "tau": 0.0599, + "v": 0.7901, + "H_a": 0.161, + "d_bind": 0.01446 + } + ] + }, + { + "step": 2880, + "tok": 11796480, + "ce": 3.4402, + "ppl": 31.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22406, + "peak_MiB": 5301, + "v_gain": 0.1136, + "loop_stats": [ + { + "tau": 0.0793, + "v": 0.8897, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8549, + "H_a": 0.17, + "d_bind": 0.02154 + }, + { + "tau": 0.0447, + "v": 0.8259, + "H_a": 0.154, + "d_bind": 0.01419 + } + ] + }, + { + "step": 2900, + "tok": 11878400, + "ce": 3.5693, + "ppl": 35.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.1129, + "loop_stats": [ + { + "tau": 0.0833, + "v": 0.9052, + "H_a": 0.214, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8859, + "H_a": 0.171, + "d_bind": 0.02311 + }, + { + "tau": 0.0464, + "v": 0.8671, + "H_a": 0.164, + "d_bind": 0.01511 + } + ] + }, + { + "step": 2920, + "tok": 11960320, + "ce": 3.6484, + "ppl": 38.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.925, + "tok_s": 22413, + "peak_MiB": 5301, + "v_gain": 0.1136, + "loop_stats": [ + { + "tau": 0.1148, + "v": 0.9079, + "H_a": 0.201, + "d_bind": 0.0 + }, + { + "tau": 0.0457, + "v": 0.8848, + "H_a": 0.176, + "d_bind": 0.0231 + }, + { + "tau": 0.0447, + "v": 0.8681, + "H_a": 0.168, + "d_bind": 0.01556 + } + ] + }, + { + "step": 2940, + "tok": 12042240, + "ce": 3.5353, + "ppl": 34.3, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.938, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.117, + "loop_stats": [ + { + "tau": 0.0919, + "v": 0.9089, + "H_a": 0.201, + "d_bind": 0.0 + }, + { + "tau": 0.0362, + "v": 0.8863, + "H_a": 0.175, + "d_bind": 0.02213 + }, + { + "tau": 0.0391, + "v": 0.8693, + "H_a": 0.157, + "d_bind": 0.01389 + } + ] + }, + { + "step": 2960, + "tok": 12124160, + "ce": 3.6012, + "ppl": 36.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.935, + "tok_s": 22438, + "peak_MiB": 5301, + "v_gain": 0.1175, + "loop_stats": [ + { + "tau": 0.0881, + "v": 0.9248, + "H_a": 0.203, + "d_bind": 0.0 + }, + { + "tau": 0.0381, + "v": 0.9008, + "H_a": 0.177, + "d_bind": 0.02404 + }, + { + "tau": 0.04, + "v": 0.8848, + "H_a": 0.172, + "d_bind": 0.01679 + } + ] + }, + { + "step": 2980, + "tok": 12206080, + "ce": 3.5231, + "ppl": 33.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.938, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.1177, + "loop_stats": [ + { + "tau": 0.0908, + "v": 0.9119, + "H_a": 0.19, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.8827, + "H_a": 0.168, + "d_bind": 0.02319 + }, + { + "tau": 0.0385, + "v": 0.8605, + "H_a": 0.153, + "d_bind": 0.01429 + } + ] + }, + { + "step": 3000, + "tok": 12288000, + "ce": 3.561, + "ppl": 35.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.93, + "tok_s": 22391, + "peak_MiB": 5301, + "v_gain": 0.1191, + "loop_stats": [ + { + "tau": 0.0972, + "v": 0.9283, + "H_a": 0.21, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.9036, + "H_a": 0.171, + "d_bind": 0.02388 + }, + { + "tau": 0.0327, + "v": 0.8855, + "H_a": 0.151, + "d_bind": 0.0144 + } + ] + }, + { + "step": 3020, + "tok": 12369920, + "ce": 3.839, + "ppl": 46.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.92, + "tok_s": 22395, + "peak_MiB": 5301, + "v_gain": 0.1191, + "loop_stats": [ + { + "tau": 0.0815, + "v": 0.8803, + "H_a": 0.213, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8493, + "H_a": 0.181, + "d_bind": 0.02443 + }, + { + "tau": 0.0259, + "v": 0.8323, + "H_a": 0.156, + "d_bind": 0.01528 + } + ] + }, + { + "step": 3040, + "tok": 12451840, + "ce": 3.0371, + "ppl": 20.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22394, + "peak_MiB": 5301, + "v_gain": 0.1213, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8835, + "H_a": 0.176, + "d_bind": 0.0 + }, + { + "tau": 0.0429, + "v": 0.8584, + "H_a": 0.168, + "d_bind": 0.02504 + }, + { + "tau": 0.0374, + "v": 0.8364, + "H_a": 0.158, + "d_bind": 0.01573 + } + ] + }, + { + "step": 3060, + "tok": 12533760, + "ce": 3.9931, + "ppl": 54.2, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.93, + "tok_s": 22397, + "peak_MiB": 5301, + "v_gain": 0.1205, + "loop_stats": [ + { + "tau": 0.0743, + "v": 0.9128, + "H_a": 0.206, + "d_bind": 0.0 + }, + { + "tau": 0.0227, + "v": 0.8836, + "H_a": 0.174, + "d_bind": 0.02399 + }, + { + "tau": 0.0203, + "v": 0.8629, + "H_a": 0.164, + "d_bind": 0.01548 + } + ] + }, + { + "step": 3080, + "tok": 12615680, + "ce": 3.8211, + "ppl": 45.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.1174, + "loop_stats": [ + { + "tau": 0.0881, + "v": 0.8964, + "H_a": 0.215, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.8616, + "H_a": 0.178, + "d_bind": 0.02379 + }, + { + "tau": 0.0265, + "v": 0.8387, + "H_a": 0.166, + "d_bind": 0.01514 + } + ] + }, + { + "step": 3100, + "tok": 12697600, + "ce": 3.7122, + "ppl": 40.9, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.915, + "tok_s": 22406, + "peak_MiB": 5301, + "v_gain": 0.1191, + "loop_stats": [ + { + "tau": 0.1032, + "v": 0.9094, + "H_a": 0.206, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.8874, + "H_a": 0.172, + "d_bind": 0.02394 + }, + { + "tau": 0.0353, + "v": 0.8725, + "H_a": 0.168, + "d_bind": 0.01633 + } + ] + }, + { + "step": 3120, + "tok": 12779520, + "ce": 3.5375, + "ppl": 34.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.922, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1214, + "loop_stats": [ + { + "tau": 0.1195, + "v": 0.8998, + "H_a": 0.188, + "d_bind": 0.0 + }, + { + "tau": 0.0506, + "v": 0.8749, + "H_a": 0.171, + "d_bind": 0.02363 + }, + { + "tau": 0.0494, + "v": 0.8526, + "H_a": 0.166, + "d_bind": 0.0161 + } + ] + }, + { + "step": 3140, + "tok": 12861440, + "ce": 3.3989, + "ppl": 29.9, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.925, + "tok_s": 22383, + "peak_MiB": 5301, + "v_gain": 0.1245, + "loop_stats": [ + { + "tau": 0.0996, + "v": 0.8938, + "H_a": 0.192, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8684, + "H_a": 0.159, + "d_bind": 0.02514 + }, + { + "tau": 0.048, + "v": 0.8472, + "H_a": 0.15, + "d_bind": 0.0153 + } + ] + }, + { + "step": 3160, + "tok": 12943360, + "ce": 3.6406, + "ppl": 38.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.92, + "tok_s": 22388, + "peak_MiB": 5301, + "v_gain": 0.1266, + "loop_stats": [ + { + "tau": 0.0974, + "v": 0.8883, + "H_a": 0.186, + "d_bind": 0.0 + }, + { + "tau": 0.0391, + "v": 0.8523, + "H_a": 0.166, + "d_bind": 0.02313 + }, + { + "tau": 0.0375, + "v": 0.8308, + "H_a": 0.154, + "d_bind": 0.01466 + } + ] + }, + { + "step": 3180, + "tok": 13025280, + "ce": 3.6802, + "ppl": 39.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.922, + "tok_s": 22395, + "peak_MiB": 5301, + "v_gain": 0.1301, + "loop_stats": [ + { + "tau": 0.0872, + "v": 0.8994, + "H_a": 0.196, + "d_bind": 0.0 + }, + { + "tau": 0.0167, + "v": 0.8708, + "H_a": 0.179, + "d_bind": 0.02382 + }, + { + "tau": 0.0163, + "v": 0.8473, + "H_a": 0.166, + "d_bind": 0.01668 + } + ] + }, + { + "step": 3200, + "tok": 13107200, + "ce": 3.8143, + "ppl": 45.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.897, + "tok_s": 22412, + "peak_MiB": 5301, + "v_gain": 0.1298, + "loop_stats": [ + { + "tau": 0.0883, + "v": 0.8884, + "H_a": 0.178, + "d_bind": 0.0 + }, + { + "tau": 0.0182, + "v": 0.8681, + "H_a": 0.164, + "d_bind": 0.02093 + }, + { + "tau": 0.0197, + "v": 0.8537, + "H_a": 0.147, + "d_bind": 0.01398 + } + ] + }, + { + "step": 3220, + "tok": 13189120, + "ce": 3.1594, + "ppl": 23.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.922, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1299, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.8442, + "H_a": 0.194, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8218, + "H_a": 0.175, + "d_bind": 0.02437 + }, + { + "tau": 0.0319, + "v": 0.8096, + "H_a": 0.164, + "d_bind": 0.01688 + } + ] + }, + { + "step": 3240, + "tok": 13271040, + "ce": 3.5736, + "ppl": 35.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.912, + "tok_s": 22382, + "peak_MiB": 5301, + "v_gain": 0.1286, + "loop_stats": [ + { + "tau": 0.0938, + "v": 0.862, + "H_a": 0.192, + "d_bind": 0.0 + }, + { + "tau": 0.0286, + "v": 0.8334, + "H_a": 0.165, + "d_bind": 0.02386 + }, + { + "tau": 0.028, + "v": 0.8159, + "H_a": 0.152, + "d_bind": 0.01618 + } + ] + }, + { + "step": 3260, + "tok": 13352960, + "ce": 3.6956, + "ppl": 40.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.91, + "tok_s": 22385, + "peak_MiB": 5301, + "v_gain": 0.1261, + "loop_stats": [ + { + "tau": 0.102, + "v": 0.8544, + "H_a": 0.195, + "d_bind": 0.0 + }, + { + "tau": 0.0307, + "v": 0.821, + "H_a": 0.168, + "d_bind": 0.02183 + }, + { + "tau": 0.0298, + "v": 0.8002, + "H_a": 0.161, + "d_bind": 0.0144 + } + ] + }, + { + "step": 3280, + "tok": 13434880, + "ce": 3.1976, + "ppl": 24.5, + "pref_transfer_pron": 0.875, + "pref_train_sv": 0.938, + "tok_s": 22388, + "peak_MiB": 5301, + "v_gain": 0.1276, + "loop_stats": [ + { + "tau": 0.1148, + "v": 0.8554, + "H_a": 0.192, + "d_bind": 0.0 + }, + { + "tau": 0.0219, + "v": 0.827, + "H_a": 0.17, + "d_bind": 0.02259 + }, + { + "tau": 0.0235, + "v": 0.8062, + "H_a": 0.171, + "d_bind": 0.01563 + } + ] + }, + { + "step": 3300, + "tok": 13516800, + "ce": 3.6169, + "ppl": 37.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.927, + "tok_s": 22391, + "peak_MiB": 5301, + "v_gain": 0.1288, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8918, + "H_a": 0.19, + "d_bind": 0.0 + }, + { + "tau": 0.0178, + "v": 0.8767, + "H_a": 0.161, + "d_bind": 0.02174 + }, + { + "tau": 0.0195, + "v": 0.8683, + "H_a": 0.157, + "d_bind": 0.01535 + } + ] + }, + { + "step": 3320, + "tok": 13598720, + "ce": 3.3148, + "ppl": 27.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.13, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.879, + "H_a": 0.194, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.8595, + "H_a": 0.164, + "d_bind": 0.02304 + }, + { + "tau": 0.0395, + "v": 0.8487, + "H_a": 0.158, + "d_bind": 0.01572 + } + ] + }, + { + "step": 3340, + "tok": 13680640, + "ce": 3.3086, + "ppl": 27.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.945, + "tok_s": 22403, + "peak_MiB": 5301, + "v_gain": 0.1322, + "loop_stats": [ + { + "tau": 0.1085, + "v": 0.8673, + "H_a": 0.189, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.8301, + "H_a": 0.165, + "d_bind": 0.02197 + }, + { + "tau": 0.0392, + "v": 0.8123, + "H_a": 0.159, + "d_bind": 0.01538 + } + ] + }, + { + "step": 3360, + "tok": 13762560, + "ce": 3.9042, + "ppl": 49.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.94, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1338, + "loop_stats": [ + { + "tau": 0.0819, + "v": 0.9042, + "H_a": 0.231, + "d_bind": 0.0 + }, + { + "tau": 0.016, + "v": 0.8784, + "H_a": 0.183, + "d_bind": 0.02644 + }, + { + "tau": 0.0138, + "v": 0.8595, + "H_a": 0.169, + "d_bind": 0.01636 + } + ] + }, + { + "step": 3380, + "tok": 13844480, + "ce": 3.9888, + "ppl": 54.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.927, + "tok_s": 22380, + "peak_MiB": 5301, + "v_gain": 0.1347, + "loop_stats": [ + { + "tau": 0.0735, + "v": 0.8832, + "H_a": 0.223, + "d_bind": 0.0 + }, + { + "tau": 0.0153, + "v": 0.8464, + "H_a": 0.2, + "d_bind": 0.02755 + }, + { + "tau": 0.0181, + "v": 0.8208, + "H_a": 0.191, + "d_bind": 0.01927 + } + ] + }, + { + "step": 3400, + "tok": 13926400, + "ce": 3.6661, + "ppl": 39.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22383, + "peak_MiB": 5301, + "v_gain": 0.1346, + "loop_stats": [ + { + "tau": 0.0849, + "v": 0.8339, + "H_a": 0.211, + "d_bind": 0.0 + }, + { + "tau": 0.0241, + "v": 0.8108, + "H_a": 0.166, + "d_bind": 0.02447 + }, + { + "tau": 0.0269, + "v": 0.7903, + "H_a": 0.166, + "d_bind": 0.01584 + } + ] + }, + { + "step": 3420, + "tok": 14008320, + "ce": 3.3957, + "ppl": 29.8, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.938, + "tok_s": 22386, + "peak_MiB": 5301, + "v_gain": 0.1355, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8326, + "H_a": 0.196, + "d_bind": 0.0 + }, + { + "tau": 0.0227, + "v": 0.8164, + "H_a": 0.17, + "d_bind": 0.02361 + }, + { + "tau": 0.0254, + "v": 0.8049, + "H_a": 0.157, + "d_bind": 0.01587 + } + ] + }, + { + "step": 3440, + "tok": 14090240, + "ce": 3.3425, + "ppl": 28.3, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.933, + "tok_s": 22404, + "peak_MiB": 5301, + "v_gain": 0.1383, + "loop_stats": [ + { + "tau": 0.1009, + "v": 0.7978, + "H_a": 0.177, + "d_bind": 0.0 + }, + { + "tau": 0.0224, + "v": 0.7836, + "H_a": 0.163, + "d_bind": 0.02471 + }, + { + "tau": 0.0323, + "v": 0.7713, + "H_a": 0.149, + "d_bind": 0.01687 + } + ] + }, + { + "step": 3460, + "tok": 14172160, + "ce": 3.3625, + "ppl": 28.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.927, + "tok_s": 22397, + "peak_MiB": 5301, + "v_gain": 0.1375, + "loop_stats": [ + { + "tau": 0.1066, + "v": 0.7709, + "H_a": 0.162, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.7557, + "H_a": 0.167, + "d_bind": 0.02339 + }, + { + "tau": 0.0504, + "v": 0.7426, + "H_a": 0.15, + "d_bind": 0.01644 + } + ] + }, + { + "step": 3480, + "tok": 14254080, + "ce": 3.4869, + "ppl": 32.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22371, + "peak_MiB": 5301, + "v_gain": 0.1364, + "loop_stats": [ + { + "tau": 0.0737, + "v": 0.7907, + "H_a": 0.183, + "d_bind": 0.0 + }, + { + "tau": 0.0224, + "v": 0.7636, + "H_a": 0.17, + "d_bind": 0.02324 + }, + { + "tau": 0.0249, + "v": 0.749, + "H_a": 0.151, + "d_bind": 0.01658 + } + ] + }, + { + "step": 3500, + "tok": 14336000, + "ce": 3.3755, + "ppl": 29.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.927, + "tok_s": 22374, + "peak_MiB": 5301, + "v_gain": 0.1383, + "loop_stats": [ + { + "tau": 0.099, + "v": 0.8332, + "H_a": 0.195, + "d_bind": 0.0 + }, + { + "tau": 0.0301, + "v": 0.8175, + "H_a": 0.168, + "d_bind": 0.02366 + }, + { + "tau": 0.0361, + "v": 0.8109, + "H_a": 0.165, + "d_bind": 0.01758 + } + ] + }, + { + "step": 3520, + "tok": 14417920, + "ce": 3.226, + "ppl": 25.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1402, + "loop_stats": [ + { + "tau": 0.118, + "v": 0.7758, + "H_a": 0.198, + "d_bind": 0.0 + }, + { + "tau": 0.0327, + "v": 0.751, + "H_a": 0.176, + "d_bind": 0.02547 + }, + { + "tau": 0.0478, + "v": 0.7296, + "H_a": 0.162, + "d_bind": 0.01783 + } + ] + }, + { + "step": 3540, + "tok": 14499840, + "ce": 3.5094, + "ppl": 33.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22380, + "peak_MiB": 5301, + "v_gain": 0.1392, + "loop_stats": [ + { + "tau": 0.0807, + "v": 0.8063, + "H_a": 0.216, + "d_bind": 0.0 + }, + { + "tau": 0.0226, + "v": 0.783, + "H_a": 0.173, + "d_bind": 0.02509 + }, + { + "tau": 0.0283, + "v": 0.7683, + "H_a": 0.158, + "d_bind": 0.01652 + } + ] + }, + { + "step": 3560, + "tok": 14581760, + "ce": 3.3324, + "ppl": 28.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.917, + "tok_s": 22398, + "peak_MiB": 5301, + "v_gain": 0.1397, + "loop_stats": [ + { + "tau": 0.1018, + "v": 0.7532, + "H_a": 0.197, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.7337, + "H_a": 0.159, + "d_bind": 0.02449 + }, + { + "tau": 0.0375, + "v": 0.7272, + "H_a": 0.143, + "d_bind": 0.01565 + } + ] + }, + { + "step": 3580, + "tok": 14663680, + "ce": 3.3058, + "ppl": 27.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.92, + "tok_s": 22390, + "peak_MiB": 5301, + "v_gain": 0.1411, + "loop_stats": [ + { + "tau": 0.0987, + "v": 0.7901, + "H_a": 0.202, + "d_bind": 0.0 + }, + { + "tau": 0.0363, + "v": 0.7592, + "H_a": 0.163, + "d_bind": 0.03034 + }, + { + "tau": 0.0369, + "v": 0.7379, + "H_a": 0.152, + "d_bind": 0.01685 + } + ] + }, + { + "step": 3600, + "tok": 14745600, + "ce": 3.6494, + "ppl": 38.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.933, + "tok_s": 22363, + "peak_MiB": 5301, + "v_gain": 0.1437, + "loop_stats": [ + { + "tau": 0.0758, + "v": 0.8623, + "H_a": 0.238, + "d_bind": 0.0 + }, + { + "tau": 0.0227, + "v": 0.8584, + "H_a": 0.193, + "d_bind": 0.0303 + }, + { + "tau": 0.0242, + "v": 0.8522, + "H_a": 0.181, + "d_bind": 0.01834 + } + ] + }, + { + "step": 3620, + "tok": 14827520, + "ce": 3.7303, + "ppl": 41.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.927, + "tok_s": 22366, + "peak_MiB": 5301, + "v_gain": 0.1432, + "loop_stats": [ + { + "tau": 0.0768, + "v": 0.8516, + "H_a": 0.224, + "d_bind": 0.0 + }, + { + "tau": 0.0294, + "v": 0.8402, + "H_a": 0.191, + "d_bind": 0.02798 + }, + { + "tau": 0.0291, + "v": 0.8322, + "H_a": 0.171, + "d_bind": 0.01853 + } + ] + }, + { + "step": 3640, + "tok": 14909440, + "ce": 3.5768, + "ppl": 35.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.933, + "tok_s": 22369, + "peak_MiB": 5301, + "v_gain": 0.1439, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.85, + "H_a": 0.199, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8332, + "H_a": 0.179, + "d_bind": 0.02876 + }, + { + "tau": 0.0438, + "v": 0.8199, + "H_a": 0.156, + "d_bind": 0.01757 + } + ] + }, + { + "step": 3660, + "tok": 14991360, + "ce": 3.0423, + "ppl": 21.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.912, + "tok_s": 22372, + "peak_MiB": 5301, + "v_gain": 0.146, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.8496, + "H_a": 0.197, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.842, + "H_a": 0.189, + "d_bind": 0.02588 + }, + { + "tau": 0.0323, + "v": 0.8365, + "H_a": 0.187, + "d_bind": 0.02086 + } + ] + }, + { + "step": 3680, + "tok": 15073280, + "ce": 3.2774, + "ppl": 26.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22388, + "peak_MiB": 5301, + "v_gain": 0.1449, + "loop_stats": [ + { + "tau": 0.1065, + "v": 0.8949, + "H_a": 0.2, + "d_bind": 0.0 + }, + { + "tau": 0.0509, + "v": 0.8823, + "H_a": 0.193, + "d_bind": 0.02897 + }, + { + "tau": 0.0528, + "v": 0.8746, + "H_a": 0.178, + "d_bind": 0.01851 + } + ] + }, + { + "step": 3700, + "tok": 15155200, + "ce": 3.6674, + "ppl": 39.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22381, + "peak_MiB": 5301, + "v_gain": 0.1437, + "loop_stats": [ + { + "tau": 0.0813, + "v": 0.9026, + "H_a": 0.203, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.8818, + "H_a": 0.167, + "d_bind": 0.0248 + }, + { + "tau": 0.0311, + "v": 0.8753, + "H_a": 0.16, + "d_bind": 0.01643 + } + ] + }, + { + "step": 3720, + "tok": 15237120, + "ce": 3.1815, + "ppl": 24.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22358, + "peak_MiB": 5301, + "v_gain": 0.1452, + "loop_stats": [ + { + "tau": 0.0965, + "v": 0.8505, + "H_a": 0.185, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.8343, + "H_a": 0.166, + "d_bind": 0.02709 + }, + { + "tau": 0.043, + "v": 0.8206, + "H_a": 0.153, + "d_bind": 0.01788 + } + ] + }, + { + "step": 3740, + "tok": 15319040, + "ce": 3.0811, + "ppl": 21.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.927, + "tok_s": 22361, + "peak_MiB": 5301, + "v_gain": 0.1459, + "loop_stats": [ + { + "tau": 0.1331, + "v": 0.8126, + "H_a": 0.186, + "d_bind": 0.0 + }, + { + "tau": 0.0629, + "v": 0.7958, + "H_a": 0.162, + "d_bind": 0.0235 + }, + { + "tau": 0.0569, + "v": 0.7859, + "H_a": 0.147, + "d_bind": 0.01571 + } + ] + }, + { + "step": 3760, + "tok": 15400960, + "ce": 3.7099, + "ppl": 40.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.917, + "tok_s": 22364, + "peak_MiB": 5301, + "v_gain": 0.1455, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.8904, + "H_a": 0.186, + "d_bind": 0.0 + }, + { + "tau": 0.0432, + "v": 0.8726, + "H_a": 0.157, + "d_bind": 0.02435 + }, + { + "tau": 0.038, + "v": 0.8638, + "H_a": 0.146, + "d_bind": 0.01587 + } + ] + }, + { + "step": 3780, + "tok": 15482880, + "ce": 3.5182, + "ppl": 33.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.93, + "tok_s": 22368, + "peak_MiB": 5301, + "v_gain": 0.1457, + "loop_stats": [ + { + "tau": 0.1154, + "v": 0.8911, + "H_a": 0.171, + "d_bind": 0.0 + }, + { + "tau": 0.0517, + "v": 0.8664, + "H_a": 0.153, + "d_bind": 0.02278 + }, + { + "tau": 0.048, + "v": 0.8575, + "H_a": 0.151, + "d_bind": 0.01717 + } + ] + }, + { + "step": 3800, + "tok": 15564800, + "ce": 3.5775, + "ppl": 35.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.925, + "tok_s": 22384, + "peak_MiB": 5301, + "v_gain": 0.1468, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.8697, + "H_a": 0.17, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8404, + "H_a": 0.152, + "d_bind": 0.02197 + }, + { + "tau": 0.0393, + "v": 0.8315, + "H_a": 0.151, + "d_bind": 0.01564 + } + ] + }, + { + "step": 3820, + "tok": 15646720, + "ce": 3.4163, + "ppl": 30.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.922, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1459, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.8632, + "H_a": 0.159, + "d_bind": 0.0 + }, + { + "tau": 0.0519, + "v": 0.8362, + "H_a": 0.162, + "d_bind": 0.02292 + }, + { + "tau": 0.0488, + "v": 0.8208, + "H_a": 0.154, + "d_bind": 0.01699 + } + ] + }, + { + "step": 3840, + "tok": 15728640, + "ce": 3.3272, + "ppl": 27.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.92, + "tok_s": 22354, + "peak_MiB": 5301, + "v_gain": 0.1452, + "loop_stats": [ + { + "tau": 0.0935, + "v": 0.8747, + "H_a": 0.17, + "d_bind": 0.0 + }, + { + "tau": 0.0307, + "v": 0.8584, + "H_a": 0.167, + "d_bind": 0.02558 + }, + { + "tau": 0.0315, + "v": 0.8519, + "H_a": 0.16, + "d_bind": 0.01749 + } + ] + }, + { + "step": 3860, + "tok": 15810560, + "ce": 3.4465, + "ppl": 31.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22356, + "peak_MiB": 5301, + "v_gain": 0.1493, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.8811, + "H_a": 0.18, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8616, + "H_a": 0.156, + "d_bind": 0.02446 + }, + { + "tau": 0.0354, + "v": 0.8516, + "H_a": 0.15, + "d_bind": 0.01747 + } + ] + }, + { + "step": 3880, + "tok": 15892480, + "ce": 3.7427, + "ppl": 42.2, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.94, + "tok_s": 22359, + "peak_MiB": 5301, + "v_gain": 0.1509, + "loop_stats": [ + { + "tau": 0.0602, + "v": 0.8757, + "H_a": 0.204, + "d_bind": 0.0 + }, + { + "tau": 0.0172, + "v": 0.8547, + "H_a": 0.172, + "d_bind": 0.02901 + }, + { + "tau": 0.0156, + "v": 0.8453, + "H_a": 0.154, + "d_bind": 0.01778 + } + ] + }, + { + "step": 3900, + "tok": 15974400, + "ce": 3.0444, + "ppl": 21.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.925, + "tok_s": 22362, + "peak_MiB": 5301, + "v_gain": 0.1505, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.8455, + "H_a": 0.178, + "d_bind": 0.0 + }, + { + "tau": 0.058, + "v": 0.8306, + "H_a": 0.168, + "d_bind": 0.02353 + }, + { + "tau": 0.0567, + "v": 0.8259, + "H_a": 0.154, + "d_bind": 0.01679 + } + ] + }, + { + "step": 3920, + "tok": 16056320, + "ce": 3.2966, + "ppl": 27.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1543, + "loop_stats": [ + { + "tau": 0.1154, + "v": 0.877, + "H_a": 0.189, + "d_bind": 0.0 + }, + { + "tau": 0.0528, + "v": 0.8675, + "H_a": 0.168, + "d_bind": 0.02526 + }, + { + "tau": 0.0515, + "v": 0.8654, + "H_a": 0.155, + "d_bind": 0.01842 + } + ] + }, + { + "step": 3940, + "tok": 16138240, + "ce": 3.3992, + "ppl": 29.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.922, + "tok_s": 22371, + "peak_MiB": 5301, + "v_gain": 0.1551, + "loop_stats": [ + { + "tau": 0.1008, + "v": 0.8849, + "H_a": 0.174, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.8718, + "H_a": 0.158, + "d_bind": 0.02385 + }, + { + "tau": 0.0411, + "v": 0.8703, + "H_a": 0.149, + "d_bind": 0.0169 + } + ] + }, + { + "step": 3960, + "tok": 16220160, + "ce": 3.5818, + "ppl": 35.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22349, + "peak_MiB": 5301, + "v_gain": 0.1577, + "loop_stats": [ + { + "tau": 0.1051, + "v": 0.8602, + "H_a": 0.18, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.8448, + "H_a": 0.158, + "d_bind": 0.02575 + }, + { + "tau": 0.0351, + "v": 0.8447, + "H_a": 0.149, + "d_bind": 0.01719 + } + ] + }, + { + "step": 3980, + "tok": 16302080, + "ce": 3.3937, + "ppl": 29.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22349, + "peak_MiB": 5301, + "v_gain": 0.1608, + "loop_stats": [ + { + "tau": 0.1147, + "v": 0.8823, + "H_a": 0.212, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8665, + "H_a": 0.175, + "d_bind": 0.0273 + }, + { + "tau": 0.0448, + "v": 0.8553, + "H_a": 0.171, + "d_bind": 0.0184 + } + ] + }, + { + "step": 4000, + "tok": 16384000, + "ce": 3.2623, + "ppl": 26.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.947, + "tok_s": 22353, + "peak_MiB": 5301, + "v_gain": 0.1615, + "loop_stats": [ + { + "tau": 0.136, + "v": 0.8659, + "H_a": 0.205, + "d_bind": 0.0 + }, + { + "tau": 0.046, + "v": 0.8545, + "H_a": 0.158, + "d_bind": 0.02643 + }, + { + "tau": 0.0451, + "v": 0.8503, + "H_a": 0.152, + "d_bind": 0.01647 + } + ] + }, + { + "step": 4020, + "tok": 16465920, + "ce": 3.3637, + "ppl": 28.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.938, + "tok_s": 22356, + "peak_MiB": 5301, + "v_gain": 0.1599, + "loop_stats": [ + { + "tau": 0.0947, + "v": 0.9029, + "H_a": 0.21, + "d_bind": 0.0 + }, + { + "tau": 0.0234, + "v": 0.8861, + "H_a": 0.175, + "d_bind": 0.02798 + }, + { + "tau": 0.0281, + "v": 0.879, + "H_a": 0.165, + "d_bind": 0.01878 + } + ] + }, + { + "step": 4040, + "tok": 16547840, + "ce": 2.9547, + "ppl": 19.2, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.927, + "tok_s": 22372, + "peak_MiB": 5301, + "v_gain": 0.1576, + "loop_stats": [ + { + "tau": 0.1375, + "v": 0.8597, + "H_a": 0.172, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8331, + "H_a": 0.161, + "d_bind": 0.02559 + }, + { + "tau": 0.0407, + "v": 0.821, + "H_a": 0.145, + "d_bind": 0.01618 + } + ] + }, + { + "step": 4060, + "tok": 16629760, + "ce": 3.4294, + "ppl": 30.9, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.917, + "tok_s": 22366, + "peak_MiB": 5301, + "v_gain": 0.1572, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.9008, + "H_a": 0.212, + "d_bind": 0.0 + }, + { + "tau": 0.0434, + "v": 0.8685, + "H_a": 0.166, + "d_bind": 0.02572 + }, + { + "tau": 0.0467, + "v": 0.851, + "H_a": 0.151, + "d_bind": 0.016 + } + ] + }, + { + "step": 4080, + "tok": 16711680, + "ce": 3.2857, + "ppl": 26.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22345, + "peak_MiB": 5301, + "v_gain": 0.1584, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8793, + "H_a": 0.194, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8542, + "H_a": 0.157, + "d_bind": 0.02556 + }, + { + "tau": 0.0447, + "v": 0.8436, + "H_a": 0.132, + "d_bind": 0.01627 + } + ] + }, + { + "step": 4100, + "tok": 16793600, + "ce": 3.928, + "ppl": 50.8, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.938, + "tok_s": 22348, + "peak_MiB": 5301, + "v_gain": 0.1575, + "loop_stats": [ + { + "tau": 0.0786, + "v": 0.9123, + "H_a": 0.189, + "d_bind": 0.0 + }, + { + "tau": 0.0282, + "v": 0.8895, + "H_a": 0.17, + "d_bind": 0.03012 + }, + { + "tau": 0.0328, + "v": 0.8779, + "H_a": 0.152, + "d_bind": 0.01861 + } + ] + }, + { + "step": 4120, + "tok": 16875520, + "ce": 3.3861, + "ppl": 29.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.93, + "tok_s": 22351, + "peak_MiB": 5301, + "v_gain": 0.1546, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.8909, + "H_a": 0.176, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.8771, + "H_a": 0.158, + "d_bind": 0.02645 + }, + { + "tau": 0.0534, + "v": 0.8723, + "H_a": 0.148, + "d_bind": 0.0184 + } + ] + }, + { + "step": 4140, + "tok": 16957440, + "ce": 2.9938, + "ppl": 20.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.92, + "tok_s": 22354, + "peak_MiB": 5301, + "v_gain": 0.1571, + "loop_stats": [ + { + "tau": 0.1426, + "v": 0.8438, + "H_a": 0.175, + "d_bind": 0.0 + }, + { + "tau": 0.0545, + "v": 0.8155, + "H_a": 0.16, + "d_bind": 0.02547 + }, + { + "tau": 0.0585, + "v": 0.8002, + "H_a": 0.144, + "d_bind": 0.0175 + } + ] + }, + { + "step": 4160, + "tok": 17039360, + "ce": 3.5841, + "ppl": 36.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.912, + "tok_s": 22370, + "peak_MiB": 5301, + "v_gain": 0.158, + "loop_stats": [ + { + "tau": 0.0771, + "v": 0.8847, + "H_a": 0.188, + "d_bind": 0.0 + }, + { + "tau": 0.0318, + "v": 0.8648, + "H_a": 0.163, + "d_bind": 0.0279 + }, + { + "tau": 0.0341, + "v": 0.8575, + "H_a": 0.157, + "d_bind": 0.01997 + } + ] + }, + { + "step": 4180, + "tok": 17121280, + "ce": 3.8246, + "ppl": 45.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22364, + "peak_MiB": 5301, + "v_gain": 0.1576, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.8827, + "H_a": 0.183, + "d_bind": 0.0 + }, + { + "tau": 0.0261, + "v": 0.8587, + "H_a": 0.152, + "d_bind": 0.02688 + }, + { + "tau": 0.0287, + "v": 0.8458, + "H_a": 0.149, + "d_bind": 0.01788 + } + ] + }, + { + "step": 4200, + "tok": 17203200, + "ce": 3.3601, + "ppl": 28.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22344, + "peak_MiB": 5301, + "v_gain": 0.1572, + "loop_stats": [ + { + "tau": 0.0938, + "v": 0.8571, + "H_a": 0.174, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8357, + "H_a": 0.152, + "d_bind": 0.02739 + }, + { + "tau": 0.0371, + "v": 0.8291, + "H_a": 0.134, + "d_bind": 0.01603 + } + ] + }, + { + "step": 4220, + "tok": 17285120, + "ce": 3.3422, + "ppl": 28.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22347, + "peak_MiB": 5301, + "v_gain": 0.1565, + "loop_stats": [ + { + "tau": 0.1275, + "v": 0.8712, + "H_a": 0.172, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8561, + "H_a": 0.159, + "d_bind": 0.02712 + }, + { + "tau": 0.0507, + "v": 0.8485, + "H_a": 0.16, + "d_bind": 0.01966 + } + ] + }, + { + "step": 4240, + "tok": 17367040, + "ce": 3.1978, + "ppl": 24.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.93, + "tok_s": 22351, + "peak_MiB": 5301, + "v_gain": 0.1551, + "loop_stats": [ + { + "tau": 0.1191, + "v": 0.856, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0537, + "v": 0.8527, + "H_a": 0.143, + "d_bind": 0.0237 + }, + { + "tau": 0.0556, + "v": 0.8517, + "H_a": 0.137, + "d_bind": 0.0159 + } + ] + }, + { + "step": 4260, + "tok": 17448960, + "ce": 3.7242, + "ppl": 41.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.912, + "tok_s": 22355, + "peak_MiB": 5301, + "v_gain": 0.1529, + "loop_stats": [ + { + "tau": 0.0972, + "v": 0.8458, + "H_a": 0.179, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.827, + "H_a": 0.16, + "d_bind": 0.02622 + }, + { + "tau": 0.0248, + "v": 0.8244, + "H_a": 0.157, + "d_bind": 0.01958 + } + ] + }, + { + "step": 4280, + "tok": 17530880, + "ce": 3.0129, + "ppl": 20.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22370, + "peak_MiB": 5301, + "v_gain": 0.1554, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8152, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0381, + "v": 0.8135, + "H_a": 0.126, + "d_bind": 0.02214 + }, + { + "tau": 0.0381, + "v": 0.814, + "H_a": 0.114, + "d_bind": 0.01384 + } + ] + }, + { + "step": 4300, + "tok": 17612800, + "ce": 3.4754, + "ppl": 32.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.925, + "tok_s": 22365, + "peak_MiB": 5301, + "v_gain": 0.156, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.8613, + "H_a": 0.187, + "d_bind": 0.0 + }, + { + "tau": 0.0453, + "v": 0.8544, + "H_a": 0.152, + "d_bind": 0.02691 + }, + { + "tau": 0.0443, + "v": 0.8508, + "H_a": 0.144, + "d_bind": 0.01834 + } + ] + }, + { + "step": 4320, + "tok": 17694720, + "ce": 2.8725, + "ppl": 17.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.93, + "tok_s": 22345, + "peak_MiB": 5301, + "v_gain": 0.1547, + "loop_stats": [ + { + "tau": 0.1337, + "v": 0.8359, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0723, + "v": 0.838, + "H_a": 0.146, + "d_bind": 0.02473 + }, + { + "tau": 0.0739, + "v": 0.8428, + "H_a": 0.139, + "d_bind": 0.0179 + } + ] + }, + { + "step": 4340, + "tok": 17776640, + "ce": 3.5421, + "ppl": 34.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.925, + "tok_s": 22348, + "peak_MiB": 5301, + "v_gain": 0.1544, + "loop_stats": [ + { + "tau": 0.0937, + "v": 0.8777, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8663, + "H_a": 0.148, + "d_bind": 0.02482 + }, + { + "tau": 0.0359, + "v": 0.8622, + "H_a": 0.145, + "d_bind": 0.01762 + } + ] + }, + { + "step": 4360, + "tok": 17858560, + "ce": 3.5102, + "ppl": 33.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.922, + "tok_s": 22352, + "peak_MiB": 5301, + "v_gain": 0.1522, + "loop_stats": [ + { + "tau": 0.0941, + "v": 0.8828, + "H_a": 0.163, + "d_bind": 0.0 + }, + { + "tau": 0.0379, + "v": 0.8794, + "H_a": 0.157, + "d_bind": 0.02584 + }, + { + "tau": 0.0393, + "v": 0.8771, + "H_a": 0.143, + "d_bind": 0.01879 + } + ] + }, + { + "step": 4380, + "tok": 17940480, + "ce": 3.8714, + "ppl": 48.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.93, + "tok_s": 22355, + "peak_MiB": 5301, + "v_gain": 0.1528, + "loop_stats": [ + { + "tau": 0.0746, + "v": 0.8996, + "H_a": 0.163, + "d_bind": 0.0 + }, + { + "tau": 0.024, + "v": 0.8849, + "H_a": 0.154, + "d_bind": 0.02528 + }, + { + "tau": 0.0248, + "v": 0.8773, + "H_a": 0.143, + "d_bind": 0.01788 + } + ] + }, + { + "step": 4400, + "tok": 18022400, + "ce": 3.082, + "ppl": 21.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.925, + "tok_s": 22371, + "peak_MiB": 5301, + "v_gain": 0.1569, + "loop_stats": [ + { + "tau": 0.1235, + "v": 0.8752, + "H_a": 0.143, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8666, + "H_a": 0.131, + "d_bind": 0.02504 + }, + { + "tau": 0.0476, + "v": 0.8666, + "H_a": 0.12, + "d_bind": 0.01644 + } + ] + }, + { + "step": 4420, + "tok": 18104320, + "ce": 3.3671, + "ppl": 29.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.917, + "tok_s": 22366, + "peak_MiB": 5301, + "v_gain": 0.159, + "loop_stats": [ + { + "tau": 0.1264, + "v": 0.8864, + "H_a": 0.144, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.8697, + "H_a": 0.134, + "d_bind": 0.02393 + }, + { + "tau": 0.0403, + "v": 0.8637, + "H_a": 0.131, + "d_bind": 0.01752 + } + ] + }, + { + "step": 4440, + "tok": 18186240, + "ce": 3.2646, + "ppl": 26.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22347, + "peak_MiB": 5301, + "v_gain": 0.1576, + "loop_stats": [ + { + "tau": 0.1391, + "v": 0.8412, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0569, + "v": 0.8191, + "H_a": 0.142, + "d_bind": 0.02264 + }, + { + "tau": 0.0504, + "v": 0.8139, + "H_a": 0.136, + "d_bind": 0.01583 + } + ] + }, + { + "step": 4460, + "tok": 18268160, + "ce": 3.1309, + "ppl": 22.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.922, + "tok_s": 22350, + "peak_MiB": 5301, + "v_gain": 0.159, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.8591, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0623, + "v": 0.8652, + "H_a": 0.141, + "d_bind": 0.02585 + }, + { + "tau": 0.0589, + "v": 0.8697, + "H_a": 0.128, + "d_bind": 0.01733 + } + ] + }, + { + "step": 4480, + "tok": 18350080, + "ce": 3.9534, + "ppl": 52.1, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.938, + "tok_s": 22354, + "peak_MiB": 5301, + "v_gain": 0.1622, + "loop_stats": [ + { + "tau": 0.0757, + "v": 0.8468, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0304, + "v": 0.8267, + "H_a": 0.135, + "d_bind": 0.02369 + }, + { + "tau": 0.0305, + "v": 0.818, + "H_a": 0.129, + "d_bind": 0.01596 + } + ] + }, + { + "step": 4500, + "tok": 18432000, + "ce": 3.1643, + "ppl": 23.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.91, + "tok_s": 22358, + "peak_MiB": 5301, + "v_gain": 0.1654, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.8563, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0487, + "v": 0.8411, + "H_a": 0.13, + "d_bind": 0.02319 + }, + { + "tau": 0.0514, + "v": 0.8366, + "H_a": 0.125, + "d_bind": 0.01615 + } + ] + }, + { + "step": 4520, + "tok": 18513920, + "ce": 3.6253, + "ppl": 37.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.938, + "tok_s": 22373, + "peak_MiB": 5301, + "v_gain": 0.1686, + "loop_stats": [ + { + "tau": 0.0892, + "v": 0.8836, + "H_a": 0.183, + "d_bind": 0.0 + }, + { + "tau": 0.0218, + "v": 0.8771, + "H_a": 0.162, + "d_bind": 0.0272 + }, + { + "tau": 0.0261, + "v": 0.8761, + "H_a": 0.144, + "d_bind": 0.01737 + } + ] + }, + { + "step": 4540, + "tok": 18595840, + "ce": 3.6651, + "ppl": 39.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22369, + "peak_MiB": 5301, + "v_gain": 0.1701, + "loop_stats": [ + { + "tau": 0.0813, + "v": 0.907, + "H_a": 0.182, + "d_bind": 0.0 + }, + { + "tau": 0.0235, + "v": 0.898, + "H_a": 0.163, + "d_bind": 0.02646 + }, + { + "tau": 0.0188, + "v": 0.8991, + "H_a": 0.145, + "d_bind": 0.01782 + } + ] + }, + { + "step": 4560, + "tok": 18677760, + "ce": 3.598, + "ppl": 36.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.927, + "tok_s": 22352, + "peak_MiB": 5301, + "v_gain": 0.1716, + "loop_stats": [ + { + "tau": 0.0862, + "v": 0.863, + "H_a": 0.168, + "d_bind": 0.0 + }, + { + "tau": 0.0244, + "v": 0.8467, + "H_a": 0.15, + "d_bind": 0.02562 + }, + { + "tau": 0.0228, + "v": 0.8458, + "H_a": 0.133, + "d_bind": 0.01592 + } + ] + }, + { + "step": 4580, + "tok": 18759680, + "ce": 3.7646, + "ppl": 43.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.927, + "tok_s": 22356, + "peak_MiB": 5301, + "v_gain": 0.1721, + "loop_stats": [ + { + "tau": 0.0943, + "v": 0.8757, + "H_a": 0.168, + "d_bind": 0.0 + }, + { + "tau": 0.0232, + "v": 0.8504, + "H_a": 0.145, + "d_bind": 0.02548 + }, + { + "tau": 0.0246, + "v": 0.8401, + "H_a": 0.134, + "d_bind": 0.01525 + } + ] + }, + { + "step": 4600, + "tok": 18841600, + "ce": 3.5057, + "ppl": 33.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.927, + "tok_s": 22359, + "peak_MiB": 5301, + "v_gain": 0.1728, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.8829, + "H_a": 0.139, + "d_bind": 0.0 + }, + { + "tau": 0.0257, + "v": 0.8527, + "H_a": 0.135, + "d_bind": 0.02541 + }, + { + "tau": 0.0261, + "v": 0.8338, + "H_a": 0.122, + "d_bind": 0.01677 + } + ] + }, + { + "step": 4620, + "tok": 18923520, + "ce": 2.9291, + "ppl": 18.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.925, + "tok_s": 22363, + "peak_MiB": 5301, + "v_gain": 0.1732, + "loop_stats": [ + { + "tau": 0.1332, + "v": 0.8788, + "H_a": 0.129, + "d_bind": 0.0 + }, + { + "tau": 0.0366, + "v": 0.8757, + "H_a": 0.122, + "d_bind": 0.02092 + }, + { + "tau": 0.0396, + "v": 0.8766, + "H_a": 0.107, + "d_bind": 0.01416 + } + ] + }, + { + "step": 4640, + "tok": 19005440, + "ce": 2.8132, + "ppl": 16.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.938, + "tok_s": 22378, + "peak_MiB": 5301, + "v_gain": 0.1728, + "loop_stats": [ + { + "tau": 0.1371, + "v": 0.8589, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0498, + "v": 0.8359, + "H_a": 0.114, + "d_bind": 0.02146 + }, + { + "tau": 0.0476, + "v": 0.831, + "H_a": 0.111, + "d_bind": 0.01437 + } + ] + }, + { + "step": 4660, + "tok": 19087360, + "ce": 3.3017, + "ppl": 27.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22374, + "peak_MiB": 5301, + "v_gain": 0.1723, + "loop_stats": [ + { + "tau": 0.1007, + "v": 0.8758, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.8525, + "H_a": 0.138, + "d_bind": 0.02603 + }, + { + "tau": 0.036, + "v": 0.8442, + "H_a": 0.123, + "d_bind": 0.01688 + } + ] + }, + { + "step": 4680, + "tok": 19169280, + "ce": 3.1434, + "ppl": 23.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22357, + "peak_MiB": 5301, + "v_gain": 0.1707, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.8912, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.868, + "H_a": 0.121, + "d_bind": 0.02231 + }, + { + "tau": 0.0428, + "v": 0.8556, + "H_a": 0.115, + "d_bind": 0.01522 + } + ] + }, + { + "step": 4700, + "tok": 19251200, + "ce": 3.6053, + "ppl": 36.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22361, + "peak_MiB": 5301, + "v_gain": 0.1709, + "loop_stats": [ + { + "tau": 0.0988, + "v": 0.8879, + "H_a": 0.139, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8647, + "H_a": 0.133, + "d_bind": 0.02511 + }, + { + "tau": 0.0287, + "v": 0.8542, + "H_a": 0.129, + "d_bind": 0.01697 + } + ] + }, + { + "step": 4720, + "tok": 19333120, + "ce": 3.9814, + "ppl": 53.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22365, + "peak_MiB": 5301, + "v_gain": 0.1684, + "loop_stats": [ + { + "tau": 0.0863, + "v": 0.8819, + "H_a": 0.158, + "d_bind": 0.0 + }, + { + "tau": 0.0317, + "v": 0.8615, + "H_a": 0.151, + "d_bind": 0.02595 + }, + { + "tau": 0.0298, + "v": 0.8539, + "H_a": 0.14, + "d_bind": 0.0171 + } + ] + }, + { + "step": 4740, + "tok": 19415040, + "ce": 3.0863, + "ppl": 21.9, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22369, + "peak_MiB": 5301, + "v_gain": 0.1712, + "loop_stats": [ + { + "tau": 0.118, + "v": 0.88, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.854, + "H_a": 0.126, + "d_bind": 0.02634 + }, + { + "tau": 0.0432, + "v": 0.838, + "H_a": 0.109, + "d_bind": 0.01513 + } + ] + }, + { + "step": 4760, + "tok": 19496960, + "ce": 3.2066, + "ppl": 24.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.912, + "tok_s": 22383, + "peak_MiB": 5301, + "v_gain": 0.1722, + "loop_stats": [ + { + "tau": 0.1372, + "v": 0.8742, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0609, + "v": 0.858, + "H_a": 0.136, + "d_bind": 0.02571 + }, + { + "tau": 0.0625, + "v": 0.8516, + "H_a": 0.121, + "d_bind": 0.01585 + } + ] + }, + { + "step": 4780, + "tok": 19578880, + "ce": 3.8853, + "ppl": 48.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22379, + "peak_MiB": 5301, + "v_gain": 0.1718, + "loop_stats": [ + { + "tau": 0.0603, + "v": 0.9054, + "H_a": 0.164, + "d_bind": 0.0 + }, + { + "tau": 0.0167, + "v": 0.8853, + "H_a": 0.148, + "d_bind": 0.02565 + }, + { + "tau": 0.0185, + "v": 0.8781, + "H_a": 0.14, + "d_bind": 0.01732 + } + ] + }, + { + "step": 4800, + "tok": 19660800, + "ce": 3.0109, + "ppl": 20.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22362, + "peak_MiB": 5301, + "v_gain": 0.1716, + "loop_stats": [ + { + "tau": 0.1183, + "v": 0.8804, + "H_a": 0.143, + "d_bind": 0.0 + }, + { + "tau": 0.046, + "v": 0.8576, + "H_a": 0.13, + "d_bind": 0.02381 + }, + { + "tau": 0.0484, + "v": 0.8487, + "H_a": 0.117, + "d_bind": 0.01506 + } + ] + }, + { + "step": 4820, + "tok": 19742720, + "ce": 3.6846, + "ppl": 39.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.927, + "tok_s": 22365, + "peak_MiB": 5301, + "v_gain": 0.1708, + "loop_stats": [ + { + "tau": 0.056, + "v": 0.9173, + "H_a": 0.171, + "d_bind": 0.0 + }, + { + "tau": 0.0221, + "v": 0.8969, + "H_a": 0.161, + "d_bind": 0.03088 + }, + { + "tau": 0.0228, + "v": 0.8901, + "H_a": 0.144, + "d_bind": 0.01832 + } + ] + }, + { + "step": 4840, + "tok": 19824640, + "ce": 3.4336, + "ppl": 31.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22369, + "peak_MiB": 5301, + "v_gain": 0.1743, + "loop_stats": [ + { + "tau": 0.0945, + "v": 0.9149, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0291, + "v": 0.9042, + "H_a": 0.146, + "d_bind": 0.02578 + }, + { + "tau": 0.0312, + "v": 0.9013, + "H_a": 0.126, + "d_bind": 0.01645 + } + ] + }, + { + "step": 4860, + "tok": 19906560, + "ce": 3.443, + "ppl": 31.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22374, + "peak_MiB": 5301, + "v_gain": 0.1746, + "loop_stats": [ + { + "tau": 0.0991, + "v": 0.9035, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.889, + "H_a": 0.142, + "d_bind": 0.02616 + }, + { + "tau": 0.0374, + "v": 0.8866, + "H_a": 0.125, + "d_bind": 0.0152 + } + ] + }, + { + "step": 4880, + "tok": 19988480, + "ce": 3.5756, + "ppl": 35.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.922, + "tok_s": 22388, + "peak_MiB": 5301, + "v_gain": 0.1754, + "loop_stats": [ + { + "tau": 0.1004, + "v": 0.9053, + "H_a": 0.171, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.8992, + "H_a": 0.159, + "d_bind": 0.02906 + }, + { + "tau": 0.0233, + "v": 0.903, + "H_a": 0.143, + "d_bind": 0.01831 + } + ] + }, + { + "step": 4900, + "tok": 20070400, + "ce": 3.0962, + "ppl": 22.1, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.935, + "tok_s": 22384, + "peak_MiB": 5301, + "v_gain": 0.1799, + "loop_stats": [ + { + "tau": 0.1244, + "v": 0.9218, + "H_a": 0.161, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.9047, + "H_a": 0.132, + "d_bind": 0.02602 + }, + { + "tau": 0.0403, + "v": 0.8974, + "H_a": 0.123, + "d_bind": 0.015 + } + ] + }, + { + "step": 4920, + "tok": 20152320, + "ce": 3.4128, + "ppl": 30.3, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.915, + "tok_s": 22367, + "peak_MiB": 5301, + "v_gain": 0.1784, + "loop_stats": [ + { + "tau": 0.1349, + "v": 0.8766, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8443, + "H_a": 0.129, + "d_bind": 0.02184 + }, + { + "tau": 0.0462, + "v": 0.8369, + "H_a": 0.132, + "d_bind": 0.01775 + } + ] + }, + { + "step": 4940, + "tok": 20234240, + "ce": 3.5532, + "ppl": 34.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22371, + "peak_MiB": 5301, + "v_gain": 0.1777, + "loop_stats": [ + { + "tau": 0.1023, + "v": 0.9134, + "H_a": 0.166, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.8864, + "H_a": 0.154, + "d_bind": 0.02886 + }, + { + "tau": 0.0234, + "v": 0.8803, + "H_a": 0.143, + "d_bind": 0.02144 + } + ] + }, + { + "step": 4960, + "tok": 20316160, + "ce": 3.8498, + "ppl": 47.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.907, + "tok_s": 22375, + "peak_MiB": 5301, + "v_gain": 0.1801, + "loop_stats": [ + { + "tau": 0.0893, + "v": 0.9133, + "H_a": 0.173, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.8912, + "H_a": 0.156, + "d_bind": 0.02887 + }, + { + "tau": 0.0332, + "v": 0.8876, + "H_a": 0.142, + "d_bind": 0.02019 + } + ] + }, + { + "step": 4980, + "tok": 20398080, + "ce": 2.9544, + "ppl": 19.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22379, + "peak_MiB": 5301, + "v_gain": 0.1834, + "loop_stats": [ + { + "tau": 0.1325, + "v": 0.9177, + "H_a": 0.139, + "d_bind": 0.0 + }, + { + "tau": 0.0445, + "v": 0.893, + "H_a": 0.133, + "d_bind": 0.02472 + }, + { + "tau": 0.0445, + "v": 0.8884, + "H_a": 0.112, + "d_bind": 0.01596 + } + ] + }, + { + "step": 5000, + "tok": 20480000, + "ce": 3.2149, + "ppl": 24.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.91, + "tok_s": 22393, + "peak_MiB": 5301, + "v_gain": 0.1822, + "loop_stats": [ + { + "tau": 0.1389, + "v": 0.9145, + "H_a": 0.143, + "d_bind": 0.0 + }, + { + "tau": 0.0545, + "v": 0.8875, + "H_a": 0.132, + "d_bind": 0.02535 + }, + { + "tau": 0.0554, + "v": 0.8854, + "H_a": 0.119, + "d_bind": 0.01592 + } + ] + }, + { + "step": 5020, + "tok": 20561920, + "ce": 3.3101, + "ppl": 27.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.917, + "tok_s": 22389, + "peak_MiB": 5301, + "v_gain": 0.1821, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.9038, + "H_a": 0.154, + "d_bind": 0.0 + }, + { + "tau": 0.0303, + "v": 0.8868, + "H_a": 0.14, + "d_bind": 0.02832 + }, + { + "tau": 0.0312, + "v": 0.8844, + "H_a": 0.131, + "d_bind": 0.01783 + } + ] + }, + { + "step": 5040, + "tok": 20643840, + "ce": 3.392, + "ppl": 29.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22373, + "peak_MiB": 5301, + "v_gain": 0.1792, + "loop_stats": [ + { + "tau": 0.1055, + "v": 0.9014, + "H_a": 0.143, + "d_bind": 0.0 + }, + { + "tau": 0.0366, + "v": 0.8898, + "H_a": 0.128, + "d_bind": 0.02436 + }, + { + "tau": 0.0387, + "v": 0.8947, + "H_a": 0.12, + "d_bind": 0.01622 + } + ] + }, + { + "step": 5060, + "tok": 20725760, + "ce": 3.3006, + "ppl": 27.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.907, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1798, + "loop_stats": [ + { + "tau": 0.1066, + "v": 0.8894, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.8718, + "H_a": 0.134, + "d_bind": 0.02448 + }, + { + "tau": 0.0484, + "v": 0.8687, + "H_a": 0.123, + "d_bind": 0.01676 + } + ] + }, + { + "step": 5080, + "tok": 20807680, + "ce": 3.2382, + "ppl": 25.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.93, + "tok_s": 22380, + "peak_MiB": 5301, + "v_gain": 0.1813, + "loop_stats": [ + { + "tau": 0.109, + "v": 0.8993, + "H_a": 0.15, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8871, + "H_a": 0.135, + "d_bind": 0.02525 + }, + { + "tau": 0.0359, + "v": 0.8861, + "H_a": 0.122, + "d_bind": 0.01645 + } + ] + }, + { + "step": 5100, + "tok": 20889600, + "ce": 2.9168, + "ppl": 18.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22384, + "peak_MiB": 5301, + "v_gain": 0.1803, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.8727, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0573, + "v": 0.87, + "H_a": 0.133, + "d_bind": 0.02471 + }, + { + "tau": 0.0569, + "v": 0.8735, + "H_a": 0.111, + "d_bind": 0.01479 + } + ] + }, + { + "step": 5120, + "tok": 20971520, + "ce": 2.9976, + "ppl": 20.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22397, + "peak_MiB": 5301, + "v_gain": 0.1822, + "loop_stats": [ + { + "tau": 0.1145, + "v": 0.8598, + "H_a": 0.147, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8511, + "H_a": 0.135, + "d_bind": 0.02773 + }, + { + "tau": 0.0482, + "v": 0.8516, + "H_a": 0.122, + "d_bind": 0.01697 + } + ] + }, + { + "step": 5140, + "tok": 21053440, + "ce": 3.1282, + "ppl": 22.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.915, + "tok_s": 22393, + "peak_MiB": 5301, + "v_gain": 0.1788, + "loop_stats": [ + { + "tau": 0.1149, + "v": 0.8782, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.873, + "H_a": 0.123, + "d_bind": 0.02373 + }, + { + "tau": 0.0486, + "v": 0.8732, + "H_a": 0.114, + "d_bind": 0.01603 + } + ] + }, + { + "step": 5160, + "tok": 21135360, + "ce": 3.2245, + "ppl": 25.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.922, + "tok_s": 22377, + "peak_MiB": 5301, + "v_gain": 0.1746, + "loop_stats": [ + { + "tau": 0.0997, + "v": 0.8694, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.8603, + "H_a": 0.156, + "d_bind": 0.02878 + }, + { + "tau": 0.0331, + "v": 0.8603, + "H_a": 0.145, + "d_bind": 0.02065 + } + ] + }, + { + "step": 5180, + "tok": 21217280, + "ce": 3.0759, + "ppl": 21.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.933, + "tok_s": 22380, + "peak_MiB": 5301, + "v_gain": 0.1737, + "loop_stats": [ + { + "tau": 0.1165, + "v": 0.8563, + "H_a": 0.163, + "d_bind": 0.0 + }, + { + "tau": 0.0573, + "v": 0.8582, + "H_a": 0.147, + "d_bind": 0.03031 + }, + { + "tau": 0.0586, + "v": 0.8648, + "H_a": 0.13, + "d_bind": 0.01845 + } + ] + }, + { + "step": 5200, + "tok": 21299200, + "ce": 3.6417, + "ppl": 38.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.925, + "tok_s": 22383, + "peak_MiB": 5301, + "v_gain": 0.174, + "loop_stats": [ + { + "tau": 0.096, + "v": 0.8605, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8531, + "H_a": 0.147, + "d_bind": 0.02745 + }, + { + "tau": 0.0339, + "v": 0.8561, + "H_a": 0.125, + "d_bind": 0.01857 + } + ] + }, + { + "step": 5220, + "tok": 21381120, + "ce": 3.6686, + "ppl": 39.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.915, + "tok_s": 22387, + "peak_MiB": 5301, + "v_gain": 0.1768, + "loop_stats": [ + { + "tau": 0.0985, + "v": 0.8771, + "H_a": 0.155, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8779, + "H_a": 0.131, + "d_bind": 0.02634 + }, + { + "tau": 0.0313, + "v": 0.8836, + "H_a": 0.122, + "d_bind": 0.01718 + } + ] + }, + { + "step": 5240, + "tok": 21463040, + "ce": 3.4651, + "ppl": 32.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.925, + "tok_s": 22400, + "peak_MiB": 5301, + "v_gain": 0.1759, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.8768, + "H_a": 0.166, + "d_bind": 0.0 + }, + { + "tau": 0.0496, + "v": 0.8822, + "H_a": 0.15, + "d_bind": 0.02986 + }, + { + "tau": 0.0502, + "v": 0.8872, + "H_a": 0.14, + "d_bind": 0.01933 + } + ] + }, + { + "step": 5260, + "tok": 21544960, + "ce": 3.1957, + "ppl": 24.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22396, + "peak_MiB": 5301, + "v_gain": 0.1772, + "loop_stats": [ + { + "tau": 0.0965, + "v": 0.8844, + "H_a": 0.17, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.884, + "H_a": 0.14, + "d_bind": 0.02881 + }, + { + "tau": 0.0404, + "v": 0.8843, + "H_a": 0.13, + "d_bind": 0.0181 + } + ] + }, + { + "step": 5280, + "tok": 21626880, + "ce": 3.3392, + "ppl": 28.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22381, + "peak_MiB": 5301, + "v_gain": 0.1783, + "loop_stats": [ + { + "tau": 0.1139, + "v": 0.8551, + "H_a": 0.159, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.8299, + "H_a": 0.135, + "d_bind": 0.02935 + }, + { + "tau": 0.0397, + "v": 0.8171, + "H_a": 0.128, + "d_bind": 0.01883 + } + ] + }, + { + "step": 5300, + "tok": 21708800, + "ce": 3.3745, + "ppl": 29.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.927, + "tok_s": 22384, + "peak_MiB": 5301, + "v_gain": 0.1751, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.8893, + "H_a": 0.155, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.8738, + "H_a": 0.139, + "d_bind": 0.0294 + }, + { + "tau": 0.0396, + "v": 0.87, + "H_a": 0.131, + "d_bind": 0.01861 + } + ] + }, + { + "step": 5320, + "tok": 21790720, + "ce": 3.2635, + "ppl": 26.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22387, + "peak_MiB": 5301, + "v_gain": 0.1753, + "loop_stats": [ + { + "tau": 0.1078, + "v": 0.8728, + "H_a": 0.16, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.8653, + "H_a": 0.138, + "d_bind": 0.03183 + }, + { + "tau": 0.046, + "v": 0.8642, + "H_a": 0.13, + "d_bind": 0.01822 + } + ] + }, + { + "step": 5340, + "tok": 21872640, + "ce": 3.605, + "ppl": 36.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.922, + "tok_s": 22391, + "peak_MiB": 5301, + "v_gain": 0.1782, + "loop_stats": [ + { + "tau": 0.0899, + "v": 0.8794, + "H_a": 0.158, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.8758, + "H_a": 0.154, + "d_bind": 0.0333 + }, + { + "tau": 0.0285, + "v": 0.8769, + "H_a": 0.133, + "d_bind": 0.02036 + } + ] + }, + { + "step": 5360, + "tok": 21954560, + "ce": 2.8912, + "ppl": 18.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.935, + "tok_s": 22404, + "peak_MiB": 5301, + "v_gain": 0.1805, + "loop_stats": [ + { + "tau": 0.1473, + "v": 0.8821, + "H_a": 0.149, + "d_bind": 0.0 + }, + { + "tau": 0.0613, + "v": 0.8815, + "H_a": 0.132, + "d_bind": 0.03029 + }, + { + "tau": 0.0621, + "v": 0.8789, + "H_a": 0.115, + "d_bind": 0.01524 + } + ] + }, + { + "step": 5380, + "tok": 22036480, + "ce": 3.6302, + "ppl": 37.7, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.915, + "tok_s": 22400, + "peak_MiB": 5301, + "v_gain": 0.1811, + "loop_stats": [ + { + "tau": 0.0661, + "v": 0.8592, + "H_a": 0.161, + "d_bind": 0.0 + }, + { + "tau": 0.0272, + "v": 0.8431, + "H_a": 0.146, + "d_bind": 0.03546 + }, + { + "tau": 0.0305, + "v": 0.8322, + "H_a": 0.129, + "d_bind": 0.01873 + } + ] + }, + { + "step": 5400, + "tok": 22118400, + "ce": 2.8539, + "ppl": 17.4, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.917, + "tok_s": 22385, + "peak_MiB": 5301, + "v_gain": 0.1795, + "loop_stats": [ + { + "tau": 0.1491, + "v": 0.8553, + "H_a": 0.139, + "d_bind": 0.0 + }, + { + "tau": 0.0625, + "v": 0.8522, + "H_a": 0.127, + "d_bind": 0.02746 + }, + { + "tau": 0.0631, + "v": 0.8531, + "H_a": 0.117, + "d_bind": 0.01687 + } + ] + }, + { + "step": 5420, + "tok": 22200320, + "ce": 3.2992, + "ppl": 27.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.917, + "tok_s": 22388, + "peak_MiB": 5301, + "v_gain": 0.1783, + "loop_stats": [ + { + "tau": 0.135, + "v": 0.8363, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0469, + "v": 0.8235, + "H_a": 0.139, + "d_bind": 0.02774 + }, + { + "tau": 0.0443, + "v": 0.8209, + "H_a": 0.13, + "d_bind": 0.01799 + } + ] + }, + { + "step": 5440, + "tok": 22282240, + "ce": 3.6106, + "ppl": 37.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.92, + "tok_s": 22392, + "peak_MiB": 5301, + "v_gain": 0.1785, + "loop_stats": [ + { + "tau": 0.0842, + "v": 0.8764, + "H_a": 0.165, + "d_bind": 0.0 + }, + { + "tau": 0.0331, + "v": 0.8702, + "H_a": 0.15, + "d_bind": 0.03303 + }, + { + "tau": 0.0306, + "v": 0.8745, + "H_a": 0.131, + "d_bind": 0.01997 + } + ] + }, + { + "step": 5460, + "tok": 22364160, + "ce": 3.3989, + "ppl": 29.9, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.917, + "tok_s": 22395, + "peak_MiB": 5301, + "v_gain": 0.1791, + "loop_stats": [ + { + "tau": 0.0948, + "v": 0.8809, + "H_a": 0.16, + "d_bind": 0.0 + }, + { + "tau": 0.0332, + "v": 0.8768, + "H_a": 0.145, + "d_bind": 0.03202 + }, + { + "tau": 0.0319, + "v": 0.8797, + "H_a": 0.127, + "d_bind": 0.01998 + } + ] + }, + { + "step": 5480, + "tok": 22446080, + "ce": 3.0931, + "ppl": 22.0, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.922, + "tok_s": 22408, + "peak_MiB": 5301, + "v_gain": 0.1751, + "loop_stats": [ + { + "tau": 0.1236, + "v": 0.8416, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0559, + "v": 0.8365, + "H_a": 0.13, + "d_bind": 0.02917 + }, + { + "tau": 0.053, + "v": 0.8386, + "H_a": 0.123, + "d_bind": 0.01977 + } + ] + }, + { + "step": 5500, + "tok": 22528000, + "ce": 3.6196, + "ppl": 37.3, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.922, + "tok_s": 22404, + "peak_MiB": 5301, + "v_gain": 0.1745, + "loop_stats": [ + { + "tau": 0.0873, + "v": 0.8573, + "H_a": 0.145, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8424, + "H_a": 0.152, + "d_bind": 0.03175 + }, + { + "tau": 0.0388, + "v": 0.8452, + "H_a": 0.13, + "d_bind": 0.02061 + } + ] + }, + { + "step": 5520, + "tok": 22609920, + "ce": 3.0918, + "ppl": 22.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.922, + "tok_s": 22389, + "peak_MiB": 5301, + "v_gain": 0.1765, + "loop_stats": [ + { + "tau": 0.1038, + "v": 0.8634, + "H_a": 0.129, + "d_bind": 0.0 + }, + { + "tau": 0.0323, + "v": 0.8642, + "H_a": 0.127, + "d_bind": 0.02658 + }, + { + "tau": 0.0323, + "v": 0.8709, + "H_a": 0.12, + "d_bind": 0.01845 + } + ] + }, + { + "step": 5540, + "tok": 22691840, + "ce": 3.141, + "ppl": 23.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.927, + "tok_s": 22392, + "peak_MiB": 5301, + "v_gain": 0.1791, + "loop_stats": [ + { + "tau": 0.1079, + "v": 0.8787, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8785, + "H_a": 0.137, + "d_bind": 0.02943 + }, + { + "tau": 0.0406, + "v": 0.878, + "H_a": 0.118, + "d_bind": 0.01733 + } + ] + }, + { + "step": 5560, + "tok": 22773760, + "ce": 3.3251, + "ppl": 27.8, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.927, + "tok_s": 22395, + "peak_MiB": 5301, + "v_gain": 0.1764, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.8815, + "H_a": 0.162, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.8643, + "H_a": 0.144, + "d_bind": 0.03142 + }, + { + "tau": 0.0311, + "v": 0.8567, + "H_a": 0.13, + "d_bind": 0.01767 + } + ] + }, + { + "step": 5580, + "tok": 22855680, + "ce": 3.5917, + "ppl": 36.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22399, + "peak_MiB": 5301, + "v_gain": 0.1733, + "loop_stats": [ + { + "tau": 0.085, + "v": 0.8957, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8922, + "H_a": 0.15, + "d_bind": 0.03488 + }, + { + "tau": 0.0326, + "v": 0.8918, + "H_a": 0.126, + "d_bind": 0.0196 + } + ] + }, + { + "step": 5600, + "tok": 22937600, + "ce": 3.022, + "ppl": 20.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22411, + "peak_MiB": 5301, + "v_gain": 0.1755, + "loop_stats": [ + { + "tau": 0.0914, + "v": 0.8867, + "H_a": 0.149, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.881, + "H_a": 0.144, + "d_bind": 0.03372 + }, + { + "tau": 0.0375, + "v": 0.8817, + "H_a": 0.123, + "d_bind": 0.01945 + } + ] + }, + { + "step": 5620, + "tok": 23019520, + "ce": 3.2256, + "ppl": 25.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1729, + "loop_stats": [ + { + "tau": 0.1044, + "v": 0.8897, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.8824, + "H_a": 0.144, + "d_bind": 0.0342 + }, + { + "tau": 0.0461, + "v": 0.8804, + "H_a": 0.118, + "d_bind": 0.01919 + } + ] + }, + { + "step": 5640, + "tok": 23101440, + "ce": 3.739, + "ppl": 42.1, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.93, + "tok_s": 22393, + "peak_MiB": 5301, + "v_gain": 0.1737, + "loop_stats": [ + { + "tau": 0.0739, + "v": 0.9198, + "H_a": 0.155, + "d_bind": 0.0 + }, + { + "tau": 0.0254, + "v": 0.912, + "H_a": 0.153, + "d_bind": 0.03327 + }, + { + "tau": 0.0226, + "v": 0.9139, + "H_a": 0.138, + "d_bind": 0.01951 + } + ] + }, + { + "step": 5660, + "tok": 23183360, + "ce": 3.3508, + "ppl": 28.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.915, + "tok_s": 22396, + "peak_MiB": 5301, + "v_gain": 0.1746, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.9289, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.9219, + "H_a": 0.137, + "d_bind": 0.03496 + }, + { + "tau": 0.0465, + "v": 0.917, + "H_a": 0.123, + "d_bind": 0.01789 + } + ] + }, + { + "step": 5680, + "tok": 23265280, + "ce": 3.1255, + "ppl": 22.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.925, + "tok_s": 22400, + "peak_MiB": 5301, + "v_gain": 0.1761, + "loop_stats": [ + { + "tau": 0.1072, + "v": 0.9174, + "H_a": 0.154, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.9043, + "H_a": 0.141, + "d_bind": 0.03885 + }, + { + "tau": 0.0571, + "v": 0.893, + "H_a": 0.116, + "d_bind": 0.01688 + } + ] + }, + { + "step": 5700, + "tok": 23347200, + "ce": 3.3476, + "ppl": 28.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.917, + "tok_s": 22403, + "peak_MiB": 5301, + "v_gain": 0.1742, + "loop_stats": [ + { + "tau": 0.1044, + "v": 0.9207, + "H_a": 0.151, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.9036, + "H_a": 0.138, + "d_bind": 0.03219 + }, + { + "tau": 0.0491, + "v": 0.8933, + "H_a": 0.125, + "d_bind": 0.01799 + } + ] + }, + { + "step": 5720, + "tok": 23429120, + "ce": 3.4816, + "ppl": 32.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.927, + "tok_s": 22415, + "peak_MiB": 5301, + "v_gain": 0.174, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8981, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8725, + "H_a": 0.129, + "d_bind": 0.03455 + }, + { + "tau": 0.0475, + "v": 0.8621, + "H_a": 0.107, + "d_bind": 0.01697 + } + ] + }, + { + "step": 5740, + "tok": 23511040, + "ce": 3.4012, + "ppl": 30.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22411, + "peak_MiB": 5301, + "v_gain": 0.1725, + "loop_stats": [ + { + "tau": 0.098, + "v": 0.9026, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8822, + "H_a": 0.133, + "d_bind": 0.03036 + }, + { + "tau": 0.0377, + "v": 0.8712, + "H_a": 0.119, + "d_bind": 0.0171 + } + ] + }, + { + "step": 5760, + "tok": 23592960, + "ce": 3.0374, + "ppl": 20.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22397, + "peak_MiB": 5301, + "v_gain": 0.1757, + "loop_stats": [ + { + "tau": 0.1135, + "v": 0.9126, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0513, + "v": 0.9147, + "H_a": 0.11, + "d_bind": 0.02613 + }, + { + "tau": 0.0551, + "v": 0.915, + "H_a": 0.092, + "d_bind": 0.01379 + } + ] + }, + { + "step": 5780, + "tok": 23674880, + "ce": 2.7141, + "ppl": 15.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22400, + "peak_MiB": 5301, + "v_gain": 0.1754, + "loop_stats": [ + { + "tau": 0.1371, + "v": 0.9017, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0587, + "v": 0.8989, + "H_a": 0.109, + "d_bind": 0.02564 + }, + { + "tau": 0.0529, + "v": 0.8961, + "H_a": 0.103, + "d_bind": 0.01603 + } + ] + }, + { + "step": 5800, + "tok": 23756800, + "ce": 3.4401, + "ppl": 31.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22403, + "peak_MiB": 5301, + "v_gain": 0.1757, + "loop_stats": [ + { + "tau": 0.0938, + "v": 0.9229, + "H_a": 0.144, + "d_bind": 0.0 + }, + { + "tau": 0.0496, + "v": 0.9226, + "H_a": 0.124, + "d_bind": 0.03127 + }, + { + "tau": 0.0496, + "v": 0.9225, + "H_a": 0.112, + "d_bind": 0.01618 + } + ] + }, + { + "step": 5820, + "tok": 23838720, + "ce": 3.3098, + "ppl": 27.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1757, + "loop_stats": [ + { + "tau": 0.0932, + "v": 0.9287, + "H_a": 0.153, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.9274, + "H_a": 0.133, + "d_bind": 0.03058 + }, + { + "tau": 0.0343, + "v": 0.9263, + "H_a": 0.113, + "d_bind": 0.01674 + } + ] + }, + { + "step": 5840, + "tok": 23920640, + "ce": 3.9627, + "ppl": 52.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.1756, + "loop_stats": [ + { + "tau": 0.0794, + "v": 0.8723, + "H_a": 0.151, + "d_bind": 0.0 + }, + { + "tau": 0.0239, + "v": 0.8461, + "H_a": 0.135, + "d_bind": 0.03197 + }, + { + "tau": 0.0185, + "v": 0.8386, + "H_a": 0.123, + "d_bind": 0.0198 + } + ] + }, + { + "step": 5860, + "tok": 24002560, + "ce": 3.3904, + "ppl": 29.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22415, + "peak_MiB": 5301, + "v_gain": 0.1753, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.9034, + "H_a": 0.162, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.901, + "H_a": 0.125, + "d_bind": 0.02739 + }, + { + "tau": 0.0384, + "v": 0.9019, + "H_a": 0.115, + "d_bind": 0.01744 + } + ] + }, + { + "step": 5880, + "tok": 24084480, + "ce": 3.1605, + "ppl": 23.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22400, + "peak_MiB": 5301, + "v_gain": 0.1751, + "loop_stats": [ + { + "tau": 0.1023, + "v": 0.9006, + "H_a": 0.168, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.8983, + "H_a": 0.131, + "d_bind": 0.03214 + }, + { + "tau": 0.0439, + "v": 0.9011, + "H_a": 0.114, + "d_bind": 0.01736 + } + ] + }, + { + "step": 5900, + "tok": 24166400, + "ce": 3.372, + "ppl": 29.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22403, + "peak_MiB": 5301, + "v_gain": 0.178, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.9139, + "H_a": 0.142, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.9045, + "H_a": 0.124, + "d_bind": 0.02904 + }, + { + "tau": 0.0412, + "v": 0.9005, + "H_a": 0.112, + "d_bind": 0.01775 + } + ] + }, + { + "step": 5920, + "tok": 24248320, + "ce": 3.1978, + "ppl": 24.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.933, + "tok_s": 22406, + "peak_MiB": 5301, + "v_gain": 0.1807, + "loop_stats": [ + { + "tau": 0.1178, + "v": 0.8806, + "H_a": 0.147, + "d_bind": 0.0 + }, + { + "tau": 0.0379, + "v": 0.874, + "H_a": 0.12, + "d_bind": 0.02903 + }, + { + "tau": 0.0381, + "v": 0.8754, + "H_a": 0.114, + "d_bind": 0.01705 + } + ] + }, + { + "step": 5940, + "tok": 24330240, + "ce": 3.2008, + "ppl": 24.6, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.922, + "tok_s": 22409, + "peak_MiB": 5301, + "v_gain": 0.1809, + "loop_stats": [ + { + "tau": 0.1046, + "v": 0.8805, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0402, + "v": 0.8783, + "H_a": 0.129, + "d_bind": 0.02939 + }, + { + "tau": 0.0424, + "v": 0.8805, + "H_a": 0.111, + "d_bind": 0.0168 + } + ] + }, + { + "step": 5960, + "tok": 24412160, + "ce": 3.3356, + "ppl": 28.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.922, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.1844, + "loop_stats": [ + { + "tau": 0.1083, + "v": 0.8863, + "H_a": 0.144, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8807, + "H_a": 0.135, + "d_bind": 0.03026 + }, + { + "tau": 0.0439, + "v": 0.8819, + "H_a": 0.116, + "d_bind": 0.01877 + } + ] + }, + { + "step": 5980, + "tok": 24494080, + "ce": 3.0998, + "ppl": 22.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.922, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.1866, + "loop_stats": [ + { + "tau": 0.1204, + "v": 0.8783, + "H_a": 0.143, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.8739, + "H_a": 0.124, + "d_bind": 0.03012 + }, + { + "tau": 0.0395, + "v": 0.876, + "H_a": 0.113, + "d_bind": 0.01624 + } + ] + }, + { + "step": 6000, + "tok": 24576000, + "ce": 3.074, + "ppl": 21.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.912, + "tok_s": 22403, + "peak_MiB": 5301, + "v_gain": 0.1846, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.8874, + "H_a": 0.145, + "d_bind": 0.0 + }, + { + "tau": 0.0435, + "v": 0.8899, + "H_a": 0.13, + "d_bind": 0.03154 + }, + { + "tau": 0.0455, + "v": 0.8961, + "H_a": 0.114, + "d_bind": 0.01709 + } + ] + }, + { + "step": 6020, + "tok": 24657920, + "ce": 3.3465, + "ppl": 28.4, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.925, + "tok_s": 22406, + "peak_MiB": 5301, + "v_gain": 0.1862, + "loop_stats": [ + { + "tau": 0.0946, + "v": 0.8875, + "H_a": 0.157, + "d_bind": 0.0 + }, + { + "tau": 0.0341, + "v": 0.8939, + "H_a": 0.134, + "d_bind": 0.03159 + }, + { + "tau": 0.0334, + "v": 0.8988, + "H_a": 0.113, + "d_bind": 0.01747 + } + ] + }, + { + "step": 6040, + "tok": 24739840, + "ce": 3.6848, + "ppl": 39.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.922, + "tok_s": 22409, + "peak_MiB": 5301, + "v_gain": 0.1869, + "loop_stats": [ + { + "tau": 0.0869, + "v": 0.9065, + "H_a": 0.158, + "d_bind": 0.0 + }, + { + "tau": 0.0254, + "v": 0.8973, + "H_a": 0.13, + "d_bind": 0.03118 + }, + { + "tau": 0.0254, + "v": 0.894, + "H_a": 0.122, + "d_bind": 0.01842 + } + ] + }, + { + "step": 6060, + "tok": 24821760, + "ce": 3.4631, + "ppl": 31.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.91, + "tok_s": 22411, + "peak_MiB": 5301, + "v_gain": 0.1884, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.8726, + "H_a": 0.156, + "d_bind": 0.0 + }, + { + "tau": 0.0325, + "v": 0.8591, + "H_a": 0.131, + "d_bind": 0.03066 + }, + { + "tau": 0.0339, + "v": 0.8574, + "H_a": 0.122, + "d_bind": 0.01844 + } + ] + }, + { + "step": 6080, + "tok": 24903680, + "ce": 2.6714, + "ppl": 14.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.1877, + "loop_stats": [ + { + "tau": 0.1357, + "v": 0.8848, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0656, + "v": 0.8707, + "H_a": 0.109, + "d_bind": 0.02708 + }, + { + "tau": 0.0637, + "v": 0.8685, + "H_a": 0.1, + "d_bind": 0.01617 + } + ] + }, + { + "step": 6100, + "tok": 24985600, + "ce": 3.1735, + "ppl": 23.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.922, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.1851, + "loop_stats": [ + { + "tau": 0.1311, + "v": 0.8764, + "H_a": 0.121, + "d_bind": 0.0 + }, + { + "tau": 0.0634, + "v": 0.8602, + "H_a": 0.115, + "d_bind": 0.02865 + }, + { + "tau": 0.0616, + "v": 0.8556, + "H_a": 0.101, + "d_bind": 0.01781 + } + ] + }, + { + "step": 6120, + "tok": 25067520, + "ce": 3.2962, + "ppl": 27.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22405, + "peak_MiB": 5301, + "v_gain": 0.1859, + "loop_stats": [ + { + "tau": 0.0988, + "v": 0.9246, + "H_a": 0.148, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.9207, + "H_a": 0.125, + "d_bind": 0.03316 + }, + { + "tau": 0.039, + "v": 0.9168, + "H_a": 0.112, + "d_bind": 0.01903 + } + ] + }, + { + "step": 6140, + "tok": 25149440, + "ce": 2.9626, + "ppl": 19.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.925, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1871, + "loop_stats": [ + { + "tau": 0.1165, + "v": 0.8842, + "H_a": 0.142, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8672, + "H_a": 0.12, + "d_bind": 0.03244 + }, + { + "tau": 0.0449, + "v": 0.86, + "H_a": 0.106, + "d_bind": 0.01815 + } + ] + }, + { + "step": 6160, + "tok": 25231360, + "ce": 3.6171, + "ppl": 37.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.907, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.1851, + "loop_stats": [ + { + "tau": 0.0926, + "v": 0.9162, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0287, + "v": 0.9047, + "H_a": 0.134, + "d_bind": 0.03362 + }, + { + "tau": 0.0264, + "v": 0.906, + "H_a": 0.116, + "d_bind": 0.01855 + } + ] + }, + { + "step": 6180, + "tok": 25313280, + "ce": 3.8732, + "ppl": 48.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.92, + "tok_s": 22412, + "peak_MiB": 5301, + "v_gain": 0.1839, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.9204, + "H_a": 0.139, + "d_bind": 0.0 + }, + { + "tau": 0.0259, + "v": 0.9023, + "H_a": 0.135, + "d_bind": 0.03136 + }, + { + "tau": 0.0244, + "v": 0.9015, + "H_a": 0.123, + "d_bind": 0.02012 + } + ] + }, + { + "step": 6200, + "tok": 25395200, + "ce": 3.087, + "ppl": 21.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.927, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.1832, + "loop_stats": [ + { + "tau": 0.1392, + "v": 0.8874, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8691, + "H_a": 0.114, + "d_bind": 0.03079 + }, + { + "tau": 0.0431, + "v": 0.8624, + "H_a": 0.106, + "d_bind": 0.01843 + } + ] + }, + { + "step": 6220, + "tok": 25477120, + "ce": 3.494, + "ppl": 32.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.933, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.183, + "loop_stats": [ + { + "tau": 0.0936, + "v": 0.932, + "H_a": 0.149, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.9263, + "H_a": 0.134, + "d_bind": 0.03331 + }, + { + "tau": 0.0323, + "v": 0.9248, + "H_a": 0.117, + "d_bind": 0.0194 + } + ] + }, + { + "step": 6240, + "tok": 25559040, + "ce": 2.9262, + "ppl": 18.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22405, + "peak_MiB": 5301, + "v_gain": 0.1833, + "loop_stats": [ + { + "tau": 0.1202, + "v": 0.9141, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0399, + "v": 0.9017, + "H_a": 0.12, + "d_bind": 0.03151 + }, + { + "tau": 0.0368, + "v": 0.8945, + "H_a": 0.114, + "d_bind": 0.01979 + } + ] + }, + { + "step": 6260, + "tok": 25640960, + "ce": 3.7162, + "ppl": 41.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1815, + "loop_stats": [ + { + "tau": 0.0847, + "v": 0.9052, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.9063, + "H_a": 0.13, + "d_bind": 0.03764 + }, + { + "tau": 0.03, + "v": 0.9124, + "H_a": 0.12, + "d_bind": 0.01989 + } + ] + }, + { + "step": 6280, + "tok": 25722880, + "ce": 3.016, + "ppl": 20.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22409, + "peak_MiB": 5301, + "v_gain": 0.182, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.9049, + "H_a": 0.142, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.9119, + "H_a": 0.117, + "d_bind": 0.03648 + }, + { + "tau": 0.0332, + "v": 0.9141, + "H_a": 0.106, + "d_bind": 0.0189 + } + ] + }, + { + "step": 6300, + "tok": 25804800, + "ce": 3.552, + "ppl": 34.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22412, + "peak_MiB": 5301, + "v_gain": 0.1796, + "loop_stats": [ + { + "tau": 0.0786, + "v": 0.9325, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.9344, + "H_a": 0.128, + "d_bind": 0.03406 + }, + { + "tau": 0.0301, + "v": 0.9358, + "H_a": 0.11, + "d_bind": 0.01893 + } + ] + }, + { + "step": 6320, + "tok": 25886720, + "ce": 3.4756, + "ppl": 32.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.1795, + "loop_stats": [ + { + "tau": 0.0865, + "v": 0.9028, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.028, + "v": 0.8949, + "H_a": 0.118, + "d_bind": 0.03208 + }, + { + "tau": 0.0249, + "v": 0.8905, + "H_a": 0.108, + "d_bind": 0.01757 + } + ] + }, + { + "step": 6340, + "tok": 25968640, + "ce": 3.395, + "ppl": 29.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.93, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.1802, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.9172, + "H_a": 0.144, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.9127, + "H_a": 0.118, + "d_bind": 0.03253 + }, + { + "tau": 0.0439, + "v": 0.9096, + "H_a": 0.106, + "d_bind": 0.01806 + } + ] + }, + { + "step": 6360, + "tok": 26050560, + "ce": 2.9689, + "ppl": 19.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.925, + "tok_s": 22406, + "peak_MiB": 5301, + "v_gain": 0.1819, + "loop_stats": [ + { + "tau": 0.1328, + "v": 0.8687, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0461, + "v": 0.8508, + "H_a": 0.099, + "d_bind": 0.03003 + }, + { + "tau": 0.0434, + "v": 0.8512, + "H_a": 0.095, + "d_bind": 0.0158 + } + ] + }, + { + "step": 6380, + "tok": 26132480, + "ce": 3.4362, + "ppl": 31.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.927, + "tok_s": 22408, + "peak_MiB": 5301, + "v_gain": 0.1815, + "loop_stats": [ + { + "tau": 0.0931, + "v": 0.9231, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.9225, + "H_a": 0.123, + "d_bind": 0.02976 + }, + { + "tau": 0.0234, + "v": 0.9215, + "H_a": 0.122, + "d_bind": 0.02064 + } + ] + }, + { + "step": 6400, + "tok": 26214400, + "ce": 3.3761, + "ppl": 29.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.925, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.1806, + "loop_stats": [ + { + "tau": 0.0868, + "v": 0.94, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0276, + "v": 0.9374, + "H_a": 0.122, + "d_bind": 0.03417 + }, + { + "tau": 0.0227, + "v": 0.9365, + "H_a": 0.108, + "d_bind": 0.01978 + } + ] + }, + { + "step": 6420, + "tok": 26296320, + "ce": 2.9277, + "ppl": 18.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.935, + "tok_s": 22413, + "peak_MiB": 5301, + "v_gain": 0.1826, + "loop_stats": [ + { + "tau": 0.1321, + "v": 0.9043, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0562, + "v": 0.8919, + "H_a": 0.117, + "d_bind": 0.03371 + }, + { + "tau": 0.0503, + "v": 0.891, + "H_a": 0.096, + "d_bind": 0.01645 + } + ] + }, + { + "step": 6440, + "tok": 26378240, + "ce": 3.1503, + "ppl": 23.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.1878, + "loop_stats": [ + { + "tau": 0.1206, + "v": 0.9134, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.9019, + "H_a": 0.115, + "d_bind": 0.03334 + }, + { + "tau": 0.0408, + "v": 0.8991, + "H_a": 0.097, + "d_bind": 0.01681 + } + ] + }, + { + "step": 6460, + "tok": 26460160, + "ce": 3.1842, + "ppl": 24.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.938, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.1899, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.8827, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0515, + "v": 0.8578, + "H_a": 0.122, + "d_bind": 0.03501 + }, + { + "tau": 0.0494, + "v": 0.8527, + "H_a": 0.109, + "d_bind": 0.01799 + } + ] + }, + { + "step": 6480, + "tok": 26542080, + "ce": 3.3643, + "ppl": 28.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22407, + "peak_MiB": 5301, + "v_gain": 0.1901, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.8933, + "H_a": 0.147, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.88, + "H_a": 0.114, + "d_bind": 0.03543 + }, + { + "tau": 0.0347, + "v": 0.8765, + "H_a": 0.098, + "d_bind": 0.01709 + } + ] + }, + { + "step": 6500, + "tok": 26624000, + "ce": 2.7989, + "ppl": 16.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22409, + "peak_MiB": 5301, + "v_gain": 0.1921, + "loop_stats": [ + { + "tau": 0.1235, + "v": 0.8809, + "H_a": 0.137, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.872, + "H_a": 0.106, + "d_bind": 0.03575 + }, + { + "tau": 0.0393, + "v": 0.8726, + "H_a": 0.099, + "d_bind": 0.01789 + } + ] + }, + { + "step": 6520, + "tok": 26705920, + "ce": 3.4371, + "ppl": 31.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.927, + "tok_s": 22411, + "peak_MiB": 5301, + "v_gain": 0.1898, + "loop_stats": [ + { + "tau": 0.0926, + "v": 0.933, + "H_a": 0.133, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.9244, + "H_a": 0.116, + "d_bind": 0.03523 + }, + { + "tau": 0.0359, + "v": 0.9229, + "H_a": 0.102, + "d_bind": 0.01789 + } + ] + }, + { + "step": 6540, + "tok": 26787840, + "ce": 2.9615, + "ppl": 19.3, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.927, + "tok_s": 22414, + "peak_MiB": 5301, + "v_gain": 0.1887, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.8878, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0415, + "v": 0.8844, + "H_a": 0.111, + "d_bind": 0.03171 + }, + { + "tau": 0.0382, + "v": 0.8821, + "H_a": 0.099, + "d_bind": 0.01817 + } + ] + }, + { + "step": 6560, + "tok": 26869760, + "ce": 3.3709, + "ppl": 29.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.1891, + "loop_stats": [ + { + "tau": 0.0988, + "v": 0.8747, + "H_a": 0.152, + "d_bind": 0.0 + }, + { + "tau": 0.0303, + "v": 0.8734, + "H_a": 0.12, + "d_bind": 0.03531 + }, + { + "tau": 0.0293, + "v": 0.8733, + "H_a": 0.105, + "d_bind": 0.01754 + } + ] + }, + { + "step": 6580, + "tok": 26951680, + "ce": 3.264, + "ppl": 26.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.89, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.188, + "loop_stats": [ + { + "tau": 0.1205, + "v": 0.8848, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.037, + "v": 0.8837, + "H_a": 0.108, + "d_bind": 0.03468 + }, + { + "tau": 0.0315, + "v": 0.881, + "H_a": 0.096, + "d_bind": 0.01783 + } + ] + }, + { + "step": 6600, + "tok": 27033600, + "ce": 3.5232, + "ppl": 33.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.922, + "tok_s": 22408, + "peak_MiB": 5301, + "v_gain": 0.1855, + "loop_stats": [ + { + "tau": 0.1093, + "v": 0.9088, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.9163, + "H_a": 0.112, + "d_bind": 0.03399 + }, + { + "tau": 0.0415, + "v": 0.9213, + "H_a": 0.097, + "d_bind": 0.01651 + } + ] + }, + { + "step": 6620, + "tok": 27115520, + "ce": 3.2702, + "ppl": 26.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.915, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.1838, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.8604, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.8771, + "H_a": 0.106, + "d_bind": 0.03218 + }, + { + "tau": 0.0329, + "v": 0.892, + "H_a": 0.096, + "d_bind": 0.01726 + } + ] + }, + { + "step": 6640, + "tok": 27197440, + "ce": 3.2076, + "ppl": 24.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.912, + "tok_s": 22413, + "peak_MiB": 5301, + "v_gain": 0.186, + "loop_stats": [ + { + "tau": 0.1078, + "v": 0.8604, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8511, + "H_a": 0.108, + "d_bind": 0.03317 + }, + { + "tau": 0.0361, + "v": 0.8538, + "H_a": 0.099, + "d_bind": 0.02032 + } + ] + }, + { + "step": 6660, + "tok": 27279360, + "ce": 3.5185, + "ppl": 33.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.922, + "tok_s": 22415, + "peak_MiB": 5301, + "v_gain": 0.1877, + "loop_stats": [ + { + "tau": 0.0851, + "v": 0.8802, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.874, + "H_a": 0.106, + "d_bind": 0.0318 + }, + { + "tau": 0.0289, + "v": 0.8815, + "H_a": 0.1, + "d_bind": 0.01918 + } + ] + }, + { + "step": 6680, + "tok": 27361280, + "ce": 3.7085, + "ppl": 40.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.927, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.1915, + "loop_stats": [ + { + "tau": 0.0805, + "v": 0.8913, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.037, + "v": 0.8855, + "H_a": 0.119, + "d_bind": 0.03536 + }, + { + "tau": 0.0347, + "v": 0.8873, + "H_a": 0.105, + "d_bind": 0.01945 + } + ] + }, + { + "step": 6700, + "tok": 27443200, + "ce": 3.3524, + "ppl": 28.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.925, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.1927, + "loop_stats": [ + { + "tau": 0.0927, + "v": 0.8668, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.8805, + "H_a": 0.105, + "d_bind": 0.03032 + }, + { + "tau": 0.0361, + "v": 0.887, + "H_a": 0.105, + "d_bind": 0.01718 + } + ] + }, + { + "step": 6720, + "tok": 27525120, + "ce": 3.2219, + "ppl": 25.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.912, + "tok_s": 22410, + "peak_MiB": 5301, + "v_gain": 0.1923, + "loop_stats": [ + { + "tau": 0.1215, + "v": 0.8492, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.8414, + "H_a": 0.103, + "d_bind": 0.03263 + }, + { + "tau": 0.0378, + "v": 0.842, + "H_a": 0.098, + "d_bind": 0.01916 + } + ] + }, + { + "step": 6740, + "tok": 27607040, + "ce": 3.7274, + "ppl": 41.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.91, + "tok_s": 22412, + "peak_MiB": 5301, + "v_gain": 0.1918, + "loop_stats": [ + { + "tau": 0.0731, + "v": 0.8549, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8587, + "H_a": 0.113, + "d_bind": 0.03142 + }, + { + "tau": 0.0325, + "v": 0.8685, + "H_a": 0.111, + "d_bind": 0.01972 + } + ] + }, + { + "step": 6760, + "tok": 27688960, + "ce": 3.4523, + "ppl": 31.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.92, + "tok_s": 22414, + "peak_MiB": 5301, + "v_gain": 0.1896, + "loop_stats": [ + { + "tau": 0.095, + "v": 0.8325, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.8477, + "H_a": 0.119, + "d_bind": 0.03454 + }, + { + "tau": 0.0268, + "v": 0.8646, + "H_a": 0.109, + "d_bind": 0.01937 + } + ] + }, + { + "step": 6780, + "tok": 27770880, + "ce": 3.1744, + "ppl": 23.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.922, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.1887, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.8361, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0272, + "v": 0.8529, + "H_a": 0.121, + "d_bind": 0.03533 + }, + { + "tau": 0.026, + "v": 0.8663, + "H_a": 0.099, + "d_bind": 0.02017 + } + ] + }, + { + "step": 6800, + "tok": 27852800, + "ce": 3.0178, + "ppl": 20.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.917, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.1883, + "loop_stats": [ + { + "tau": 0.1181, + "v": 0.8272, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.831, + "H_a": 0.103, + "d_bind": 0.03444 + }, + { + "tau": 0.0379, + "v": 0.8407, + "H_a": 0.09, + "d_bind": 0.01711 + } + ] + }, + { + "step": 6820, + "tok": 27934720, + "ce": 3.372, + "ppl": 29.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.92, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.1894, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.8471, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.035, + "v": 0.8482, + "H_a": 0.1, + "d_bind": 0.02871 + }, + { + "tau": 0.0346, + "v": 0.852, + "H_a": 0.097, + "d_bind": 0.01947 + } + ] + }, + { + "step": 6840, + "tok": 28016640, + "ce": 2.9948, + "ppl": 20.0, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.922, + "tok_s": 22411, + "peak_MiB": 5301, + "v_gain": 0.1874, + "loop_stats": [ + { + "tau": 0.1211, + "v": 0.7886, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0402, + "v": 0.7834, + "H_a": 0.114, + "d_bind": 0.03443 + }, + { + "tau": 0.035, + "v": 0.7951, + "H_a": 0.096, + "d_bind": 0.01857 + } + ] + }, + { + "step": 6860, + "tok": 28098560, + "ce": 2.8038, + "ppl": 16.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.915, + "tok_s": 22413, + "peak_MiB": 5301, + "v_gain": 0.1862, + "loop_stats": [ + { + "tau": 0.1326, + "v": 0.8115, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8059, + "H_a": 0.104, + "d_bind": 0.03267 + }, + { + "tau": 0.0404, + "v": 0.8103, + "H_a": 0.098, + "d_bind": 0.01839 + } + ] + }, + { + "step": 6880, + "tok": 28180480, + "ce": 3.1632, + "ppl": 23.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.922, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.1862, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.8289, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.8354, + "H_a": 0.113, + "d_bind": 0.03461 + }, + { + "tau": 0.036, + "v": 0.8425, + "H_a": 0.105, + "d_bind": 0.01988 + } + ] + }, + { + "step": 6900, + "tok": 28262400, + "ce": 3.5138, + "ppl": 33.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.922, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.1877, + "loop_stats": [ + { + "tau": 0.1069, + "v": 0.8208, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8304, + "H_a": 0.096, + "d_bind": 0.03095 + }, + { + "tau": 0.0289, + "v": 0.842, + "H_a": 0.097, + "d_bind": 0.01747 + } + ] + }, + { + "step": 6920, + "tok": 28344320, + "ce": 3.1759, + "ppl": 23.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.93, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.19, + "loop_stats": [ + { + "tau": 0.1149, + "v": 0.8214, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.8276, + "H_a": 0.103, + "d_bind": 0.03207 + }, + { + "tau": 0.0363, + "v": 0.8384, + "H_a": 0.098, + "d_bind": 0.01838 + } + ] + }, + { + "step": 6940, + "tok": 28426240, + "ce": 3.25, + "ppl": 25.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.925, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.1879, + "loop_stats": [ + { + "tau": 0.1188, + "v": 0.8466, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0399, + "v": 0.8482, + "H_a": 0.11, + "d_bind": 0.02998 + }, + { + "tau": 0.0394, + "v": 0.8572, + "H_a": 0.106, + "d_bind": 0.01868 + } + ] + }, + { + "step": 6960, + "tok": 28508160, + "ce": 2.7802, + "ppl": 16.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.92, + "tok_s": 22412, + "peak_MiB": 5301, + "v_gain": 0.1904, + "loop_stats": [ + { + "tau": 0.1275, + "v": 0.8301, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0545, + "v": 0.8358, + "H_a": 0.104, + "d_bind": 0.03229 + }, + { + "tau": 0.0534, + "v": 0.8503, + "H_a": 0.086, + "d_bind": 0.01751 + } + ] + }, + { + "step": 6980, + "tok": 28590080, + "ce": 3.3763, + "ppl": 29.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.915, + "tok_s": 22414, + "peak_MiB": 5301, + "v_gain": 0.1914, + "loop_stats": [ + { + "tau": 0.089, + "v": 0.8636, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.045, + "v": 0.8781, + "H_a": 0.127, + "d_bind": 0.03538 + }, + { + "tau": 0.0457, + "v": 0.8916, + "H_a": 0.107, + "d_bind": 0.01997 + } + ] + }, + { + "step": 7000, + "tok": 28672000, + "ce": 3.0242, + "ppl": 20.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.917, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.1919, + "loop_stats": [ + { + "tau": 0.1237, + "v": 0.8551, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0536, + "v": 0.8529, + "H_a": 0.112, + "d_bind": 0.03629 + }, + { + "tau": 0.0494, + "v": 0.8553, + "H_a": 0.1, + "d_bind": 0.01831 + } + ] + }, + { + "step": 7020, + "tok": 28753920, + "ce": 3.2912, + "ppl": 26.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.912, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.1913, + "loop_stats": [ + { + "tau": 0.1001, + "v": 0.846, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8494, + "H_a": 0.108, + "d_bind": 0.03514 + }, + { + "tau": 0.0387, + "v": 0.8579, + "H_a": 0.094, + "d_bind": 0.01778 + } + ] + }, + { + "step": 7040, + "tok": 28835840, + "ce": 3.1046, + "ppl": 22.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.1929, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.8285, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8327, + "H_a": 0.115, + "d_bind": 0.03349 + }, + { + "tau": 0.0372, + "v": 0.8456, + "H_a": 0.101, + "d_bind": 0.01818 + } + ] + }, + { + "step": 7060, + "tok": 28917760, + "ce": 3.4795, + "ppl": 32.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.925, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.195, + "loop_stats": [ + { + "tau": 0.0933, + "v": 0.8342, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0402, + "v": 0.8363, + "H_a": 0.113, + "d_bind": 0.03482 + }, + { + "tau": 0.0387, + "v": 0.8463, + "H_a": 0.111, + "d_bind": 0.0194 + } + ] + }, + { + "step": 7080, + "tok": 28999680, + "ce": 3.3009, + "ppl": 27.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.935, + "tok_s": 22414, + "peak_MiB": 5301, + "v_gain": 0.1983, + "loop_stats": [ + { + "tau": 0.1092, + "v": 0.8472, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0427, + "v": 0.8542, + "H_a": 0.119, + "d_bind": 0.03115 + }, + { + "tau": 0.0424, + "v": 0.8627, + "H_a": 0.107, + "d_bind": 0.01944 + } + ] + }, + { + "step": 7100, + "tok": 29081600, + "ce": 3.554, + "ppl": 35.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.1989, + "loop_stats": [ + { + "tau": 0.1004, + "v": 0.853, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.866, + "H_a": 0.112, + "d_bind": 0.03106 + }, + { + "tau": 0.0397, + "v": 0.8769, + "H_a": 0.104, + "d_bind": 0.01997 + } + ] + }, + { + "step": 7120, + "tok": 29163520, + "ce": 3.4749, + "ppl": 32.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2032, + "loop_stats": [ + { + "tau": 0.1018, + "v": 0.7941, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8054, + "H_a": 0.109, + "d_bind": 0.03186 + }, + { + "tau": 0.047, + "v": 0.8188, + "H_a": 0.093, + "d_bind": 0.01696 + } + ] + }, + { + "step": 7140, + "tok": 29245440, + "ce": 3.5244, + "ppl": 33.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2085, + "loop_stats": [ + { + "tau": 0.0911, + "v": 0.82, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.8291, + "H_a": 0.116, + "d_bind": 0.03365 + }, + { + "tau": 0.0276, + "v": 0.8427, + "H_a": 0.101, + "d_bind": 0.01761 + } + ] + }, + { + "step": 7160, + "tok": 29327360, + "ce": 3.9162, + "ppl": 50.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2118, + "loop_stats": [ + { + "tau": 0.0759, + "v": 0.8338, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0097, + "v": 0.8444, + "H_a": 0.113, + "d_bind": 0.03273 + }, + { + "tau": 0.0109, + "v": 0.8575, + "H_a": 0.102, + "d_bind": 0.01973 + } + ] + }, + { + "step": 7180, + "tok": 29409280, + "ce": 2.8957, + "ppl": 18.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.2134, + "loop_stats": [ + { + "tau": 0.1306, + "v": 0.8244, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.8162, + "H_a": 0.108, + "d_bind": 0.03751 + }, + { + "tau": 0.0414, + "v": 0.8249, + "H_a": 0.095, + "d_bind": 0.01838 + } + ] + }, + { + "step": 7200, + "tok": 29491200, + "ce": 3.8102, + "ppl": 45.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.92, + "tok_s": 22415, + "peak_MiB": 5301, + "v_gain": 0.2122, + "loop_stats": [ + { + "tau": 0.0736, + "v": 0.8331, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0189, + "v": 0.8321, + "H_a": 0.13, + "d_bind": 0.03463 + }, + { + "tau": 0.0204, + "v": 0.8379, + "H_a": 0.113, + "d_bind": 0.02195 + } + ] + }, + { + "step": 7220, + "tok": 29573120, + "ce": 2.8272, + "ppl": 16.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2096, + "loop_stats": [ + { + "tau": 0.1297, + "v": 0.8207, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0518, + "v": 0.8226, + "H_a": 0.104, + "d_bind": 0.03742 + }, + { + "tau": 0.0559, + "v": 0.8263, + "H_a": 0.092, + "d_bind": 0.01891 + } + ] + }, + { + "step": 7240, + "tok": 29655040, + "ce": 3.3534, + "ppl": 28.6, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.922, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2122, + "loop_stats": [ + { + "tau": 0.0896, + "v": 0.8041, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0177, + "v": 0.8187, + "H_a": 0.117, + "d_bind": 0.03327 + }, + { + "tau": 0.024, + "v": 0.8324, + "H_a": 0.106, + "d_bind": 0.02118 + } + ] + }, + { + "step": 7260, + "tok": 29736960, + "ce": 3.3037, + "ppl": 27.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.92, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2167, + "loop_stats": [ + { + "tau": 0.0969, + "v": 0.8189, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0231, + "v": 0.8297, + "H_a": 0.112, + "d_bind": 0.03709 + }, + { + "tau": 0.0198, + "v": 0.841, + "H_a": 0.1, + "d_bind": 0.01978 + } + ] + }, + { + "step": 7280, + "tok": 29818880, + "ce": 3.5554, + "ppl": 35.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.2126, + "loop_stats": [ + { + "tau": 0.0801, + "v": 0.7934, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.809, + "H_a": 0.112, + "d_bind": 0.03348 + }, + { + "tau": 0.0313, + "v": 0.823, + "H_a": 0.104, + "d_bind": 0.01976 + } + ] + }, + { + "step": 7300, + "tok": 29900800, + "ce": 3.1257, + "ppl": 22.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.925, + "tok_s": 22427, + "peak_MiB": 5301, + "v_gain": 0.2109, + "loop_stats": [ + { + "tau": 0.121, + "v": 0.7977, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8237, + "H_a": 0.098, + "d_bind": 0.03004 + }, + { + "tau": 0.0352, + "v": 0.8441, + "H_a": 0.09, + "d_bind": 0.01733 + } + ] + }, + { + "step": 7320, + "tok": 29982720, + "ce": 3.0467, + "ppl": 21.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.933, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.2104, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.7883, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.8115, + "H_a": 0.103, + "d_bind": 0.0313 + }, + { + "tau": 0.0358, + "v": 0.8241, + "H_a": 0.091, + "d_bind": 0.0181 + } + ] + }, + { + "step": 7340, + "tok": 30064640, + "ce": 3.5525, + "ppl": 34.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.935, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2071, + "loop_stats": [ + { + "tau": 0.0908, + "v": 0.7974, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8036, + "H_a": 0.109, + "d_bind": 0.03294 + }, + { + "tau": 0.0256, + "v": 0.8144, + "H_a": 0.102, + "d_bind": 0.02081 + } + ] + }, + { + "step": 7360, + "tok": 30146560, + "ce": 3.3466, + "ppl": 28.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.912, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2053, + "loop_stats": [ + { + "tau": 0.1021, + "v": 0.7935, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8084, + "H_a": 0.109, + "d_bind": 0.03326 + }, + { + "tau": 0.0341, + "v": 0.8206, + "H_a": 0.088, + "d_bind": 0.01768 + } + ] + }, + { + "step": 7380, + "tok": 30228480, + "ce": 3.233, + "ppl": 25.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.2077, + "loop_stats": [ + { + "tau": 0.1142, + "v": 0.8312, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8552, + "H_a": 0.103, + "d_bind": 0.03161 + }, + { + "tau": 0.033, + "v": 0.8749, + "H_a": 0.095, + "d_bind": 0.0181 + } + ] + }, + { + "step": 7400, + "tok": 30310400, + "ce": 3.3582, + "ppl": 28.7, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.933, + "tok_s": 22431, + "peak_MiB": 5301, + "v_gain": 0.2045, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.7979, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.045, + "v": 0.8052, + "H_a": 0.098, + "d_bind": 0.03113 + }, + { + "tau": 0.041, + "v": 0.8161, + "H_a": 0.092, + "d_bind": 0.01685 + } + ] + }, + { + "step": 7420, + "tok": 30392320, + "ce": 3.3114, + "ppl": 27.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.922, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.206, + "loop_stats": [ + { + "tau": 0.1144, + "v": 0.7877, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0313, + "v": 0.7906, + "H_a": 0.096, + "d_bind": 0.03118 + }, + { + "tau": 0.0274, + "v": 0.8, + "H_a": 0.092, + "d_bind": 0.01769 + } + ] + }, + { + "step": 7440, + "tok": 30474240, + "ce": 2.8955, + "ppl": 18.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.92, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2066, + "loop_stats": [ + { + "tau": 0.1081, + "v": 0.7612, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0287, + "v": 0.7761, + "H_a": 0.095, + "d_bind": 0.02889 + }, + { + "tau": 0.0262, + "v": 0.7933, + "H_a": 0.083, + "d_bind": 0.0174 + } + ] + }, + { + "step": 7460, + "tok": 30556160, + "ce": 2.9574, + "ppl": 19.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.907, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2093, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.7384, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.031, + "v": 0.7565, + "H_a": 0.108, + "d_bind": 0.03261 + }, + { + "tau": 0.0274, + "v": 0.7748, + "H_a": 0.094, + "d_bind": 0.01826 + } + ] + }, + { + "step": 7480, + "tok": 30638080, + "ce": 3.0285, + "ppl": 20.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.927, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2075, + "loop_stats": [ + { + "tau": 0.1161, + "v": 0.771, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.7873, + "H_a": 0.117, + "d_bind": 0.03445 + }, + { + "tau": 0.039, + "v": 0.8087, + "H_a": 0.09, + "d_bind": 0.02043 + } + ] + }, + { + "step": 7500, + "tok": 30720000, + "ce": 3.1417, + "ppl": 23.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.92, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.2066, + "loop_stats": [ + { + "tau": 0.1166, + "v": 0.7919, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.802, + "H_a": 0.104, + "d_bind": 0.02963 + }, + { + "tau": 0.0455, + "v": 0.8149, + "H_a": 0.091, + "d_bind": 0.01785 + } + ] + }, + { + "step": 7520, + "tok": 30801920, + "ce": 3.0492, + "ppl": 21.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.2063, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.8107, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8235, + "H_a": 0.112, + "d_bind": 0.03458 + }, + { + "tau": 0.0423, + "v": 0.8344, + "H_a": 0.098, + "d_bind": 0.01932 + } + ] + }, + { + "step": 7540, + "tok": 30883840, + "ce": 3.2532, + "ppl": 25.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.2079, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.8121, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.8148, + "H_a": 0.113, + "d_bind": 0.03258 + }, + { + "tau": 0.0313, + "v": 0.8236, + "H_a": 0.11, + "d_bind": 0.02279 + } + ] + }, + { + "step": 7560, + "tok": 30965760, + "ce": 3.0689, + "ppl": 21.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2078, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.7803, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.8049, + "H_a": 0.106, + "d_bind": 0.03579 + }, + { + "tau": 0.0286, + "v": 0.8254, + "H_a": 0.091, + "d_bind": 0.01956 + } + ] + }, + { + "step": 7580, + "tok": 31047680, + "ce": 3.3698, + "ppl": 29.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.93, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2096, + "loop_stats": [ + { + "tau": 0.0968, + "v": 0.7575, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.7715, + "H_a": 0.114, + "d_bind": 0.03678 + }, + { + "tau": 0.0236, + "v": 0.7908, + "H_a": 0.097, + "d_bind": 0.02055 + } + ] + }, + { + "step": 7600, + "tok": 31129600, + "ce": 3.4992, + "ppl": 33.1, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.927, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2081, + "loop_stats": [ + { + "tau": 0.0786, + "v": 0.775, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.7798, + "H_a": 0.118, + "d_bind": 0.03464 + }, + { + "tau": 0.0201, + "v": 0.7942, + "H_a": 0.113, + "d_bind": 0.02031 + } + ] + }, + { + "step": 7620, + "tok": 31211520, + "ce": 3.2316, + "ppl": 25.3, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.93, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.2066, + "loop_stats": [ + { + "tau": 0.0969, + "v": 0.7717, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8026, + "H_a": 0.107, + "d_bind": 0.03298 + }, + { + "tau": 0.032, + "v": 0.8265, + "H_a": 0.089, + "d_bind": 0.01884 + } + ] + }, + { + "step": 7640, + "tok": 31293440, + "ce": 3.1905, + "ppl": 24.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.2056, + "loop_stats": [ + { + "tau": 0.1217, + "v": 0.7686, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0453, + "v": 0.7817, + "H_a": 0.102, + "d_bind": 0.03531 + }, + { + "tau": 0.0436, + "v": 0.7946, + "H_a": 0.093, + "d_bind": 0.01726 + } + ] + }, + { + "step": 7660, + "tok": 31375360, + "ce": 3.4492, + "ppl": 31.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.205, + "loop_stats": [ + { + "tau": 0.0896, + "v": 0.8079, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.022, + "v": 0.8288, + "H_a": 0.118, + "d_bind": 0.03262 + }, + { + "tau": 0.0197, + "v": 0.843, + "H_a": 0.109, + "d_bind": 0.02244 + } + ] + }, + { + "step": 7680, + "tok": 31457280, + "ce": 3.2712, + "ppl": 26.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2075, + "loop_stats": [ + { + "tau": 0.1057, + "v": 0.8313, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8458, + "H_a": 0.111, + "d_bind": 0.03214 + }, + { + "tau": 0.0348, + "v": 0.8582, + "H_a": 0.1, + "d_bind": 0.01897 + } + ] + }, + { + "step": 7700, + "tok": 31539200, + "ce": 2.9473, + "ppl": 19.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.93, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2088, + "loop_stats": [ + { + "tau": 0.1294, + "v": 0.8133, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8391, + "H_a": 0.098, + "d_bind": 0.02904 + }, + { + "tau": 0.035, + "v": 0.8559, + "H_a": 0.084, + "d_bind": 0.01874 + } + ] + }, + { + "step": 7720, + "tok": 31621120, + "ce": 2.7961, + "ppl": 16.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.935, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.2091, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.7889, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.7991, + "H_a": 0.087, + "d_bind": 0.02713 + }, + { + "tau": 0.0469, + "v": 0.8156, + "H_a": 0.079, + "d_bind": 0.01587 + } + ] + }, + { + "step": 7740, + "tok": 31703040, + "ce": 3.621, + "ppl": 37.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.925, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.2089, + "loop_stats": [ + { + "tau": 0.0774, + "v": 0.7966, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.023, + "v": 0.8121, + "H_a": 0.104, + "d_bind": 0.02961 + }, + { + "tau": 0.0215, + "v": 0.8304, + "H_a": 0.099, + "d_bind": 0.019 + } + ] + }, + { + "step": 7760, + "tok": 31784960, + "ce": 3.4748, + "ppl": 32.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.933, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.207, + "loop_stats": [ + { + "tau": 0.0896, + "v": 0.7721, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0323, + "v": 0.7823, + "H_a": 0.106, + "d_bind": 0.03059 + }, + { + "tau": 0.0328, + "v": 0.793, + "H_a": 0.095, + "d_bind": 0.01867 + } + ] + }, + { + "step": 7780, + "tok": 31866880, + "ce": 2.5416, + "ppl": 12.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2072, + "loop_stats": [ + { + "tau": 0.1375, + "v": 0.7462, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0502, + "v": 0.7689, + "H_a": 0.086, + "d_bind": 0.02699 + }, + { + "tau": 0.0489, + "v": 0.7874, + "H_a": 0.076, + "d_bind": 0.01543 + } + ] + }, + { + "step": 7800, + "tok": 31948800, + "ce": 3.2973, + "ppl": 27.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2062, + "loop_stats": [ + { + "tau": 0.1044, + "v": 0.8037, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8248, + "H_a": 0.104, + "d_bind": 0.02923 + }, + { + "tau": 0.0388, + "v": 0.8433, + "H_a": 0.096, + "d_bind": 0.01873 + } + ] + }, + { + "step": 7820, + "tok": 32030720, + "ce": 3.2555, + "ppl": 25.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.933, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.206, + "loop_stats": [ + { + "tau": 0.1178, + "v": 0.7327, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0537, + "v": 0.7641, + "H_a": 0.1, + "d_bind": 0.02781 + }, + { + "tau": 0.0527, + "v": 0.7851, + "H_a": 0.095, + "d_bind": 0.01826 + } + ] + }, + { + "step": 7840, + "tok": 32112640, + "ce": 2.6748, + "ppl": 14.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2077, + "loop_stats": [ + { + "tau": 0.143, + "v": 0.7802, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0565, + "v": 0.7925, + "H_a": 0.095, + "d_bind": 0.02958 + }, + { + "tau": 0.0529, + "v": 0.8039, + "H_a": 0.08, + "d_bind": 0.01681 + } + ] + }, + { + "step": 7860, + "tok": 32194560, + "ce": 3.0909, + "ppl": 22.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.2079, + "loop_stats": [ + { + "tau": 0.1283, + "v": 0.8354, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.8431, + "H_a": 0.103, + "d_bind": 0.03159 + }, + { + "tau": 0.0508, + "v": 0.8497, + "H_a": 0.097, + "d_bind": 0.01982 + } + ] + }, + { + "step": 7880, + "tok": 32276480, + "ce": 3.6382, + "ppl": 38.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22432, + "peak_MiB": 5301, + "v_gain": 0.2081, + "loop_stats": [ + { + "tau": 0.0766, + "v": 0.8243, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0172, + "v": 0.8369, + "H_a": 0.108, + "d_bind": 0.03207 + }, + { + "tau": 0.0173, + "v": 0.8488, + "H_a": 0.098, + "d_bind": 0.01932 + } + ] + }, + { + "step": 7900, + "tok": 32358400, + "ce": 2.9993, + "ppl": 20.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2085, + "loop_stats": [ + { + "tau": 0.1178, + "v": 0.7976, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.7996, + "H_a": 0.107, + "d_bind": 0.03604 + }, + { + "tau": 0.0351, + "v": 0.8104, + "H_a": 0.088, + "d_bind": 0.01767 + } + ] + }, + { + "step": 7920, + "tok": 32440320, + "ce": 3.2212, + "ppl": 25.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2084, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.8287, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8451, + "H_a": 0.105, + "d_bind": 0.03627 + }, + { + "tau": 0.0272, + "v": 0.8596, + "H_a": 0.096, + "d_bind": 0.01933 + } + ] + }, + { + "step": 7940, + "tok": 32522240, + "ce": 3.486, + "ppl": 32.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2082, + "loop_stats": [ + { + "tau": 0.085, + "v": 0.8298, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0152, + "v": 0.8576, + "H_a": 0.114, + "d_bind": 0.03668 + }, + { + "tau": 0.0126, + "v": 0.8772, + "H_a": 0.101, + "d_bind": 0.02019 + } + ] + }, + { + "step": 7960, + "tok": 32604160, + "ce": 3.8203, + "ppl": 45.6, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.92, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2073, + "loop_stats": [ + { + "tau": 0.0769, + "v": 0.8301, + "H_a": 0.133, + "d_bind": 0.0 + }, + { + "tau": 0.0146, + "v": 0.8538, + "H_a": 0.115, + "d_bind": 0.03463 + }, + { + "tau": 0.0123, + "v": 0.8674, + "H_a": 0.103, + "d_bind": 0.01986 + } + ] + }, + { + "step": 7980, + "tok": 32686080, + "ce": 3.3422, + "ppl": 28.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22423, + "peak_MiB": 5301, + "v_gain": 0.2072, + "loop_stats": [ + { + "tau": 0.0999, + "v": 0.8587, + "H_a": 0.137, + "d_bind": 0.0 + }, + { + "tau": 0.0199, + "v": 0.8732, + "H_a": 0.118, + "d_bind": 0.03872 + }, + { + "tau": 0.0143, + "v": 0.8799, + "H_a": 0.102, + "d_bind": 0.01938 + } + ] + }, + { + "step": 8000, + "tok": 32768000, + "ce": 3.1245, + "ppl": 22.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22431, + "peak_MiB": 5301, + "v_gain": 0.2058, + "loop_stats": [ + { + "tau": 0.1192, + "v": 0.8489, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.8696, + "H_a": 0.101, + "d_bind": 0.03448 + }, + { + "tau": 0.0393, + "v": 0.8764, + "H_a": 0.089, + "d_bind": 0.01751 + } + ] + }, + { + "step": 8020, + "tok": 32849920, + "ce": 2.9616, + "ppl": 19.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2049, + "loop_stats": [ + { + "tau": 0.1325, + "v": 0.8417, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.8459, + "H_a": 0.118, + "d_bind": 0.03552 + }, + { + "tau": 0.0276, + "v": 0.8538, + "H_a": 0.097, + "d_bind": 0.01953 + } + ] + }, + { + "step": 8040, + "tok": 32931840, + "ce": 3.4432, + "ppl": 31.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2063, + "loop_stats": [ + { + "tau": 0.0803, + "v": 0.816, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0243, + "v": 0.8303, + "H_a": 0.121, + "d_bind": 0.03554 + }, + { + "tau": 0.0204, + "v": 0.8473, + "H_a": 0.107, + "d_bind": 0.021 + } + ] + }, + { + "step": 8060, + "tok": 33013760, + "ce": 2.9843, + "ppl": 19.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.912, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2054, + "loop_stats": [ + { + "tau": 0.1276, + "v": 0.8258, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0462, + "v": 0.84, + "H_a": 0.098, + "d_bind": 0.03393 + }, + { + "tau": 0.0416, + "v": 0.8536, + "H_a": 0.091, + "d_bind": 0.01763 + } + ] + }, + { + "step": 8080, + "tok": 33095680, + "ce": 3.3019, + "ppl": 27.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.925, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2061, + "loop_stats": [ + { + "tau": 0.1216, + "v": 0.8487, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0549, + "v": 0.8617, + "H_a": 0.098, + "d_bind": 0.03297 + }, + { + "tau": 0.0523, + "v": 0.8722, + "H_a": 0.088, + "d_bind": 0.01761 + } + ] + }, + { + "step": 8100, + "tok": 33177600, + "ce": 3.1812, + "ppl": 24.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.925, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2075, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8411, + "H_a": 0.132, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8407, + "H_a": 0.113, + "d_bind": 0.03477 + }, + { + "tau": 0.0387, + "v": 0.8522, + "H_a": 0.098, + "d_bind": 0.01968 + } + ] + }, + { + "step": 8120, + "tok": 33259520, + "ce": 3.1279, + "ppl": 22.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.917, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.2084, + "loop_stats": [ + { + "tau": 0.1188, + "v": 0.8572, + "H_a": 0.132, + "d_bind": 0.0 + }, + { + "tau": 0.0332, + "v": 0.8546, + "H_a": 0.116, + "d_bind": 0.03356 + }, + { + "tau": 0.028, + "v": 0.8548, + "H_a": 0.101, + "d_bind": 0.0201 + } + ] + }, + { + "step": 8140, + "tok": 33341440, + "ce": 3.2025, + "ppl": 24.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.922, + "tok_s": 22427, + "peak_MiB": 5301, + "v_gain": 0.2084, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.8721, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.8477, + "H_a": 0.118, + "d_bind": 0.03456 + }, + { + "tau": 0.0375, + "v": 0.8462, + "H_a": 0.105, + "d_bind": 0.0206 + } + ] + }, + { + "step": 8160, + "tok": 33423360, + "ce": 2.8494, + "ppl": 17.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2117, + "loop_stats": [ + { + "tau": 0.1428, + "v": 0.8733, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8683, + "H_a": 0.107, + "d_bind": 0.03685 + }, + { + "tau": 0.0422, + "v": 0.8681, + "H_a": 0.092, + "d_bind": 0.01765 + } + ] + }, + { + "step": 8180, + "tok": 33505280, + "ce": 3.4099, + "ppl": 30.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2131, + "loop_stats": [ + { + "tau": 0.0999, + "v": 0.8625, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8601, + "H_a": 0.114, + "d_bind": 0.03235 + }, + { + "tau": 0.0268, + "v": 0.8589, + "H_a": 0.102, + "d_bind": 0.02086 + } + ] + }, + { + "step": 8200, + "tok": 33587200, + "ce": 2.6785, + "ppl": 14.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.938, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2128, + "loop_stats": [ + { + "tau": 0.1359, + "v": 0.89, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8853, + "H_a": 0.116, + "d_bind": 0.03809 + }, + { + "tau": 0.0446, + "v": 0.8781, + "H_a": 0.094, + "d_bind": 0.01813 + } + ] + }, + { + "step": 8220, + "tok": 33669120, + "ce": 3.527, + "ppl": 34.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.927, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2128, + "loop_stats": [ + { + "tau": 0.1034, + "v": 0.8546, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0313, + "v": 0.854, + "H_a": 0.104, + "d_bind": 0.0317 + }, + { + "tau": 0.0262, + "v": 0.8521, + "H_a": 0.098, + "d_bind": 0.01924 + } + ] + }, + { + "step": 8240, + "tok": 33751040, + "ce": 3.1036, + "ppl": 22.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.2146, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.8873, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8919, + "H_a": 0.109, + "d_bind": 0.03228 + }, + { + "tau": 0.023, + "v": 0.8814, + "H_a": 0.092, + "d_bind": 0.01888 + } + ] + }, + { + "step": 8260, + "tok": 33832960, + "ce": 3.3598, + "ppl": 28.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22427, + "peak_MiB": 5301, + "v_gain": 0.2182, + "loop_stats": [ + { + "tau": 0.0879, + "v": 0.8732, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0208, + "v": 0.8849, + "H_a": 0.11, + "d_bind": 0.03502 + }, + { + "tau": 0.0176, + "v": 0.8835, + "H_a": 0.087, + "d_bind": 0.01818 + } + ] + }, + { + "step": 8280, + "tok": 33914880, + "ce": 2.8476, + "ppl": 17.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.2196, + "loop_stats": [ + { + "tau": 0.1259, + "v": 0.8862, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.8906, + "H_a": 0.099, + "d_bind": 0.03866 + }, + { + "tau": 0.0451, + "v": 0.8877, + "H_a": 0.08, + "d_bind": 0.01632 + } + ] + }, + { + "step": 8300, + "tok": 33996800, + "ce": 3.3091, + "ppl": 27.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2173, + "loop_stats": [ + { + "tau": 0.1159, + "v": 0.8689, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0434, + "v": 0.8678, + "H_a": 0.103, + "d_bind": 0.03251 + }, + { + "tau": 0.0375, + "v": 0.8747, + "H_a": 0.089, + "d_bind": 0.01822 + } + ] + }, + { + "step": 8320, + "tok": 34078720, + "ce": 3.5713, + "ppl": 35.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2155, + "loop_stats": [ + { + "tau": 0.1077, + "v": 0.895, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.9014, + "H_a": 0.107, + "d_bind": 0.03187 + }, + { + "tau": 0.0253, + "v": 0.9111, + "H_a": 0.095, + "d_bind": 0.02022 + } + ] + }, + { + "step": 8340, + "tok": 34160640, + "ce": 3.2719, + "ppl": 26.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.935, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2172, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.8637, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.8596, + "H_a": 0.103, + "d_bind": 0.03659 + }, + { + "tau": 0.0366, + "v": 0.8658, + "H_a": 0.085, + "d_bind": 0.01804 + } + ] + }, + { + "step": 8360, + "tok": 34242560, + "ce": 3.6568, + "ppl": 38.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.933, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2181, + "loop_stats": [ + { + "tau": 0.0819, + "v": 0.8764, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.023, + "v": 0.8631, + "H_a": 0.113, + "d_bind": 0.03464 + }, + { + "tau": 0.0215, + "v": 0.8662, + "H_a": 0.101, + "d_bind": 0.02127 + } + ] + }, + { + "step": 8380, + "tok": 34324480, + "ce": 2.695, + "ppl": 14.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.927, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.2181, + "loop_stats": [ + { + "tau": 0.1481, + "v": 0.8885, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.859, + "H_a": 0.109, + "d_bind": 0.03776 + }, + { + "tau": 0.0445, + "v": 0.8573, + "H_a": 0.087, + "d_bind": 0.0191 + } + ] + }, + { + "step": 8400, + "tok": 34406400, + "ce": 3.49, + "ppl": 32.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.922, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.2145, + "loop_stats": [ + { + "tau": 0.0908, + "v": 0.8412, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0302, + "v": 0.8446, + "H_a": 0.115, + "d_bind": 0.03726 + }, + { + "tau": 0.0259, + "v": 0.8564, + "H_a": 0.109, + "d_bind": 0.02179 + } + ] + }, + { + "step": 8420, + "tok": 34488320, + "ce": 3.1603, + "ppl": 23.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.927, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2123, + "loop_stats": [ + { + "tau": 0.0948, + "v": 0.8219, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8173, + "H_a": 0.102, + "d_bind": 0.03529 + }, + { + "tau": 0.0321, + "v": 0.8211, + "H_a": 0.093, + "d_bind": 0.02031 + } + ] + }, + { + "step": 8440, + "tok": 34570240, + "ce": 3.0802, + "ppl": 21.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2142, + "loop_stats": [ + { + "tau": 0.0916, + "v": 0.8766, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8685, + "H_a": 0.113, + "d_bind": 0.03836 + }, + { + "tau": 0.0326, + "v": 0.8773, + "H_a": 0.099, + "d_bind": 0.02172 + } + ] + }, + { + "step": 8460, + "tok": 34652160, + "ce": 2.8722, + "ppl": 17.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.22, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.892, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0278, + "v": 0.8917, + "H_a": 0.109, + "d_bind": 0.03482 + }, + { + "tau": 0.0269, + "v": 0.9, + "H_a": 0.097, + "d_bind": 0.02071 + } + ] + }, + { + "step": 8480, + "tok": 34734080, + "ce": 3.0131, + "ppl": 20.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2202, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.8736, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0533, + "v": 0.8505, + "H_a": 0.101, + "d_bind": 0.03783 + }, + { + "tau": 0.0474, + "v": 0.8514, + "H_a": 0.089, + "d_bind": 0.01888 + } + ] + }, + { + "step": 8500, + "tok": 34816000, + "ce": 3.413, + "ppl": 30.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.22, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.8931, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8799, + "H_a": 0.108, + "d_bind": 0.0357 + }, + { + "tau": 0.0379, + "v": 0.8819, + "H_a": 0.091, + "d_bind": 0.01962 + } + ] + }, + { + "step": 8520, + "tok": 34897920, + "ce": 3.3821, + "ppl": 29.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.933, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.2227, + "loop_stats": [ + { + "tau": 0.0994, + "v": 0.8647, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0384, + "v": 0.8544, + "H_a": 0.118, + "d_bind": 0.03361 + }, + { + "tau": 0.0339, + "v": 0.8635, + "H_a": 0.091, + "d_bind": 0.01961 + } + ] + }, + { + "step": 8540, + "tok": 34979840, + "ce": 3.0136, + "ppl": 20.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2254, + "loop_stats": [ + { + "tau": 0.0957, + "v": 0.8948, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0305, + "v": 0.8917, + "H_a": 0.096, + "d_bind": 0.03346 + }, + { + "tau": 0.0279, + "v": 0.8934, + "H_a": 0.086, + "d_bind": 0.01899 + } + ] + }, + { + "step": 8560, + "tok": 35061760, + "ce": 2.9537, + "ppl": 19.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.942, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2214, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.8742, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8626, + "H_a": 0.102, + "d_bind": 0.03451 + }, + { + "tau": 0.0345, + "v": 0.856, + "H_a": 0.094, + "d_bind": 0.01953 + } + ] + }, + { + "step": 8580, + "tok": 35143680, + "ce": 2.7919, + "ppl": 16.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.224, + "loop_stats": [ + { + "tau": 0.1134, + "v": 0.9047, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.9037, + "H_a": 0.091, + "d_bind": 0.03519 + }, + { + "tau": 0.0361, + "v": 0.9117, + "H_a": 0.079, + "d_bind": 0.01779 + } + ] + }, + { + "step": 8600, + "tok": 35225600, + "ce": 3.1683, + "ppl": 23.8, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2254, + "loop_stats": [ + { + "tau": 0.1233, + "v": 0.8963, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8892, + "H_a": 0.104, + "d_bind": 0.03476 + }, + { + "tau": 0.0401, + "v": 0.8828, + "H_a": 0.099, + "d_bind": 0.01973 + } + ] + }, + { + "step": 8620, + "tok": 35307520, + "ce": 3.1297, + "ppl": 22.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.2249, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.8813, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8859, + "H_a": 0.103, + "d_bind": 0.03488 + }, + { + "tau": 0.0307, + "v": 0.8889, + "H_a": 0.095, + "d_bind": 0.02009 + } + ] + }, + { + "step": 8640, + "tok": 35389440, + "ce": 3.4472, + "ppl": 31.4, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.93, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.2223, + "loop_stats": [ + { + "tau": 0.092, + "v": 0.8728, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.8768, + "H_a": 0.105, + "d_bind": 0.03363 + }, + { + "tau": 0.0324, + "v": 0.8768, + "H_a": 0.096, + "d_bind": 0.01956 + } + ] + }, + { + "step": 8660, + "tok": 35471360, + "ce": 3.5731, + "ppl": 35.6, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.92, + "tok_s": 22417, + "peak_MiB": 5301, + "v_gain": 0.219, + "loop_stats": [ + { + "tau": 0.0865, + "v": 0.8851, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0196, + "v": 0.8983, + "H_a": 0.108, + "d_bind": 0.03456 + }, + { + "tau": 0.0157, + "v": 0.9169, + "H_a": 0.093, + "d_bind": 0.02106 + } + ] + }, + { + "step": 8680, + "tok": 35553280, + "ce": 2.9283, + "ppl": 18.7, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.94, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2225, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.889, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0291, + "v": 0.8994, + "H_a": 0.103, + "d_bind": 0.03521 + }, + { + "tau": 0.0245, + "v": 0.9112, + "H_a": 0.09, + "d_bind": 0.0199 + } + ] + }, + { + "step": 8700, + "tok": 35635200, + "ce": 2.9844, + "ppl": 19.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2245, + "loop_stats": [ + { + "tau": 0.1294, + "v": 0.8727, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8836, + "H_a": 0.103, + "d_bind": 0.03535 + }, + { + "tau": 0.0347, + "v": 0.8911, + "H_a": 0.091, + "d_bind": 0.02006 + } + ] + }, + { + "step": 8720, + "tok": 35717120, + "ce": 3.3842, + "ppl": 29.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.922, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2246, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.8826, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0276, + "v": 0.8886, + "H_a": 0.118, + "d_bind": 0.03387 + }, + { + "tau": 0.0246, + "v": 0.8923, + "H_a": 0.101, + "d_bind": 0.02121 + } + ] + }, + { + "step": 8740, + "tok": 35799040, + "ce": 3.5133, + "ppl": 33.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.2251, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8574, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0273, + "v": 0.8512, + "H_a": 0.111, + "d_bind": 0.03348 + }, + { + "tau": 0.0236, + "v": 0.8607, + "H_a": 0.094, + "d_bind": 0.02269 + } + ] + }, + { + "step": 8760, + "tok": 35880960, + "ce": 3.6428, + "ppl": 38.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.226, + "loop_stats": [ + { + "tau": 0.0607, + "v": 0.8362, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0169, + "v": 0.8392, + "H_a": 0.113, + "d_bind": 0.03487 + }, + { + "tau": 0.0151, + "v": 0.846, + "H_a": 0.105, + "d_bind": 0.02118 + } + ] + }, + { + "step": 8780, + "tok": 35962880, + "ce": 3.1032, + "ppl": 22.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2279, + "loop_stats": [ + { + "tau": 0.1195, + "v": 0.8766, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8809, + "H_a": 0.098, + "d_bind": 0.03389 + }, + { + "tau": 0.0405, + "v": 0.8832, + "H_a": 0.084, + "d_bind": 0.02039 + } + ] + }, + { + "step": 8800, + "tok": 36044800, + "ce": 2.4359, + "ppl": 11.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22419, + "peak_MiB": 5301, + "v_gain": 0.2301, + "loop_stats": [ + { + "tau": 0.1547, + "v": 0.8702, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0614, + "v": 0.8724, + "H_a": 0.097, + "d_bind": 0.03302 + }, + { + "tau": 0.0597, + "v": 0.8729, + "H_a": 0.086, + "d_bind": 0.02064 + } + ] + }, + { + "step": 8820, + "tok": 36126720, + "ce": 3.4563, + "ppl": 31.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.935, + "tok_s": 22421, + "peak_MiB": 5301, + "v_gain": 0.2307, + "loop_stats": [ + { + "tau": 0.0909, + "v": 0.8462, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0302, + "v": 0.846, + "H_a": 0.107, + "d_bind": 0.03282 + }, + { + "tau": 0.0287, + "v": 0.8531, + "H_a": 0.098, + "d_bind": 0.02183 + } + ] + }, + { + "step": 8840, + "tok": 36208640, + "ce": 3.029, + "ppl": 20.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2298, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.8576, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0363, + "v": 0.843, + "H_a": 0.104, + "d_bind": 0.03217 + }, + { + "tau": 0.0353, + "v": 0.8475, + "H_a": 0.091, + "d_bind": 0.02022 + } + ] + }, + { + "step": 8860, + "tok": 36290560, + "ce": 3.3014, + "ppl": 27.2, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.94, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.23, + "loop_stats": [ + { + "tau": 0.1254, + "v": 0.8688, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8626, + "H_a": 0.098, + "d_bind": 0.03288 + }, + { + "tau": 0.0434, + "v": 0.8644, + "H_a": 0.085, + "d_bind": 0.01975 + } + ] + }, + { + "step": 8880, + "tok": 36372480, + "ce": 3.3454, + "ppl": 28.4, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.942, + "tok_s": 22416, + "peak_MiB": 5301, + "v_gain": 0.2326, + "loop_stats": [ + { + "tau": 0.0894, + "v": 0.8805, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8851, + "H_a": 0.102, + "d_bind": 0.03239 + }, + { + "tau": 0.0291, + "v": 0.8942, + "H_a": 0.09, + "d_bind": 0.0205 + } + ] + }, + { + "step": 8900, + "tok": 36454400, + "ce": 2.9337, + "ppl": 18.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2325, + "loop_stats": [ + { + "tau": 0.1297, + "v": 0.8463, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8237, + "H_a": 0.105, + "d_bind": 0.03329 + }, + { + "tau": 0.0324, + "v": 0.8277, + "H_a": 0.093, + "d_bind": 0.02193 + } + ] + }, + { + "step": 8920, + "tok": 36536320, + "ce": 3.5138, + "ppl": 33.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2343, + "loop_stats": [ + { + "tau": 0.1003, + "v": 0.8363, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0245, + "v": 0.8399, + "H_a": 0.108, + "d_bind": 0.03399 + }, + { + "tau": 0.0211, + "v": 0.8562, + "H_a": 0.091, + "d_bind": 0.0205 + } + ] + }, + { + "step": 8940, + "tok": 36618240, + "ce": 3.2899, + "ppl": 26.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2347, + "loop_stats": [ + { + "tau": 0.1006, + "v": 0.8698, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.023, + "v": 0.8719, + "H_a": 0.109, + "d_bind": 0.03319 + }, + { + "tau": 0.0228, + "v": 0.8693, + "H_a": 0.094, + "d_bind": 0.0211 + } + ] + }, + { + "step": 8960, + "tok": 36700160, + "ce": 3.039, + "ppl": 20.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.945, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2319, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.8474, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.8368, + "H_a": 0.108, + "d_bind": 0.04045 + }, + { + "tau": 0.0356, + "v": 0.8431, + "H_a": 0.087, + "d_bind": 0.02067 + } + ] + }, + { + "step": 8980, + "tok": 36782080, + "ce": 2.6731, + "ppl": 14.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22427, + "peak_MiB": 5301, + "v_gain": 0.2267, + "loop_stats": [ + { + "tau": 0.1401, + "v": 0.8808, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8539, + "H_a": 0.104, + "d_bind": 0.03754 + }, + { + "tau": 0.0377, + "v": 0.8568, + "H_a": 0.072, + "d_bind": 0.019 + } + ] + }, + { + "step": 9000, + "tok": 36864000, + "ce": 2.882, + "ppl": 17.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22418, + "peak_MiB": 5301, + "v_gain": 0.2254, + "loop_stats": [ + { + "tau": 0.1323, + "v": 0.8644, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8502, + "H_a": 0.105, + "d_bind": 0.03661 + }, + { + "tau": 0.0414, + "v": 0.8594, + "H_a": 0.09, + "d_bind": 0.02145 + } + ] + }, + { + "step": 9020, + "tok": 36945920, + "ce": 2.8964, + "ppl": 18.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2279, + "loop_stats": [ + { + "tau": 0.1247, + "v": 0.8396, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8239, + "H_a": 0.099, + "d_bind": 0.03471 + }, + { + "tau": 0.0441, + "v": 0.8283, + "H_a": 0.081, + "d_bind": 0.02045 + } + ] + }, + { + "step": 9040, + "tok": 37027840, + "ce": 2.9332, + "ppl": 18.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2314, + "loop_stats": [ + { + "tau": 0.1232, + "v": 0.9053, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8972, + "H_a": 0.112, + "d_bind": 0.03868 + }, + { + "tau": 0.0355, + "v": 0.8921, + "H_a": 0.093, + "d_bind": 0.02131 + } + ] + }, + { + "step": 9060, + "tok": 37109760, + "ce": 3.4562, + "ppl": 31.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.2321, + "loop_stats": [ + { + "tau": 0.0852, + "v": 0.8947, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0187, + "v": 0.8823, + "H_a": 0.114, + "d_bind": 0.03972 + }, + { + "tau": 0.0162, + "v": 0.8907, + "H_a": 0.09, + "d_bind": 0.02221 + } + ] + }, + { + "step": 9080, + "tok": 37191680, + "ce": 2.6315, + "ppl": 13.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22432, + "peak_MiB": 5301, + "v_gain": 0.2294, + "loop_stats": [ + { + "tau": 0.1326, + "v": 0.9227, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.042, + "v": 0.9017, + "H_a": 0.101, + "d_bind": 0.03675 + }, + { + "tau": 0.0445, + "v": 0.8901, + "H_a": 0.078, + "d_bind": 0.02078 + } + ] + }, + { + "step": 9100, + "tok": 37273600, + "ce": 3.1372, + "ppl": 23.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2293, + "loop_stats": [ + { + "tau": 0.086, + "v": 0.8865, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0322, + "v": 0.8693, + "H_a": 0.113, + "d_bind": 0.03845 + }, + { + "tau": 0.0324, + "v": 0.8675, + "H_a": 0.087, + "d_bind": 0.02012 + } + ] + }, + { + "step": 9120, + "tok": 37355520, + "ce": 3.1591, + "ppl": 23.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.93, + "tok_s": 22420, + "peak_MiB": 5301, + "v_gain": 0.2264, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.9017, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8854, + "H_a": 0.109, + "d_bind": 0.03831 + }, + { + "tau": 0.0406, + "v": 0.881, + "H_a": 0.082, + "d_bind": 0.01864 + } + ] + }, + { + "step": 9140, + "tok": 37437440, + "ce": 3.5316, + "ppl": 34.2, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2254, + "loop_stats": [ + { + "tau": 0.0809, + "v": 0.878, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0202, + "v": 0.8856, + "H_a": 0.118, + "d_bind": 0.03548 + }, + { + "tau": 0.0185, + "v": 0.8945, + "H_a": 0.097, + "d_bind": 0.02161 + } + ] + }, + { + "step": 9160, + "tok": 37519360, + "ce": 3.4318, + "ppl": 30.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.2261, + "loop_stats": [ + { + "tau": 0.0887, + "v": 0.9083, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.8983, + "H_a": 0.108, + "d_bind": 0.03772 + }, + { + "tau": 0.0279, + "v": 0.9033, + "H_a": 0.097, + "d_bind": 0.02024 + } + ] + }, + { + "step": 9180, + "tok": 37601280, + "ce": 2.6857, + "ppl": 14.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.933, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.2279, + "loop_stats": [ + { + "tau": 0.1381, + "v": 0.9108, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8829, + "H_a": 0.115, + "d_bind": 0.03847 + }, + { + "tau": 0.0461, + "v": 0.8778, + "H_a": 0.086, + "d_bind": 0.01974 + } + ] + }, + { + "step": 9200, + "tok": 37683200, + "ce": 3.0163, + "ppl": 20.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22434, + "peak_MiB": 5301, + "v_gain": 0.2269, + "loop_stats": [ + { + "tau": 0.1196, + "v": 0.8843, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8645, + "H_a": 0.103, + "d_bind": 0.03737 + }, + { + "tau": 0.0332, + "v": 0.8672, + "H_a": 0.081, + "d_bind": 0.01843 + } + ] + }, + { + "step": 9220, + "tok": 37765120, + "ce": 3.0684, + "ppl": 21.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22431, + "peak_MiB": 5301, + "v_gain": 0.2245, + "loop_stats": [ + { + "tau": 0.111, + "v": 0.9125, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0572, + "v": 0.8834, + "H_a": 0.11, + "d_bind": 0.0398 + }, + { + "tau": 0.0558, + "v": 0.8805, + "H_a": 0.086, + "d_bind": 0.01896 + } + ] + }, + { + "step": 9240, + "tok": 37847040, + "ce": 2.8269, + "ppl": 16.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.962, + "tok_s": 22422, + "peak_MiB": 5301, + "v_gain": 0.2264, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.9186, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.056, + "v": 0.8979, + "H_a": 0.111, + "d_bind": 0.03523 + }, + { + "tau": 0.054, + "v": 0.8872, + "H_a": 0.088, + "d_bind": 0.02116 + } + ] + }, + { + "step": 9260, + "tok": 37928960, + "ce": 3.4626, + "ppl": 31.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22424, + "peak_MiB": 5301, + "v_gain": 0.2252, + "loop_stats": [ + { + "tau": 0.1211, + "v": 0.9251, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0421, + "v": 0.9118, + "H_a": 0.105, + "d_bind": 0.03708 + }, + { + "tau": 0.0383, + "v": 0.9087, + "H_a": 0.09, + "d_bind": 0.01954 + } + ] + }, + { + "step": 9280, + "tok": 38010880, + "ce": 3.2374, + "ppl": 25.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.227, + "loop_stats": [ + { + "tau": 0.1088, + "v": 0.9305, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.919, + "H_a": 0.095, + "d_bind": 0.03692 + }, + { + "tau": 0.0426, + "v": 0.9129, + "H_a": 0.084, + "d_bind": 0.0176 + } + ] + }, + { + "step": 9300, + "tok": 38092800, + "ce": 3.3679, + "ppl": 29.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2267, + "loop_stats": [ + { + "tau": 0.0812, + "v": 0.9282, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.9265, + "H_a": 0.106, + "d_bind": 0.03699 + }, + { + "tau": 0.0326, + "v": 0.9286, + "H_a": 0.094, + "d_bind": 0.01877 + } + ] + }, + { + "step": 9320, + "tok": 38174720, + "ce": 2.6975, + "ppl": 14.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22436, + "peak_MiB": 5301, + "v_gain": 0.2271, + "loop_stats": [ + { + "tau": 0.1361, + "v": 0.929, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0484, + "v": 0.9148, + "H_a": 0.116, + "d_bind": 0.04069 + }, + { + "tau": 0.0449, + "v": 0.908, + "H_a": 0.083, + "d_bind": 0.02131 + } + ] + }, + { + "step": 9340, + "tok": 38256640, + "ce": 3.1074, + "ppl": 22.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.2264, + "loop_stats": [ + { + "tau": 0.1194, + "v": 0.8956, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8702, + "H_a": 0.106, + "d_bind": 0.03771 + }, + { + "tau": 0.0455, + "v": 0.8675, + "H_a": 0.082, + "d_bind": 0.01875 + } + ] + }, + { + "step": 9360, + "tok": 38338560, + "ce": 3.0657, + "ppl": 21.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22425, + "peak_MiB": 5301, + "v_gain": 0.2244, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.9149, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8971, + "H_a": 0.108, + "d_bind": 0.03869 + }, + { + "tau": 0.042, + "v": 0.8934, + "H_a": 0.088, + "d_bind": 0.01989 + } + ] + }, + { + "step": 9380, + "tok": 38420480, + "ce": 3.0719, + "ppl": 21.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.942, + "tok_s": 22427, + "peak_MiB": 5301, + "v_gain": 0.2265, + "loop_stats": [ + { + "tau": 0.1252, + "v": 0.9055, + "H_a": 0.137, + "d_bind": 0.0 + }, + { + "tau": 0.0474, + "v": 0.8864, + "H_a": 0.107, + "d_bind": 0.03851 + }, + { + "tau": 0.0435, + "v": 0.8809, + "H_a": 0.087, + "d_bind": 0.01945 + } + ] + }, + { + "step": 9400, + "tok": 38502400, + "ce": 3.6156, + "ppl": 37.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.93, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2282, + "loop_stats": [ + { + "tau": 0.0836, + "v": 0.9006, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0233, + "v": 0.8917, + "H_a": 0.1, + "d_bind": 0.03194 + }, + { + "tau": 0.0182, + "v": 0.8931, + "H_a": 0.09, + "d_bind": 0.01925 + } + ] + }, + { + "step": 9420, + "tok": 38584320, + "ce": 2.9192, + "ppl": 18.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.2308, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.9177, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8997, + "H_a": 0.102, + "d_bind": 0.03753 + }, + { + "tau": 0.0386, + "v": 0.9011, + "H_a": 0.084, + "d_bind": 0.01829 + } + ] + }, + { + "step": 9440, + "tok": 38666240, + "ce": 3.4393, + "ppl": 31.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22438, + "peak_MiB": 5301, + "v_gain": 0.234, + "loop_stats": [ + { + "tau": 0.0898, + "v": 0.929, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0246, + "v": 0.9163, + "H_a": 0.108, + "d_bind": 0.03782 + }, + { + "tau": 0.0234, + "v": 0.9079, + "H_a": 0.094, + "d_bind": 0.02075 + } + ] + }, + { + "step": 9460, + "tok": 38748160, + "ce": 3.2418, + "ppl": 25.6, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22435, + "peak_MiB": 5301, + "v_gain": 0.2328, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.8976, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0288, + "v": 0.8737, + "H_a": 0.104, + "d_bind": 0.0376 + }, + { + "tau": 0.0253, + "v": 0.87, + "H_a": 0.082, + "d_bind": 0.01893 + } + ] + }, + { + "step": 9480, + "tok": 38830080, + "ce": 3.343, + "ppl": 28.3, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.94, + "tok_s": 22426, + "peak_MiB": 5301, + "v_gain": 0.2327, + "loop_stats": [ + { + "tau": 0.0821, + "v": 0.9072, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0204, + "v": 0.8969, + "H_a": 0.103, + "d_bind": 0.03673 + }, + { + "tau": 0.0181, + "v": 0.8974, + "H_a": 0.088, + "d_bind": 0.01986 + } + ] + }, + { + "step": 9500, + "tok": 38912000, + "ce": 2.8409, + "ppl": 17.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.2343, + "loop_stats": [ + { + "tau": 0.119, + "v": 0.9188, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.8987, + "H_a": 0.087, + "d_bind": 0.03959 + }, + { + "tau": 0.0481, + "v": 0.8969, + "H_a": 0.071, + "d_bind": 0.0166 + } + ] + }, + { + "step": 9520, + "tok": 38993920, + "ce": 3.1128, + "ppl": 22.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.2341, + "loop_stats": [ + { + "tau": 0.1268, + "v": 0.8901, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8722, + "H_a": 0.099, + "d_bind": 0.03646 + }, + { + "tau": 0.0373, + "v": 0.8716, + "H_a": 0.081, + "d_bind": 0.01806 + } + ] + }, + { + "step": 9540, + "tok": 39075840, + "ce": 3.005, + "ppl": 20.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.927, + "tok_s": 22431, + "peak_MiB": 5301, + "v_gain": 0.2308, + "loop_stats": [ + { + "tau": 0.1308, + "v": 0.8961, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8563, + "H_a": 0.102, + "d_bind": 0.03661 + }, + { + "tau": 0.0437, + "v": 0.8522, + "H_a": 0.079, + "d_bind": 0.01932 + } + ] + }, + { + "step": 9560, + "tok": 39157760, + "ce": 2.9339, + "ppl": 18.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22438, + "peak_MiB": 5301, + "v_gain": 0.2303, + "loop_stats": [ + { + "tau": 0.1143, + "v": 0.8839, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.8739, + "H_a": 0.105, + "d_bind": 0.03852 + }, + { + "tau": 0.0373, + "v": 0.8781, + "H_a": 0.082, + "d_bind": 0.01862 + } + ] + }, + { + "step": 9580, + "tok": 39239680, + "ce": 3.0958, + "ppl": 22.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22436, + "peak_MiB": 5301, + "v_gain": 0.2315, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.9001, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0354, + "v": 0.8757, + "H_a": 0.103, + "d_bind": 0.03932 + }, + { + "tau": 0.032, + "v": 0.8805, + "H_a": 0.086, + "d_bind": 0.02015 + } + ] + }, + { + "step": 9600, + "tok": 39321600, + "ce": 3.3679, + "ppl": 29.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.933, + "tok_s": 22428, + "peak_MiB": 5301, + "v_gain": 0.226, + "loop_stats": [ + { + "tau": 0.1215, + "v": 0.9025, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8861, + "H_a": 0.094, + "d_bind": 0.03358 + }, + { + "tau": 0.0289, + "v": 0.8863, + "H_a": 0.085, + "d_bind": 0.01787 + } + ] + }, + { + "step": 9620, + "tok": 39403520, + "ce": 2.564, + "ppl": 13.0, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.938, + "tok_s": 22429, + "peak_MiB": 5301, + "v_gain": 0.227, + "loop_stats": [ + { + "tau": 0.1412, + "v": 0.8876, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0662, + "v": 0.8578, + "H_a": 0.094, + "d_bind": 0.0418 + }, + { + "tau": 0.0623, + "v": 0.8594, + "H_a": 0.073, + "d_bind": 0.01561 + } + ] + }, + { + "step": 9640, + "tok": 39485440, + "ce": 3.6303, + "ppl": 37.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22431, + "peak_MiB": 5301, + "v_gain": 0.2295, + "loop_stats": [ + { + "tau": 0.0827, + "v": 0.8967, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.02, + "v": 0.8785, + "H_a": 0.104, + "d_bind": 0.03517 + }, + { + "tau": 0.0148, + "v": 0.8818, + "H_a": 0.089, + "d_bind": 0.01844 + } + ] + }, + { + "step": 9660, + "tok": 39567360, + "ce": 2.8371, + "ppl": 17.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.938, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.2305, + "loop_stats": [ + { + "tau": 0.1378, + "v": 0.9103, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0587, + "v": 0.8581, + "H_a": 0.1, + "d_bind": 0.04125 + }, + { + "tau": 0.0558, + "v": 0.8375, + "H_a": 0.084, + "d_bind": 0.01865 + } + ] + }, + { + "step": 9680, + "tok": 39649280, + "ce": 3.4273, + "ppl": 30.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22440, + "peak_MiB": 5301, + "v_gain": 0.2302, + "loop_stats": [ + { + "tau": 0.0963, + "v": 0.884, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8621, + "H_a": 0.1, + "d_bind": 0.03588 + }, + { + "tau": 0.0218, + "v": 0.8589, + "H_a": 0.085, + "d_bind": 0.01865 + } + ] + }, + { + "step": 9700, + "tok": 39731200, + "ce": 3.0702, + "ppl": 21.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22438, + "peak_MiB": 5301, + "v_gain": 0.2298, + "loop_stats": [ + { + "tau": 0.1055, + "v": 0.9172, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0465, + "v": 0.894, + "H_a": 0.101, + "d_bind": 0.0393 + }, + { + "tau": 0.0354, + "v": 0.8903, + "H_a": 0.081, + "d_bind": 0.01893 + } + ] + }, + { + "step": 9720, + "tok": 39813120, + "ce": 3.4689, + "ppl": 32.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22430, + "peak_MiB": 5301, + "v_gain": 0.233, + "loop_stats": [ + { + "tau": 0.0921, + "v": 0.9301, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.9227, + "H_a": 0.099, + "d_bind": 0.03479 + }, + { + "tau": 0.029, + "v": 0.9269, + "H_a": 0.088, + "d_bind": 0.01864 + } + ] + }, + { + "step": 9740, + "tok": 39895040, + "ce": 3.5114, + "ppl": 33.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22432, + "peak_MiB": 5301, + "v_gain": 0.2339, + "loop_stats": [ + { + "tau": 0.1025, + "v": 0.895, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0292, + "v": 0.875, + "H_a": 0.102, + "d_bind": 0.03519 + }, + { + "tau": 0.0234, + "v": 0.8787, + "H_a": 0.092, + "d_bind": 0.01828 + } + ] + }, + { + "step": 9760, + "tok": 39976960, + "ce": 3.557, + "ppl": 35.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.95, + "tok_s": 22433, + "peak_MiB": 5301, + "v_gain": 0.2335, + "loop_stats": [ + { + "tau": 0.0856, + "v": 0.9002, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.8962, + "H_a": 0.109, + "d_bind": 0.03833 + }, + { + "tau": 0.024, + "v": 0.9075, + "H_a": 0.087, + "d_bind": 0.01971 + } + ] + }, + { + "step": 9780, + "tok": 40058880, + "ce": 3.34, + "ppl": 28.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22435, + "peak_MiB": 5301, + "v_gain": 0.2324, + "loop_stats": [ + { + "tau": 0.0958, + "v": 0.9158, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.04, + "v": 0.9066, + "H_a": 0.101, + "d_bind": 0.03792 + }, + { + "tau": 0.0323, + "v": 0.9102, + "H_a": 0.082, + "d_bind": 0.01821 + } + ] + }, + { + "step": 9800, + "tok": 40140800, + "ce": 3.0061, + "ppl": 20.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22442, + "peak_MiB": 5301, + "v_gain": 0.2351, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.9124, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.8932, + "H_a": 0.103, + "d_bind": 0.03847 + }, + { + "tau": 0.0406, + "v": 0.8931, + "H_a": 0.08, + "d_bind": 0.01855 + } + ] + }, + { + "step": 9820, + "tok": 40222720, + "ce": 3.4437, + "ppl": 31.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22440, + "peak_MiB": 5301, + "v_gain": 0.2377, + "loop_stats": [ + { + "tau": 0.0984, + "v": 0.9058, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.8923, + "H_a": 0.104, + "d_bind": 0.03958 + }, + { + "tau": 0.0274, + "v": 0.896, + "H_a": 0.083, + "d_bind": 0.01782 + } + ] + }, + { + "step": 9840, + "tok": 40304640, + "ce": 3.2419, + "ppl": 25.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.947, + "tok_s": 22432, + "peak_MiB": 5301, + "v_gain": 0.2392, + "loop_stats": [ + { + "tau": 0.0991, + "v": 0.9183, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.8905, + "H_a": 0.098, + "d_bind": 0.03814 + }, + { + "tau": 0.031, + "v": 0.8857, + "H_a": 0.085, + "d_bind": 0.01861 + } + ] + }, + { + "step": 9860, + "tok": 40386560, + "ce": 2.898, + "ppl": 18.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22434, + "peak_MiB": 5301, + "v_gain": 0.2426, + "loop_stats": [ + { + "tau": 0.1407, + "v": 0.8823, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8515, + "H_a": 0.099, + "d_bind": 0.03903 + }, + { + "tau": 0.0378, + "v": 0.8552, + "H_a": 0.081, + "d_bind": 0.01708 + } + ] + }, + { + "step": 9880, + "tok": 40468480, + "ce": 2.8218, + "ppl": 16.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22435, + "peak_MiB": 5301, + "v_gain": 0.2436, + "loop_stats": [ + { + "tau": 0.1331, + "v": 0.888, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0456, + "v": 0.861, + "H_a": 0.094, + "d_bind": 0.03956 + }, + { + "tau": 0.0375, + "v": 0.8636, + "H_a": 0.076, + "d_bind": 0.01795 + } + ] + }, + { + "step": 9900, + "tok": 40550400, + "ce": 3.2465, + "ppl": 25.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22437, + "peak_MiB": 5301, + "v_gain": 0.241, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.9001, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.8877, + "H_a": 0.094, + "d_bind": 0.03474 + }, + { + "tau": 0.0355, + "v": 0.8901, + "H_a": 0.079, + "d_bind": 0.01765 + } + ] + }, + { + "step": 9920, + "tok": 40632320, + "ce": 3.2238, + "ppl": 25.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22444, + "peak_MiB": 5301, + "v_gain": 0.2415, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.8892, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0286, + "v": 0.8734, + "H_a": 0.1, + "d_bind": 0.03923 + }, + { + "tau": 0.0204, + "v": 0.877, + "H_a": 0.08, + "d_bind": 0.01742 + } + ] + }, + { + "step": 9940, + "tok": 40714240, + "ce": 3.4991, + "ppl": 33.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.938, + "tok_s": 22442, + "peak_MiB": 5301, + "v_gain": 0.2429, + "loop_stats": [ + { + "tau": 0.0822, + "v": 0.9108, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.8993, + "H_a": 0.103, + "d_bind": 0.03761 + }, + { + "tau": 0.0225, + "v": 0.9003, + "H_a": 0.09, + "d_bind": 0.01939 + } + ] + }, + { + "step": 9960, + "tok": 40796160, + "ce": 2.8219, + "ppl": 16.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22434, + "peak_MiB": 5301, + "v_gain": 0.2446, + "loop_stats": [ + { + "tau": 0.1161, + "v": 0.8935, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8692, + "H_a": 0.1, + "d_bind": 0.03431 + }, + { + "tau": 0.0271, + "v": 0.8697, + "H_a": 0.087, + "d_bind": 0.01866 + } + ] + }, + { + "step": 9980, + "tok": 40878080, + "ce": 3.0877, + "ppl": 21.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22435, + "peak_MiB": 5301, + "v_gain": 0.2423, + "loop_stats": [ + { + "tau": 0.0985, + "v": 0.8907, + "H_a": 0.121, + "d_bind": 0.0 + }, + { + "tau": 0.0277, + "v": 0.8768, + "H_a": 0.102, + "d_bind": 0.0354 + }, + { + "tau": 0.0197, + "v": 0.878, + "H_a": 0.081, + "d_bind": 0.0202 + } + ] + }, + { + "step": 10000, + "tok": 40960000, + "ce": 2.8186, + "ppl": 16.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22437, + "peak_MiB": 5301, + "v_gain": 0.2425, + "loop_stats": [ + { + "tau": 0.1218, + "v": 0.878, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.049, + "v": 0.8679, + "H_a": 0.097, + "d_bind": 0.03539 + }, + { + "tau": 0.0379, + "v": 0.873, + "H_a": 0.079, + "d_bind": 0.01886 + } + ] + }, + { + "step": 10020, + "tok": 41041920, + "ce": 2.5006, + "ppl": 12.2, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.94, + "tok_s": 22438, + "peak_MiB": 5301, + "v_gain": 0.2419, + "loop_stats": [ + { + "tau": 0.1463, + "v": 0.8728, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0701, + "v": 0.8597, + "H_a": 0.087, + "d_bind": 0.03665 + }, + { + "tau": 0.0575, + "v": 0.859, + "H_a": 0.068, + "d_bind": 0.01622 + } + ] + }, + { + "step": 10040, + "tok": 41123840, + "ce": 2.6424, + "ppl": 14.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.938, + "tok_s": 22445, + "peak_MiB": 5301, + "v_gain": 0.2402, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.9049, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8846, + "H_a": 0.102, + "d_bind": 0.03384 + }, + { + "tau": 0.0412, + "v": 0.8846, + "H_a": 0.073, + "d_bind": 0.01905 + } + ] + }, + { + "step": 10060, + "tok": 41205760, + "ce": 2.9538, + "ppl": 19.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.95, + "tok_s": 22443, + "peak_MiB": 5301, + "v_gain": 0.2409, + "loop_stats": [ + { + "tau": 0.1232, + "v": 0.8961, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0457, + "v": 0.8833, + "H_a": 0.092, + "d_bind": 0.0364 + }, + { + "tau": 0.0415, + "v": 0.888, + "H_a": 0.072, + "d_bind": 0.01795 + } + ] + }, + { + "step": 10080, + "tok": 41287680, + "ce": 2.8801, + "ppl": 17.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22435, + "peak_MiB": 5301, + "v_gain": 0.2412, + "loop_stats": [ + { + "tau": 0.1347, + "v": 0.896, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0586, + "v": 0.8618, + "H_a": 0.094, + "d_bind": 0.03322 + }, + { + "tau": 0.0533, + "v": 0.8591, + "H_a": 0.07, + "d_bind": 0.01859 + } + ] + }, + { + "step": 10100, + "tok": 41369600, + "ce": 3.4434, + "ppl": 31.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.933, + "tok_s": 22437, + "peak_MiB": 5301, + "v_gain": 0.2405, + "loop_stats": [ + { + "tau": 0.101, + "v": 0.9028, + "H_a": 0.129, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.8903, + "H_a": 0.106, + "d_bind": 0.03458 + }, + { + "tau": 0.0303, + "v": 0.8881, + "H_a": 0.08, + "d_bind": 0.01914 + } + ] + }, + { + "step": 10120, + "tok": 41451520, + "ce": 3.0026, + "ppl": 20.1, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.933, + "tok_s": 22439, + "peak_MiB": 5301, + "v_gain": 0.2428, + "loop_stats": [ + { + "tau": 0.1166, + "v": 0.9083, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0479, + "v": 0.8924, + "H_a": 0.095, + "d_bind": 0.0339 + }, + { + "tau": 0.0365, + "v": 0.8885, + "H_a": 0.081, + "d_bind": 0.01975 + } + ] + }, + { + "step": 10140, + "tok": 41533440, + "ce": 3.2585, + "ppl": 26.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.933, + "tok_s": 22440, + "peak_MiB": 5301, + "v_gain": 0.2462, + "loop_stats": [ + { + "tau": 0.085, + "v": 0.8704, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8767, + "H_a": 0.102, + "d_bind": 0.03384 + }, + { + "tau": 0.0246, + "v": 0.888, + "H_a": 0.087, + "d_bind": 0.01879 + } + ] + }, + { + "step": 10160, + "tok": 41615360, + "ce": 3.0181, + "ppl": 20.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.935, + "tok_s": 22447, + "peak_MiB": 5301, + "v_gain": 0.2505, + "loop_stats": [ + { + "tau": 0.104, + "v": 0.8863, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0445, + "v": 0.8683, + "H_a": 0.094, + "d_bind": 0.03321 + }, + { + "tau": 0.0356, + "v": 0.8688, + "H_a": 0.074, + "d_bind": 0.01705 + } + ] + }, + { + "step": 10180, + "tok": 41697280, + "ce": 3.385, + "ppl": 29.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22445, + "peak_MiB": 5301, + "v_gain": 0.2518, + "loop_stats": [ + { + "tau": 0.089, + "v": 0.9068, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.8896, + "H_a": 0.092, + "d_bind": 0.03324 + }, + { + "tau": 0.0336, + "v": 0.8881, + "H_a": 0.072, + "d_bind": 0.01699 + } + ] + }, + { + "step": 10200, + "tok": 41779200, + "ce": 3.2371, + "ppl": 25.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22437, + "peak_MiB": 5301, + "v_gain": 0.2519, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.8913, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.8819, + "H_a": 0.098, + "d_bind": 0.03408 + }, + { + "tau": 0.0416, + "v": 0.8875, + "H_a": 0.073, + "d_bind": 0.01973 + } + ] + }, + { + "step": 10220, + "tok": 41861120, + "ce": 3.3969, + "ppl": 29.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.942, + "tok_s": 22439, + "peak_MiB": 5301, + "v_gain": 0.2513, + "loop_stats": [ + { + "tau": 0.0912, + "v": 0.8821, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0332, + "v": 0.8875, + "H_a": 0.11, + "d_bind": 0.03539 + }, + { + "tau": 0.0316, + "v": 0.8946, + "H_a": 0.085, + "d_bind": 0.01978 + } + ] + }, + { + "step": 10240, + "tok": 41943040, + "ce": 3.1845, + "ppl": 24.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22440, + "peak_MiB": 5301, + "v_gain": 0.2455, + "loop_stats": [ + { + "tau": 0.1183, + "v": 0.876, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8563, + "H_a": 0.092, + "d_bind": 0.03457 + }, + { + "tau": 0.0489, + "v": 0.852, + "H_a": 0.081, + "d_bind": 0.01698 + } + ] + }, + { + "step": 10260, + "tok": 42024960, + "ce": 2.9609, + "ppl": 19.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22442, + "peak_MiB": 5301, + "v_gain": 0.2472, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.9035, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8812, + "H_a": 0.094, + "d_bind": 0.03774 + }, + { + "tau": 0.0386, + "v": 0.8771, + "H_a": 0.072, + "d_bind": 0.01761 + } + ] + }, + { + "step": 10280, + "tok": 42106880, + "ce": 3.3183, + "ppl": 27.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2493, + "loop_stats": [ + { + "tau": 0.1287, + "v": 0.91, + "H_a": 0.121, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.8914, + "H_a": 0.099, + "d_bind": 0.03636 + }, + { + "tau": 0.0372, + "v": 0.8943, + "H_a": 0.084, + "d_bind": 0.01897 + } + ] + }, + { + "step": 10300, + "tok": 42188800, + "ce": 3.2875, + "ppl": 26.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22447, + "peak_MiB": 5301, + "v_gain": 0.2506, + "loop_stats": [ + { + "tau": 0.1004, + "v": 0.897, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.8708, + "H_a": 0.098, + "d_bind": 0.03668 + }, + { + "tau": 0.0369, + "v": 0.8655, + "H_a": 0.083, + "d_bind": 0.01854 + } + ] + }, + { + "step": 10320, + "tok": 42270720, + "ce": 2.9931, + "ppl": 19.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22439, + "peak_MiB": 5301, + "v_gain": 0.2495, + "loop_stats": [ + { + "tau": 0.1072, + "v": 0.8503, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8372, + "H_a": 0.099, + "d_bind": 0.03629 + }, + { + "tau": 0.041, + "v": 0.8447, + "H_a": 0.079, + "d_bind": 0.01802 + } + ] + }, + { + "step": 10340, + "tok": 42352640, + "ce": 3.0638, + "ppl": 21.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22441, + "peak_MiB": 5301, + "v_gain": 0.2478, + "loop_stats": [ + { + "tau": 0.1403, + "v": 0.8857, + "H_a": 0.136, + "d_bind": 0.0 + }, + { + "tau": 0.0645, + "v": 0.8535, + "H_a": 0.092, + "d_bind": 0.03754 + }, + { + "tau": 0.0577, + "v": 0.8525, + "H_a": 0.078, + "d_bind": 0.01705 + } + ] + }, + { + "step": 10360, + "tok": 42434560, + "ce": 3.2158, + "ppl": 24.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22442, + "peak_MiB": 5301, + "v_gain": 0.2483, + "loop_stats": [ + { + "tau": 0.0912, + "v": 0.8873, + "H_a": 0.133, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8776, + "H_a": 0.1, + "d_bind": 0.03712 + }, + { + "tau": 0.0267, + "v": 0.8793, + "H_a": 0.08, + "d_bind": 0.01946 + } + ] + }, + { + "step": 10380, + "tok": 42516480, + "ce": 2.9481, + "ppl": 19.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22444, + "peak_MiB": 5301, + "v_gain": 0.2466, + "loop_stats": [ + { + "tau": 0.1184, + "v": 0.8735, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0554, + "v": 0.8631, + "H_a": 0.093, + "d_bind": 0.03698 + }, + { + "tau": 0.0491, + "v": 0.8628, + "H_a": 0.072, + "d_bind": 0.01788 + } + ] + }, + { + "step": 10400, + "tok": 42598400, + "ce": 3.0877, + "ppl": 21.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2465, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8927, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8843, + "H_a": 0.09, + "d_bind": 0.03788 + }, + { + "tau": 0.0305, + "v": 0.8827, + "H_a": 0.079, + "d_bind": 0.01899 + } + ] + }, + { + "step": 10420, + "tok": 42680320, + "ce": 3.2818, + "ppl": 26.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2492, + "loop_stats": [ + { + "tau": 0.0969, + "v": 0.9142, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0385, + "v": 0.9106, + "H_a": 0.105, + "d_bind": 0.03688 + }, + { + "tau": 0.027, + "v": 0.9113, + "H_a": 0.087, + "d_bind": 0.01883 + } + ] + }, + { + "step": 10440, + "tok": 42762240, + "ce": 3.4472, + "ppl": 31.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22441, + "peak_MiB": 5301, + "v_gain": 0.2499, + "loop_stats": [ + { + "tau": 0.0855, + "v": 0.8925, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.0214, + "v": 0.8848, + "H_a": 0.104, + "d_bind": 0.03796 + }, + { + "tau": 0.0168, + "v": 0.8884, + "H_a": 0.081, + "d_bind": 0.02011 + } + ] + }, + { + "step": 10460, + "tok": 42844160, + "ce": 2.4486, + "ppl": 11.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22443, + "peak_MiB": 5301, + "v_gain": 0.2486, + "loop_stats": [ + { + "tau": 0.1498, + "v": 0.8988, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0625, + "v": 0.8696, + "H_a": 0.09, + "d_bind": 0.04396 + }, + { + "tau": 0.0565, + "v": 0.8756, + "H_a": 0.069, + "d_bind": 0.01651 + } + ] + }, + { + "step": 10480, + "tok": 42926080, + "ce": 3.2463, + "ppl": 25.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22444, + "peak_MiB": 5301, + "v_gain": 0.246, + "loop_stats": [ + { + "tau": 0.0968, + "v": 0.8837, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0476, + "v": 0.8666, + "H_a": 0.106, + "d_bind": 0.04141 + }, + { + "tau": 0.0412, + "v": 0.8615, + "H_a": 0.082, + "d_bind": 0.01783 + } + ] + }, + { + "step": 10500, + "tok": 43008000, + "ce": 3.2323, + "ppl": 25.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22446, + "peak_MiB": 5301, + "v_gain": 0.2457, + "loop_stats": [ + { + "tau": 0.0615, + "v": 0.8808, + "H_a": 0.132, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.8906, + "H_a": 0.11, + "d_bind": 0.03683 + }, + { + "tau": 0.0339, + "v": 0.898, + "H_a": 0.09, + "d_bind": 0.02133 + } + ] + }, + { + "step": 10520, + "tok": 43089920, + "ce": 3.1266, + "ppl": 22.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22452, + "peak_MiB": 5301, + "v_gain": 0.2475, + "loop_stats": [ + { + "tau": 0.1136, + "v": 0.8806, + "H_a": 0.134, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8825, + "H_a": 0.095, + "d_bind": 0.03929 + }, + { + "tau": 0.0462, + "v": 0.891, + "H_a": 0.087, + "d_bind": 0.01845 + } + ] + }, + { + "step": 10540, + "tok": 43171840, + "ce": 3.0886, + "ppl": 21.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2457, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.8991, + "H_a": 0.141, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8842, + "H_a": 0.096, + "d_bind": 0.04043 + }, + { + "tau": 0.0398, + "v": 0.8881, + "H_a": 0.08, + "d_bind": 0.01733 + } + ] + }, + { + "step": 10560, + "tok": 43253760, + "ce": 3.0056, + "ppl": 20.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22443, + "peak_MiB": 5301, + "v_gain": 0.2456, + "loop_stats": [ + { + "tau": 0.1229, + "v": 0.9184, + "H_a": 0.132, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.8909, + "H_a": 0.094, + "d_bind": 0.04179 + }, + { + "tau": 0.0459, + "v": 0.8876, + "H_a": 0.078, + "d_bind": 0.0174 + } + ] + }, + { + "step": 10580, + "tok": 43335680, + "ce": 3.6732, + "ppl": 39.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.938, + "tok_s": 22444, + "peak_MiB": 5301, + "v_gain": 0.2476, + "loop_stats": [ + { + "tau": 0.0873, + "v": 0.8991, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0315, + "v": 0.8867, + "H_a": 0.108, + "d_bind": 0.04008 + }, + { + "tau": 0.0277, + "v": 0.8879, + "H_a": 0.082, + "d_bind": 0.0203 + } + ] + }, + { + "step": 10600, + "tok": 43417600, + "ce": 2.9592, + "ppl": 19.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22446, + "peak_MiB": 5301, + "v_gain": 0.2478, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.904, + "H_a": 0.133, + "d_bind": 0.0 + }, + { + "tau": 0.0588, + "v": 0.8741, + "H_a": 0.091, + "d_bind": 0.0419 + }, + { + "tau": 0.0508, + "v": 0.8728, + "H_a": 0.076, + "d_bind": 0.01813 + } + ] + }, + { + "step": 10620, + "tok": 43499520, + "ce": 3.3416, + "ppl": 28.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22448, + "peak_MiB": 5301, + "v_gain": 0.2514, + "loop_stats": [ + { + "tau": 0.1191, + "v": 0.9095, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0566, + "v": 0.8876, + "H_a": 0.09, + "d_bind": 0.03989 + }, + { + "tau": 0.0501, + "v": 0.8895, + "H_a": 0.079, + "d_bind": 0.01829 + } + ] + }, + { + "step": 10640, + "tok": 43581440, + "ce": 3.3746, + "ppl": 29.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22454, + "peak_MiB": 5301, + "v_gain": 0.2511, + "loop_stats": [ + { + "tau": 0.0935, + "v": 0.8913, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.8804, + "H_a": 0.105, + "d_bind": 0.03767 + }, + { + "tau": 0.0382, + "v": 0.8802, + "H_a": 0.083, + "d_bind": 0.01981 + } + ] + }, + { + "step": 10660, + "tok": 43663360, + "ce": 3.3684, + "ppl": 29.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22452, + "peak_MiB": 5301, + "v_gain": 0.2528, + "loop_stats": [ + { + "tau": 0.0924, + "v": 0.8876, + "H_a": 0.136, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.8646, + "H_a": 0.099, + "d_bind": 0.03935 + }, + { + "tau": 0.0328, + "v": 0.8575, + "H_a": 0.08, + "d_bind": 0.01885 + } + ] + }, + { + "step": 10680, + "tok": 43745280, + "ce": 3.2039, + "ppl": 24.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22444, + "peak_MiB": 5301, + "v_gain": 0.2531, + "loop_stats": [ + { + "tau": 0.0943, + "v": 0.9189, + "H_a": 0.146, + "d_bind": 0.0 + }, + { + "tau": 0.0233, + "v": 0.9009, + "H_a": 0.106, + "d_bind": 0.04017 + }, + { + "tau": 0.017, + "v": 0.8979, + "H_a": 0.084, + "d_bind": 0.02088 + } + ] + }, + { + "step": 10700, + "tok": 43827200, + "ce": 2.9002, + "ppl": 18.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22446, + "peak_MiB": 5301, + "v_gain": 0.2521, + "loop_stats": [ + { + "tau": 0.1335, + "v": 0.8906, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0611, + "v": 0.8764, + "H_a": 0.094, + "d_bind": 0.03886 + }, + { + "tau": 0.0524, + "v": 0.8781, + "H_a": 0.076, + "d_bind": 0.01719 + } + ] + }, + { + "step": 10720, + "tok": 43909120, + "ce": 2.8634, + "ppl": 17.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22448, + "peak_MiB": 5301, + "v_gain": 0.25, + "loop_stats": [ + { + "tau": 0.1021, + "v": 0.8974, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8882, + "H_a": 0.1, + "d_bind": 0.04062 + }, + { + "tau": 0.041, + "v": 0.8927, + "H_a": 0.081, + "d_bind": 0.01915 + } + ] + }, + { + "step": 10740, + "tok": 43991040, + "ce": 3.2265, + "ppl": 25.2, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2499, + "loop_stats": [ + { + "tau": 0.0986, + "v": 0.8847, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.877, + "H_a": 0.091, + "d_bind": 0.04069 + }, + { + "tau": 0.0364, + "v": 0.885, + "H_a": 0.075, + "d_bind": 0.01704 + } + ] + }, + { + "step": 10760, + "tok": 44072960, + "ce": 3.018, + "ppl": 20.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.917, + "tok_s": 22456, + "peak_MiB": 5301, + "v_gain": 0.2502, + "loop_stats": [ + { + "tau": 0.1354, + "v": 0.9327, + "H_a": 0.135, + "d_bind": 0.0 + }, + { + "tau": 0.062, + "v": 0.9139, + "H_a": 0.094, + "d_bind": 0.04077 + }, + { + "tau": 0.0522, + "v": 0.9137, + "H_a": 0.075, + "d_bind": 0.01776 + } + ] + }, + { + "step": 10780, + "tok": 44154880, + "ce": 2.8216, + "ppl": 16.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22454, + "peak_MiB": 5301, + "v_gain": 0.2499, + "loop_stats": [ + { + "tau": 0.1275, + "v": 0.9143, + "H_a": 0.127, + "d_bind": 0.0 + }, + { + "tau": 0.0524, + "v": 0.8837, + "H_a": 0.101, + "d_bind": 0.03946 + }, + { + "tau": 0.0417, + "v": 0.88, + "H_a": 0.075, + "d_bind": 0.01873 + } + ] + }, + { + "step": 10800, + "tok": 44236800, + "ce": 3.0894, + "ppl": 22.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22446, + "peak_MiB": 5301, + "v_gain": 0.2512, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8711, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0597, + "v": 0.8424, + "H_a": 0.095, + "d_bind": 0.03738 + }, + { + "tau": 0.0491, + "v": 0.8485, + "H_a": 0.078, + "d_bind": 0.01859 + } + ] + }, + { + "step": 10820, + "tok": 44318720, + "ce": 3.0512, + "ppl": 21.1, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.945, + "tok_s": 22448, + "peak_MiB": 5301, + "v_gain": 0.2531, + "loop_stats": [ + { + "tau": 0.1301, + "v": 0.8913, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0421, + "v": 0.8671, + "H_a": 0.1, + "d_bind": 0.03835 + }, + { + "tau": 0.0359, + "v": 0.8639, + "H_a": 0.078, + "d_bind": 0.01864 + } + ] + }, + { + "step": 10840, + "tok": 44400640, + "ce": 2.7713, + "ppl": 16.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2508, + "loop_stats": [ + { + "tau": 0.1124, + "v": 0.9188, + "H_a": 0.133, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.9029, + "H_a": 0.093, + "d_bind": 0.03984 + }, + { + "tau": 0.0372, + "v": 0.9011, + "H_a": 0.079, + "d_bind": 0.01708 + } + ] + }, + { + "step": 10860, + "tok": 44482560, + "ce": 2.8458, + "ppl": 17.2, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2504, + "loop_stats": [ + { + "tau": 0.135, + "v": 0.9088, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0603, + "v": 0.8775, + "H_a": 0.092, + "d_bind": 0.04225 + }, + { + "tau": 0.0568, + "v": 0.8787, + "H_a": 0.072, + "d_bind": 0.0169 + } + ] + }, + { + "step": 10880, + "tok": 44564480, + "ce": 3.4149, + "ppl": 30.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22457, + "peak_MiB": 5301, + "v_gain": 0.2513, + "loop_stats": [ + { + "tau": 0.0925, + "v": 0.8793, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0319, + "v": 0.8635, + "H_a": 0.1, + "d_bind": 0.03952 + }, + { + "tau": 0.0272, + "v": 0.8661, + "H_a": 0.082, + "d_bind": 0.02049 + } + ] + }, + { + "step": 10900, + "tok": 44646400, + "ce": 3.4228, + "ppl": 30.7, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.95, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.2552, + "loop_stats": [ + { + "tau": 0.0788, + "v": 0.8885, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0289, + "v": 0.8766, + "H_a": 0.102, + "d_bind": 0.03761 + }, + { + "tau": 0.0237, + "v": 0.8803, + "H_a": 0.083, + "d_bind": 0.01877 + } + ] + }, + { + "step": 10920, + "tok": 44728320, + "ce": 2.8521, + "ppl": 17.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22448, + "peak_MiB": 5301, + "v_gain": 0.2565, + "loop_stats": [ + { + "tau": 0.1171, + "v": 0.878, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8591, + "H_a": 0.092, + "d_bind": 0.03898 + }, + { + "tau": 0.035, + "v": 0.8602, + "H_a": 0.074, + "d_bind": 0.01841 + } + ] + }, + { + "step": 10940, + "tok": 44810240, + "ce": 2.5395, + "ppl": 12.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2562, + "loop_stats": [ + { + "tau": 0.1375, + "v": 0.8832, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0532, + "v": 0.8577, + "H_a": 0.081, + "d_bind": 0.03867 + }, + { + "tau": 0.0491, + "v": 0.8514, + "H_a": 0.066, + "d_bind": 0.01668 + } + ] + }, + { + "step": 10960, + "tok": 44892160, + "ce": 2.8516, + "ppl": 17.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2585, + "loop_stats": [ + { + "tau": 0.1314, + "v": 0.9031, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0501, + "v": 0.8792, + "H_a": 0.086, + "d_bind": 0.03945 + }, + { + "tau": 0.047, + "v": 0.8733, + "H_a": 0.072, + "d_bind": 0.01636 + } + ] + }, + { + "step": 10980, + "tok": 44974080, + "ce": 3.0494, + "ppl": 21.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.965, + "tok_s": 22452, + "peak_MiB": 5301, + "v_gain": 0.2606, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.8992, + "H_a": 0.13, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8751, + "H_a": 0.089, + "d_bind": 0.04037 + }, + { + "tau": 0.0272, + "v": 0.8716, + "H_a": 0.072, + "d_bind": 0.01712 + } + ] + }, + { + "step": 11000, + "tok": 45056000, + "ce": 2.9103, + "ppl": 18.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2629, + "loop_stats": [ + { + "tau": 0.1342, + "v": 0.9086, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0541, + "v": 0.8998, + "H_a": 0.085, + "d_bind": 0.03896 + }, + { + "tau": 0.0474, + "v": 0.9029, + "H_a": 0.068, + "d_bind": 0.0169 + } + ] + }, + { + "step": 11020, + "tok": 45137920, + "ce": 3.4121, + "ppl": 30.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22457, + "peak_MiB": 5301, + "v_gain": 0.2672, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8831, + "H_a": 0.131, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.8634, + "H_a": 0.099, + "d_bind": 0.03852 + }, + { + "tau": 0.0299, + "v": 0.8614, + "H_a": 0.078, + "d_bind": 0.01949 + } + ] + }, + { + "step": 11040, + "tok": 45219840, + "ce": 3.4129, + "ppl": 30.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22449, + "peak_MiB": 5301, + "v_gain": 0.2696, + "loop_stats": [ + { + "tau": 0.0872, + "v": 0.8993, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0225, + "v": 0.8743, + "H_a": 0.099, + "d_bind": 0.03848 + }, + { + "tau": 0.019, + "v": 0.8694, + "H_a": 0.083, + "d_bind": 0.02084 + } + ] + }, + { + "step": 11060, + "tok": 45301760, + "ce": 3.1284, + "ppl": 22.8, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.938, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2749, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8692, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8448, + "H_a": 0.085, + "d_bind": 0.04253 + }, + { + "tau": 0.0426, + "v": 0.8447, + "H_a": 0.067, + "d_bind": 0.01623 + } + ] + }, + { + "step": 11080, + "tok": 45383680, + "ce": 3.2645, + "ppl": 26.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22452, + "peak_MiB": 5301, + "v_gain": 0.2782, + "loop_stats": [ + { + "tau": 0.1106, + "v": 0.8775, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8537, + "H_a": 0.092, + "d_bind": 0.03956 + }, + { + "tau": 0.0308, + "v": 0.8542, + "H_a": 0.076, + "d_bind": 0.01782 + } + ] + }, + { + "step": 11100, + "tok": 45465600, + "ce": 3.1433, + "ppl": 23.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22454, + "peak_MiB": 5301, + "v_gain": 0.2776, + "loop_stats": [ + { + "tau": 0.106, + "v": 0.9111, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8884, + "H_a": 0.088, + "d_bind": 0.04105 + }, + { + "tau": 0.0337, + "v": 0.8859, + "H_a": 0.072, + "d_bind": 0.01852 + } + ] + }, + { + "step": 11120, + "tok": 45547520, + "ce": 3.1088, + "ppl": 22.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2738, + "loop_stats": [ + { + "tau": 0.1185, + "v": 0.8959, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0554, + "v": 0.8755, + "H_a": 0.097, + "d_bind": 0.04086 + }, + { + "tau": 0.0467, + "v": 0.8752, + "H_a": 0.073, + "d_bind": 0.01891 + } + ] + }, + { + "step": 11140, + "tok": 45629440, + "ce": 2.9686, + "ppl": 19.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22457, + "peak_MiB": 5301, + "v_gain": 0.274, + "loop_stats": [ + { + "tau": 0.1066, + "v": 0.9187, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.9075, + "H_a": 0.096, + "d_bind": 0.04065 + }, + { + "tau": 0.0249, + "v": 0.9129, + "H_a": 0.078, + "d_bind": 0.01977 + } + ] + }, + { + "step": 11160, + "tok": 45711360, + "ce": 3.0133, + "ppl": 20.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22450, + "peak_MiB": 5301, + "v_gain": 0.2745, + "loop_stats": [ + { + "tau": 0.1141, + "v": 0.8989, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0473, + "v": 0.8729, + "H_a": 0.099, + "d_bind": 0.04277 + }, + { + "tau": 0.0424, + "v": 0.8746, + "H_a": 0.076, + "d_bind": 0.01948 + } + ] + }, + { + "step": 11180, + "tok": 45793280, + "ce": 3.3789, + "ppl": 29.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2781, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.9061, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0366, + "v": 0.8816, + "H_a": 0.106, + "d_bind": 0.04117 + }, + { + "tau": 0.0298, + "v": 0.8778, + "H_a": 0.081, + "d_bind": 0.0198 + } + ] + }, + { + "step": 11200, + "tok": 45875200, + "ce": 2.9311, + "ppl": 18.7, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22453, + "peak_MiB": 5301, + "v_gain": 0.2751, + "loop_stats": [ + { + "tau": 0.123, + "v": 0.9174, + "H_a": 0.126, + "d_bind": 0.0 + }, + { + "tau": 0.0581, + "v": 0.9025, + "H_a": 0.097, + "d_bind": 0.04316 + }, + { + "tau": 0.0543, + "v": 0.9, + "H_a": 0.08, + "d_bind": 0.01932 + } + ] + }, + { + "step": 11220, + "tok": 45957120, + "ce": 3.3454, + "ppl": 28.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.276, + "loop_stats": [ + { + "tau": 0.1077, + "v": 0.9155, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8913, + "H_a": 0.096, + "d_bind": 0.04298 + }, + { + "tau": 0.0346, + "v": 0.8896, + "H_a": 0.077, + "d_bind": 0.01977 + } + ] + }, + { + "step": 11240, + "tok": 46039040, + "ce": 3.0308, + "ppl": 20.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.2781, + "loop_stats": [ + { + "tau": 0.1181, + "v": 0.9271, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0561, + "v": 0.9102, + "H_a": 0.092, + "d_bind": 0.04293 + }, + { + "tau": 0.0471, + "v": 0.9091, + "H_a": 0.074, + "d_bind": 0.01788 + } + ] + }, + { + "step": 11260, + "tok": 46120960, + "ce": 3.1656, + "ppl": 23.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2769, + "loop_stats": [ + { + "tau": 0.1156, + "v": 0.9088, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8889, + "H_a": 0.091, + "d_bind": 0.03898 + }, + { + "tau": 0.0435, + "v": 0.8867, + "H_a": 0.078, + "d_bind": 0.0195 + } + ] + }, + { + "step": 11280, + "tok": 46202880, + "ce": 3.1778, + "ppl": 24.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22451, + "peak_MiB": 5301, + "v_gain": 0.2781, + "loop_stats": [ + { + "tau": 0.1099, + "v": 0.9107, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0321, + "v": 0.8919, + "H_a": 0.096, + "d_bind": 0.03923 + }, + { + "tau": 0.0244, + "v": 0.8915, + "H_a": 0.081, + "d_bind": 0.02044 + } + ] + }, + { + "step": 11300, + "tok": 46284800, + "ce": 3.1818, + "ppl": 24.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22453, + "peak_MiB": 5301, + "v_gain": 0.2743, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8867, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8823, + "H_a": 0.096, + "d_bind": 0.03867 + }, + { + "tau": 0.0381, + "v": 0.8865, + "H_a": 0.076, + "d_bind": 0.01962 + } + ] + }, + { + "step": 11320, + "tok": 46366720, + "ce": 2.9139, + "ppl": 18.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22454, + "peak_MiB": 5301, + "v_gain": 0.273, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8975, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.8773, + "H_a": 0.089, + "d_bind": 0.04142 + }, + { + "tau": 0.0476, + "v": 0.8771, + "H_a": 0.072, + "d_bind": 0.01659 + } + ] + }, + { + "step": 11340, + "tok": 46448640, + "ce": 3.3111, + "ppl": 27.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.933, + "tok_s": 22456, + "peak_MiB": 5301, + "v_gain": 0.2727, + "loop_stats": [ + { + "tau": 0.0947, + "v": 0.9211, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.9037, + "H_a": 0.09, + "d_bind": 0.03804 + }, + { + "tau": 0.0295, + "v": 0.8967, + "H_a": 0.072, + "d_bind": 0.01811 + } + ] + }, + { + "step": 11360, + "tok": 46530560, + "ce": 2.6591, + "ppl": 14.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2712, + "loop_stats": [ + { + "tau": 0.1394, + "v": 0.871, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0565, + "v": 0.8478, + "H_a": 0.092, + "d_bind": 0.04129 + }, + { + "tau": 0.0528, + "v": 0.8448, + "H_a": 0.073, + "d_bind": 0.01909 + } + ] + }, + { + "step": 11380, + "tok": 46612480, + "ce": 3.0444, + "ppl": 21.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.2746, + "loop_stats": [ + { + "tau": 0.108, + "v": 0.9136, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8971, + "H_a": 0.087, + "d_bind": 0.04092 + }, + { + "tau": 0.0401, + "v": 0.8962, + "H_a": 0.076, + "d_bind": 0.01915 + } + ] + }, + { + "step": 11400, + "tok": 46694400, + "ce": 3.2917, + "ppl": 26.9, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22453, + "peak_MiB": 5301, + "v_gain": 0.2799, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.8875, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0399, + "v": 0.8782, + "H_a": 0.095, + "d_bind": 0.0396 + }, + { + "tau": 0.0339, + "v": 0.8753, + "H_a": 0.078, + "d_bind": 0.01962 + } + ] + }, + { + "step": 11420, + "tok": 46776320, + "ce": 3.3123, + "ppl": 27.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.2838, + "loop_stats": [ + { + "tau": 0.0839, + "v": 0.9003, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0309, + "v": 0.8816, + "H_a": 0.086, + "d_bind": 0.04051 + }, + { + "tau": 0.0253, + "v": 0.8737, + "H_a": 0.071, + "d_bind": 0.0186 + } + ] + }, + { + "step": 11440, + "tok": 46858240, + "ce": 2.9839, + "ppl": 19.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22456, + "peak_MiB": 5301, + "v_gain": 0.2858, + "loop_stats": [ + { + "tau": 0.1257, + "v": 0.9038, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8883, + "H_a": 0.089, + "d_bind": 0.04174 + }, + { + "tau": 0.0384, + "v": 0.8884, + "H_a": 0.065, + "d_bind": 0.01739 + } + ] + }, + { + "step": 11460, + "tok": 46940160, + "ce": 2.7399, + "ppl": 15.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.2805, + "loop_stats": [ + { + "tau": 0.13, + "v": 0.8979, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.066, + "v": 0.8708, + "H_a": 0.082, + "d_bind": 0.04372 + }, + { + "tau": 0.0517, + "v": 0.8768, + "H_a": 0.071, + "d_bind": 0.0176 + } + ] + }, + { + "step": 11480, + "tok": 47022080, + "ce": 3.2542, + "ppl": 25.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.2792, + "loop_stats": [ + { + "tau": 0.0994, + "v": 0.8845, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8598, + "H_a": 0.088, + "d_bind": 0.0395 + }, + { + "tau": 0.0311, + "v": 0.8584, + "H_a": 0.075, + "d_bind": 0.01858 + } + ] + }, + { + "step": 11500, + "tok": 47104000, + "ce": 3.1076, + "ppl": 22.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2833, + "loop_stats": [ + { + "tau": 0.1075, + "v": 0.8965, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8823, + "H_a": 0.092, + "d_bind": 0.04072 + }, + { + "tau": 0.0282, + "v": 0.8789, + "H_a": 0.078, + "d_bind": 0.01993 + } + ] + }, + { + "step": 11520, + "tok": 47185920, + "ce": 3.4006, + "ppl": 30.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22455, + "peak_MiB": 5301, + "v_gain": 0.2843, + "loop_stats": [ + { + "tau": 0.0983, + "v": 0.8732, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0284, + "v": 0.8565, + "H_a": 0.088, + "d_bind": 0.04191 + }, + { + "tau": 0.0318, + "v": 0.8635, + "H_a": 0.069, + "d_bind": 0.01941 + } + ] + }, + { + "step": 11540, + "tok": 47267840, + "ce": 2.6143, + "ppl": 13.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.957, + "tok_s": 22456, + "peak_MiB": 5301, + "v_gain": 0.2856, + "loop_stats": [ + { + "tau": 0.1249, + "v": 0.8852, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8569, + "H_a": 0.086, + "d_bind": 0.04562 + }, + { + "tau": 0.0458, + "v": 0.8546, + "H_a": 0.075, + "d_bind": 0.01831 + } + ] + }, + { + "step": 11560, + "tok": 47349760, + "ce": 3.3648, + "ppl": 28.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.2829, + "loop_stats": [ + { + "tau": 0.074, + "v": 0.8679, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0162, + "v": 0.8615, + "H_a": 0.096, + "d_bind": 0.03644 + }, + { + "tau": 0.0137, + "v": 0.8663, + "H_a": 0.078, + "d_bind": 0.02016 + } + ] + }, + { + "step": 11580, + "tok": 47431680, + "ce": 2.8144, + "ppl": 16.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2831, + "loop_stats": [ + { + "tau": 0.1002, + "v": 0.9225, + "H_a": 0.119, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.8957, + "H_a": 0.096, + "d_bind": 0.04452 + }, + { + "tau": 0.0428, + "v": 0.8958, + "H_a": 0.074, + "d_bind": 0.02069 + } + ] + }, + { + "step": 11600, + "tok": 47513600, + "ce": 2.9812, + "ppl": 19.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.2822, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.9049, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0532, + "v": 0.8865, + "H_a": 0.087, + "d_bind": 0.04095 + }, + { + "tau": 0.048, + "v": 0.8867, + "H_a": 0.074, + "d_bind": 0.01903 + } + ] + }, + { + "step": 11620, + "tok": 47595520, + "ce": 3.4919, + "ppl": 32.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.2846, + "loop_stats": [ + { + "tau": 0.0758, + "v": 0.9039, + "H_a": 0.128, + "d_bind": 0.0 + }, + { + "tau": 0.0194, + "v": 0.8871, + "H_a": 0.097, + "d_bind": 0.04123 + }, + { + "tau": 0.0184, + "v": 0.887, + "H_a": 0.08, + "d_bind": 0.02075 + } + ] + }, + { + "step": 11640, + "tok": 47677440, + "ce": 2.8235, + "ppl": 16.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22456, + "peak_MiB": 5301, + "v_gain": 0.2881, + "loop_stats": [ + { + "tau": 0.13, + "v": 0.8759, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.8521, + "H_a": 0.088, + "d_bind": 0.04255 + }, + { + "tau": 0.0342, + "v": 0.8496, + "H_a": 0.072, + "d_bind": 0.02142 + } + ] + }, + { + "step": 11660, + "tok": 47759360, + "ce": 2.5378, + "ppl": 12.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.2885, + "loop_stats": [ + { + "tau": 0.1469, + "v": 0.8627, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0673, + "v": 0.8492, + "H_a": 0.083, + "d_bind": 0.0474 + }, + { + "tau": 0.0652, + "v": 0.8457, + "H_a": 0.062, + "d_bind": 0.01665 + } + ] + }, + { + "step": 11680, + "tok": 47841280, + "ce": 3.2622, + "ppl": 26.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2895, + "loop_stats": [ + { + "tau": 0.1096, + "v": 0.85, + "H_a": 0.117, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8477, + "H_a": 0.089, + "d_bind": 0.04235 + }, + { + "tau": 0.0334, + "v": 0.8547, + "H_a": 0.075, + "d_bind": 0.01841 + } + ] + }, + { + "step": 11700, + "tok": 47923200, + "ce": 3.0801, + "ppl": 21.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.292, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.8711, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8637, + "H_a": 0.086, + "d_bind": 0.04368 + }, + { + "tau": 0.0441, + "v": 0.8677, + "H_a": 0.074, + "d_bind": 0.01916 + } + ] + }, + { + "step": 11720, + "tok": 48005120, + "ce": 3.2287, + "ppl": 25.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.295, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.8582, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.8526, + "H_a": 0.089, + "d_bind": 0.04039 + }, + { + "tau": 0.0286, + "v": 0.8571, + "H_a": 0.071, + "d_bind": 0.01939 + } + ] + }, + { + "step": 11740, + "tok": 48087040, + "ce": 2.9156, + "ppl": 18.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.2972, + "loop_stats": [ + { + "tau": 0.1216, + "v": 0.869, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0518, + "v": 0.855, + "H_a": 0.082, + "d_bind": 0.0415 + }, + { + "tau": 0.048, + "v": 0.8506, + "H_a": 0.067, + "d_bind": 0.01758 + } + ] + }, + { + "step": 11760, + "tok": 48168960, + "ce": 2.7429, + "ppl": 15.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22457, + "peak_MiB": 5301, + "v_gain": 0.2973, + "loop_stats": [ + { + "tau": 0.1292, + "v": 0.8632, + "H_a": 0.123, + "d_bind": 0.0 + }, + { + "tau": 0.0509, + "v": 0.852, + "H_a": 0.085, + "d_bind": 0.04582 + }, + { + "tau": 0.0479, + "v": 0.8479, + "H_a": 0.069, + "d_bind": 0.01837 + } + ] + }, + { + "step": 11780, + "tok": 48250880, + "ce": 2.7625, + "ppl": 15.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2959, + "loop_stats": [ + { + "tau": 0.1173, + "v": 0.8733, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0502, + "v": 0.8411, + "H_a": 0.082, + "d_bind": 0.04644 + }, + { + "tau": 0.0488, + "v": 0.8319, + "H_a": 0.067, + "d_bind": 0.01768 + } + ] + }, + { + "step": 11800, + "tok": 48332800, + "ce": 2.9438, + "ppl": 19.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.2977, + "loop_stats": [ + { + "tau": 0.1026, + "v": 0.8477, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.8379, + "H_a": 0.084, + "d_bind": 0.04357 + }, + { + "tau": 0.0404, + "v": 0.8437, + "H_a": 0.069, + "d_bind": 0.01859 + } + ] + }, + { + "step": 11820, + "tok": 48414720, + "ce": 3.2531, + "ppl": 25.9, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.933, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.2975, + "loop_stats": [ + { + "tau": 0.1133, + "v": 0.8774, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8565, + "H_a": 0.086, + "d_bind": 0.0379 + }, + { + "tau": 0.0408, + "v": 0.8579, + "H_a": 0.072, + "d_bind": 0.01901 + } + ] + }, + { + "step": 11840, + "tok": 48496640, + "ce": 2.7072, + "ppl": 15.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.2959, + "loop_stats": [ + { + "tau": 0.1293, + "v": 0.8863, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.873, + "H_a": 0.09, + "d_bind": 0.04271 + }, + { + "tau": 0.0317, + "v": 0.8761, + "H_a": 0.077, + "d_bind": 0.02054 + } + ] + }, + { + "step": 11860, + "tok": 48578560, + "ce": 3.0644, + "ppl": 21.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.93, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.2908, + "loop_stats": [ + { + "tau": 0.1238, + "v": 0.8676, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.052, + "v": 0.8639, + "H_a": 0.085, + "d_bind": 0.03744 + }, + { + "tau": 0.049, + "v": 0.8651, + "H_a": 0.074, + "d_bind": 0.0181 + } + ] + }, + { + "step": 11880, + "tok": 48660480, + "ce": 2.7892, + "ppl": 16.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.2887, + "loop_stats": [ + { + "tau": 0.1278, + "v": 0.8484, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0578, + "v": 0.8409, + "H_a": 0.091, + "d_bind": 0.04353 + }, + { + "tau": 0.0538, + "v": 0.8393, + "H_a": 0.074, + "d_bind": 0.01864 + } + ] + }, + { + "step": 11900, + "tok": 48742400, + "ce": 3.1061, + "ppl": 22.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2899, + "loop_stats": [ + { + "tau": 0.1115, + "v": 0.8718, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.8615, + "H_a": 0.085, + "d_bind": 0.03805 + }, + { + "tau": 0.0493, + "v": 0.8549, + "H_a": 0.074, + "d_bind": 0.01789 + } + ] + }, + { + "step": 11920, + "tok": 48824320, + "ce": 3.0846, + "ppl": 21.9, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.947, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.2913, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.8682, + "H_a": 0.12, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.8678, + "H_a": 0.093, + "d_bind": 0.04156 + }, + { + "tau": 0.0337, + "v": 0.8727, + "H_a": 0.076, + "d_bind": 0.01798 + } + ] + }, + { + "step": 11940, + "tok": 48906240, + "ce": 3.489, + "ppl": 32.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2921, + "loop_stats": [ + { + "tau": 0.1058, + "v": 0.851, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8306, + "H_a": 0.089, + "d_bind": 0.04211 + }, + { + "tau": 0.0381, + "v": 0.8304, + "H_a": 0.073, + "d_bind": 0.01893 + } + ] + }, + { + "step": 11960, + "tok": 48988160, + "ce": 3.2128, + "ppl": 24.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.2929, + "loop_stats": [ + { + "tau": 0.0998, + "v": 0.8515, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.8355, + "H_a": 0.093, + "d_bind": 0.04026 + }, + { + "tau": 0.0361, + "v": 0.8354, + "H_a": 0.078, + "d_bind": 0.02006 + } + ] + }, + { + "step": 11980, + "tok": 49070080, + "ce": 2.9438, + "ppl": 19.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.2924, + "loop_stats": [ + { + "tau": 0.123, + "v": 0.8582, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0616, + "v": 0.8465, + "H_a": 0.082, + "d_bind": 0.04297 + }, + { + "tau": 0.058, + "v": 0.8481, + "H_a": 0.071, + "d_bind": 0.01693 + } + ] + }, + { + "step": 12000, + "tok": 49152000, + "ce": 3.036, + "ppl": 20.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.2939, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.8592, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.8641, + "H_a": 0.087, + "d_bind": 0.04085 + }, + { + "tau": 0.0365, + "v": 0.8738, + "H_a": 0.075, + "d_bind": 0.01972 + } + ] + }, + { + "step": 12020, + "tok": 49233920, + "ce": 2.9563, + "ppl": 19.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.96, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.2959, + "loop_stats": [ + { + "tau": 0.1013, + "v": 0.8498, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.8468, + "H_a": 0.083, + "d_bind": 0.04024 + }, + { + "tau": 0.0474, + "v": 0.8519, + "H_a": 0.064, + "d_bind": 0.01775 + } + ] + }, + { + "step": 12040, + "tok": 49315840, + "ce": 2.9586, + "ppl": 19.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2937, + "loop_stats": [ + { + "tau": 0.1061, + "v": 0.838, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8394, + "H_a": 0.086, + "d_bind": 0.0438 + }, + { + "tau": 0.0318, + "v": 0.8454, + "H_a": 0.071, + "d_bind": 0.01794 + } + ] + }, + { + "step": 12060, + "tok": 49397760, + "ce": 3.4257, + "ppl": 30.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.2931, + "loop_stats": [ + { + "tau": 0.1062, + "v": 0.8586, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0282, + "v": 0.8543, + "H_a": 0.085, + "d_bind": 0.03875 + }, + { + "tau": 0.0247, + "v": 0.8578, + "H_a": 0.07, + "d_bind": 0.01914 + } + ] + }, + { + "step": 12080, + "tok": 49479680, + "ce": 3.3887, + "ppl": 29.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.2914, + "loop_stats": [ + { + "tau": 0.1009, + "v": 0.8597, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0429, + "v": 0.8548, + "H_a": 0.09, + "d_bind": 0.03911 + }, + { + "tau": 0.0368, + "v": 0.8646, + "H_a": 0.074, + "d_bind": 0.01966 + } + ] + }, + { + "step": 12100, + "tok": 49561600, + "ce": 3.3094, + "ppl": 27.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.2935, + "loop_stats": [ + { + "tau": 0.0985, + "v": 0.8602, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8654, + "H_a": 0.088, + "d_bind": 0.03863 + }, + { + "tau": 0.0273, + "v": 0.8715, + "H_a": 0.067, + "d_bind": 0.01868 + } + ] + }, + { + "step": 12120, + "tok": 49643520, + "ce": 2.8701, + "ppl": 17.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.291, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.8563, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0597, + "v": 0.8439, + "H_a": 0.083, + "d_bind": 0.04465 + }, + { + "tau": 0.0561, + "v": 0.848, + "H_a": 0.07, + "d_bind": 0.01765 + } + ] + }, + { + "step": 12140, + "tok": 49725440, + "ce": 2.8745, + "ppl": 17.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2908, + "loop_stats": [ + { + "tau": 0.1208, + "v": 0.8456, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8492, + "H_a": 0.092, + "d_bind": 0.04089 + }, + { + "tau": 0.0452, + "v": 0.8505, + "H_a": 0.073, + "d_bind": 0.0204 + } + ] + }, + { + "step": 12160, + "tok": 49807360, + "ce": 2.8544, + "ppl": 17.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.2914, + "loop_stats": [ + { + "tau": 0.1106, + "v": 0.8399, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.8401, + "H_a": 0.086, + "d_bind": 0.04289 + }, + { + "tau": 0.0395, + "v": 0.847, + "H_a": 0.072, + "d_bind": 0.01857 + } + ] + }, + { + "step": 12180, + "tok": 49889280, + "ce": 3.4149, + "ppl": 30.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.962, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.2916, + "loop_stats": [ + { + "tau": 0.0765, + "v": 0.8477, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0175, + "v": 0.8453, + "H_a": 0.082, + "d_bind": 0.03735 + }, + { + "tau": 0.0155, + "v": 0.8484, + "H_a": 0.068, + "d_bind": 0.01867 + } + ] + }, + { + "step": 12200, + "tok": 49971200, + "ce": 2.8614, + "ppl": 17.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.2912, + "loop_stats": [ + { + "tau": 0.126, + "v": 0.8669, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.856, + "H_a": 0.09, + "d_bind": 0.04493 + }, + { + "tau": 0.0329, + "v": 0.8609, + "H_a": 0.068, + "d_bind": 0.02013 + } + ] + }, + { + "step": 12220, + "tok": 50053120, + "ce": 3.3427, + "ppl": 28.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.2871, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8564, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8581, + "H_a": 0.082, + "d_bind": 0.04073 + }, + { + "tau": 0.0365, + "v": 0.8707, + "H_a": 0.068, + "d_bind": 0.01804 + } + ] + }, + { + "step": 12240, + "tok": 50135040, + "ce": 3.2364, + "ppl": 25.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.2889, + "loop_stats": [ + { + "tau": 0.0979, + "v": 0.8615, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0226, + "v": 0.8641, + "H_a": 0.089, + "d_bind": 0.04245 + }, + { + "tau": 0.0171, + "v": 0.8739, + "H_a": 0.074, + "d_bind": 0.02016 + } + ] + }, + { + "step": 12260, + "tok": 50216960, + "ce": 3.0926, + "ppl": 22.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.2891, + "loop_stats": [ + { + "tau": 0.1057, + "v": 0.8446, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.835, + "H_a": 0.088, + "d_bind": 0.04379 + }, + { + "tau": 0.032, + "v": 0.8366, + "H_a": 0.068, + "d_bind": 0.01892 + } + ] + }, + { + "step": 12280, + "tok": 50298880, + "ce": 3.0985, + "ppl": 22.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.962, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.2828, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.8574, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.04, + "v": 0.8455, + "H_a": 0.077, + "d_bind": 0.03908 + }, + { + "tau": 0.0374, + "v": 0.8457, + "H_a": 0.066, + "d_bind": 0.01763 + } + ] + }, + { + "step": 12300, + "tok": 50380800, + "ce": 3.2571, + "ppl": 26.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.2792, + "loop_stats": [ + { + "tau": 0.1339, + "v": 0.8456, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0608, + "v": 0.8263, + "H_a": 0.086, + "d_bind": 0.04239 + }, + { + "tau": 0.0554, + "v": 0.828, + "H_a": 0.071, + "d_bind": 0.01824 + } + ] + }, + { + "step": 12320, + "tok": 50462720, + "ce": 3.4766, + "ppl": 32.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.96, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.2821, + "loop_stats": [ + { + "tau": 0.0901, + "v": 0.8734, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0299, + "v": 0.8748, + "H_a": 0.089, + "d_bind": 0.04222 + }, + { + "tau": 0.0254, + "v": 0.8856, + "H_a": 0.073, + "d_bind": 0.02066 + } + ] + }, + { + "step": 12340, + "tok": 50544640, + "ce": 2.3988, + "ppl": 11.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.957, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.2826, + "loop_stats": [ + { + "tau": 0.1405, + "v": 0.8434, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0777, + "v": 0.8397, + "H_a": 0.074, + "d_bind": 0.04795 + }, + { + "tau": 0.074, + "v": 0.8392, + "H_a": 0.065, + "d_bind": 0.01823 + } + ] + }, + { + "step": 12360, + "tok": 50626560, + "ce": 3.1472, + "ppl": 23.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.965, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.286, + "loop_stats": [ + { + "tau": 0.1096, + "v": 0.8685, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0556, + "v": 0.8606, + "H_a": 0.091, + "d_bind": 0.04194 + }, + { + "tau": 0.0481, + "v": 0.864, + "H_a": 0.073, + "d_bind": 0.0201 + } + ] + }, + { + "step": 12380, + "tok": 50708480, + "ce": 3.2932, + "ppl": 26.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.962, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.288, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.8801, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8589, + "H_a": 0.09, + "d_bind": 0.03873 + }, + { + "tau": 0.0357, + "v": 0.8522, + "H_a": 0.076, + "d_bind": 0.01983 + } + ] + }, + { + "step": 12400, + "tok": 50790400, + "ce": 3.2753, + "ppl": 26.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.2867, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.841, + "H_a": 0.118, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.825, + "H_a": 0.088, + "d_bind": 0.0423 + }, + { + "tau": 0.0359, + "v": 0.8208, + "H_a": 0.071, + "d_bind": 0.01868 + } + ] + }, + { + "step": 12420, + "tok": 50872320, + "ce": 2.9613, + "ppl": 19.3, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.942, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.2877, + "loop_stats": [ + { + "tau": 0.1357, + "v": 0.8853, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0522, + "v": 0.8626, + "H_a": 0.078, + "d_bind": 0.04439 + }, + { + "tau": 0.0507, + "v": 0.8577, + "H_a": 0.065, + "d_bind": 0.01801 + } + ] + }, + { + "step": 12440, + "tok": 50954240, + "ce": 2.8384, + "ppl": 17.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.2917, + "loop_stats": [ + { + "tau": 0.1171, + "v": 0.8725, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0534, + "v": 0.8646, + "H_a": 0.085, + "d_bind": 0.04275 + }, + { + "tau": 0.0473, + "v": 0.8652, + "H_a": 0.068, + "d_bind": 0.01786 + } + ] + }, + { + "step": 12460, + "tok": 51036160, + "ce": 3.7163, + "ppl": 41.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.962, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.2949, + "loop_stats": [ + { + "tau": 0.0841, + "v": 0.8808, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0223, + "v": 0.8722, + "H_a": 0.084, + "d_bind": 0.037 + }, + { + "tau": 0.0182, + "v": 0.8729, + "H_a": 0.067, + "d_bind": 0.01831 + } + ] + }, + { + "step": 12480, + "tok": 51118080, + "ce": 2.7166, + "ppl": 15.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.2968, + "loop_stats": [ + { + "tau": 0.1379, + "v": 0.8483, + "H_a": 0.124, + "d_bind": 0.0 + }, + { + "tau": 0.0579, + "v": 0.8373, + "H_a": 0.081, + "d_bind": 0.04319 + }, + { + "tau": 0.0555, + "v": 0.8443, + "H_a": 0.068, + "d_bind": 0.01695 + } + ] + }, + { + "step": 12500, + "tok": 51200000, + "ce": 3.0228, + "ppl": 20.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.2988, + "loop_stats": [ + { + "tau": 0.1123, + "v": 0.8589, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0432, + "v": 0.8459, + "H_a": 0.085, + "d_bind": 0.04249 + }, + { + "tau": 0.042, + "v": 0.8468, + "H_a": 0.064, + "d_bind": 0.0183 + } + ] + }, + { + "step": 12520, + "tok": 51281920, + "ce": 3.0149, + "ppl": 20.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.2977, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8217, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.8213, + "H_a": 0.087, + "d_bind": 0.04217 + }, + { + "tau": 0.0358, + "v": 0.8329, + "H_a": 0.07, + "d_bind": 0.01871 + } + ] + }, + { + "step": 12540, + "tok": 51363840, + "ce": 2.9406, + "ppl": 18.9, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.94, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.2995, + "loop_stats": [ + { + "tau": 0.1259, + "v": 0.8668, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.8655, + "H_a": 0.08, + "d_bind": 0.03933 + }, + { + "tau": 0.0444, + "v": 0.872, + "H_a": 0.062, + "d_bind": 0.01771 + } + ] + }, + { + "step": 12560, + "tok": 51445760, + "ce": 2.98, + "ppl": 19.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3028, + "loop_stats": [ + { + "tau": 0.1059, + "v": 0.8848, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.8872, + "H_a": 0.081, + "d_bind": 0.0355 + }, + { + "tau": 0.025, + "v": 0.8992, + "H_a": 0.068, + "d_bind": 0.01747 + } + ] + }, + { + "step": 12580, + "tok": 51527680, + "ce": 2.922, + "ppl": 18.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.96, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.302, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.8827, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8765, + "H_a": 0.084, + "d_bind": 0.03834 + }, + { + "tau": 0.0303, + "v": 0.8796, + "H_a": 0.071, + "d_bind": 0.01932 + } + ] + }, + { + "step": 12600, + "tok": 51609600, + "ce": 3.2929, + "ppl": 26.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3007, + "loop_stats": [ + { + "tau": 0.089, + "v": 0.8652, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0313, + "v": 0.8643, + "H_a": 0.082, + "d_bind": 0.03596 + }, + { + "tau": 0.019, + "v": 0.8659, + "H_a": 0.068, + "d_bind": 0.01804 + } + ] + }, + { + "step": 12620, + "tok": 51691520, + "ce": 2.6528, + "ppl": 14.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3027, + "loop_stats": [ + { + "tau": 0.1221, + "v": 0.837, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8388, + "H_a": 0.083, + "d_bind": 0.04142 + }, + { + "tau": 0.042, + "v": 0.8448, + "H_a": 0.067, + "d_bind": 0.01793 + } + ] + }, + { + "step": 12640, + "tok": 51773440, + "ce": 3.2498, + "ppl": 25.8, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.942, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3002, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.8481, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0385, + "v": 0.8477, + "H_a": 0.069, + "d_bind": 0.03899 + }, + { + "tau": 0.032, + "v": 0.8575, + "H_a": 0.062, + "d_bind": 0.01608 + } + ] + }, + { + "step": 12660, + "tok": 51855360, + "ce": 2.7871, + "ppl": 16.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3006, + "loop_stats": [ + { + "tau": 0.1355, + "v": 0.8281, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0556, + "v": 0.8208, + "H_a": 0.076, + "d_bind": 0.04174 + }, + { + "tau": 0.0494, + "v": 0.8222, + "H_a": 0.064, + "d_bind": 0.01753 + } + ] + }, + { + "step": 12680, + "tok": 51937280, + "ce": 3.0729, + "ppl": 21.6, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.301, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8602, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8602, + "H_a": 0.076, + "d_bind": 0.04132 + }, + { + "tau": 0.0418, + "v": 0.8666, + "H_a": 0.063, + "d_bind": 0.01903 + } + ] + }, + { + "step": 12700, + "tok": 52019200, + "ce": 3.5664, + "ppl": 35.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.2999, + "loop_stats": [ + { + "tau": 0.0807, + "v": 0.8427, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8516, + "H_a": 0.077, + "d_bind": 0.03756 + }, + { + "tau": 0.0221, + "v": 0.8569, + "H_a": 0.065, + "d_bind": 0.01636 + } + ] + }, + { + "step": 12720, + "tok": 52101120, + "ce": 2.7547, + "ppl": 15.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.301, + "loop_stats": [ + { + "tau": 0.1205, + "v": 0.8453, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8346, + "H_a": 0.083, + "d_bind": 0.04216 + }, + { + "tau": 0.035, + "v": 0.8335, + "H_a": 0.062, + "d_bind": 0.01785 + } + ] + }, + { + "step": 12740, + "tok": 52183040, + "ce": 2.6608, + "ppl": 14.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3003, + "loop_stats": [ + { + "tau": 0.141, + "v": 0.8494, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0617, + "v": 0.8464, + "H_a": 0.072, + "d_bind": 0.04329 + }, + { + "tau": 0.0562, + "v": 0.8509, + "H_a": 0.06, + "d_bind": 0.01808 + } + ] + }, + { + "step": 12760, + "tok": 52264960, + "ce": 2.9093, + "ppl": 18.3, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.942, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3012, + "loop_stats": [ + { + "tau": 0.1164, + "v": 0.8597, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8611, + "H_a": 0.079, + "d_bind": 0.04301 + }, + { + "tau": 0.0451, + "v": 0.8624, + "H_a": 0.068, + "d_bind": 0.0174 + } + ] + }, + { + "step": 12780, + "tok": 52346880, + "ce": 3.1384, + "ppl": 23.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3011, + "loop_stats": [ + { + "tau": 0.1033, + "v": 0.8665, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.042, + "v": 0.8643, + "H_a": 0.085, + "d_bind": 0.04191 + }, + { + "tau": 0.0334, + "v": 0.8668, + "H_a": 0.066, + "d_bind": 0.0193 + } + ] + }, + { + "step": 12800, + "tok": 52428800, + "ce": 3.1256, + "ppl": 22.8, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.95, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3042, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.8578, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0363, + "v": 0.8588, + "H_a": 0.086, + "d_bind": 0.03954 + }, + { + "tau": 0.0301, + "v": 0.8662, + "H_a": 0.071, + "d_bind": 0.02031 + } + ] + }, + { + "step": 12820, + "tok": 52510720, + "ce": 3.4193, + "ppl": 30.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3063, + "loop_stats": [ + { + "tau": 0.0834, + "v": 0.8422, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0194, + "v": 0.8319, + "H_a": 0.086, + "d_bind": 0.03892 + }, + { + "tau": 0.0164, + "v": 0.8309, + "H_a": 0.07, + "d_bind": 0.01951 + } + ] + }, + { + "step": 12840, + "tok": 52592640, + "ce": 2.8858, + "ppl": 17.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3088, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8533, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.8573, + "H_a": 0.079, + "d_bind": 0.04178 + }, + { + "tau": 0.0283, + "v": 0.871, + "H_a": 0.063, + "d_bind": 0.01784 + } + ] + }, + { + "step": 12860, + "tok": 52674560, + "ce": 2.7585, + "ppl": 15.8, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.95, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3082, + "loop_stats": [ + { + "tau": 0.1384, + "v": 0.8493, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0491, + "v": 0.8331, + "H_a": 0.081, + "d_bind": 0.04705 + }, + { + "tau": 0.0473, + "v": 0.8324, + "H_a": 0.061, + "d_bind": 0.01753 + } + ] + }, + { + "step": 12880, + "tok": 52756480, + "ce": 2.8816, + "ppl": 17.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3045, + "loop_stats": [ + { + "tau": 0.1221, + "v": 0.8352, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8288, + "H_a": 0.081, + "d_bind": 0.04072 + }, + { + "tau": 0.0369, + "v": 0.8339, + "H_a": 0.065, + "d_bind": 0.01845 + } + ] + }, + { + "step": 12900, + "tok": 52838400, + "ce": 3.0648, + "ppl": 21.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3077, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.8483, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8445, + "H_a": 0.077, + "d_bind": 0.04216 + }, + { + "tau": 0.0285, + "v": 0.8511, + "H_a": 0.069, + "d_bind": 0.01752 + } + ] + }, + { + "step": 12920, + "tok": 52920320, + "ce": 2.8456, + "ppl": 17.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3078, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.8244, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8206, + "H_a": 0.084, + "d_bind": 0.0436 + }, + { + "tau": 0.0365, + "v": 0.8306, + "H_a": 0.065, + "d_bind": 0.01776 + } + ] + }, + { + "step": 12940, + "tok": 53002240, + "ce": 2.6641, + "ppl": 14.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3061, + "loop_stats": [ + { + "tau": 0.1168, + "v": 0.8277, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8219, + "H_a": 0.085, + "d_bind": 0.04312 + }, + { + "tau": 0.0403, + "v": 0.8237, + "H_a": 0.066, + "d_bind": 0.01826 + } + ] + }, + { + "step": 12960, + "tok": 53084160, + "ce": 2.9647, + "ppl": 19.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.957, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3089, + "loop_stats": [ + { + "tau": 0.1237, + "v": 0.8583, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8451, + "H_a": 0.083, + "d_bind": 0.03938 + }, + { + "tau": 0.0318, + "v": 0.848, + "H_a": 0.066, + "d_bind": 0.01792 + } + ] + }, + { + "step": 12980, + "tok": 53166080, + "ce": 2.5046, + "ppl": 12.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3084, + "loop_stats": [ + { + "tau": 0.1387, + "v": 0.8474, + "H_a": 0.125, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8409, + "H_a": 0.084, + "d_bind": 0.04488 + }, + { + "tau": 0.0339, + "v": 0.8491, + "H_a": 0.062, + "d_bind": 0.01729 + } + ] + }, + { + "step": 13000, + "tok": 53248000, + "ce": 2.3196, + "ppl": 10.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3076, + "loop_stats": [ + { + "tau": 0.1538, + "v": 0.8563, + "H_a": 0.122, + "d_bind": 0.0 + }, + { + "tau": 0.0624, + "v": 0.8496, + "H_a": 0.088, + "d_bind": 0.04473 + }, + { + "tau": 0.0495, + "v": 0.8555, + "H_a": 0.067, + "d_bind": 0.01893 + } + ] + }, + { + "step": 13020, + "tok": 53329920, + "ce": 2.4904, + "ppl": 12.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3102, + "loop_stats": [ + { + "tau": 0.1534, + "v": 0.8372, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0389, + "v": 0.8356, + "H_a": 0.082, + "d_bind": 0.04177 + }, + { + "tau": 0.0316, + "v": 0.8396, + "H_a": 0.06, + "d_bind": 0.01696 + } + ] + }, + { + "step": 13040, + "tok": 53411840, + "ce": 2.6388, + "ppl": 14.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3104, + "loop_stats": [ + { + "tau": 0.1176, + "v": 0.8786, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8756, + "H_a": 0.086, + "d_bind": 0.04216 + }, + { + "tau": 0.0301, + "v": 0.8807, + "H_a": 0.065, + "d_bind": 0.01823 + } + ] + }, + { + "step": 13060, + "tok": 53493760, + "ce": 2.8131, + "ppl": 16.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.938, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3086, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.8472, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.055, + "v": 0.8354, + "H_a": 0.085, + "d_bind": 0.04071 + }, + { + "tau": 0.0482, + "v": 0.8362, + "H_a": 0.062, + "d_bind": 0.01716 + } + ] + }, + { + "step": 13080, + "tok": 53575680, + "ce": 3.0467, + "ppl": 21.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.925, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.307, + "loop_stats": [ + { + "tau": 0.1199, + "v": 0.8745, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8691, + "H_a": 0.074, + "d_bind": 0.03919 + }, + { + "tau": 0.0298, + "v": 0.8732, + "H_a": 0.063, + "d_bind": 0.01714 + } + ] + }, + { + "step": 13100, + "tok": 53657600, + "ce": 2.495, + "ppl": 12.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3077, + "loop_stats": [ + { + "tau": 0.1725, + "v": 0.8434, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0629, + "v": 0.8279, + "H_a": 0.076, + "d_bind": 0.04204 + }, + { + "tau": 0.0573, + "v": 0.8217, + "H_a": 0.065, + "d_bind": 0.01723 + } + ] + }, + { + "step": 13120, + "tok": 53739520, + "ce": 2.6926, + "ppl": 14.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.933, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3125, + "loop_stats": [ + { + "tau": 0.1395, + "v": 0.8427, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0461, + "v": 0.8381, + "H_a": 0.078, + "d_bind": 0.03966 + }, + { + "tau": 0.0336, + "v": 0.8361, + "H_a": 0.058, + "d_bind": 0.01679 + } + ] + }, + { + "step": 13140, + "tok": 53821440, + "ce": 3.2454, + "ppl": 25.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3151, + "loop_stats": [ + { + "tau": 0.0874, + "v": 0.8806, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0244, + "v": 0.8823, + "H_a": 0.082, + "d_bind": 0.03774 + }, + { + "tau": 0.0165, + "v": 0.8894, + "H_a": 0.065, + "d_bind": 0.01816 + } + ] + }, + { + "step": 13160, + "tok": 53903360, + "ce": 3.0302, + "ppl": 20.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3133, + "loop_stats": [ + { + "tau": 0.1124, + "v": 0.858, + "H_a": 0.113, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8596, + "H_a": 0.085, + "d_bind": 0.03865 + }, + { + "tau": 0.0282, + "v": 0.8637, + "H_a": 0.063, + "d_bind": 0.01855 + } + ] + }, + { + "step": 13180, + "tok": 53985280, + "ce": 3.1058, + "ppl": 22.3, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.94, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3167, + "loop_stats": [ + { + "tau": 0.1359, + "v": 0.8443, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8385, + "H_a": 0.078, + "d_bind": 0.03991 + }, + { + "tau": 0.0321, + "v": 0.845, + "H_a": 0.069, + "d_bind": 0.01845 + } + ] + }, + { + "step": 13200, + "tok": 54067200, + "ce": 2.571, + "ppl": 13.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3173, + "loop_stats": [ + { + "tau": 0.1647, + "v": 0.8649, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.86, + "H_a": 0.082, + "d_bind": 0.03946 + }, + { + "tau": 0.0395, + "v": 0.8652, + "H_a": 0.059, + "d_bind": 0.0187 + } + ] + }, + { + "step": 13220, + "tok": 54149120, + "ce": 2.7124, + "ppl": 15.1, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.93, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3155, + "loop_stats": [ + { + "tau": 0.1271, + "v": 0.8437, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0415, + "v": 0.8297, + "H_a": 0.08, + "d_bind": 0.03988 + }, + { + "tau": 0.0356, + "v": 0.829, + "H_a": 0.068, + "d_bind": 0.01841 + } + ] + }, + { + "step": 13240, + "tok": 54231040, + "ce": 3.2016, + "ppl": 24.6, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.947, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3165, + "loop_stats": [ + { + "tau": 0.0953, + "v": 0.8719, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.8597, + "H_a": 0.083, + "d_bind": 0.03787 + }, + { + "tau": 0.0272, + "v": 0.8634, + "H_a": 0.063, + "d_bind": 0.01971 + } + ] + }, + { + "step": 13260, + "tok": 54312960, + "ce": 3.3165, + "ppl": 27.6, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.93, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3144, + "loop_stats": [ + { + "tau": 0.0725, + "v": 0.9032, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0294, + "v": 0.9011, + "H_a": 0.076, + "d_bind": 0.03471 + }, + { + "tau": 0.0199, + "v": 0.9013, + "H_a": 0.067, + "d_bind": 0.01809 + } + ] + }, + { + "step": 13280, + "tok": 54394880, + "ce": 2.894, + "ppl": 18.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3153, + "loop_stats": [ + { + "tau": 0.1163, + "v": 0.8867, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.8779, + "H_a": 0.081, + "d_bind": 0.03904 + }, + { + "tau": 0.0331, + "v": 0.8794, + "H_a": 0.07, + "d_bind": 0.01911 + } + ] + }, + { + "step": 13300, + "tok": 54476800, + "ce": 2.771, + "ppl": 16.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3196, + "loop_stats": [ + { + "tau": 0.1272, + "v": 0.8751, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8711, + "H_a": 0.08, + "d_bind": 0.03847 + }, + { + "tau": 0.031, + "v": 0.8773, + "H_a": 0.069, + "d_bind": 0.01884 + } + ] + }, + { + "step": 13320, + "tok": 54558720, + "ce": 2.8944, + "ppl": 18.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3196, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.8834, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.8713, + "H_a": 0.081, + "d_bind": 0.03949 + }, + { + "tau": 0.0336, + "v": 0.8684, + "H_a": 0.067, + "d_bind": 0.01861 + } + ] + }, + { + "step": 13340, + "tok": 54640640, + "ce": 2.6423, + "ppl": 14.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.933, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3208, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.8685, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8673, + "H_a": 0.083, + "d_bind": 0.0387 + }, + { + "tau": 0.0457, + "v": 0.8707, + "H_a": 0.062, + "d_bind": 0.01994 + } + ] + }, + { + "step": 13360, + "tok": 54722560, + "ce": 2.9019, + "ppl": 18.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3232, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8623, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0317, + "v": 0.8426, + "H_a": 0.086, + "d_bind": 0.03819 + }, + { + "tau": 0.0272, + "v": 0.8396, + "H_a": 0.067, + "d_bind": 0.02029 + } + ] + }, + { + "step": 13380, + "tok": 54804480, + "ce": 3.1728, + "ppl": 23.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3243, + "loop_stats": [ + { + "tau": 0.1074, + "v": 0.9069, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.8966, + "H_a": 0.081, + "d_bind": 0.03923 + }, + { + "tau": 0.0165, + "v": 0.892, + "H_a": 0.062, + "d_bind": 0.01679 + } + ] + }, + { + "step": 13400, + "tok": 54886400, + "ce": 3.3033, + "ppl": 27.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3254, + "loop_stats": [ + { + "tau": 0.083, + "v": 0.8943, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0203, + "v": 0.883, + "H_a": 0.083, + "d_bind": 0.0399 + }, + { + "tau": 0.0154, + "v": 0.8779, + "H_a": 0.063, + "d_bind": 0.01863 + } + ] + }, + { + "step": 13420, + "tok": 54968320, + "ce": 2.7577, + "ppl": 15.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3243, + "loop_stats": [ + { + "tau": 0.1286, + "v": 0.8842, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.8656, + "H_a": 0.085, + "d_bind": 0.04025 + }, + { + "tau": 0.0261, + "v": 0.862, + "H_a": 0.062, + "d_bind": 0.01676 + } + ] + }, + { + "step": 13440, + "tok": 55050240, + "ce": 3.1552, + "ppl": 23.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3279, + "loop_stats": [ + { + "tau": 0.1057, + "v": 0.8844, + "H_a": 0.111, + "d_bind": 0.0 + }, + { + "tau": 0.0217, + "v": 0.8769, + "H_a": 0.082, + "d_bind": 0.04216 + }, + { + "tau": 0.0174, + "v": 0.8779, + "H_a": 0.064, + "d_bind": 0.01913 + } + ] + }, + { + "step": 13460, + "tok": 55132160, + "ce": 3.1295, + "ppl": 22.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3254, + "loop_stats": [ + { + "tau": 0.1123, + "v": 0.8691, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0369, + "v": 0.8595, + "H_a": 0.075, + "d_bind": 0.03793 + }, + { + "tau": 0.0273, + "v": 0.8596, + "H_a": 0.064, + "d_bind": 0.01799 + } + ] + }, + { + "step": 13480, + "tok": 55214080, + "ce": 2.9491, + "ppl": 19.1, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.94, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3224, + "loop_stats": [ + { + "tau": 0.1088, + "v": 0.8927, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0301, + "v": 0.8834, + "H_a": 0.082, + "d_bind": 0.03609 + }, + { + "tau": 0.0212, + "v": 0.8839, + "H_a": 0.067, + "d_bind": 0.01943 + } + ] + }, + { + "step": 13500, + "tok": 55296000, + "ce": 3.0973, + "ppl": 22.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3204, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8942, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.8932, + "H_a": 0.083, + "d_bind": 0.03519 + }, + { + "tau": 0.0208, + "v": 0.8953, + "H_a": 0.064, + "d_bind": 0.01929 + } + ] + }, + { + "step": 13520, + "tok": 55377920, + "ce": 3.0556, + "ppl": 21.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3166, + "loop_stats": [ + { + "tau": 0.1034, + "v": 0.9003, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0396, + "v": 0.8949, + "H_a": 0.078, + "d_bind": 0.03743 + }, + { + "tau": 0.0319, + "v": 0.8963, + "H_a": 0.066, + "d_bind": 0.01926 + } + ] + }, + { + "step": 13540, + "tok": 55459840, + "ce": 2.9858, + "ppl": 19.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3177, + "loop_stats": [ + { + "tau": 0.1296, + "v": 0.8758, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8716, + "H_a": 0.086, + "d_bind": 0.03677 + }, + { + "tau": 0.0382, + "v": 0.8714, + "H_a": 0.064, + "d_bind": 0.01827 + } + ] + }, + { + "step": 13560, + "tok": 55541760, + "ce": 3.1113, + "ppl": 22.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3161, + "loop_stats": [ + { + "tau": 0.1171, + "v": 0.8949, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8906, + "H_a": 0.077, + "d_bind": 0.03768 + }, + { + "tau": 0.0288, + "v": 0.8913, + "H_a": 0.065, + "d_bind": 0.01809 + } + ] + }, + { + "step": 13580, + "tok": 55623680, + "ce": 2.6881, + "ppl": 14.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.314, + "loop_stats": [ + { + "tau": 0.1219, + "v": 0.8761, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8688, + "H_a": 0.078, + "d_bind": 0.03884 + }, + { + "tau": 0.0469, + "v": 0.8705, + "H_a": 0.064, + "d_bind": 0.01975 + } + ] + }, + { + "step": 13600, + "tok": 55705600, + "ce": 2.7364, + "ppl": 15.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3154, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.88, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0446, + "v": 0.8698, + "H_a": 0.087, + "d_bind": 0.03716 + }, + { + "tau": 0.0409, + "v": 0.8702, + "H_a": 0.065, + "d_bind": 0.01968 + } + ] + }, + { + "step": 13620, + "tok": 55787520, + "ce": 3.0455, + "ppl": 21.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3123, + "loop_stats": [ + { + "tau": 0.1181, + "v": 0.8808, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0495, + "v": 0.8641, + "H_a": 0.086, + "d_bind": 0.0378 + }, + { + "tau": 0.0468, + "v": 0.8606, + "H_a": 0.067, + "d_bind": 0.01917 + } + ] + }, + { + "step": 13640, + "tok": 55869440, + "ce": 3.1511, + "ppl": 23.4, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.947, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3136, + "loop_stats": [ + { + "tau": 0.1154, + "v": 0.8704, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8596, + "H_a": 0.086, + "d_bind": 0.03684 + }, + { + "tau": 0.0292, + "v": 0.8605, + "H_a": 0.069, + "d_bind": 0.01901 + } + ] + }, + { + "step": 13660, + "tok": 55951360, + "ce": 3.1425, + "ppl": 23.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.938, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3132, + "loop_stats": [ + { + "tau": 0.1041, + "v": 0.8518, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8395, + "H_a": 0.079, + "d_bind": 0.04081 + }, + { + "tau": 0.0448, + "v": 0.8383, + "H_a": 0.067, + "d_bind": 0.01934 + } + ] + }, + { + "step": 13680, + "tok": 56033280, + "ce": 2.6496, + "ppl": 14.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3156, + "loop_stats": [ + { + "tau": 0.1165, + "v": 0.8576, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.8451, + "H_a": 0.08, + "d_bind": 0.03911 + }, + { + "tau": 0.0302, + "v": 0.8487, + "H_a": 0.066, + "d_bind": 0.01931 + } + ] + }, + { + "step": 13700, + "tok": 56115200, + "ce": 3.2943, + "ppl": 27.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.316, + "loop_stats": [ + { + "tau": 0.105, + "v": 0.8777, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.8747, + "H_a": 0.079, + "d_bind": 0.0392 + }, + { + "tau": 0.0302, + "v": 0.8759, + "H_a": 0.068, + "d_bind": 0.01938 + } + ] + }, + { + "step": 13720, + "tok": 56197120, + "ce": 2.8602, + "ppl": 17.5, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.952, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3151, + "loop_stats": [ + { + "tau": 0.0943, + "v": 0.862, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8547, + "H_a": 0.083, + "d_bind": 0.04125 + }, + { + "tau": 0.0325, + "v": 0.8555, + "H_a": 0.068, + "d_bind": 0.02025 + } + ] + }, + { + "step": 13740, + "tok": 56279040, + "ce": 3.0113, + "ppl": 20.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.96, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3128, + "loop_stats": [ + { + "tau": 0.0926, + "v": 0.8777, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0476, + "v": 0.8641, + "H_a": 0.078, + "d_bind": 0.03761 + }, + { + "tau": 0.0418, + "v": 0.8637, + "H_a": 0.062, + "d_bind": 0.01977 + } + ] + }, + { + "step": 13760, + "tok": 56360960, + "ce": 3.0054, + "ppl": 20.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3088, + "loop_stats": [ + { + "tau": 0.1083, + "v": 0.8764, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0496, + "v": 0.8683, + "H_a": 0.077, + "d_bind": 0.03989 + }, + { + "tau": 0.0461, + "v": 0.8678, + "H_a": 0.061, + "d_bind": 0.0181 + } + ] + }, + { + "step": 13780, + "tok": 56442880, + "ce": 2.8064, + "ppl": 16.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3074, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.8623, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.844, + "H_a": 0.078, + "d_bind": 0.03983 + }, + { + "tau": 0.0394, + "v": 0.8412, + "H_a": 0.066, + "d_bind": 0.02115 + } + ] + }, + { + "step": 13800, + "tok": 56524800, + "ce": 2.7412, + "ppl": 15.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3085, + "loop_stats": [ + { + "tau": 0.1088, + "v": 0.8771, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0518, + "v": 0.8744, + "H_a": 0.084, + "d_bind": 0.03978 + }, + { + "tau": 0.0461, + "v": 0.8856, + "H_a": 0.071, + "d_bind": 0.02149 + } + ] + }, + { + "step": 13820, + "tok": 56606720, + "ce": 3.0015, + "ppl": 20.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.942, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.309, + "loop_stats": [ + { + "tau": 0.1136, + "v": 0.8993, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8896, + "H_a": 0.08, + "d_bind": 0.03815 + }, + { + "tau": 0.0406, + "v": 0.889, + "H_a": 0.068, + "d_bind": 0.02045 + } + ] + }, + { + "step": 13840, + "tok": 56688640, + "ce": 3.0122, + "ppl": 20.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.31, + "loop_stats": [ + { + "tau": 0.1012, + "v": 0.9076, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.8879, + "H_a": 0.073, + "d_bind": 0.04038 + }, + { + "tau": 0.0408, + "v": 0.8835, + "H_a": 0.059, + "d_bind": 0.0186 + } + ] + }, + { + "step": 13860, + "tok": 56770560, + "ce": 2.9955, + "ppl": 20.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3097, + "loop_stats": [ + { + "tau": 0.1162, + "v": 0.8853, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8818, + "H_a": 0.082, + "d_bind": 0.03884 + }, + { + "tau": 0.0334, + "v": 0.8846, + "H_a": 0.059, + "d_bind": 0.01851 + } + ] + }, + { + "step": 13880, + "tok": 56852480, + "ce": 3.085, + "ppl": 21.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.952, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3048, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.8599, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0537, + "v": 0.843, + "H_a": 0.075, + "d_bind": 0.03923 + }, + { + "tau": 0.0506, + "v": 0.8451, + "H_a": 0.061, + "d_bind": 0.01866 + } + ] + }, + { + "step": 13900, + "tok": 56934400, + "ce": 2.9413, + "ppl": 18.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3068, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.8582, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.8558, + "H_a": 0.082, + "d_bind": 0.03638 + }, + { + "tau": 0.0457, + "v": 0.8612, + "H_a": 0.065, + "d_bind": 0.01881 + } + ] + }, + { + "step": 13920, + "tok": 57016320, + "ce": 3.3471, + "ppl": 28.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.308, + "loop_stats": [ + { + "tau": 0.1037, + "v": 0.8473, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8325, + "H_a": 0.086, + "d_bind": 0.03812 + }, + { + "tau": 0.0403, + "v": 0.8378, + "H_a": 0.067, + "d_bind": 0.01946 + } + ] + }, + { + "step": 13940, + "tok": 57098240, + "ce": 3.4354, + "ppl": 31.0, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.952, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3102, + "loop_stats": [ + { + "tau": 0.0857, + "v": 0.8546, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8435, + "H_a": 0.085, + "d_bind": 0.04265 + }, + { + "tau": 0.0327, + "v": 0.8422, + "H_a": 0.067, + "d_bind": 0.02063 + } + ] + }, + { + "step": 13960, + "tok": 57180160, + "ce": 3.4103, + "ppl": 30.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3114, + "loop_stats": [ + { + "tau": 0.0835, + "v": 0.8816, + "H_a": 0.112, + "d_bind": 0.0 + }, + { + "tau": 0.0269, + "v": 0.8749, + "H_a": 0.093, + "d_bind": 0.03891 + }, + { + "tau": 0.024, + "v": 0.8793, + "H_a": 0.076, + "d_bind": 0.02179 + } + ] + }, + { + "step": 13980, + "tok": 57262080, + "ce": 3.2358, + "ppl": 25.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3098, + "loop_stats": [ + { + "tau": 0.0854, + "v": 0.8857, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8668, + "H_a": 0.091, + "d_bind": 0.04112 + }, + { + "tau": 0.0326, + "v": 0.8656, + "H_a": 0.069, + "d_bind": 0.0221 + } + ] + }, + { + "step": 14000, + "tok": 57344000, + "ce": 3.5655, + "ppl": 35.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3099, + "loop_stats": [ + { + "tau": 0.0798, + "v": 0.8791, + "H_a": 0.115, + "d_bind": 0.0 + }, + { + "tau": 0.0277, + "v": 0.8745, + "H_a": 0.09, + "d_bind": 0.03972 + }, + { + "tau": 0.0254, + "v": 0.8768, + "H_a": 0.065, + "d_bind": 0.02035 + } + ] + }, + { + "step": 14020, + "tok": 57425920, + "ce": 3.1705, + "ppl": 23.8, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.93, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3104, + "loop_stats": [ + { + "tau": 0.0954, + "v": 0.8853, + "H_a": 0.114, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8806, + "H_a": 0.089, + "d_bind": 0.04232 + }, + { + "tau": 0.0378, + "v": 0.8828, + "H_a": 0.072, + "d_bind": 0.02058 + } + ] + }, + { + "step": 14040, + "tok": 57507840, + "ce": 2.9769, + "ppl": 19.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3127, + "loop_stats": [ + { + "tau": 0.096, + "v": 0.8865, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.885, + "H_a": 0.081, + "d_bind": 0.036 + }, + { + "tau": 0.0389, + "v": 0.8948, + "H_a": 0.065, + "d_bind": 0.01842 + } + ] + }, + { + "step": 14060, + "tok": 57589760, + "ce": 3.1376, + "ppl": 23.0, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.952, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3128, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8773, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0496, + "v": 0.8635, + "H_a": 0.078, + "d_bind": 0.03879 + }, + { + "tau": 0.0476, + "v": 0.8627, + "H_a": 0.059, + "d_bind": 0.01894 + } + ] + }, + { + "step": 14080, + "tok": 57671680, + "ce": 2.7253, + "ppl": 15.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3088, + "loop_stats": [ + { + "tau": 0.1245, + "v": 0.8657, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0485, + "v": 0.8561, + "H_a": 0.085, + "d_bind": 0.04027 + }, + { + "tau": 0.0457, + "v": 0.8514, + "H_a": 0.061, + "d_bind": 0.01996 + } + ] + }, + { + "step": 14100, + "tok": 57753600, + "ce": 2.7466, + "ppl": 15.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3076, + "loop_stats": [ + { + "tau": 0.1407, + "v": 0.8749, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0476, + "v": 0.8541, + "H_a": 0.085, + "d_bind": 0.03656 + }, + { + "tau": 0.0405, + "v": 0.8526, + "H_a": 0.064, + "d_bind": 0.0212 + } + ] + }, + { + "step": 14120, + "tok": 57835520, + "ce": 2.454, + "ppl": 11.6, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.955, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3076, + "loop_stats": [ + { + "tau": 0.1385, + "v": 0.856, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.8347, + "H_a": 0.085, + "d_bind": 0.03821 + }, + { + "tau": 0.0425, + "v": 0.8289, + "H_a": 0.067, + "d_bind": 0.02247 + } + ] + }, + { + "step": 14140, + "tok": 57917440, + "ce": 2.8828, + "ppl": 17.9, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.945, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3061, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.8531, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0547, + "v": 0.8341, + "H_a": 0.085, + "d_bind": 0.03772 + }, + { + "tau": 0.0438, + "v": 0.8293, + "H_a": 0.063, + "d_bind": 0.01982 + } + ] + }, + { + "step": 14160, + "tok": 57999360, + "ce": 2.961, + "ppl": 19.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3077, + "loop_stats": [ + { + "tau": 0.1233, + "v": 0.8575, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0491, + "v": 0.8429, + "H_a": 0.084, + "d_bind": 0.0394 + }, + { + "tau": 0.0398, + "v": 0.839, + "H_a": 0.066, + "d_bind": 0.02049 + } + ] + }, + { + "step": 14180, + "tok": 58081280, + "ce": 3.0678, + "ppl": 21.5, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.95, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3085, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8172, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.7857, + "H_a": 0.081, + "d_bind": 0.03655 + }, + { + "tau": 0.0302, + "v": 0.7854, + "H_a": 0.061, + "d_bind": 0.01944 + } + ] + }, + { + "step": 14200, + "tok": 58163200, + "ce": 2.6181, + "ppl": 13.7, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.935, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3087, + "loop_stats": [ + { + "tau": 0.1408, + "v": 0.8332, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0621, + "v": 0.7938, + "H_a": 0.086, + "d_bind": 0.04193 + }, + { + "tau": 0.0537, + "v": 0.7914, + "H_a": 0.062, + "d_bind": 0.02045 + } + ] + }, + { + "step": 14220, + "tok": 58245120, + "ce": 2.4896, + "ppl": 12.1, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.933, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3109, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.8315, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8128, + "H_a": 0.084, + "d_bind": 0.0345 + }, + { + "tau": 0.0349, + "v": 0.8094, + "H_a": 0.073, + "d_bind": 0.02398 + } + ] + }, + { + "step": 14240, + "tok": 58327040, + "ce": 2.6709, + "ppl": 14.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.93, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3141, + "loop_stats": [ + { + "tau": 0.1344, + "v": 0.8174, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.7922, + "H_a": 0.082, + "d_bind": 0.04287 + }, + { + "tau": 0.0427, + "v": 0.791, + "H_a": 0.066, + "d_bind": 0.02063 + } + ] + }, + { + "step": 14260, + "tok": 58408960, + "ce": 3.1292, + "ppl": 22.9, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.945, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.318, + "loop_stats": [ + { + "tau": 0.0976, + "v": 0.8028, + "H_a": 0.116, + "d_bind": 0.0 + }, + { + "tau": 0.0292, + "v": 0.7828, + "H_a": 0.081, + "d_bind": 0.03978 + }, + { + "tau": 0.0239, + "v": 0.7793, + "H_a": 0.063, + "d_bind": 0.01913 + } + ] + }, + { + "step": 14280, + "tok": 58490880, + "ce": 3.0013, + "ppl": 20.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3171, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.7567, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0516, + "v": 0.7525, + "H_a": 0.077, + "d_bind": 0.03576 + }, + { + "tau": 0.0449, + "v": 0.7618, + "H_a": 0.064, + "d_bind": 0.02209 + } + ] + }, + { + "step": 14300, + "tok": 58572800, + "ce": 3.1582, + "ppl": 23.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3197, + "loop_stats": [ + { + "tau": 0.1039, + "v": 0.7907, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0406, + "v": 0.7945, + "H_a": 0.075, + "d_bind": 0.03562 + }, + { + "tau": 0.0363, + "v": 0.8055, + "H_a": 0.063, + "d_bind": 0.01964 + } + ] + }, + { + "step": 14320, + "tok": 58654720, + "ce": 3.152, + "ppl": 23.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.323, + "loop_stats": [ + { + "tau": 0.0868, + "v": 0.7327, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0316, + "v": 0.7253, + "H_a": 0.084, + "d_bind": 0.03955 + }, + { + "tau": 0.0291, + "v": 0.7274, + "H_a": 0.066, + "d_bind": 0.02101 + } + ] + }, + { + "step": 14340, + "tok": 58736640, + "ce": 3.1811, + "ppl": 24.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3271, + "loop_stats": [ + { + "tau": 0.097, + "v": 0.7849, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0369, + "v": 0.7577, + "H_a": 0.082, + "d_bind": 0.0375 + }, + { + "tau": 0.0299, + "v": 0.7567, + "H_a": 0.07, + "d_bind": 0.02166 + } + ] + }, + { + "step": 14360, + "tok": 58818560, + "ce": 3.3495, + "ppl": 28.5, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.95, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3277, + "loop_stats": [ + { + "tau": 0.1059, + "v": 0.7677, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.77, + "H_a": 0.078, + "d_bind": 0.03455 + }, + { + "tau": 0.0322, + "v": 0.7775, + "H_a": 0.065, + "d_bind": 0.02134 + } + ] + }, + { + "step": 14380, + "tok": 58900480, + "ce": 3.4568, + "ppl": 31.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3271, + "loop_stats": [ + { + "tau": 0.1048, + "v": 0.7902, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0392, + "v": 0.7725, + "H_a": 0.08, + "d_bind": 0.03503 + }, + { + "tau": 0.0389, + "v": 0.7717, + "H_a": 0.069, + "d_bind": 0.02144 + } + ] + }, + { + "step": 14400, + "tok": 58982400, + "ce": 3.3449, + "ppl": 28.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3243, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.7667, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.757, + "H_a": 0.078, + "d_bind": 0.0369 + }, + { + "tau": 0.0343, + "v": 0.7618, + "H_a": 0.064, + "d_bind": 0.01862 + } + ] + }, + { + "step": 14420, + "tok": 59064320, + "ce": 2.8728, + "ppl": 17.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.323, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.7784, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.7483, + "H_a": 0.081, + "d_bind": 0.03723 + }, + { + "tau": 0.0475, + "v": 0.7357, + "H_a": 0.067, + "d_bind": 0.02177 + } + ] + }, + { + "step": 14440, + "tok": 59146240, + "ce": 3.3887, + "ppl": 29.6, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.95, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3246, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.8241, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0231, + "v": 0.8183, + "H_a": 0.077, + "d_bind": 0.03418 + }, + { + "tau": 0.0211, + "v": 0.8226, + "H_a": 0.062, + "d_bind": 0.02158 + } + ] + }, + { + "step": 14460, + "tok": 59228160, + "ce": 3.1444, + "ppl": 23.2, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.95, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3264, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.833, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.8257, + "H_a": 0.077, + "d_bind": 0.03889 + }, + { + "tau": 0.0398, + "v": 0.8279, + "H_a": 0.064, + "d_bind": 0.01887 + } + ] + }, + { + "step": 14480, + "tok": 59310080, + "ce": 3.1043, + "ppl": 22.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3285, + "loop_stats": [ + { + "tau": 0.0959, + "v": 0.8014, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.7783, + "H_a": 0.08, + "d_bind": 0.03587 + }, + { + "tau": 0.0258, + "v": 0.7794, + "H_a": 0.073, + "d_bind": 0.02167 + } + ] + }, + { + "step": 14500, + "tok": 59392000, + "ce": 3.1713, + "ppl": 23.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.957, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3319, + "loop_stats": [ + { + "tau": 0.0779, + "v": 0.755, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0225, + "v": 0.7274, + "H_a": 0.083, + "d_bind": 0.03935 + }, + { + "tau": 0.0209, + "v": 0.7251, + "H_a": 0.068, + "d_bind": 0.02343 + } + ] + }, + { + "step": 14520, + "tok": 59473920, + "ce": 2.982, + "ppl": 19.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3334, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.7821, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.7633, + "H_a": 0.075, + "d_bind": 0.03696 + }, + { + "tau": 0.0492, + "v": 0.7666, + "H_a": 0.067, + "d_bind": 0.01965 + } + ] + }, + { + "step": 14540, + "tok": 59555840, + "ce": 3.1859, + "ppl": 24.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3307, + "loop_stats": [ + { + "tau": 0.1078, + "v": 0.7737, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0421, + "v": 0.751, + "H_a": 0.076, + "d_bind": 0.0368 + }, + { + "tau": 0.0395, + "v": 0.7474, + "H_a": 0.065, + "d_bind": 0.0221 + } + ] + }, + { + "step": 14560, + "tok": 59637760, + "ce": 2.7045, + "ppl": 14.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.935, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3305, + "loop_stats": [ + { + "tau": 0.1457, + "v": 0.7639, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0521, + "v": 0.7328, + "H_a": 0.073, + "d_bind": 0.03503 + }, + { + "tau": 0.0504, + "v": 0.7272, + "H_a": 0.063, + "d_bind": 0.02111 + } + ] + }, + { + "step": 14580, + "tok": 59719680, + "ce": 2.5227, + "ppl": 12.5, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.955, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3275, + "loop_stats": [ + { + "tau": 0.1143, + "v": 0.8223, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.7889, + "H_a": 0.08, + "d_bind": 0.03595 + }, + { + "tau": 0.0401, + "v": 0.781, + "H_a": 0.065, + "d_bind": 0.0212 + } + ] + }, + { + "step": 14600, + "tok": 59801600, + "ce": 3.6822, + "ppl": 39.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3284, + "loop_stats": [ + { + "tau": 0.0824, + "v": 0.793, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.7707, + "H_a": 0.085, + "d_bind": 0.0364 + }, + { + "tau": 0.0292, + "v": 0.7681, + "H_a": 0.072, + "d_bind": 0.02078 + } + ] + }, + { + "step": 14620, + "tok": 59883520, + "ce": 2.9504, + "ppl": 19.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3299, + "loop_stats": [ + { + "tau": 0.1208, + "v": 0.8184, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.8052, + "H_a": 0.086, + "d_bind": 0.03858 + }, + { + "tau": 0.0381, + "v": 0.7967, + "H_a": 0.07, + "d_bind": 0.02345 + } + ] + }, + { + "step": 14640, + "tok": 59965440, + "ce": 2.8754, + "ppl": 17.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3325, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.8183, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0446, + "v": 0.8032, + "H_a": 0.088, + "d_bind": 0.03864 + }, + { + "tau": 0.0404, + "v": 0.7995, + "H_a": 0.072, + "d_bind": 0.02138 + } + ] + }, + { + "step": 14660, + "tok": 60047360, + "ce": 3.3401, + "ppl": 28.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3353, + "loop_stats": [ + { + "tau": 0.0982, + "v": 0.807, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0362, + "v": 0.7735, + "H_a": 0.084, + "d_bind": 0.03674 + }, + { + "tau": 0.0328, + "v": 0.7647, + "H_a": 0.066, + "d_bind": 0.02067 + } + ] + }, + { + "step": 14680, + "tok": 60129280, + "ce": 3.1598, + "ppl": 23.6, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.94, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.334, + "loop_stats": [ + { + "tau": 0.1145, + "v": 0.7624, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0542, + "v": 0.724, + "H_a": 0.087, + "d_bind": 0.04093 + }, + { + "tau": 0.0506, + "v": 0.7071, + "H_a": 0.067, + "d_bind": 0.02163 + } + ] + }, + { + "step": 14700, + "tok": 60211200, + "ce": 3.2092, + "ppl": 24.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3366, + "loop_stats": [ + { + "tau": 0.088, + "v": 0.8268, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8113, + "H_a": 0.078, + "d_bind": 0.03718 + }, + { + "tau": 0.022, + "v": 0.8071, + "H_a": 0.068, + "d_bind": 0.02073 + } + ] + }, + { + "step": 14720, + "tok": 60293120, + "ce": 3.166, + "ppl": 23.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3354, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.7676, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.7479, + "H_a": 0.08, + "d_bind": 0.03908 + }, + { + "tau": 0.044, + "v": 0.7427, + "H_a": 0.068, + "d_bind": 0.022 + } + ] + }, + { + "step": 14740, + "tok": 60375040, + "ce": 2.6621, + "ppl": 14.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3335, + "loop_stats": [ + { + "tau": 0.1381, + "v": 0.8106, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.7945, + "H_a": 0.076, + "d_bind": 0.03813 + }, + { + "tau": 0.0512, + "v": 0.7986, + "H_a": 0.065, + "d_bind": 0.02178 + } + ] + }, + { + "step": 14760, + "tok": 60456960, + "ce": 2.9861, + "ppl": 19.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3354, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.8128, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.7893, + "H_a": 0.086, + "d_bind": 0.03968 + }, + { + "tau": 0.0317, + "v": 0.7795, + "H_a": 0.065, + "d_bind": 0.02045 + } + ] + }, + { + "step": 14780, + "tok": 60538880, + "ce": 3.269, + "ppl": 26.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3342, + "loop_stats": [ + { + "tau": 0.0982, + "v": 0.7855, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.7437, + "H_a": 0.078, + "d_bind": 0.04069 + }, + { + "tau": 0.0269, + "v": 0.7321, + "H_a": 0.067, + "d_bind": 0.02239 + } + ] + }, + { + "step": 14800, + "tok": 60620800, + "ce": 3.1729, + "ppl": 23.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3325, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.807, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0261, + "v": 0.7803, + "H_a": 0.084, + "d_bind": 0.03845 + }, + { + "tau": 0.0235, + "v": 0.7748, + "H_a": 0.063, + "d_bind": 0.02079 + } + ] + }, + { + "step": 14820, + "tok": 60702720, + "ce": 2.9415, + "ppl": 18.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3291, + "loop_stats": [ + { + "tau": 0.1144, + "v": 0.8054, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.7746, + "H_a": 0.073, + "d_bind": 0.04029 + }, + { + "tau": 0.0388, + "v": 0.7741, + "H_a": 0.061, + "d_bind": 0.02171 + } + ] + }, + { + "step": 14840, + "tok": 60784640, + "ce": 3.0427, + "ppl": 21.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3306, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.8662, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0408, + "v": 0.8439, + "H_a": 0.08, + "d_bind": 0.03892 + }, + { + "tau": 0.0353, + "v": 0.8336, + "H_a": 0.064, + "d_bind": 0.02115 + } + ] + }, + { + "step": 14860, + "tok": 60866560, + "ce": 3.1714, + "ppl": 23.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3336, + "loop_stats": [ + { + "tau": 0.0886, + "v": 0.8422, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0316, + "v": 0.8259, + "H_a": 0.081, + "d_bind": 0.03599 + }, + { + "tau": 0.0276, + "v": 0.8222, + "H_a": 0.064, + "d_bind": 0.02352 + } + ] + }, + { + "step": 14880, + "tok": 60948480, + "ce": 2.9957, + "ppl": 20.0, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.947, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3366, + "loop_stats": [ + { + "tau": 0.1031, + "v": 0.8679, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0168, + "v": 0.8524, + "H_a": 0.081, + "d_bind": 0.03287 + }, + { + "tau": 0.0105, + "v": 0.8517, + "H_a": 0.065, + "d_bind": 0.02189 + } + ] + }, + { + "step": 14900, + "tok": 61030400, + "ce": 2.9286, + "ppl": 18.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3352, + "loop_stats": [ + { + "tau": 0.0959, + "v": 0.8165, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.7848, + "H_a": 0.087, + "d_bind": 0.04006 + }, + { + "tau": 0.0292, + "v": 0.7728, + "H_a": 0.07, + "d_bind": 0.02047 + } + ] + }, + { + "step": 14920, + "tok": 61112320, + "ce": 3.1982, + "ppl": 24.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3355, + "loop_stats": [ + { + "tau": 0.0982, + "v": 0.8569, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.024, + "v": 0.844, + "H_a": 0.081, + "d_bind": 0.03767 + }, + { + "tau": 0.0164, + "v": 0.8388, + "H_a": 0.07, + "d_bind": 0.02191 + } + ] + }, + { + "step": 14940, + "tok": 61194240, + "ce": 3.2101, + "ppl": 24.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.336, + "loop_stats": [ + { + "tau": 0.1109, + "v": 0.8204, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.7921, + "H_a": 0.074, + "d_bind": 0.04062 + }, + { + "tau": 0.0286, + "v": 0.7829, + "H_a": 0.065, + "d_bind": 0.01925 + } + ] + }, + { + "step": 14960, + "tok": 61276160, + "ce": 2.8868, + "ppl": 17.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3363, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8452, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.824, + "H_a": 0.083, + "d_bind": 0.04164 + }, + { + "tau": 0.0327, + "v": 0.8157, + "H_a": 0.063, + "d_bind": 0.02021 + } + ] + }, + { + "step": 14980, + "tok": 61358080, + "ce": 2.927, + "ppl": 18.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3366, + "loop_stats": [ + { + "tau": 0.1186, + "v": 0.8007, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.7756, + "H_a": 0.08, + "d_bind": 0.04139 + }, + { + "tau": 0.031, + "v": 0.7724, + "H_a": 0.059, + "d_bind": 0.01969 + } + ] + }, + { + "step": 15000, + "tok": 61440000, + "ce": 3.0404, + "ppl": 20.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3362, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.8426, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0309, + "v": 0.8201, + "H_a": 0.085, + "d_bind": 0.03921 + }, + { + "tau": 0.028, + "v": 0.8152, + "H_a": 0.069, + "d_bind": 0.02106 + } + ] + }, + { + "step": 15020, + "tok": 61521920, + "ce": 3.4667, + "ppl": 32.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.957, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3346, + "loop_stats": [ + { + "tau": 0.0793, + "v": 0.865, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0323, + "v": 0.8415, + "H_a": 0.083, + "d_bind": 0.03565 + }, + { + "tau": 0.0279, + "v": 0.8333, + "H_a": 0.064, + "d_bind": 0.02028 + } + ] + }, + { + "step": 15040, + "tok": 61603840, + "ce": 3.4832, + "ppl": 32.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.96, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3335, + "loop_stats": [ + { + "tau": 0.0775, + "v": 0.8301, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0235, + "v": 0.8115, + "H_a": 0.086, + "d_bind": 0.037 + }, + { + "tau": 0.0215, + "v": 0.8138, + "H_a": 0.068, + "d_bind": 0.02018 + } + ] + }, + { + "step": 15060, + "tok": 61685760, + "ce": 2.7879, + "ppl": 16.2, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.952, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3323, + "loop_stats": [ + { + "tau": 0.1135, + "v": 0.8452, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8252, + "H_a": 0.084, + "d_bind": 0.03993 + }, + { + "tau": 0.0337, + "v": 0.8251, + "H_a": 0.071, + "d_bind": 0.02175 + } + ] + }, + { + "step": 15080, + "tok": 61767680, + "ce": 3.2618, + "ppl": 26.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.335, + "loop_stats": [ + { + "tau": 0.0938, + "v": 0.8565, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0366, + "v": 0.8221, + "H_a": 0.083, + "d_bind": 0.03786 + }, + { + "tau": 0.0311, + "v": 0.8107, + "H_a": 0.067, + "d_bind": 0.02075 + } + ] + }, + { + "step": 15100, + "tok": 61849600, + "ce": 3.0304, + "ppl": 20.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.336, + "loop_stats": [ + { + "tau": 0.0979, + "v": 0.8324, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.8024, + "H_a": 0.084, + "d_bind": 0.04171 + }, + { + "tau": 0.0289, + "v": 0.7935, + "H_a": 0.074, + "d_bind": 0.02235 + } + ] + }, + { + "step": 15120, + "tok": 61931520, + "ce": 2.5986, + "ppl": 13.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3332, + "loop_stats": [ + { + "tau": 0.1296, + "v": 0.8724, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0383, + "v": 0.8501, + "H_a": 0.079, + "d_bind": 0.04178 + }, + { + "tau": 0.0369, + "v": 0.8424, + "H_a": 0.061, + "d_bind": 0.02066 + } + ] + }, + { + "step": 15140, + "tok": 62013440, + "ce": 2.918, + "ppl": 18.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3327, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8929, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0309, + "v": 0.8721, + "H_a": 0.085, + "d_bind": 0.03699 + }, + { + "tau": 0.0257, + "v": 0.8604, + "H_a": 0.065, + "d_bind": 0.02108 + } + ] + }, + { + "step": 15160, + "tok": 62095360, + "ce": 3.0579, + "ppl": 21.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3343, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.8801, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8545, + "H_a": 0.079, + "d_bind": 0.03833 + }, + { + "tau": 0.0371, + "v": 0.8475, + "H_a": 0.067, + "d_bind": 0.0202 + } + ] + }, + { + "step": 15180, + "tok": 62177280, + "ce": 3.2517, + "ppl": 25.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.927, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3349, + "loop_stats": [ + { + "tau": 0.0829, + "v": 0.8755, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0218, + "v": 0.8515, + "H_a": 0.078, + "d_bind": 0.03749 + }, + { + "tau": 0.0192, + "v": 0.8501, + "H_a": 0.065, + "d_bind": 0.02107 + } + ] + }, + { + "step": 15200, + "tok": 62259200, + "ce": 2.8245, + "ppl": 16.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3355, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.8653, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0368, + "v": 0.8333, + "H_a": 0.084, + "d_bind": 0.04153 + }, + { + "tau": 0.0287, + "v": 0.8287, + "H_a": 0.066, + "d_bind": 0.02093 + } + ] + }, + { + "step": 15220, + "tok": 62341120, + "ce": 2.9711, + "ppl": 19.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3393, + "loop_stats": [ + { + "tau": 0.0959, + "v": 0.8694, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0198, + "v": 0.8481, + "H_a": 0.086, + "d_bind": 0.03889 + }, + { + "tau": 0.0133, + "v": 0.8439, + "H_a": 0.064, + "d_bind": 0.02032 + } + ] + }, + { + "step": 15240, + "tok": 62423040, + "ce": 3.103, + "ppl": 22.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3378, + "loop_stats": [ + { + "tau": 0.107, + "v": 0.8837, + "H_a": 0.11, + "d_bind": 0.0 + }, + { + "tau": 0.0301, + "v": 0.8531, + "H_a": 0.084, + "d_bind": 0.03795 + }, + { + "tau": 0.0222, + "v": 0.8494, + "H_a": 0.065, + "d_bind": 0.0197 + } + ] + }, + { + "step": 15260, + "tok": 62504960, + "ce": 2.9781, + "ppl": 19.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3377, + "loop_stats": [ + { + "tau": 0.094, + "v": 0.8798, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.025, + "v": 0.8595, + "H_a": 0.072, + "d_bind": 0.0388 + }, + { + "tau": 0.0187, + "v": 0.8616, + "H_a": 0.064, + "d_bind": 0.01986 + } + ] + }, + { + "step": 15280, + "tok": 62586880, + "ce": 2.8573, + "ppl": 17.4, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3372, + "loop_stats": [ + { + "tau": 0.1292, + "v": 0.8752, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8527, + "H_a": 0.081, + "d_bind": 0.0398 + }, + { + "tau": 0.0292, + "v": 0.8491, + "H_a": 0.06, + "d_bind": 0.02202 + } + ] + }, + { + "step": 15300, + "tok": 62668800, + "ce": 2.6969, + "ppl": 14.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3346, + "loop_stats": [ + { + "tau": 0.137, + "v": 0.8799, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0543, + "v": 0.8304, + "H_a": 0.082, + "d_bind": 0.03912 + }, + { + "tau": 0.0468, + "v": 0.8183, + "H_a": 0.061, + "d_bind": 0.02341 + } + ] + }, + { + "step": 15320, + "tok": 62750720, + "ce": 3.0683, + "ppl": 21.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3355, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8805, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8582, + "H_a": 0.075, + "d_bind": 0.0358 + }, + { + "tau": 0.0314, + "v": 0.8551, + "H_a": 0.069, + "d_bind": 0.02194 + } + ] + }, + { + "step": 15340, + "tok": 62832640, + "ce": 3.1864, + "ppl": 24.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.96, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3382, + "loop_stats": [ + { + "tau": 0.1078, + "v": 0.894, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.865, + "H_a": 0.08, + "d_bind": 0.03636 + }, + { + "tau": 0.0235, + "v": 0.8534, + "H_a": 0.063, + "d_bind": 0.0201 + } + ] + }, + { + "step": 15360, + "tok": 62914560, + "ce": 3.2154, + "ppl": 24.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3378, + "loop_stats": [ + { + "tau": 0.098, + "v": 0.8848, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.8566, + "H_a": 0.082, + "d_bind": 0.03835 + }, + { + "tau": 0.0227, + "v": 0.851, + "H_a": 0.069, + "d_bind": 0.02195 + } + ] + }, + { + "step": 15380, + "tok": 62996480, + "ce": 3.0363, + "ppl": 20.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3329, + "loop_stats": [ + { + "tau": 0.1109, + "v": 0.8595, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8271, + "H_a": 0.082, + "d_bind": 0.03895 + }, + { + "tau": 0.0341, + "v": 0.8213, + "H_a": 0.072, + "d_bind": 0.02169 + } + ] + }, + { + "step": 15400, + "tok": 63078400, + "ce": 2.3078, + "ppl": 10.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.962, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3334, + "loop_stats": [ + { + "tau": 0.166, + "v": 0.8703, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0585, + "v": 0.8554, + "H_a": 0.083, + "d_bind": 0.04171 + }, + { + "tau": 0.0515, + "v": 0.8557, + "H_a": 0.062, + "d_bind": 0.02161 + } + ] + }, + { + "step": 15420, + "tok": 63160320, + "ce": 3.0563, + "ppl": 21.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3306, + "loop_stats": [ + { + "tau": 0.0977, + "v": 0.8614, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8379, + "H_a": 0.078, + "d_bind": 0.03533 + }, + { + "tau": 0.0293, + "v": 0.8344, + "H_a": 0.063, + "d_bind": 0.02174 + } + ] + }, + { + "step": 15440, + "tok": 63242240, + "ce": 3.2924, + "ppl": 26.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3303, + "loop_stats": [ + { + "tau": 0.0972, + "v": 0.8804, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8682, + "H_a": 0.083, + "d_bind": 0.03741 + }, + { + "tau": 0.0251, + "v": 0.8658, + "H_a": 0.071, + "d_bind": 0.02122 + } + ] + }, + { + "step": 15460, + "tok": 63324160, + "ce": 3.3063, + "ppl": 27.3, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.96, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3336, + "loop_stats": [ + { + "tau": 0.0914, + "v": 0.8538, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.8382, + "H_a": 0.086, + "d_bind": 0.04095 + }, + { + "tau": 0.0252, + "v": 0.8393, + "H_a": 0.069, + "d_bind": 0.02249 + } + ] + }, + { + "step": 15480, + "tok": 63406080, + "ce": 3.2476, + "ppl": 25.7, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.952, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3346, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.9032, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0275, + "v": 0.8809, + "H_a": 0.076, + "d_bind": 0.03558 + }, + { + "tau": 0.0192, + "v": 0.8722, + "H_a": 0.066, + "d_bind": 0.02048 + } + ] + }, + { + "step": 15500, + "tok": 63488000, + "ce": 3.2631, + "ppl": 26.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.96, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3344, + "loop_stats": [ + { + "tau": 0.0889, + "v": 0.8805, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0295, + "v": 0.8512, + "H_a": 0.086, + "d_bind": 0.03893 + }, + { + "tau": 0.0232, + "v": 0.845, + "H_a": 0.074, + "d_bind": 0.02276 + } + ] + }, + { + "step": 15520, + "tok": 63569920, + "ce": 2.8308, + "ppl": 17.0, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.945, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.333, + "loop_stats": [ + { + "tau": 0.1093, + "v": 0.9009, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0427, + "v": 0.8663, + "H_a": 0.091, + "d_bind": 0.0396 + }, + { + "tau": 0.0327, + "v": 0.8599, + "H_a": 0.069, + "d_bind": 0.02201 + } + ] + }, + { + "step": 15540, + "tok": 63651840, + "ce": 2.6954, + "ppl": 14.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3333, + "loop_stats": [ + { + "tau": 0.1487, + "v": 0.8648, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0583, + "v": 0.8237, + "H_a": 0.081, + "d_bind": 0.0439 + }, + { + "tau": 0.0505, + "v": 0.8103, + "H_a": 0.06, + "d_bind": 0.02066 + } + ] + }, + { + "step": 15560, + "tok": 63733760, + "ce": 2.3403, + "ppl": 10.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3316, + "loop_stats": [ + { + "tau": 0.1512, + "v": 0.8773, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0566, + "v": 0.8353, + "H_a": 0.074, + "d_bind": 0.04626 + }, + { + "tau": 0.0503, + "v": 0.8219, + "H_a": 0.065, + "d_bind": 0.01957 + } + ] + }, + { + "step": 15580, + "tok": 63815680, + "ce": 3.0245, + "ppl": 20.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3262, + "loop_stats": [ + { + "tau": 0.1193, + "v": 0.8864, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0493, + "v": 0.8636, + "H_a": 0.077, + "d_bind": 0.0434 + }, + { + "tau": 0.044, + "v": 0.8573, + "H_a": 0.064, + "d_bind": 0.01959 + } + ] + }, + { + "step": 15600, + "tok": 63897600, + "ce": 3.2996, + "ppl": 27.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3255, + "loop_stats": [ + { + "tau": 0.091, + "v": 0.8882, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.025, + "v": 0.8633, + "H_a": 0.084, + "d_bind": 0.04002 + }, + { + "tau": 0.022, + "v": 0.8534, + "H_a": 0.07, + "d_bind": 0.02168 + } + ] + }, + { + "step": 15620, + "tok": 63979520, + "ce": 3.0036, + "ppl": 20.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3275, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.9064, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8848, + "H_a": 0.081, + "d_bind": 0.03868 + }, + { + "tau": 0.0391, + "v": 0.8749, + "H_a": 0.063, + "d_bind": 0.02014 + } + ] + }, + { + "step": 15640, + "tok": 64061440, + "ce": 2.8198, + "ppl": 16.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3301, + "loop_stats": [ + { + "tau": 0.1405, + "v": 0.8888, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0576, + "v": 0.8565, + "H_a": 0.071, + "d_bind": 0.03989 + }, + { + "tau": 0.0513, + "v": 0.8508, + "H_a": 0.062, + "d_bind": 0.01928 + } + ] + }, + { + "step": 15660, + "tok": 64143360, + "ce": 2.9468, + "ppl": 19.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.331, + "loop_stats": [ + { + "tau": 0.1271, + "v": 0.8693, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.8475, + "H_a": 0.075, + "d_bind": 0.03892 + }, + { + "tau": 0.036, + "v": 0.8394, + "H_a": 0.06, + "d_bind": 0.02033 + } + ] + }, + { + "step": 15680, + "tok": 64225280, + "ce": 2.9475, + "ppl": 19.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.957, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3305, + "loop_stats": [ + { + "tau": 0.1153, + "v": 0.9112, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8913, + "H_a": 0.074, + "d_bind": 0.03988 + }, + { + "tau": 0.0292, + "v": 0.8793, + "H_a": 0.065, + "d_bind": 0.02005 + } + ] + }, + { + "step": 15700, + "tok": 64307200, + "ce": 2.7687, + "ppl": 15.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3273, + "loop_stats": [ + { + "tau": 0.0932, + "v": 0.8625, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.8293, + "H_a": 0.084, + "d_bind": 0.0403 + }, + { + "tau": 0.0249, + "v": 0.8229, + "H_a": 0.073, + "d_bind": 0.02174 + } + ] + }, + { + "step": 15720, + "tok": 64389120, + "ce": 2.9509, + "ppl": 19.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.329, + "loop_stats": [ + { + "tau": 0.1147, + "v": 0.9074, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0244, + "v": 0.8808, + "H_a": 0.085, + "d_bind": 0.03924 + }, + { + "tau": 0.0188, + "v": 0.8717, + "H_a": 0.061, + "d_bind": 0.02188 + } + ] + }, + { + "step": 15740, + "tok": 64471040, + "ce": 2.7405, + "ppl": 15.5, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.952, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3282, + "loop_stats": [ + { + "tau": 0.1344, + "v": 0.8865, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0541, + "v": 0.8565, + "H_a": 0.074, + "d_bind": 0.04014 + }, + { + "tau": 0.0488, + "v": 0.847, + "H_a": 0.058, + "d_bind": 0.02011 + } + ] + }, + { + "step": 15760, + "tok": 64552960, + "ce": 2.8326, + "ppl": 17.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3265, + "loop_stats": [ + { + "tau": 0.1221, + "v": 0.8675, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0481, + "v": 0.8355, + "H_a": 0.08, + "d_bind": 0.04124 + }, + { + "tau": 0.0425, + "v": 0.8212, + "H_a": 0.057, + "d_bind": 0.02003 + } + ] + }, + { + "step": 15780, + "tok": 64634880, + "ce": 2.8525, + "ppl": 17.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3267, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.8757, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8601, + "H_a": 0.084, + "d_bind": 0.04207 + }, + { + "tau": 0.0366, + "v": 0.8551, + "H_a": 0.063, + "d_bind": 0.02236 + } + ] + }, + { + "step": 15800, + "tok": 64716800, + "ce": 2.9817, + "ppl": 19.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.327, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.8614, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.849, + "H_a": 0.075, + "d_bind": 0.03896 + }, + { + "tau": 0.0364, + "v": 0.8452, + "H_a": 0.062, + "d_bind": 0.02082 + } + ] + }, + { + "step": 15820, + "tok": 64798720, + "ce": 3.2929, + "ppl": 26.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3299, + "loop_stats": [ + { + "tau": 0.1029, + "v": 0.844, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0323, + "v": 0.8288, + "H_a": 0.085, + "d_bind": 0.04041 + }, + { + "tau": 0.0293, + "v": 0.8223, + "H_a": 0.065, + "d_bind": 0.02093 + } + ] + }, + { + "step": 15840, + "tok": 64880640, + "ce": 3.1399, + "ppl": 23.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3319, + "loop_stats": [ + { + "tau": 0.1111, + "v": 0.8921, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0498, + "v": 0.8732, + "H_a": 0.08, + "d_bind": 0.0377 + }, + { + "tau": 0.0414, + "v": 0.8647, + "H_a": 0.066, + "d_bind": 0.02138 + } + ] + }, + { + "step": 15860, + "tok": 64962560, + "ce": 3.0424, + "ppl": 21.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3315, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.8649, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8339, + "H_a": 0.085, + "d_bind": 0.03849 + }, + { + "tau": 0.0417, + "v": 0.8264, + "H_a": 0.07, + "d_bind": 0.02161 + } + ] + }, + { + "step": 15880, + "tok": 65044480, + "ce": 2.7713, + "ppl": 16.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3335, + "loop_stats": [ + { + "tau": 0.1104, + "v": 0.8573, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8303, + "H_a": 0.087, + "d_bind": 0.04088 + }, + { + "tau": 0.0383, + "v": 0.8218, + "H_a": 0.071, + "d_bind": 0.02371 + } + ] + }, + { + "step": 15900, + "tok": 65126400, + "ce": 3.4114, + "ppl": 30.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3354, + "loop_stats": [ + { + "tau": 0.0673, + "v": 0.8558, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0204, + "v": 0.8411, + "H_a": 0.083, + "d_bind": 0.03495 + }, + { + "tau": 0.0178, + "v": 0.8434, + "H_a": 0.066, + "d_bind": 0.01945 + } + ] + }, + { + "step": 15920, + "tok": 65208320, + "ce": 2.5184, + "ppl": 12.4, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.962, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3345, + "loop_stats": [ + { + "tau": 0.1286, + "v": 0.8907, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.8508, + "H_a": 0.08, + "d_bind": 0.03995 + }, + { + "tau": 0.0398, + "v": 0.8327, + "H_a": 0.06, + "d_bind": 0.02042 + } + ] + }, + { + "step": 15940, + "tok": 65290240, + "ce": 2.9942, + "ppl": 20.0, + "pref_transfer_pron": 0.873, + "pref_train_sv": 0.95, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3323, + "loop_stats": [ + { + "tau": 0.1038, + "v": 0.8595, + "H_a": 0.106, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.8337, + "H_a": 0.085, + "d_bind": 0.04034 + }, + { + "tau": 0.0228, + "v": 0.8221, + "H_a": 0.069, + "d_bind": 0.02139 + } + ] + }, + { + "step": 15960, + "tok": 65372160, + "ce": 3.1885, + "ppl": 24.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3325, + "loop_stats": [ + { + "tau": 0.0952, + "v": 0.8436, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0379, + "v": 0.823, + "H_a": 0.082, + "d_bind": 0.03737 + }, + { + "tau": 0.0285, + "v": 0.8227, + "H_a": 0.071, + "d_bind": 0.02072 + } + ] + }, + { + "step": 15980, + "tok": 65454080, + "ce": 3.1319, + "ppl": 22.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3302, + "loop_stats": [ + { + "tau": 0.1164, + "v": 0.855, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0416, + "v": 0.8365, + "H_a": 0.077, + "d_bind": 0.03893 + }, + { + "tau": 0.0345, + "v": 0.8289, + "H_a": 0.061, + "d_bind": 0.02114 + } + ] + }, + { + "step": 16000, + "tok": 65536000, + "ce": 2.7336, + "ppl": 15.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3321, + "loop_stats": [ + { + "tau": 0.115, + "v": 0.8301, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.049, + "v": 0.8173, + "H_a": 0.08, + "d_bind": 0.04416 + }, + { + "tau": 0.0377, + "v": 0.8148, + "H_a": 0.067, + "d_bind": 0.02028 + } + ] + }, + { + "step": 16020, + "tok": 65617920, + "ce": 2.7375, + "ppl": 15.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3318, + "loop_stats": [ + { + "tau": 0.1272, + "v": 0.8256, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0509, + "v": 0.7993, + "H_a": 0.08, + "d_bind": 0.04018 + }, + { + "tau": 0.0476, + "v": 0.7867, + "H_a": 0.066, + "d_bind": 0.02093 + } + ] + }, + { + "step": 16040, + "tok": 65699840, + "ce": 2.9968, + "ppl": 20.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.947, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3301, + "loop_stats": [ + { + "tau": 0.0941, + "v": 0.8654, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0451, + "v": 0.8399, + "H_a": 0.082, + "d_bind": 0.03845 + }, + { + "tau": 0.0345, + "v": 0.8258, + "H_a": 0.068, + "d_bind": 0.02144 + } + ] + }, + { + "step": 16060, + "tok": 65781760, + "ce": 2.6899, + "ppl": 14.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.965, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3287, + "loop_stats": [ + { + "tau": 0.1389, + "v": 0.8104, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0638, + "v": 0.7919, + "H_a": 0.077, + "d_bind": 0.03889 + }, + { + "tau": 0.0576, + "v": 0.79, + "H_a": 0.062, + "d_bind": 0.01939 + } + ] + }, + { + "step": 16080, + "tok": 65863680, + "ce": 2.6997, + "ppl": 14.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.962, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3271, + "loop_stats": [ + { + "tau": 0.1279, + "v": 0.8384, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.055, + "v": 0.8238, + "H_a": 0.079, + "d_bind": 0.03858 + }, + { + "tau": 0.0488, + "v": 0.821, + "H_a": 0.061, + "d_bind": 0.02058 + } + ] + }, + { + "step": 16100, + "tok": 65945600, + "ce": 2.8214, + "ppl": 16.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.33, + "loop_stats": [ + { + "tau": 0.1123, + "v": 0.8277, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8032, + "H_a": 0.076, + "d_bind": 0.03936 + }, + { + "tau": 0.0338, + "v": 0.7963, + "H_a": 0.064, + "d_bind": 0.01887 + } + ] + }, + { + "step": 16120, + "tok": 66027520, + "ce": 2.9399, + "ppl": 18.9, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.938, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3307, + "loop_stats": [ + { + "tau": 0.109, + "v": 0.8354, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.8073, + "H_a": 0.081, + "d_bind": 0.03794 + }, + { + "tau": 0.0374, + "v": 0.8048, + "H_a": 0.065, + "d_bind": 0.02031 + } + ] + }, + { + "step": 16140, + "tok": 66109440, + "ce": 2.7908, + "ppl": 16.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3315, + "loop_stats": [ + { + "tau": 0.1284, + "v": 0.7745, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.7454, + "H_a": 0.082, + "d_bind": 0.0429 + }, + { + "tau": 0.0516, + "v": 0.7373, + "H_a": 0.066, + "d_bind": 0.0212 + } + ] + }, + { + "step": 16160, + "tok": 66191360, + "ce": 2.8213, + "ppl": 16.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.957, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3335, + "loop_stats": [ + { + "tau": 0.1202, + "v": 0.8497, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0671, + "v": 0.8271, + "H_a": 0.087, + "d_bind": 0.0396 + }, + { + "tau": 0.07, + "v": 0.8143, + "H_a": 0.069, + "d_bind": 0.02235 + } + ] + }, + { + "step": 16180, + "tok": 66273280, + "ce": 3.162, + "ppl": 23.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3327, + "loop_stats": [ + { + "tau": 0.1025, + "v": 0.816, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0516, + "v": 0.7899, + "H_a": 0.077, + "d_bind": 0.04196 + }, + { + "tau": 0.0522, + "v": 0.783, + "H_a": 0.062, + "d_bind": 0.02004 + } + ] + }, + { + "step": 16200, + "tok": 66355200, + "ce": 3.2386, + "ppl": 25.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3334, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.8546, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8251, + "H_a": 0.08, + "d_bind": 0.03808 + }, + { + "tau": 0.0373, + "v": 0.8127, + "H_a": 0.062, + "d_bind": 0.02 + } + ] + }, + { + "step": 16220, + "tok": 66437120, + "ce": 2.6084, + "ppl": 13.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3361, + "loop_stats": [ + { + "tau": 0.1274, + "v": 0.8012, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0601, + "v": 0.778, + "H_a": 0.074, + "d_bind": 0.04211 + }, + { + "tau": 0.0531, + "v": 0.7682, + "H_a": 0.064, + "d_bind": 0.02146 + } + ] + }, + { + "step": 16240, + "tok": 66519040, + "ce": 3.3614, + "ppl": 28.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.962, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3357, + "loop_stats": [ + { + "tau": 0.0893, + "v": 0.8353, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8323, + "H_a": 0.089, + "d_bind": 0.03912 + }, + { + "tau": 0.0289, + "v": 0.823, + "H_a": 0.066, + "d_bind": 0.022 + } + ] + }, + { + "step": 16260, + "tok": 66600960, + "ce": 2.6718, + "ppl": 14.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3351, + "loop_stats": [ + { + "tau": 0.1242, + "v": 0.7788, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0487, + "v": 0.7528, + "H_a": 0.081, + "d_bind": 0.03942 + }, + { + "tau": 0.0441, + "v": 0.7454, + "H_a": 0.064, + "d_bind": 0.02189 + } + ] + }, + { + "step": 16280, + "tok": 66682880, + "ce": 2.7459, + "ppl": 15.6, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3355, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.7862, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0408, + "v": 0.7627, + "H_a": 0.081, + "d_bind": 0.04182 + }, + { + "tau": 0.0384, + "v": 0.756, + "H_a": 0.067, + "d_bind": 0.02094 + } + ] + }, + { + "step": 16300, + "tok": 66764800, + "ce": 3.0686, + "ppl": 21.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3362, + "loop_stats": [ + { + "tau": 0.0943, + "v": 0.8286, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.028, + "v": 0.8118, + "H_a": 0.081, + "d_bind": 0.04052 + }, + { + "tau": 0.0242, + "v": 0.8032, + "H_a": 0.064, + "d_bind": 0.02166 + } + ] + }, + { + "step": 16320, + "tok": 66846720, + "ce": 2.9927, + "ppl": 19.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3373, + "loop_stats": [ + { + "tau": 0.1195, + "v": 0.7658, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.7608, + "H_a": 0.076, + "d_bind": 0.04315 + }, + { + "tau": 0.0383, + "v": 0.7632, + "H_a": 0.065, + "d_bind": 0.02111 + } + ] + }, + { + "step": 16340, + "tok": 66928640, + "ce": 3.2956, + "ppl": 27.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3362, + "loop_stats": [ + { + "tau": 0.0828, + "v": 0.8163, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0136, + "v": 0.811, + "H_a": 0.086, + "d_bind": 0.03891 + }, + { + "tau": 0.009, + "v": 0.8097, + "H_a": 0.072, + "d_bind": 0.02348 + } + ] + }, + { + "step": 16360, + "tok": 67010560, + "ce": 3.4629, + "ppl": 31.9, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.947, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3335, + "loop_stats": [ + { + "tau": 0.0843, + "v": 0.7746, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0322, + "v": 0.7556, + "H_a": 0.088, + "d_bind": 0.04058 + }, + { + "tau": 0.0296, + "v": 0.7539, + "H_a": 0.064, + "d_bind": 0.02246 + } + ] + }, + { + "step": 16380, + "tok": 67092480, + "ce": 2.576, + "ppl": 13.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3333, + "loop_stats": [ + { + "tau": 0.1481, + "v": 0.7797, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0591, + "v": 0.77, + "H_a": 0.087, + "d_bind": 0.04501 + }, + { + "tau": 0.0526, + "v": 0.768, + "H_a": 0.059, + "d_bind": 0.02227 + } + ] + }, + { + "step": 16400, + "tok": 67174400, + "ce": 2.9549, + "ppl": 19.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3332, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.7759, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0341, + "v": 0.7611, + "H_a": 0.081, + "d_bind": 0.04042 + }, + { + "tau": 0.0295, + "v": 0.7506, + "H_a": 0.072, + "d_bind": 0.02303 + } + ] + }, + { + "step": 16420, + "tok": 67256320, + "ce": 2.6278, + "ppl": 13.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3367, + "loop_stats": [ + { + "tau": 0.1307, + "v": 0.7922, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0451, + "v": 0.7788, + "H_a": 0.076, + "d_bind": 0.04755 + }, + { + "tau": 0.0415, + "v": 0.778, + "H_a": 0.064, + "d_bind": 0.02129 + } + ] + }, + { + "step": 16440, + "tok": 67338240, + "ce": 2.7049, + "ppl": 15.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.3426, + "loop_stats": [ + { + "tau": 0.1214, + "v": 0.8026, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.7843, + "H_a": 0.075, + "d_bind": 0.04325 + }, + { + "tau": 0.0284, + "v": 0.7744, + "H_a": 0.059, + "d_bind": 0.01911 + } + ] + }, + { + "step": 16460, + "tok": 67420160, + "ce": 2.9619, + "ppl": 19.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3467, + "loop_stats": [ + { + "tau": 0.0972, + "v": 0.827, + "H_a": 0.107, + "d_bind": 0.0 + }, + { + "tau": 0.0288, + "v": 0.8157, + "H_a": 0.077, + "d_bind": 0.04323 + }, + { + "tau": 0.0233, + "v": 0.8113, + "H_a": 0.064, + "d_bind": 0.02138 + } + ] + }, + { + "step": 16480, + "tok": 67502080, + "ce": 3.2586, + "ppl": 26.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.95, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3469, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.8105, + "H_a": 0.104, + "d_bind": 0.0 + }, + { + "tau": 0.0355, + "v": 0.7881, + "H_a": 0.078, + "d_bind": 0.04196 + }, + { + "tau": 0.0304, + "v": 0.7793, + "H_a": 0.063, + "d_bind": 0.02025 + } + ] + }, + { + "step": 16500, + "tok": 67584000, + "ce": 3.2059, + "ppl": 24.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3463, + "loop_stats": [ + { + "tau": 0.1085, + "v": 0.8272, + "H_a": 0.102, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8031, + "H_a": 0.08, + "d_bind": 0.04179 + }, + { + "tau": 0.0386, + "v": 0.7875, + "H_a": 0.067, + "d_bind": 0.0222 + } + ] + }, + { + "step": 16520, + "tok": 67665920, + "ce": 2.8783, + "ppl": 17.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.348, + "loop_stats": [ + { + "tau": 0.1162, + "v": 0.8025, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0446, + "v": 0.7876, + "H_a": 0.081, + "d_bind": 0.03896 + }, + { + "tau": 0.0395, + "v": 0.7858, + "H_a": 0.065, + "d_bind": 0.02101 + } + ] + }, + { + "step": 16540, + "tok": 67747840, + "ce": 2.7808, + "ppl": 16.1, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.955, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.348, + "loop_stats": [ + { + "tau": 0.1174, + "v": 0.8037, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.7812, + "H_a": 0.079, + "d_bind": 0.04182 + }, + { + "tau": 0.038, + "v": 0.7755, + "H_a": 0.066, + "d_bind": 0.01916 + } + ] + }, + { + "step": 16560, + "tok": 67829760, + "ce": 3.1914, + "ppl": 24.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.3506, + "loop_stats": [ + { + "tau": 0.1091, + "v": 0.7927, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.7773, + "H_a": 0.076, + "d_bind": 0.04078 + }, + { + "tau": 0.0209, + "v": 0.776, + "H_a": 0.062, + "d_bind": 0.02236 + } + ] + }, + { + "step": 16580, + "tok": 67911680, + "ce": 2.8148, + "ppl": 16.7, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.927, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.3476, + "loop_stats": [ + { + "tau": 0.1336, + "v": 0.8535, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0331, + "v": 0.8407, + "H_a": 0.078, + "d_bind": 0.03846 + }, + { + "tau": 0.0301, + "v": 0.8384, + "H_a": 0.063, + "d_bind": 0.02447 + } + ] + }, + { + "step": 16600, + "tok": 67993600, + "ce": 3.503, + "ppl": 33.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3451, + "loop_stats": [ + { + "tau": 0.0603, + "v": 0.7843, + "H_a": 0.109, + "d_bind": 0.0 + }, + { + "tau": 0.0238, + "v": 0.758, + "H_a": 0.077, + "d_bind": 0.04055 + }, + { + "tau": 0.0175, + "v": 0.7506, + "H_a": 0.065, + "d_bind": 0.02054 + } + ] + }, + { + "step": 16620, + "tok": 68075520, + "ce": 2.987, + "ppl": 19.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3412, + "loop_stats": [ + { + "tau": 0.1014, + "v": 0.8419, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0466, + "v": 0.8378, + "H_a": 0.082, + "d_bind": 0.04092 + }, + { + "tau": 0.0346, + "v": 0.8364, + "H_a": 0.071, + "d_bind": 0.02245 + } + ] + }, + { + "step": 16640, + "tok": 68157440, + "ce": 2.9568, + "ppl": 19.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3389, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.827, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8027, + "H_a": 0.073, + "d_bind": 0.0411 + }, + { + "tau": 0.046, + "v": 0.7908, + "H_a": 0.064, + "d_bind": 0.01969 + } + ] + }, + { + "step": 16660, + "tok": 68239360, + "ce": 2.9372, + "ppl": 18.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.337, + "loop_stats": [ + { + "tau": 0.1326, + "v": 0.8171, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8042, + "H_a": 0.073, + "d_bind": 0.04285 + }, + { + "tau": 0.0485, + "v": 0.7997, + "H_a": 0.06, + "d_bind": 0.02035 + } + ] + }, + { + "step": 16680, + "tok": 68321280, + "ce": 2.8865, + "ppl": 17.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22458, + "peak_MiB": 5301, + "v_gain": 0.3383, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8215, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0362, + "v": 0.8065, + "H_a": 0.076, + "d_bind": 0.04228 + }, + { + "tau": 0.0313, + "v": 0.8037, + "H_a": 0.06, + "d_bind": 0.02056 + } + ] + }, + { + "step": 16700, + "tok": 68403200, + "ce": 3.3506, + "ppl": 28.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.927, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.338, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.828, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8144, + "H_a": 0.08, + "d_bind": 0.03846 + }, + { + "tau": 0.039, + "v": 0.8124, + "H_a": 0.074, + "d_bind": 0.02289 + } + ] + }, + { + "step": 16720, + "tok": 68485120, + "ce": 3.042, + "ppl": 20.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3403, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.8378, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8143, + "H_a": 0.082, + "d_bind": 0.03995 + }, + { + "tau": 0.0299, + "v": 0.8008, + "H_a": 0.071, + "d_bind": 0.02263 + } + ] + }, + { + "step": 16740, + "tok": 68567040, + "ce": 3.3055, + "ppl": 27.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3405, + "loop_stats": [ + { + "tau": 0.0658, + "v": 0.8233, + "H_a": 0.105, + "d_bind": 0.0 + }, + { + "tau": 0.0238, + "v": 0.792, + "H_a": 0.083, + "d_bind": 0.04112 + }, + { + "tau": 0.021, + "v": 0.7722, + "H_a": 0.062, + "d_bind": 0.02121 + } + ] + }, + { + "step": 16760, + "tok": 68648960, + "ce": 3.1081, + "ppl": 22.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.342, + "loop_stats": [ + { + "tau": 0.0816, + "v": 0.8375, + "H_a": 0.108, + "d_bind": 0.0 + }, + { + "tau": 0.0258, + "v": 0.8317, + "H_a": 0.081, + "d_bind": 0.04202 + }, + { + "tau": 0.026, + "v": 0.8288, + "H_a": 0.064, + "d_bind": 0.02265 + } + ] + }, + { + "step": 16780, + "tok": 68730880, + "ce": 3.0688, + "ppl": 21.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3439, + "loop_stats": [ + { + "tau": 0.1086, + "v": 0.8189, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8058, + "H_a": 0.081, + "d_bind": 0.04353 + }, + { + "tau": 0.0395, + "v": 0.8025, + "H_a": 0.07, + "d_bind": 0.02331 + } + ] + }, + { + "step": 16800, + "tok": 68812800, + "ce": 3.4222, + "ppl": 30.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22459, + "peak_MiB": 5301, + "v_gain": 0.3431, + "loop_stats": [ + { + "tau": 0.0797, + "v": 0.792, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0282, + "v": 0.7641, + "H_a": 0.08, + "d_bind": 0.04074 + }, + { + "tau": 0.0266, + "v": 0.7551, + "H_a": 0.063, + "d_bind": 0.02223 + } + ] + }, + { + "step": 16820, + "tok": 68894720, + "ce": 3.17, + "ppl": 23.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3442, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.803, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.7791, + "H_a": 0.076, + "d_bind": 0.04123 + }, + { + "tau": 0.0307, + "v": 0.7758, + "H_a": 0.057, + "d_bind": 0.02112 + } + ] + }, + { + "step": 16840, + "tok": 68976640, + "ce": 2.8264, + "ppl": 16.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3418, + "loop_stats": [ + { + "tau": 0.1091, + "v": 0.7938, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.035, + "v": 0.7602, + "H_a": 0.08, + "d_bind": 0.04549 + }, + { + "tau": 0.0323, + "v": 0.7499, + "H_a": 0.066, + "d_bind": 0.0214 + } + ] + }, + { + "step": 16860, + "tok": 69058560, + "ce": 2.7617, + "ppl": 15.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3425, + "loop_stats": [ + { + "tau": 0.1238, + "v": 0.83, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8063, + "H_a": 0.076, + "d_bind": 0.04203 + }, + { + "tau": 0.04, + "v": 0.8, + "H_a": 0.06, + "d_bind": 0.02053 + } + ] + }, + { + "step": 16880, + "tok": 69140480, + "ce": 2.7922, + "ppl": 16.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3437, + "loop_stats": [ + { + "tau": 0.1365, + "v": 0.8162, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.7971, + "H_a": 0.075, + "d_bind": 0.04202 + }, + { + "tau": 0.0502, + "v": 0.7908, + "H_a": 0.061, + "d_bind": 0.01888 + } + ] + }, + { + "step": 16900, + "tok": 69222400, + "ce": 2.901, + "ppl": 18.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3439, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.8252, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.8029, + "H_a": 0.081, + "d_bind": 0.03954 + }, + { + "tau": 0.0369, + "v": 0.8007, + "H_a": 0.059, + "d_bind": 0.02083 + } + ] + }, + { + "step": 16920, + "tok": 69304320, + "ce": 3.2808, + "ppl": 26.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22460, + "peak_MiB": 5301, + "v_gain": 0.3434, + "loop_stats": [ + { + "tau": 0.1133, + "v": 0.8016, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0475, + "v": 0.7737, + "H_a": 0.086, + "d_bind": 0.03882 + }, + { + "tau": 0.0434, + "v": 0.7625, + "H_a": 0.071, + "d_bind": 0.02258 + } + ] + }, + { + "step": 16940, + "tok": 69386240, + "ce": 3.0717, + "ppl": 21.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22461, + "peak_MiB": 5301, + "v_gain": 0.3435, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.7767, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0217, + "v": 0.7519, + "H_a": 0.086, + "d_bind": 0.04139 + }, + { + "tau": 0.0195, + "v": 0.7461, + "H_a": 0.067, + "d_bind": 0.0215 + } + ] + }, + { + "step": 16960, + "tok": 69468160, + "ce": 2.941, + "ppl": 18.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3443, + "loop_stats": [ + { + "tau": 0.1059, + "v": 0.7935, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0506, + "v": 0.774, + "H_a": 0.079, + "d_bind": 0.04225 + }, + { + "tau": 0.0448, + "v": 0.7641, + "H_a": 0.064, + "d_bind": 0.02057 + } + ] + }, + { + "step": 16980, + "tok": 69550080, + "ce": 3.4056, + "ppl": 30.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3442, + "loop_stats": [ + { + "tau": 0.0907, + "v": 0.8206, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.7926, + "H_a": 0.083, + "d_bind": 0.04003 + }, + { + "tau": 0.0277, + "v": 0.7826, + "H_a": 0.066, + "d_bind": 0.0206 + } + ] + }, + { + "step": 17000, + "tok": 69632000, + "ce": 2.5726, + "ppl": 13.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.957, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3455, + "loop_stats": [ + { + "tau": 0.1109, + "v": 0.8355, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.801, + "H_a": 0.082, + "d_bind": 0.04423 + }, + { + "tau": 0.0328, + "v": 0.7839, + "H_a": 0.069, + "d_bind": 0.02406 + } + ] + }, + { + "step": 17020, + "tok": 69713920, + "ce": 3.4928, + "ppl": 32.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.957, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.347, + "loop_stats": [ + { + "tau": 0.1002, + "v": 0.8654, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0172, + "v": 0.8411, + "H_a": 0.076, + "d_bind": 0.03611 + }, + { + "tau": 0.0142, + "v": 0.8369, + "H_a": 0.065, + "d_bind": 0.02082 + } + ] + }, + { + "step": 17040, + "tok": 69795840, + "ce": 2.7502, + "ppl": 15.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22462, + "peak_MiB": 5301, + "v_gain": 0.3448, + "loop_stats": [ + { + "tau": 0.1297, + "v": 0.8498, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0532, + "v": 0.8241, + "H_a": 0.071, + "d_bind": 0.04008 + }, + { + "tau": 0.0468, + "v": 0.8144, + "H_a": 0.063, + "d_bind": 0.02039 + } + ] + }, + { + "step": 17060, + "tok": 69877760, + "ce": 2.6416, + "ppl": 14.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3465, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8139, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.7897, + "H_a": 0.074, + "d_bind": 0.0473 + }, + { + "tau": 0.0479, + "v": 0.782, + "H_a": 0.057, + "d_bind": 0.01997 + } + ] + }, + { + "step": 17080, + "tok": 69959680, + "ce": 3.1413, + "ppl": 23.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.933, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3487, + "loop_stats": [ + { + "tau": 0.105, + "v": 0.8296, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0188, + "v": 0.8139, + "H_a": 0.084, + "d_bind": 0.03819 + }, + { + "tau": 0.014, + "v": 0.8151, + "H_a": 0.063, + "d_bind": 0.02196 + } + ] + }, + { + "step": 17100, + "tok": 70041600, + "ce": 2.6211, + "ppl": 13.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3479, + "loop_stats": [ + { + "tau": 0.1244, + "v": 0.7813, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0457, + "v": 0.7729, + "H_a": 0.078, + "d_bind": 0.04648 + }, + { + "tau": 0.0395, + "v": 0.7741, + "H_a": 0.06, + "d_bind": 0.02168 + } + ] + }, + { + "step": 17120, + "tok": 70123520, + "ce": 3.0468, + "ppl": 21.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.947, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3468, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.8058, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.7764, + "H_a": 0.08, + "d_bind": 0.04106 + }, + { + "tau": 0.0187, + "v": 0.7705, + "H_a": 0.066, + "d_bind": 0.02055 + } + ] + }, + { + "step": 17140, + "tok": 70205440, + "ce": 2.8187, + "ppl": 16.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3482, + "loop_stats": [ + { + "tau": 0.1138, + "v": 0.8437, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.8119, + "H_a": 0.077, + "d_bind": 0.04406 + }, + { + "tau": 0.0401, + "v": 0.8044, + "H_a": 0.062, + "d_bind": 0.02067 + } + ] + }, + { + "step": 17160, + "tok": 70287360, + "ce": 2.6194, + "ppl": 13.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22463, + "peak_MiB": 5301, + "v_gain": 0.3483, + "loop_stats": [ + { + "tau": 0.135, + "v": 0.863, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.042, + "v": 0.846, + "H_a": 0.072, + "d_bind": 0.04433 + }, + { + "tau": 0.0331, + "v": 0.8357, + "H_a": 0.059, + "d_bind": 0.01973 + } + ] + }, + { + "step": 17180, + "tok": 70369280, + "ce": 3.0188, + "ppl": 20.5, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.947, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3489, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.8256, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0297, + "v": 0.797, + "H_a": 0.071, + "d_bind": 0.04269 + }, + { + "tau": 0.0251, + "v": 0.7829, + "H_a": 0.063, + "d_bind": 0.02061 + } + ] + }, + { + "step": 17200, + "tok": 70451200, + "ce": 2.5474, + "ppl": 12.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3496, + "loop_stats": [ + { + "tau": 0.1497, + "v": 0.8349, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8083, + "H_a": 0.07, + "d_bind": 0.04681 + }, + { + "tau": 0.03, + "v": 0.7954, + "H_a": 0.06, + "d_bind": 0.02025 + } + ] + }, + { + "step": 17220, + "tok": 70533120, + "ce": 3.495, + "ppl": 32.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3495, + "loop_stats": [ + { + "tau": 0.072, + "v": 0.854, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0078, + "v": 0.8429, + "H_a": 0.085, + "d_bind": 0.04051 + }, + { + "tau": 0.0036, + "v": 0.8396, + "H_a": 0.064, + "d_bind": 0.0212 + } + ] + }, + { + "step": 17240, + "tok": 70615040, + "ce": 2.9, + "ppl": 18.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3472, + "loop_stats": [ + { + "tau": 0.1264, + "v": 0.8431, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0307, + "v": 0.8298, + "H_a": 0.076, + "d_bind": 0.04455 + }, + { + "tau": 0.0243, + "v": 0.8301, + "H_a": 0.06, + "d_bind": 0.02091 + } + ] + }, + { + "step": 17260, + "tok": 70696960, + "ce": 3.3008, + "ppl": 27.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3521, + "loop_stats": [ + { + "tau": 0.0796, + "v": 0.7983, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0203, + "v": 0.7859, + "H_a": 0.075, + "d_bind": 0.03946 + }, + { + "tau": 0.0179, + "v": 0.7821, + "H_a": 0.066, + "d_bind": 0.0202 + } + ] + }, + { + "step": 17280, + "tok": 70778880, + "ce": 2.5007, + "ppl": 12.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22464, + "peak_MiB": 5301, + "v_gain": 0.3567, + "loop_stats": [ + { + "tau": 0.1378, + "v": 0.8582, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8351, + "H_a": 0.074, + "d_bind": 0.04451 + }, + { + "tau": 0.0322, + "v": 0.8261, + "H_a": 0.06, + "d_bind": 0.02258 + } + ] + }, + { + "step": 17300, + "tok": 70860800, + "ce": 3.0843, + "ppl": 21.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3586, + "loop_stats": [ + { + "tau": 0.1124, + "v": 0.8414, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8031, + "H_a": 0.072, + "d_bind": 0.04213 + }, + { + "tau": 0.0254, + "v": 0.7821, + "H_a": 0.056, + "d_bind": 0.01904 + } + ] + }, + { + "step": 17320, + "tok": 70942720, + "ce": 3.1694, + "ppl": 23.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.965, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3593, + "loop_stats": [ + { + "tau": 0.0869, + "v": 0.8401, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.022, + "v": 0.8208, + "H_a": 0.074, + "d_bind": 0.04006 + }, + { + "tau": 0.0167, + "v": 0.813, + "H_a": 0.07, + "d_bind": 0.02051 + } + ] + }, + { + "step": 17340, + "tok": 71024640, + "ce": 2.3756, + "ppl": 10.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3565, + "loop_stats": [ + { + "tau": 0.1475, + "v": 0.8602, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0384, + "v": 0.84, + "H_a": 0.073, + "d_bind": 0.04291 + }, + { + "tau": 0.0352, + "v": 0.83, + "H_a": 0.061, + "d_bind": 0.02219 + } + ] + }, + { + "step": 17360, + "tok": 71106560, + "ce": 3.2232, + "ppl": 25.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3548, + "loop_stats": [ + { + "tau": 0.1195, + "v": 0.8364, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8149, + "H_a": 0.08, + "d_bind": 0.04203 + }, + { + "tau": 0.0255, + "v": 0.806, + "H_a": 0.063, + "d_bind": 0.02167 + } + ] + }, + { + "step": 17380, + "tok": 71188480, + "ce": 2.9573, + "ppl": 19.2, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.945, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3538, + "loop_stats": [ + { + "tau": 0.0877, + "v": 0.8456, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0304, + "v": 0.8346, + "H_a": 0.071, + "d_bind": 0.04059 + }, + { + "tau": 0.0241, + "v": 0.8304, + "H_a": 0.059, + "d_bind": 0.02172 + } + ] + }, + { + "step": 17400, + "tok": 71270400, + "ce": 2.9342, + "ppl": 18.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22465, + "peak_MiB": 5301, + "v_gain": 0.3549, + "loop_stats": [ + { + "tau": 0.1041, + "v": 0.8725, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8637, + "H_a": 0.082, + "d_bind": 0.04249 + }, + { + "tau": 0.0271, + "v": 0.8567, + "H_a": 0.065, + "d_bind": 0.02223 + } + ] + }, + { + "step": 17420, + "tok": 71352320, + "ce": 2.6868, + "ppl": 14.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.962, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.3573, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.8416, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0253, + "v": 0.8259, + "H_a": 0.077, + "d_bind": 0.04112 + }, + { + "tau": 0.0199, + "v": 0.8224, + "H_a": 0.066, + "d_bind": 0.02141 + } + ] + }, + { + "step": 17440, + "tok": 71434240, + "ce": 2.9077, + "ppl": 18.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.962, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3571, + "loop_stats": [ + { + "tau": 0.1288, + "v": 0.8565, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0273, + "v": 0.8426, + "H_a": 0.078, + "d_bind": 0.04164 + }, + { + "tau": 0.0203, + "v": 0.8415, + "H_a": 0.062, + "d_bind": 0.02245 + } + ] + }, + { + "step": 17460, + "tok": 71516160, + "ce": 3.036, + "ppl": 20.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3571, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8749, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0115, + "v": 0.8375, + "H_a": 0.075, + "d_bind": 0.03917 + }, + { + "tau": 0.0068, + "v": 0.8205, + "H_a": 0.066, + "d_bind": 0.02246 + } + ] + }, + { + "step": 17480, + "tok": 71598080, + "ce": 3.1165, + "ppl": 22.6, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.957, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3553, + "loop_stats": [ + { + "tau": 0.1123, + "v": 0.8621, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8368, + "H_a": 0.073, + "d_bind": 0.04049 + }, + { + "tau": 0.021, + "v": 0.826, + "H_a": 0.067, + "d_bind": 0.02165 + } + ] + }, + { + "step": 17500, + "tok": 71680000, + "ce": 2.405, + "ppl": 11.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3545, + "loop_stats": [ + { + "tau": 0.1421, + "v": 0.8438, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0534, + "v": 0.8263, + "H_a": 0.076, + "d_bind": 0.04429 + }, + { + "tau": 0.0414, + "v": 0.8213, + "H_a": 0.059, + "d_bind": 0.0216 + } + ] + }, + { + "step": 17520, + "tok": 71761920, + "ce": 3.1643, + "ppl": 23.7, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.942, + "tok_s": 22466, + "peak_MiB": 5301, + "v_gain": 0.352, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8531, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0417, + "v": 0.8337, + "H_a": 0.077, + "d_bind": 0.04413 + }, + { + "tau": 0.0285, + "v": 0.8253, + "H_a": 0.059, + "d_bind": 0.02333 + } + ] + }, + { + "step": 17540, + "tok": 71843840, + "ce": 3.0739, + "ppl": 21.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22467, + "peak_MiB": 5301, + "v_gain": 0.3468, + "loop_stats": [ + { + "tau": 0.1155, + "v": 0.8511, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.8265, + "H_a": 0.07, + "d_bind": 0.03991 + }, + { + "tau": 0.0306, + "v": 0.8142, + "H_a": 0.057, + "d_bind": 0.02051 + } + ] + }, + { + "step": 17560, + "tok": 71925760, + "ce": 3.4127, + "ppl": 30.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3467, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8813, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8557, + "H_a": 0.074, + "d_bind": 0.03988 + }, + { + "tau": 0.0164, + "v": 0.8351, + "H_a": 0.061, + "d_bind": 0.02018 + } + ] + }, + { + "step": 17580, + "tok": 72007680, + "ce": 3.5345, + "ppl": 34.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3477, + "loop_stats": [ + { + "tau": 0.0648, + "v": 0.8462, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0213, + "v": 0.827, + "H_a": 0.067, + "d_bind": 0.03701 + }, + { + "tau": 0.0095, + "v": 0.8151, + "H_a": 0.057, + "d_bind": 0.01968 + } + ] + }, + { + "step": 17600, + "tok": 72089600, + "ce": 2.873, + "ppl": 17.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3503, + "loop_stats": [ + { + "tau": 0.124, + "v": 0.8565, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8379, + "H_a": 0.075, + "d_bind": 0.04397 + }, + { + "tau": 0.0242, + "v": 0.8285, + "H_a": 0.056, + "d_bind": 0.02015 + } + ] + }, + { + "step": 17620, + "tok": 72171520, + "ce": 3.0081, + "ppl": 20.2, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.957, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3483, + "loop_stats": [ + { + "tau": 0.0997, + "v": 0.8417, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0264, + "v": 0.8305, + "H_a": 0.075, + "d_bind": 0.04061 + }, + { + "tau": 0.0167, + "v": 0.821, + "H_a": 0.061, + "d_bind": 0.02114 + } + ] + }, + { + "step": 17640, + "tok": 72253440, + "ce": 2.9682, + "ppl": 19.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22468, + "peak_MiB": 5301, + "v_gain": 0.3478, + "loop_stats": [ + { + "tau": 0.1054, + "v": 0.8202, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8008, + "H_a": 0.071, + "d_bind": 0.04019 + }, + { + "tau": 0.0332, + "v": 0.7982, + "H_a": 0.053, + "d_bind": 0.02051 + } + ] + }, + { + "step": 17660, + "tok": 72335360, + "ce": 2.8388, + "ppl": 17.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3481, + "loop_stats": [ + { + "tau": 0.1142, + "v": 0.8634, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.8299, + "H_a": 0.07, + "d_bind": 0.04084 + }, + { + "tau": 0.0347, + "v": 0.8126, + "H_a": 0.061, + "d_bind": 0.02018 + } + ] + }, + { + "step": 17680, + "tok": 72417280, + "ce": 3.1956, + "ppl": 24.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3477, + "loop_stats": [ + { + "tau": 0.0813, + "v": 0.8459, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8331, + "H_a": 0.076, + "d_bind": 0.04088 + }, + { + "tau": 0.0181, + "v": 0.8312, + "H_a": 0.064, + "d_bind": 0.02165 + } + ] + }, + { + "step": 17700, + "tok": 72499200, + "ce": 3.0539, + "ppl": 21.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3485, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.8657, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.849, + "H_a": 0.073, + "d_bind": 0.04241 + }, + { + "tau": 0.0259, + "v": 0.85, + "H_a": 0.059, + "d_bind": 0.02176 + } + ] + }, + { + "step": 17720, + "tok": 72581120, + "ce": 3.1501, + "ppl": 23.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.962, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3449, + "loop_stats": [ + { + "tau": 0.1015, + "v": 0.8669, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8477, + "H_a": 0.07, + "d_bind": 0.03998 + }, + { + "tau": 0.0327, + "v": 0.8414, + "H_a": 0.056, + "d_bind": 0.02115 + } + ] + }, + { + "step": 17740, + "tok": 72663040, + "ce": 2.9635, + "ppl": 19.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3448, + "loop_stats": [ + { + "tau": 0.093, + "v": 0.8707, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0374, + "v": 0.8585, + "H_a": 0.071, + "d_bind": 0.03863 + }, + { + "tau": 0.029, + "v": 0.8558, + "H_a": 0.057, + "d_bind": 0.02234 + } + ] + }, + { + "step": 17760, + "tok": 72744960, + "ce": 2.8429, + "ppl": 17.2, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.957, + "tok_s": 22469, + "peak_MiB": 5301, + "v_gain": 0.3472, + "loop_stats": [ + { + "tau": 0.1327, + "v": 0.8453, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8179, + "H_a": 0.071, + "d_bind": 0.04193 + }, + { + "tau": 0.0404, + "v": 0.8134, + "H_a": 0.059, + "d_bind": 0.02071 + } + ] + }, + { + "step": 17780, + "tok": 72826880, + "ce": 2.6522, + "ppl": 14.2, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3496, + "loop_stats": [ + { + "tau": 0.1278, + "v": 0.8373, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8105, + "H_a": 0.073, + "d_bind": 0.04322 + }, + { + "tau": 0.0405, + "v": 0.7961, + "H_a": 0.057, + "d_bind": 0.02141 + } + ] + }, + { + "step": 17800, + "tok": 72908800, + "ce": 3.1866, + "ppl": 24.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3516, + "loop_stats": [ + { + "tau": 0.079, + "v": 0.8739, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.02, + "v": 0.8434, + "H_a": 0.079, + "d_bind": 0.03869 + }, + { + "tau": 0.0135, + "v": 0.836, + "H_a": 0.06, + "d_bind": 0.02212 + } + ] + }, + { + "step": 17820, + "tok": 72990720, + "ce": 3.5718, + "ppl": 35.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3507, + "loop_stats": [ + { + "tau": 0.0742, + "v": 0.845, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0196, + "v": 0.8186, + "H_a": 0.077, + "d_bind": 0.03983 + }, + { + "tau": 0.0121, + "v": 0.8083, + "H_a": 0.062, + "d_bind": 0.02073 + } + ] + }, + { + "step": 17840, + "tok": 73072640, + "ce": 3.0729, + "ppl": 21.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3507, + "loop_stats": [ + { + "tau": 0.0915, + "v": 0.8434, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.8234, + "H_a": 0.079, + "d_bind": 0.04152 + }, + { + "tau": 0.0179, + "v": 0.8125, + "H_a": 0.058, + "d_bind": 0.02109 + } + ] + }, + { + "step": 17860, + "tok": 73154560, + "ce": 3.0111, + "ppl": 20.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3488, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.8945, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8802, + "H_a": 0.072, + "d_bind": 0.03899 + }, + { + "tau": 0.0273, + "v": 0.8732, + "H_a": 0.059, + "d_bind": 0.0212 + } + ] + }, + { + "step": 17880, + "tok": 73236480, + "ce": 2.9976, + "ppl": 20.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22470, + "peak_MiB": 5301, + "v_gain": 0.3521, + "loop_stats": [ + { + "tau": 0.1011, + "v": 0.8618, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.83, + "H_a": 0.069, + "d_bind": 0.04375 + }, + { + "tau": 0.0273, + "v": 0.8105, + "H_a": 0.058, + "d_bind": 0.02044 + } + ] + }, + { + "step": 17900, + "tok": 73318400, + "ce": 2.7145, + "ppl": 15.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3537, + "loop_stats": [ + { + "tau": 0.1115, + "v": 0.8318, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.7948, + "H_a": 0.075, + "d_bind": 0.0415 + }, + { + "tau": 0.0239, + "v": 0.7842, + "H_a": 0.057, + "d_bind": 0.0218 + } + ] + }, + { + "step": 17920, + "tok": 73400320, + "ce": 3.2694, + "ppl": 26.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3576, + "loop_stats": [ + { + "tau": 0.0963, + "v": 0.8716, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8331, + "H_a": 0.077, + "d_bind": 0.04054 + }, + { + "tau": 0.0188, + "v": 0.812, + "H_a": 0.063, + "d_bind": 0.02149 + } + ] + }, + { + "step": 17940, + "tok": 73482240, + "ce": 3.0459, + "ppl": 21.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3605, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.875, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0203, + "v": 0.856, + "H_a": 0.074, + "d_bind": 0.03965 + }, + { + "tau": 0.0134, + "v": 0.85, + "H_a": 0.059, + "d_bind": 0.02238 + } + ] + }, + { + "step": 17960, + "tok": 73564160, + "ce": 3.0676, + "ppl": 21.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3627, + "loop_stats": [ + { + "tau": 0.1158, + "v": 0.8668, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8341, + "H_a": 0.069, + "d_bind": 0.0402 + }, + { + "tau": 0.0198, + "v": 0.8205, + "H_a": 0.056, + "d_bind": 0.02163 + } + ] + }, + { + "step": 17980, + "tok": 73646080, + "ce": 2.7451, + "ppl": 15.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3661, + "loop_stats": [ + { + "tau": 0.1241, + "v": 0.8646, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8369, + "H_a": 0.074, + "d_bind": 0.04058 + }, + { + "tau": 0.0274, + "v": 0.8298, + "H_a": 0.059, + "d_bind": 0.02166 + } + ] + }, + { + "step": 18000, + "tok": 73728000, + "ce": 2.5911, + "ppl": 13.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22471, + "peak_MiB": 5301, + "v_gain": 0.3681, + "loop_stats": [ + { + "tau": 0.1374, + "v": 0.8887, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.03, + "v": 0.8561, + "H_a": 0.075, + "d_bind": 0.04212 + }, + { + "tau": 0.02, + "v": 0.8459, + "H_a": 0.055, + "d_bind": 0.02082 + } + ] + }, + { + "step": 18020, + "tok": 73809920, + "ce": 3.4315, + "ppl": 30.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3652, + "loop_stats": [ + { + "tau": 0.0772, + "v": 0.872, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0176, + "v": 0.84, + "H_a": 0.072, + "d_bind": 0.0371 + }, + { + "tau": 0.0081, + "v": 0.8205, + "H_a": 0.058, + "d_bind": 0.02051 + } + ] + }, + { + "step": 18040, + "tok": 73891840, + "ce": 2.8163, + "ppl": 16.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.957, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3637, + "loop_stats": [ + { + "tau": 0.106, + "v": 0.865, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.8356, + "H_a": 0.077, + "d_bind": 0.04026 + }, + { + "tau": 0.0207, + "v": 0.8206, + "H_a": 0.067, + "d_bind": 0.02351 + } + ] + }, + { + "step": 18060, + "tok": 73973760, + "ce": 2.8909, + "ppl": 18.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3613, + "loop_stats": [ + { + "tau": 0.1143, + "v": 0.8538, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.8258, + "H_a": 0.076, + "d_bind": 0.04387 + }, + { + "tau": 0.0287, + "v": 0.811, + "H_a": 0.064, + "d_bind": 0.02199 + } + ] + }, + { + "step": 18080, + "tok": 74055680, + "ce": 3.0128, + "ppl": 20.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3592, + "loop_stats": [ + { + "tau": 0.1336, + "v": 0.8545, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.051, + "v": 0.8204, + "H_a": 0.079, + "d_bind": 0.04424 + }, + { + "tau": 0.0407, + "v": 0.808, + "H_a": 0.063, + "d_bind": 0.02084 + } + ] + }, + { + "step": 18100, + "tok": 74137600, + "ce": 3.5472, + "ppl": 34.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3588, + "loop_stats": [ + { + "tau": 0.0782, + "v": 0.8665, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0151, + "v": 0.8396, + "H_a": 0.081, + "d_bind": 0.03823 + }, + { + "tau": 0.0097, + "v": 0.8315, + "H_a": 0.069, + "d_bind": 0.02283 + } + ] + }, + { + "step": 18120, + "tok": 74219520, + "ce": 2.7648, + "ppl": 15.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.94, + "tok_s": 22472, + "peak_MiB": 5301, + "v_gain": 0.3571, + "loop_stats": [ + { + "tau": 0.1271, + "v": 0.8399, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0469, + "v": 0.7903, + "H_a": 0.076, + "d_bind": 0.04005 + }, + { + "tau": 0.0412, + "v": 0.778, + "H_a": 0.062, + "d_bind": 0.02321 + } + ] + }, + { + "step": 18140, + "tok": 74301440, + "ce": 3.3284, + "ppl": 27.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3577, + "loop_stats": [ + { + "tau": 0.094, + "v": 0.8288, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0286, + "v": 0.7934, + "H_a": 0.071, + "d_bind": 0.04035 + }, + { + "tau": 0.0233, + "v": 0.7753, + "H_a": 0.063, + "d_bind": 0.02117 + } + ] + }, + { + "step": 18160, + "tok": 74383360, + "ce": 2.9963, + "ppl": 20.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3584, + "loop_stats": [ + { + "tau": 0.1025, + "v": 0.802, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.7578, + "H_a": 0.072, + "d_bind": 0.04218 + }, + { + "tau": 0.0246, + "v": 0.7401, + "H_a": 0.064, + "d_bind": 0.02255 + } + ] + }, + { + "step": 18180, + "tok": 74465280, + "ce": 3.3848, + "ppl": 29.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3566, + "loop_stats": [ + { + "tau": 0.0915, + "v": 0.8594, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0228, + "v": 0.8232, + "H_a": 0.077, + "d_bind": 0.03884 + }, + { + "tau": 0.0181, + "v": 0.81, + "H_a": 0.062, + "d_bind": 0.02147 + } + ] + }, + { + "step": 18200, + "tok": 74547200, + "ce": 3.0459, + "ppl": 21.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3604, + "loop_stats": [ + { + "tau": 0.0838, + "v": 0.8247, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0304, + "v": 0.7788, + "H_a": 0.076, + "d_bind": 0.04461 + }, + { + "tau": 0.0239, + "v": 0.7632, + "H_a": 0.06, + "d_bind": 0.02443 + } + ] + }, + { + "step": 18220, + "tok": 74629120, + "ce": 2.8527, + "ppl": 17.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3641, + "loop_stats": [ + { + "tau": 0.1302, + "v": 0.8037, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.775, + "H_a": 0.078, + "d_bind": 0.0412 + }, + { + "tau": 0.0287, + "v": 0.7696, + "H_a": 0.058, + "d_bind": 0.02284 + } + ] + }, + { + "step": 18240, + "tok": 74711040, + "ce": 3.5693, + "ppl": 35.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22473, + "peak_MiB": 5301, + "v_gain": 0.3621, + "loop_stats": [ + { + "tau": 0.0783, + "v": 0.8788, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0304, + "v": 0.8382, + "H_a": 0.079, + "d_bind": 0.03933 + }, + { + "tau": 0.0229, + "v": 0.8195, + "H_a": 0.067, + "d_bind": 0.02418 + } + ] + }, + { + "step": 18260, + "tok": 74792960, + "ce": 2.899, + "ppl": 18.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3639, + "loop_stats": [ + { + "tau": 0.1128, + "v": 0.8234, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0313, + "v": 0.7993, + "H_a": 0.075, + "d_bind": 0.0409 + }, + { + "tau": 0.0278, + "v": 0.7944, + "H_a": 0.058, + "d_bind": 0.02298 + } + ] + }, + { + "step": 18280, + "tok": 74874880, + "ce": 3.7842, + "ppl": 44.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3669, + "loop_stats": [ + { + "tau": 0.0597, + "v": 0.847, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0086, + "v": 0.8065, + "H_a": 0.079, + "d_bind": 0.03716 + }, + { + "tau": 0.0063, + "v": 0.7915, + "H_a": 0.057, + "d_bind": 0.02126 + } + ] + }, + { + "step": 18300, + "tok": 74956800, + "ce": 2.7867, + "ppl": 16.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.952, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3703, + "loop_stats": [ + { + "tau": 0.1162, + "v": 0.8029, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.7737, + "H_a": 0.071, + "d_bind": 0.04233 + }, + { + "tau": 0.0329, + "v": 0.7642, + "H_a": 0.057, + "d_bind": 0.02094 + } + ] + }, + { + "step": 18320, + "tok": 75038720, + "ce": 3.5567, + "ppl": 35.0, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.947, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3675, + "loop_stats": [ + { + "tau": 0.0931, + "v": 0.8524, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0326, + "v": 0.8286, + "H_a": 0.07, + "d_bind": 0.03872 + }, + { + "tau": 0.0271, + "v": 0.8185, + "H_a": 0.062, + "d_bind": 0.02309 + } + ] + }, + { + "step": 18340, + "tok": 75120640, + "ce": 2.9842, + "ppl": 19.8, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.938, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3656, + "loop_stats": [ + { + "tau": 0.1176, + "v": 0.7895, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.046, + "v": 0.7604, + "H_a": 0.077, + "d_bind": 0.04381 + }, + { + "tau": 0.0408, + "v": 0.7495, + "H_a": 0.06, + "d_bind": 0.02101 + } + ] + }, + { + "step": 18360, + "tok": 75202560, + "ce": 3.1064, + "ppl": 22.3, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22474, + "peak_MiB": 5301, + "v_gain": 0.3646, + "loop_stats": [ + { + "tau": 0.0887, + "v": 0.8219, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.7947, + "H_a": 0.069, + "d_bind": 0.04008 + }, + { + "tau": 0.0226, + "v": 0.7868, + "H_a": 0.054, + "d_bind": 0.01971 + } + ] + }, + { + "step": 18380, + "tok": 75284480, + "ce": 2.8135, + "ppl": 16.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22475, + "peak_MiB": 5301, + "v_gain": 0.3653, + "loop_stats": [ + { + "tau": 0.0857, + "v": 0.8345, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0321, + "v": 0.806, + "H_a": 0.073, + "d_bind": 0.04032 + }, + { + "tau": 0.0249, + "v": 0.793, + "H_a": 0.059, + "d_bind": 0.02035 + } + ] + }, + { + "step": 18400, + "tok": 75366400, + "ce": 3.0345, + "ppl": 20.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.942, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3653, + "loop_stats": [ + { + "tau": 0.095, + "v": 0.8432, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0298, + "v": 0.805, + "H_a": 0.076, + "d_bind": 0.04411 + }, + { + "tau": 0.0217, + "v": 0.7875, + "H_a": 0.06, + "d_bind": 0.02379 + } + ] + }, + { + "step": 18420, + "tok": 75448320, + "ce": 2.9266, + "ppl": 18.7, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.947, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3654, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.8126, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.7749, + "H_a": 0.071, + "d_bind": 0.0418 + }, + { + "tau": 0.0329, + "v": 0.7615, + "H_a": 0.062, + "d_bind": 0.02246 + } + ] + }, + { + "step": 18440, + "tok": 75530240, + "ce": 3.3223, + "ppl": 27.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.366, + "loop_stats": [ + { + "tau": 0.091, + "v": 0.8773, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0381, + "v": 0.8388, + "H_a": 0.08, + "d_bind": 0.04012 + }, + { + "tau": 0.0277, + "v": 0.8131, + "H_a": 0.057, + "d_bind": 0.022 + } + ] + }, + { + "step": 18460, + "tok": 75612160, + "ce": 3.1195, + "ppl": 22.6, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.947, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3664, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.8101, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.7728, + "H_a": 0.075, + "d_bind": 0.03914 + }, + { + "tau": 0.0258, + "v": 0.7573, + "H_a": 0.057, + "d_bind": 0.02123 + } + ] + }, + { + "step": 18480, + "tok": 75694080, + "ce": 2.8473, + "ppl": 17.2, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.93, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3676, + "loop_stats": [ + { + "tau": 0.1236, + "v": 0.8495, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8247, + "H_a": 0.071, + "d_bind": 0.04182 + }, + { + "tau": 0.0323, + "v": 0.8107, + "H_a": 0.054, + "d_bind": 0.02133 + } + ] + }, + { + "step": 18500, + "tok": 75776000, + "ce": 3.3274, + "ppl": 27.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22476, + "peak_MiB": 5301, + "v_gain": 0.3667, + "loop_stats": [ + { + "tau": 0.0937, + "v": 0.8158, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0313, + "v": 0.7786, + "H_a": 0.075, + "d_bind": 0.04272 + }, + { + "tau": 0.0233, + "v": 0.7653, + "H_a": 0.061, + "d_bind": 0.02133 + } + ] + }, + { + "step": 18520, + "tok": 75857920, + "ce": 3.1076, + "ppl": 22.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3672, + "loop_stats": [ + { + "tau": 0.1145, + "v": 0.8323, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0272, + "v": 0.7865, + "H_a": 0.075, + "d_bind": 0.0401 + }, + { + "tau": 0.0222, + "v": 0.771, + "H_a": 0.066, + "d_bind": 0.02253 + } + ] + }, + { + "step": 18540, + "tok": 75939840, + "ce": 2.999, + "ppl": 20.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.367, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8498, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8163, + "H_a": 0.071, + "d_bind": 0.04553 + }, + { + "tau": 0.0326, + "v": 0.8052, + "H_a": 0.06, + "d_bind": 0.02098 + } + ] + }, + { + "step": 18560, + "tok": 76021760, + "ce": 2.5156, + "ppl": 12.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3656, + "loop_stats": [ + { + "tau": 0.1509, + "v": 0.8161, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0614, + "v": 0.807, + "H_a": 0.066, + "d_bind": 0.04571 + }, + { + "tau": 0.0542, + "v": 0.8054, + "H_a": 0.051, + "d_bind": 0.01923 + } + ] + }, + { + "step": 18580, + "tok": 76103680, + "ce": 2.9502, + "ppl": 19.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3663, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.8218, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.04, + "v": 0.7999, + "H_a": 0.069, + "d_bind": 0.04363 + }, + { + "tau": 0.0326, + "v": 0.7855, + "H_a": 0.058, + "d_bind": 0.02152 + } + ] + }, + { + "step": 18600, + "tok": 76185600, + "ce": 3.0233, + "ppl": 20.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.3656, + "loop_stats": [ + { + "tau": 0.1094, + "v": 0.8717, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.8409, + "H_a": 0.079, + "d_bind": 0.03823 + }, + { + "tau": 0.0272, + "v": 0.823, + "H_a": 0.063, + "d_bind": 0.02247 + } + ] + }, + { + "step": 18620, + "tok": 76267520, + "ce": 3.1184, + "ppl": 22.6, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.962, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3651, + "loop_stats": [ + { + "tau": 0.0958, + "v": 0.8212, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0302, + "v": 0.7954, + "H_a": 0.071, + "d_bind": 0.04076 + }, + { + "tau": 0.0248, + "v": 0.7825, + "H_a": 0.059, + "d_bind": 0.02052 + } + ] + }, + { + "step": 18640, + "tok": 76349440, + "ce": 2.8431, + "ppl": 17.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3692, + "loop_stats": [ + { + "tau": 0.1094, + "v": 0.7965, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.7637, + "H_a": 0.077, + "d_bind": 0.04113 + }, + { + "tau": 0.0229, + "v": 0.7483, + "H_a": 0.063, + "d_bind": 0.02268 + } + ] + }, + { + "step": 18660, + "tok": 76431360, + "ce": 3.2543, + "ppl": 25.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.367, + "loop_stats": [ + { + "tau": 0.0998, + "v": 0.8299, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.794, + "H_a": 0.074, + "d_bind": 0.03978 + }, + { + "tau": 0.0209, + "v": 0.7801, + "H_a": 0.063, + "d_bind": 0.02058 + } + ] + }, + { + "step": 18680, + "tok": 76513280, + "ce": 2.7063, + "ppl": 15.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3632, + "loop_stats": [ + { + "tau": 0.0901, + "v": 0.8383, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.8015, + "H_a": 0.072, + "d_bind": 0.04451 + }, + { + "tau": 0.0237, + "v": 0.7892, + "H_a": 0.063, + "d_bind": 0.02186 + } + ] + }, + { + "step": 18700, + "tok": 76595200, + "ce": 3.2732, + "ppl": 26.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3632, + "loop_stats": [ + { + "tau": 0.0872, + "v": 0.8107, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.7823, + "H_a": 0.08, + "d_bind": 0.04344 + }, + { + "tau": 0.0206, + "v": 0.7729, + "H_a": 0.065, + "d_bind": 0.02163 + } + ] + }, + { + "step": 18720, + "tok": 76677120, + "ce": 3.0012, + "ppl": 20.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.96, + "tok_s": 22477, + "peak_MiB": 5301, + "v_gain": 0.361, + "loop_stats": [ + { + "tau": 0.1015, + "v": 0.8225, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0384, + "v": 0.7845, + "H_a": 0.075, + "d_bind": 0.04641 + }, + { + "tau": 0.0313, + "v": 0.7739, + "H_a": 0.065, + "d_bind": 0.02302 + } + ] + }, + { + "step": 18740, + "tok": 76759040, + "ce": 2.6602, + "ppl": 14.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22478, + "peak_MiB": 5301, + "v_gain": 0.3586, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.8645, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0525, + "v": 0.8241, + "H_a": 0.072, + "d_bind": 0.04423 + }, + { + "tau": 0.0418, + "v": 0.8107, + "H_a": 0.062, + "d_bind": 0.02065 + } + ] + }, + { + "step": 18760, + "tok": 76840960, + "ce": 3.5507, + "ppl": 34.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.362, + "loop_stats": [ + { + "tau": 0.0683, + "v": 0.8821, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0093, + "v": 0.8524, + "H_a": 0.076, + "d_bind": 0.03836 + }, + { + "tau": 0.0045, + "v": 0.8373, + "H_a": 0.063, + "d_bind": 0.02332 + } + ] + }, + { + "step": 18780, + "tok": 76922880, + "ce": 3.3807, + "ppl": 29.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3612, + "loop_stats": [ + { + "tau": 0.0817, + "v": 0.8359, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.7936, + "H_a": 0.079, + "d_bind": 0.04178 + }, + { + "tau": 0.0235, + "v": 0.7707, + "H_a": 0.063, + "d_bind": 0.02388 + } + ] + }, + { + "step": 18800, + "tok": 77004800, + "ce": 3.2004, + "ppl": 24.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.957, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3579, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.8304, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8097, + "H_a": 0.078, + "d_bind": 0.04362 + }, + { + "tau": 0.0326, + "v": 0.8008, + "H_a": 0.066, + "d_bind": 0.02139 + } + ] + }, + { + "step": 18820, + "tok": 77086720, + "ce": 3.0579, + "ppl": 21.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3573, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8235, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.7868, + "H_a": 0.077, + "d_bind": 0.04413 + }, + { + "tau": 0.0338, + "v": 0.7789, + "H_a": 0.061, + "d_bind": 0.02224 + } + ] + }, + { + "step": 18840, + "tok": 77168640, + "ce": 2.9151, + "ppl": 18.5, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.947, + "tok_s": 22479, + "peak_MiB": 5301, + "v_gain": 0.3605, + "loop_stats": [ + { + "tau": 0.1161, + "v": 0.8312, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.049, + "v": 0.8066, + "H_a": 0.08, + "d_bind": 0.0429 + }, + { + "tau": 0.0382, + "v": 0.7942, + "H_a": 0.064, + "d_bind": 0.02415 + } + ] + }, + { + "step": 18860, + "tok": 77250560, + "ce": 2.9498, + "ppl": 19.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3616, + "loop_stats": [ + { + "tau": 0.1174, + "v": 0.8569, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.8221, + "H_a": 0.075, + "d_bind": 0.04075 + }, + { + "tau": 0.0241, + "v": 0.8061, + "H_a": 0.057, + "d_bind": 0.02222 + } + ] + }, + { + "step": 18880, + "tok": 77332480, + "ce": 2.866, + "ppl": 17.6, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.942, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3584, + "loop_stats": [ + { + "tau": 0.1023, + "v": 0.8584, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0511, + "v": 0.8324, + "H_a": 0.088, + "d_bind": 0.04675 + }, + { + "tau": 0.0347, + "v": 0.8178, + "H_a": 0.069, + "d_bind": 0.02593 + } + ] + }, + { + "step": 18900, + "tok": 77414400, + "ce": 2.8776, + "ppl": 17.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.357, + "loop_stats": [ + { + "tau": 0.1033, + "v": 0.8239, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.7857, + "H_a": 0.081, + "d_bind": 0.04362 + }, + { + "tau": 0.0321, + "v": 0.7697, + "H_a": 0.064, + "d_bind": 0.02304 + } + ] + }, + { + "step": 18920, + "tok": 77496320, + "ce": 2.9881, + "ppl": 19.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3553, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.8246, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.811, + "H_a": 0.078, + "d_bind": 0.04421 + }, + { + "tau": 0.0393, + "v": 0.8085, + "H_a": 0.059, + "d_bind": 0.02474 + } + ] + }, + { + "step": 18940, + "tok": 77578240, + "ce": 2.6835, + "ppl": 14.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3565, + "loop_stats": [ + { + "tau": 0.1269, + "v": 0.7931, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0581, + "v": 0.7694, + "H_a": 0.081, + "d_bind": 0.04563 + }, + { + "tau": 0.0488, + "v": 0.7641, + "H_a": 0.061, + "d_bind": 0.02297 + } + ] + }, + { + "step": 18960, + "tok": 77660160, + "ce": 3.3361, + "ppl": 28.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22480, + "peak_MiB": 5301, + "v_gain": 0.3583, + "loop_stats": [ + { + "tau": 0.0897, + "v": 0.845, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.8355, + "H_a": 0.079, + "d_bind": 0.0413 + }, + { + "tau": 0.0307, + "v": 0.8287, + "H_a": 0.062, + "d_bind": 0.0228 + } + ] + }, + { + "step": 18980, + "tok": 77742080, + "ce": 2.6218, + "ppl": 13.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3597, + "loop_stats": [ + { + "tau": 0.1411, + "v": 0.8294, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0624, + "v": 0.8198, + "H_a": 0.073, + "d_bind": 0.04219 + }, + { + "tau": 0.0568, + "v": 0.8131, + "H_a": 0.053, + "d_bind": 0.02272 + } + ] + }, + { + "step": 19000, + "tok": 77824000, + "ce": 2.4145, + "ppl": 11.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3624, + "loop_stats": [ + { + "tau": 0.1351, + "v": 0.7693, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0634, + "v": 0.7481, + "H_a": 0.077, + "d_bind": 0.04829 + }, + { + "tau": 0.0578, + "v": 0.7395, + "H_a": 0.065, + "d_bind": 0.02436 + } + ] + }, + { + "step": 19020, + "tok": 77905920, + "ce": 2.9248, + "ppl": 18.6, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3642, + "loop_stats": [ + { + "tau": 0.0801, + "v": 0.8192, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.7822, + "H_a": 0.084, + "d_bind": 0.04164 + }, + { + "tau": 0.0199, + "v": 0.7652, + "H_a": 0.07, + "d_bind": 0.02416 + } + ] + }, + { + "step": 19040, + "tok": 77987840, + "ce": 3.0869, + "ppl": 21.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3626, + "loop_stats": [ + { + "tau": 0.1133, + "v": 0.8163, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.7947, + "H_a": 0.08, + "d_bind": 0.0408 + }, + { + "tau": 0.0393, + "v": 0.789, + "H_a": 0.062, + "d_bind": 0.0231 + } + ] + }, + { + "step": 19060, + "tok": 78069760, + "ce": 2.8592, + "ppl": 17.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3613, + "loop_stats": [ + { + "tau": 0.1144, + "v": 0.8023, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.7837, + "H_a": 0.08, + "d_bind": 0.0453 + }, + { + "tau": 0.0435, + "v": 0.7713, + "H_a": 0.061, + "d_bind": 0.02481 + } + ] + }, + { + "step": 19080, + "tok": 78151680, + "ce": 2.9763, + "ppl": 19.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22481, + "peak_MiB": 5301, + "v_gain": 0.3606, + "loop_stats": [ + { + "tau": 0.1021, + "v": 0.7893, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.7668, + "H_a": 0.08, + "d_bind": 0.04371 + }, + { + "tau": 0.0424, + "v": 0.751, + "H_a": 0.065, + "d_bind": 0.02351 + } + ] + }, + { + "step": 19100, + "tok": 78233600, + "ce": 2.8103, + "ppl": 16.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3625, + "loop_stats": [ + { + "tau": 0.1062, + "v": 0.7978, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.7579, + "H_a": 0.08, + "d_bind": 0.04389 + }, + { + "tau": 0.0355, + "v": 0.7461, + "H_a": 0.057, + "d_bind": 0.02272 + } + ] + }, + { + "step": 19120, + "tok": 78315520, + "ce": 2.6944, + "ppl": 14.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.972, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3673, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.7629, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.7457, + "H_a": 0.079, + "d_bind": 0.04505 + }, + { + "tau": 0.0294, + "v": 0.7433, + "H_a": 0.057, + "d_bind": 0.02322 + } + ] + }, + { + "step": 19140, + "tok": 78397440, + "ce": 2.5189, + "ppl": 12.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3692, + "loop_stats": [ + { + "tau": 0.1315, + "v": 0.8164, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.7906, + "H_a": 0.073, + "d_bind": 0.04462 + }, + { + "tau": 0.0306, + "v": 0.7826, + "H_a": 0.063, + "d_bind": 0.02232 + } + ] + }, + { + "step": 19160, + "tok": 78479360, + "ce": 2.9802, + "ppl": 19.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.3693, + "loop_stats": [ + { + "tau": 0.1104, + "v": 0.8224, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0322, + "v": 0.7953, + "H_a": 0.07, + "d_bind": 0.04349 + }, + { + "tau": 0.0219, + "v": 0.7847, + "H_a": 0.062, + "d_bind": 0.02346 + } + ] + }, + { + "step": 19180, + "tok": 78561280, + "ce": 2.6926, + "ppl": 14.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3685, + "loop_stats": [ + { + "tau": 0.1322, + "v": 0.794, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.7568, + "H_a": 0.074, + "d_bind": 0.04465 + }, + { + "tau": 0.0415, + "v": 0.7419, + "H_a": 0.056, + "d_bind": 0.02119 + } + ] + }, + { + "step": 19200, + "tok": 78643200, + "ce": 3.0664, + "ppl": 21.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22482, + "peak_MiB": 5301, + "v_gain": 0.3657, + "loop_stats": [ + { + "tau": 0.1045, + "v": 0.818, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.7854, + "H_a": 0.071, + "d_bind": 0.04413 + }, + { + "tau": 0.0291, + "v": 0.7685, + "H_a": 0.059, + "d_bind": 0.02266 + } + ] + }, + { + "step": 19220, + "tok": 78725120, + "ce": 3.078, + "ppl": 21.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.957, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3684, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.8257, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0302, + "v": 0.7953, + "H_a": 0.076, + "d_bind": 0.04114 + }, + { + "tau": 0.0223, + "v": 0.7797, + "H_a": 0.063, + "d_bind": 0.02473 + } + ] + }, + { + "step": 19240, + "tok": 78807040, + "ce": 3.2222, + "ppl": 25.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3678, + "loop_stats": [ + { + "tau": 0.074, + "v": 0.8416, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0327, + "v": 0.8055, + "H_a": 0.076, + "d_bind": 0.04331 + }, + { + "tau": 0.0152, + "v": 0.7856, + "H_a": 0.061, + "d_bind": 0.02342 + } + ] + }, + { + "step": 19260, + "tok": 78888960, + "ce": 3.0794, + "ppl": 21.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.957, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3683, + "loop_stats": [ + { + "tau": 0.0854, + "v": 0.8458, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0257, + "v": 0.8204, + "H_a": 0.077, + "d_bind": 0.04213 + }, + { + "tau": 0.0148, + "v": 0.8069, + "H_a": 0.063, + "d_bind": 0.0245 + } + ] + }, + { + "step": 19280, + "tok": 78970880, + "ce": 3.1921, + "ppl": 24.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3694, + "loop_stats": [ + { + "tau": 0.1124, + "v": 0.8028, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.7685, + "H_a": 0.078, + "d_bind": 0.04515 + }, + { + "tau": 0.0255, + "v": 0.7492, + "H_a": 0.061, + "d_bind": 0.02213 + } + ] + }, + { + "step": 19300, + "tok": 79052800, + "ce": 2.6854, + "ppl": 14.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.3681, + "loop_stats": [ + { + "tau": 0.1238, + "v": 0.8285, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8009, + "H_a": 0.077, + "d_bind": 0.04227 + }, + { + "tau": 0.036, + "v": 0.7887, + "H_a": 0.064, + "d_bind": 0.02561 + } + ] + }, + { + "step": 19320, + "tok": 79134720, + "ce": 3.1131, + "ppl": 22.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22483, + "peak_MiB": 5301, + "v_gain": 0.3681, + "loop_stats": [ + { + "tau": 0.0898, + "v": 0.7923, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.7445, + "H_a": 0.073, + "d_bind": 0.04313 + }, + { + "tau": 0.0208, + "v": 0.725, + "H_a": 0.057, + "d_bind": 0.02382 + } + ] + }, + { + "step": 19340, + "tok": 79216640, + "ce": 3.409, + "ppl": 30.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.962, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.364, + "loop_stats": [ + { + "tau": 0.0859, + "v": 0.7946, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0297, + "v": 0.7746, + "H_a": 0.08, + "d_bind": 0.0427 + }, + { + "tau": 0.0214, + "v": 0.7686, + "H_a": 0.063, + "d_bind": 0.02242 + } + ] + }, + { + "step": 19360, + "tok": 79298560, + "ce": 2.9352, + "ppl": 18.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3633, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.8381, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8, + "H_a": 0.08, + "d_bind": 0.044 + }, + { + "tau": 0.0291, + "v": 0.7847, + "H_a": 0.059, + "d_bind": 0.02475 + } + ] + }, + { + "step": 19380, + "tok": 79380480, + "ce": 2.6238, + "ppl": 13.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3622, + "loop_stats": [ + { + "tau": 0.1307, + "v": 0.8602, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8402, + "H_a": 0.072, + "d_bind": 0.04481 + }, + { + "tau": 0.0373, + "v": 0.837, + "H_a": 0.06, + "d_bind": 0.02387 + } + ] + }, + { + "step": 19400, + "tok": 79462400, + "ce": 2.7617, + "ppl": 15.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3613, + "loop_stats": [ + { + "tau": 0.1283, + "v": 0.7757, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.74, + "H_a": 0.069, + "d_bind": 0.04519 + }, + { + "tau": 0.0375, + "v": 0.7179, + "H_a": 0.058, + "d_bind": 0.02297 + } + ] + }, + { + "step": 19420, + "tok": 79544320, + "ce": 3.0322, + "ppl": 20.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.965, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3612, + "loop_stats": [ + { + "tau": 0.093, + "v": 0.8268, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.7772, + "H_a": 0.072, + "d_bind": 0.04341 + }, + { + "tau": 0.0252, + "v": 0.7546, + "H_a": 0.059, + "d_bind": 0.02431 + } + ] + }, + { + "step": 19440, + "tok": 79626240, + "ce": 3.209, + "ppl": 24.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.96, + "tok_s": 22484, + "peak_MiB": 5301, + "v_gain": 0.3609, + "loop_stats": [ + { + "tau": 0.0948, + "v": 0.8464, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.8306, + "H_a": 0.075, + "d_bind": 0.04081 + }, + { + "tau": 0.0207, + "v": 0.8227, + "H_a": 0.063, + "d_bind": 0.02367 + } + ] + }, + { + "step": 19460, + "tok": 79708160, + "ce": 2.8975, + "ppl": 18.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3621, + "loop_stats": [ + { + "tau": 0.1236, + "v": 0.8188, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0498, + "v": 0.7871, + "H_a": 0.074, + "d_bind": 0.0404 + }, + { + "tau": 0.0303, + "v": 0.7778, + "H_a": 0.062, + "d_bind": 0.02391 + } + ] + }, + { + "step": 19480, + "tok": 79790080, + "ce": 3.4074, + "ppl": 30.2, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.36, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.8024, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0354, + "v": 0.7721, + "H_a": 0.071, + "d_bind": 0.04011 + }, + { + "tau": 0.0233, + "v": 0.7579, + "H_a": 0.057, + "d_bind": 0.02194 + } + ] + }, + { + "step": 19500, + "tok": 79872000, + "ce": 2.6628, + "ppl": 14.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.952, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.3564, + "loop_stats": [ + { + "tau": 0.1314, + "v": 0.7866, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.7534, + "H_a": 0.074, + "d_bind": 0.04192 + }, + { + "tau": 0.0372, + "v": 0.7407, + "H_a": 0.059, + "d_bind": 0.02522 + } + ] + }, + { + "step": 19520, + "tok": 79953920, + "ce": 2.3908, + "ppl": 10.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3572, + "loop_stats": [ + { + "tau": 0.1353, + "v": 0.7933, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0524, + "v": 0.7732, + "H_a": 0.073, + "d_bind": 0.0451 + }, + { + "tau": 0.0375, + "v": 0.7665, + "H_a": 0.059, + "d_bind": 0.0239 + } + ] + }, + { + "step": 19540, + "tok": 80035840, + "ce": 3.14, + "ppl": 23.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.358, + "loop_stats": [ + { + "tau": 0.1152, + "v": 0.8039, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0376, + "v": 0.7672, + "H_a": 0.076, + "d_bind": 0.04137 + }, + { + "tau": 0.0305, + "v": 0.7495, + "H_a": 0.064, + "d_bind": 0.02322 + } + ] + }, + { + "step": 19560, + "tok": 80117760, + "ce": 3.3779, + "ppl": 29.3, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.952, + "tok_s": 22485, + "peak_MiB": 5301, + "v_gain": 0.3579, + "loop_stats": [ + { + "tau": 0.0983, + "v": 0.8441, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8118, + "H_a": 0.072, + "d_bind": 0.03979 + }, + { + "tau": 0.0168, + "v": 0.7932, + "H_a": 0.058, + "d_bind": 0.02295 + } + ] + }, + { + "step": 19580, + "tok": 80199680, + "ce": 2.4624, + "ppl": 11.7, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.947, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.358, + "loop_stats": [ + { + "tau": 0.1484, + "v": 0.7942, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0622, + "v": 0.7728, + "H_a": 0.07, + "d_bind": 0.04291 + }, + { + "tau": 0.0522, + "v": 0.7673, + "H_a": 0.053, + "d_bind": 0.02237 + } + ] + }, + { + "step": 19600, + "tok": 80281600, + "ce": 2.8414, + "ppl": 17.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.3605, + "loop_stats": [ + { + "tau": 0.1405, + "v": 0.7978, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0637, + "v": 0.7705, + "H_a": 0.074, + "d_bind": 0.04156 + }, + { + "tau": 0.0503, + "v": 0.7615, + "H_a": 0.059, + "d_bind": 0.02374 + } + ] + }, + { + "step": 19620, + "tok": 80363520, + "ce": 2.8526, + "ppl": 17.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3616, + "loop_stats": [ + { + "tau": 0.1108, + "v": 0.826, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0451, + "v": 0.7977, + "H_a": 0.073, + "d_bind": 0.03924 + }, + { + "tau": 0.0314, + "v": 0.7843, + "H_a": 0.06, + "d_bind": 0.02303 + } + ] + }, + { + "step": 19640, + "tok": 80445440, + "ce": 2.939, + "ppl": 18.9, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.957, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3633, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.8333, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.7979, + "H_a": 0.074, + "d_bind": 0.03894 + }, + { + "tau": 0.0283, + "v": 0.7781, + "H_a": 0.063, + "d_bind": 0.02391 + } + ] + }, + { + "step": 19660, + "tok": 80527360, + "ce": 2.9027, + "ppl": 18.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3649, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.8305, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.793, + "H_a": 0.077, + "d_bind": 0.04144 + }, + { + "tau": 0.0271, + "v": 0.7763, + "H_a": 0.066, + "d_bind": 0.02489 + } + ] + }, + { + "step": 19680, + "tok": 80609280, + "ce": 3.0721, + "ppl": 21.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22486, + "peak_MiB": 5301, + "v_gain": 0.3683, + "loop_stats": [ + { + "tau": 0.1013, + "v": 0.8095, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0391, + "v": 0.781, + "H_a": 0.075, + "d_bind": 0.04471 + }, + { + "tau": 0.026, + "v": 0.7657, + "H_a": 0.061, + "d_bind": 0.0226 + } + ] + }, + { + "step": 19700, + "tok": 80691200, + "ce": 3.2009, + "ppl": 24.6, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.955, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.373, + "loop_stats": [ + { + "tau": 0.0804, + "v": 0.831, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0291, + "v": 0.8063, + "H_a": 0.077, + "d_bind": 0.04416 + }, + { + "tau": 0.0186, + "v": 0.801, + "H_a": 0.058, + "d_bind": 0.02209 + } + ] + }, + { + "step": 19720, + "tok": 80773120, + "ce": 2.9968, + "ppl": 20.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3718, + "loop_stats": [ + { + "tau": 0.1079, + "v": 0.8126, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.7921, + "H_a": 0.074, + "d_bind": 0.04513 + }, + { + "tau": 0.0257, + "v": 0.7856, + "H_a": 0.057, + "d_bind": 0.02225 + } + ] + }, + { + "step": 19740, + "tok": 80855040, + "ce": 2.6802, + "ppl": 14.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3704, + "loop_stats": [ + { + "tau": 0.1299, + "v": 0.8105, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.7939, + "H_a": 0.08, + "d_bind": 0.04507 + }, + { + "tau": 0.0341, + "v": 0.7863, + "H_a": 0.064, + "d_bind": 0.02447 + } + ] + }, + { + "step": 19760, + "tok": 80936960, + "ce": 2.7926, + "ppl": 16.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3674, + "loop_stats": [ + { + "tau": 0.1185, + "v": 0.8265, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0506, + "v": 0.8044, + "H_a": 0.072, + "d_bind": 0.04355 + }, + { + "tau": 0.0379, + "v": 0.7894, + "H_a": 0.054, + "d_bind": 0.02189 + } + ] + }, + { + "step": 19780, + "tok": 81018880, + "ce": 2.9735, + "ppl": 19.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3666, + "loop_stats": [ + { + "tau": 0.1134, + "v": 0.8565, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8211, + "H_a": 0.077, + "d_bind": 0.0417 + }, + { + "tau": 0.0287, + "v": 0.7991, + "H_a": 0.063, + "d_bind": 0.02316 + } + ] + }, + { + "step": 19800, + "tok": 81100800, + "ce": 2.8757, + "ppl": 17.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22487, + "peak_MiB": 5301, + "v_gain": 0.3637, + "loop_stats": [ + { + "tau": 0.1128, + "v": 0.8444, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.8059, + "H_a": 0.075, + "d_bind": 0.04285 + }, + { + "tau": 0.0215, + "v": 0.7868, + "H_a": 0.064, + "d_bind": 0.02407 + } + ] + }, + { + "step": 19820, + "tok": 81182720, + "ce": 3.3426, + "ppl": 28.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3653, + "loop_stats": [ + { + "tau": 0.109, + "v": 0.8287, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.024, + "v": 0.7883, + "H_a": 0.077, + "d_bind": 0.03908 + }, + { + "tau": 0.0167, + "v": 0.7708, + "H_a": 0.061, + "d_bind": 0.02253 + } + ] + }, + { + "step": 19840, + "tok": 81264640, + "ce": 2.913, + "ppl": 18.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3676, + "loop_stats": [ + { + "tau": 0.1111, + "v": 0.8662, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0354, + "v": 0.8329, + "H_a": 0.075, + "d_bind": 0.04135 + }, + { + "tau": 0.0112, + "v": 0.8125, + "H_a": 0.058, + "d_bind": 0.02293 + } + ] + }, + { + "step": 19860, + "tok": 81346560, + "ce": 2.9879, + "ppl": 19.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3672, + "loop_stats": [ + { + "tau": 0.1205, + "v": 0.8357, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8035, + "H_a": 0.073, + "d_bind": 0.03903 + }, + { + "tau": 0.0208, + "v": 0.7931, + "H_a": 0.057, + "d_bind": 0.02231 + } + ] + }, + { + "step": 19880, + "tok": 81428480, + "ce": 3.2969, + "ppl": 27.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.363, + "loop_stats": [ + { + "tau": 0.0897, + "v": 0.8545, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8215, + "H_a": 0.077, + "d_bind": 0.04155 + }, + { + "tau": 0.023, + "v": 0.7954, + "H_a": 0.066, + "d_bind": 0.02406 + } + ] + }, + { + "step": 19900, + "tok": 81510400, + "ce": 2.7693, + "ppl": 15.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3617, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.8354, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8029, + "H_a": 0.069, + "d_bind": 0.04495 + }, + { + "tau": 0.0258, + "v": 0.7868, + "H_a": 0.064, + "d_bind": 0.02299 + } + ] + }, + { + "step": 19920, + "tok": 81592320, + "ce": 2.9208, + "ppl": 18.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3643, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.8227, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.7979, + "H_a": 0.076, + "d_bind": 0.04287 + }, + { + "tau": 0.0325, + "v": 0.7833, + "H_a": 0.064, + "d_bind": 0.02124 + } + ] + }, + { + "step": 19940, + "tok": 81674240, + "ce": 2.9686, + "ppl": 19.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3688, + "loop_stats": [ + { + "tau": 0.1251, + "v": 0.8441, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0341, + "v": 0.8116, + "H_a": 0.07, + "d_bind": 0.04028 + }, + { + "tau": 0.0173, + "v": 0.7966, + "H_a": 0.062, + "d_bind": 0.02174 + } + ] + }, + { + "step": 19960, + "tok": 81756160, + "ce": 2.721, + "ppl": 15.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3714, + "loop_stats": [ + { + "tau": 0.1156, + "v": 0.7894, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.047, + "v": 0.7648, + "H_a": 0.072, + "d_bind": 0.0479 + }, + { + "tau": 0.0361, + "v": 0.7556, + "H_a": 0.059, + "d_bind": 0.02137 + } + ] + }, + { + "step": 19980, + "tok": 81838080, + "ce": 3.2757, + "ppl": 26.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3748, + "loop_stats": [ + { + "tau": 0.0866, + "v": 0.8165, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0225, + "v": 0.7805, + "H_a": 0.069, + "d_bind": 0.04097 + }, + { + "tau": 0.0144, + "v": 0.7604, + "H_a": 0.053, + "d_bind": 0.01999 + } + ] + }, + { + "step": 20000, + "tok": 81920000, + "ce": 3.2076, + "ppl": 24.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.952, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3736, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.8711, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.8377, + "H_a": 0.078, + "d_bind": 0.04183 + }, + { + "tau": 0.0165, + "v": 0.8199, + "H_a": 0.056, + "d_bind": 0.02165 + } + ] + }, + { + "step": 20020, + "tok": 82001920, + "ce": 3.0619, + "ppl": 21.4, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.952, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3725, + "loop_stats": [ + { + "tau": 0.11, + "v": 0.8294, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.7868, + "H_a": 0.075, + "d_bind": 0.04116 + }, + { + "tau": 0.0193, + "v": 0.7638, + "H_a": 0.055, + "d_bind": 0.02173 + } + ] + }, + { + "step": 20040, + "tok": 82083840, + "ce": 2.9494, + "ppl": 19.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3704, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.828, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.7919, + "H_a": 0.076, + "d_bind": 0.04493 + }, + { + "tau": 0.0261, + "v": 0.7758, + "H_a": 0.061, + "d_bind": 0.02324 + } + ] + }, + { + "step": 20060, + "tok": 82165760, + "ce": 2.8208, + "ppl": 16.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3683, + "loop_stats": [ + { + "tau": 0.109, + "v": 0.8675, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8276, + "H_a": 0.068, + "d_bind": 0.0408 + }, + { + "tau": 0.0287, + "v": 0.809, + "H_a": 0.056, + "d_bind": 0.02097 + } + ] + }, + { + "step": 20080, + "tok": 82247680, + "ce": 2.9788, + "ppl": 19.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.367, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.8324, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.7918, + "H_a": 0.075, + "d_bind": 0.04352 + }, + { + "tau": 0.0363, + "v": 0.7694, + "H_a": 0.06, + "d_bind": 0.02149 + } + ] + }, + { + "step": 20100, + "tok": 82329600, + "ce": 3.1234, + "ppl": 22.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.962, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3666, + "loop_stats": [ + { + "tau": 0.0969, + "v": 0.8404, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.7935, + "H_a": 0.076, + "d_bind": 0.04427 + }, + { + "tau": 0.024, + "v": 0.7668, + "H_a": 0.065, + "d_bind": 0.02287 + } + ] + }, + { + "step": 20120, + "tok": 82411520, + "ce": 2.696, + "ppl": 14.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.367, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.8113, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0537, + "v": 0.7682, + "H_a": 0.081, + "d_bind": 0.04741 + }, + { + "tau": 0.0375, + "v": 0.7477, + "H_a": 0.063, + "d_bind": 0.02081 + } + ] + }, + { + "step": 20140, + "tok": 82493440, + "ce": 2.8173, + "ppl": 16.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3671, + "loop_stats": [ + { + "tau": 0.1203, + "v": 0.8396, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.8076, + "H_a": 0.076, + "d_bind": 0.04641 + }, + { + "tau": 0.032, + "v": 0.8027, + "H_a": 0.063, + "d_bind": 0.02281 + } + ] + }, + { + "step": 20160, + "tok": 82575360, + "ce": 2.4887, + "ppl": 12.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3675, + "loop_stats": [ + { + "tau": 0.1353, + "v": 0.8675, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0523, + "v": 0.8453, + "H_a": 0.071, + "d_bind": 0.04371 + }, + { + "tau": 0.0357, + "v": 0.8355, + "H_a": 0.062, + "d_bind": 0.02291 + } + ] + }, + { + "step": 20180, + "tok": 82657280, + "ce": 2.6691, + "ppl": 14.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3689, + "loop_stats": [ + { + "tau": 0.1266, + "v": 0.8231, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0446, + "v": 0.7935, + "H_a": 0.077, + "d_bind": 0.04624 + }, + { + "tau": 0.0353, + "v": 0.7748, + "H_a": 0.062, + "d_bind": 0.02341 + } + ] + }, + { + "step": 20200, + "tok": 82739200, + "ce": 3.2324, + "ppl": 25.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3659, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.8317, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8031, + "H_a": 0.073, + "d_bind": 0.04513 + }, + { + "tau": 0.0319, + "v": 0.7963, + "H_a": 0.068, + "d_bind": 0.02394 + } + ] + }, + { + "step": 20220, + "tok": 82821120, + "ce": 2.9047, + "ppl": 18.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3652, + "loop_stats": [ + { + "tau": 0.1041, + "v": 0.8775, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8318, + "H_a": 0.076, + "d_bind": 0.04624 + }, + { + "tau": 0.0301, + "v": 0.8102, + "H_a": 0.06, + "d_bind": 0.02222 + } + ] + }, + { + "step": 20240, + "tok": 82903040, + "ce": 2.4158, + "ppl": 11.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.368, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.8601, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0524, + "v": 0.8175, + "H_a": 0.073, + "d_bind": 0.04784 + }, + { + "tau": 0.0435, + "v": 0.7993, + "H_a": 0.063, + "d_bind": 0.02161 + } + ] + }, + { + "step": 20260, + "tok": 82984960, + "ce": 3.3186, + "ppl": 27.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3686, + "loop_stats": [ + { + "tau": 0.0788, + "v": 0.8851, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0315, + "v": 0.8437, + "H_a": 0.079, + "d_bind": 0.03862 + }, + { + "tau": 0.0158, + "v": 0.82, + "H_a": 0.066, + "d_bind": 0.02326 + } + ] + }, + { + "step": 20280, + "tok": 83066880, + "ce": 3.0772, + "ppl": 21.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3711, + "loop_stats": [ + { + "tau": 0.1032, + "v": 0.8522, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.028, + "v": 0.804, + "H_a": 0.073, + "d_bind": 0.04374 + }, + { + "tau": 0.0238, + "v": 0.7812, + "H_a": 0.056, + "d_bind": 0.02143 + } + ] + }, + { + "step": 20300, + "tok": 83148800, + "ce": 2.2237, + "ppl": 9.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3685, + "loop_stats": [ + { + "tau": 0.1287, + "v": 0.8354, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.7902, + "H_a": 0.077, + "d_bind": 0.05161 + }, + { + "tau": 0.0425, + "v": 0.7755, + "H_a": 0.061, + "d_bind": 0.0235 + } + ] + }, + { + "step": 20320, + "tok": 83230720, + "ce": 2.5717, + "ppl": 13.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.369, + "loop_stats": [ + { + "tau": 0.1264, + "v": 0.8073, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.056, + "v": 0.7767, + "H_a": 0.079, + "d_bind": 0.04611 + }, + { + "tau": 0.043, + "v": 0.7697, + "H_a": 0.06, + "d_bind": 0.0229 + } + ] + }, + { + "step": 20340, + "tok": 83312640, + "ce": 3.3861, + "ppl": 29.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.938, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3713, + "loop_stats": [ + { + "tau": 0.086, + "v": 0.815, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0258, + "v": 0.7769, + "H_a": 0.077, + "d_bind": 0.04429 + }, + { + "tau": 0.0194, + "v": 0.7586, + "H_a": 0.065, + "d_bind": 0.02325 + } + ] + }, + { + "step": 20360, + "tok": 83394560, + "ce": 3.0897, + "ppl": 22.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3748, + "loop_stats": [ + { + "tau": 0.0903, + "v": 0.8569, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.026, + "v": 0.8216, + "H_a": 0.075, + "d_bind": 0.04449 + }, + { + "tau": 0.0199, + "v": 0.8071, + "H_a": 0.059, + "d_bind": 0.02229 + } + ] + }, + { + "step": 20380, + "tok": 83476480, + "ce": 3.0345, + "ppl": 20.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3765, + "loop_stats": [ + { + "tau": 0.1134, + "v": 0.8225, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.7995, + "H_a": 0.068, + "d_bind": 0.04271 + }, + { + "tau": 0.0213, + "v": 0.7907, + "H_a": 0.06, + "d_bind": 0.02215 + } + ] + }, + { + "step": 20400, + "tok": 83558400, + "ce": 2.8476, + "ppl": 17.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22488, + "peak_MiB": 5301, + "v_gain": 0.3723, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.8567, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0541, + "v": 0.8131, + "H_a": 0.074, + "d_bind": 0.04563 + }, + { + "tau": 0.031, + "v": 0.7972, + "H_a": 0.064, + "d_bind": 0.02186 + } + ] + }, + { + "step": 20420, + "tok": 83640320, + "ce": 2.903, + "ppl": 18.2, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.947, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3701, + "loop_stats": [ + { + "tau": 0.1316, + "v": 0.8825, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8524, + "H_a": 0.069, + "d_bind": 0.04229 + }, + { + "tau": 0.041, + "v": 0.8448, + "H_a": 0.057, + "d_bind": 0.02272 + } + ] + }, + { + "step": 20440, + "tok": 83722240, + "ce": 2.7529, + "ppl": 15.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3726, + "loop_stats": [ + { + "tau": 0.1219, + "v": 0.8138, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0416, + "v": 0.7868, + "H_a": 0.074, + "d_bind": 0.04637 + }, + { + "tau": 0.0282, + "v": 0.7806, + "H_a": 0.065, + "d_bind": 0.02178 + } + ] + }, + { + "step": 20460, + "tok": 83804160, + "ce": 3.0503, + "ppl": 21.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3743, + "loop_stats": [ + { + "tau": 0.0827, + "v": 0.8451, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0302, + "v": 0.805, + "H_a": 0.077, + "d_bind": 0.04684 + }, + { + "tau": 0.0193, + "v": 0.788, + "H_a": 0.063, + "d_bind": 0.02311 + } + ] + }, + { + "step": 20480, + "tok": 83886080, + "ce": 2.8597, + "ppl": 17.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.94, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3745, + "loop_stats": [ + { + "tau": 0.0924, + "v": 0.8043, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.7563, + "H_a": 0.081, + "d_bind": 0.04487 + }, + { + "tau": 0.0205, + "v": 0.7447, + "H_a": 0.066, + "d_bind": 0.02392 + } + ] + }, + { + "step": 20500, + "tok": 83968000, + "ce": 2.9389, + "ppl": 18.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3709, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.8141, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.78, + "H_a": 0.071, + "d_bind": 0.04521 + }, + { + "tau": 0.0337, + "v": 0.768, + "H_a": 0.055, + "d_bind": 0.0213 + } + ] + }, + { + "step": 20520, + "tok": 84049920, + "ce": 2.4842, + "ppl": 12.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22489, + "peak_MiB": 5301, + "v_gain": 0.3683, + "loop_stats": [ + { + "tau": 0.1293, + "v": 0.8436, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.8094, + "H_a": 0.068, + "d_bind": 0.04338 + }, + { + "tau": 0.0376, + "v": 0.798, + "H_a": 0.062, + "d_bind": 0.02301 + } + ] + }, + { + "step": 20540, + "tok": 84131840, + "ce": 2.5347, + "ppl": 12.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3686, + "loop_stats": [ + { + "tau": 0.1217, + "v": 0.8327, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0544, + "v": 0.792, + "H_a": 0.071, + "d_bind": 0.04429 + }, + { + "tau": 0.0322, + "v": 0.7794, + "H_a": 0.059, + "d_bind": 0.02414 + } + ] + }, + { + "step": 20560, + "tok": 84213760, + "ce": 3.011, + "ppl": 20.3, + "pref_transfer_pron": 0.875, + "pref_train_sv": 0.938, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3686, + "loop_stats": [ + { + "tau": 0.1093, + "v": 0.8673, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.049, + "v": 0.8301, + "H_a": 0.069, + "d_bind": 0.04224 + }, + { + "tau": 0.0196, + "v": 0.8135, + "H_a": 0.064, + "d_bind": 0.02216 + } + ] + }, + { + "step": 20580, + "tok": 84295680, + "ce": 2.5048, + "ppl": 12.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3672, + "loop_stats": [ + { + "tau": 0.1308, + "v": 0.8222, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0486, + "v": 0.7946, + "H_a": 0.073, + "d_bind": 0.04347 + }, + { + "tau": 0.0354, + "v": 0.7845, + "H_a": 0.054, + "d_bind": 0.02287 + } + ] + }, + { + "step": 20600, + "tok": 84377600, + "ce": 2.9062, + "ppl": 18.3, + "pref_transfer_pron": 0.875, + "pref_train_sv": 0.955, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3652, + "loop_stats": [ + { + "tau": 0.111, + "v": 0.8753, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8437, + "H_a": 0.075, + "d_bind": 0.0411 + }, + { + "tau": 0.0305, + "v": 0.8304, + "H_a": 0.062, + "d_bind": 0.02251 + } + ] + }, + { + "step": 20620, + "tok": 84459520, + "ce": 3.1901, + "ppl": 24.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3665, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.8797, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8388, + "H_a": 0.073, + "d_bind": 0.04309 + }, + { + "tau": 0.0319, + "v": 0.816, + "H_a": 0.057, + "d_bind": 0.02346 + } + ] + }, + { + "step": 20640, + "tok": 84541440, + "ce": 2.7469, + "ppl": 15.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22490, + "peak_MiB": 5301, + "v_gain": 0.3675, + "loop_stats": [ + { + "tau": 0.1119, + "v": 0.8351, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.8002, + "H_a": 0.076, + "d_bind": 0.04095 + }, + { + "tau": 0.034, + "v": 0.7855, + "H_a": 0.065, + "d_bind": 0.02263 + } + ] + }, + { + "step": 20660, + "tok": 84623360, + "ce": 2.5395, + "ppl": 12.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3656, + "loop_stats": [ + { + "tau": 0.1285, + "v": 0.8471, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.057, + "v": 0.8217, + "H_a": 0.075, + "d_bind": 0.04341 + }, + { + "tau": 0.0477, + "v": 0.8042, + "H_a": 0.061, + "d_bind": 0.02241 + } + ] + }, + { + "step": 20680, + "tok": 84705280, + "ce": 2.9363, + "ppl": 18.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3659, + "loop_stats": [ + { + "tau": 0.1109, + "v": 0.8315, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.7947, + "H_a": 0.068, + "d_bind": 0.04168 + }, + { + "tau": 0.0398, + "v": 0.7793, + "H_a": 0.063, + "d_bind": 0.02054 + } + ] + }, + { + "step": 20700, + "tok": 84787200, + "ce": 3.1592, + "ppl": 23.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.957, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3675, + "loop_stats": [ + { + "tau": 0.1057, + "v": 0.8508, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.036, + "v": 0.8187, + "H_a": 0.069, + "d_bind": 0.03809 + }, + { + "tau": 0.0271, + "v": 0.8148, + "H_a": 0.062, + "d_bind": 0.02201 + } + ] + }, + { + "step": 20720, + "tok": 84869120, + "ce": 2.6916, + "ppl": 14.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.371, + "loop_stats": [ + { + "tau": 0.1326, + "v": 0.8448, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0487, + "v": 0.8226, + "H_a": 0.071, + "d_bind": 0.04364 + }, + { + "tau": 0.0374, + "v": 0.8144, + "H_a": 0.064, + "d_bind": 0.02252 + } + ] + }, + { + "step": 20740, + "tok": 84951040, + "ce": 2.956, + "ppl": 19.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.967, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3714, + "loop_stats": [ + { + "tau": 0.1214, + "v": 0.8201, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0529, + "v": 0.7916, + "H_a": 0.078, + "d_bind": 0.0432 + }, + { + "tau": 0.0367, + "v": 0.7813, + "H_a": 0.06, + "d_bind": 0.02392 + } + ] + }, + { + "step": 20760, + "tok": 85032960, + "ce": 2.8617, + "ppl": 17.5, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.95, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3701, + "loop_stats": [ + { + "tau": 0.1158, + "v": 0.8373, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8054, + "H_a": 0.081, + "d_bind": 0.04314 + }, + { + "tau": 0.0384, + "v": 0.7962, + "H_a": 0.064, + "d_bind": 0.02327 + } + ] + }, + { + "step": 20780, + "tok": 85114880, + "ce": 3.1846, + "ppl": 24.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3694, + "loop_stats": [ + { + "tau": 0.0962, + "v": 0.85, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.825, + "H_a": 0.078, + "d_bind": 0.041 + }, + { + "tau": 0.022, + "v": 0.812, + "H_a": 0.064, + "d_bind": 0.02329 + } + ] + }, + { + "step": 20800, + "tok": 85196800, + "ce": 2.7216, + "ppl": 15.2, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.947, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3694, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.8079, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0465, + "v": 0.7754, + "H_a": 0.072, + "d_bind": 0.04498 + }, + { + "tau": 0.0371, + "v": 0.7621, + "H_a": 0.065, + "d_bind": 0.02181 + } + ] + }, + { + "step": 20820, + "tok": 85278720, + "ce": 2.7198, + "ppl": 15.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3692, + "loop_stats": [ + { + "tau": 0.1305, + "v": 0.804, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0519, + "v": 0.7786, + "H_a": 0.072, + "d_bind": 0.0456 + }, + { + "tau": 0.0471, + "v": 0.7752, + "H_a": 0.056, + "d_bind": 0.02209 + } + ] + }, + { + "step": 20840, + "tok": 85360640, + "ce": 2.9616, + "ppl": 19.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3692, + "loop_stats": [ + { + "tau": 0.0885, + "v": 0.8422, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.8106, + "H_a": 0.072, + "d_bind": 0.04237 + }, + { + "tau": 0.0279, + "v": 0.798, + "H_a": 0.061, + "d_bind": 0.0219 + } + ] + }, + { + "step": 20860, + "tok": 85442560, + "ce": 3.0263, + "ppl": 20.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.368, + "loop_stats": [ + { + "tau": 0.1042, + "v": 0.8579, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0484, + "v": 0.827, + "H_a": 0.074, + "d_bind": 0.0468 + }, + { + "tau": 0.038, + "v": 0.8217, + "H_a": 0.06, + "d_bind": 0.02303 + } + ] + }, + { + "step": 20880, + "tok": 85524480, + "ce": 2.9289, + "ppl": 18.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22491, + "peak_MiB": 5301, + "v_gain": 0.3685, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.8729, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0435, + "v": 0.8438, + "H_a": 0.076, + "d_bind": 0.04022 + }, + { + "tau": 0.0316, + "v": 0.8296, + "H_a": 0.059, + "d_bind": 0.02333 + } + ] + }, + { + "step": 20900, + "tok": 85606400, + "ce": 3.0653, + "ppl": 21.4, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.962, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3696, + "loop_stats": [ + { + "tau": 0.0869, + "v": 0.8691, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.834, + "H_a": 0.077, + "d_bind": 0.04284 + }, + { + "tau": 0.0291, + "v": 0.8192, + "H_a": 0.059, + "d_bind": 0.02278 + } + ] + }, + { + "step": 20920, + "tok": 85688320, + "ce": 2.9458, + "ppl": 19.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.368, + "loop_stats": [ + { + "tau": 0.1081, + "v": 0.8357, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8058, + "H_a": 0.075, + "d_bind": 0.04308 + }, + { + "tau": 0.0365, + "v": 0.7916, + "H_a": 0.06, + "d_bind": 0.02046 + } + ] + }, + { + "step": 20940, + "tok": 85770240, + "ce": 2.5846, + "ppl": 13.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3667, + "loop_stats": [ + { + "tau": 0.1579, + "v": 0.8293, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0589, + "v": 0.7938, + "H_a": 0.074, + "d_bind": 0.04558 + }, + { + "tau": 0.0525, + "v": 0.7891, + "H_a": 0.065, + "d_bind": 0.02183 + } + ] + }, + { + "step": 20960, + "tok": 85852160, + "ce": 2.8579, + "ppl": 17.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3679, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.8447, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8004, + "H_a": 0.069, + "d_bind": 0.04493 + }, + { + "tau": 0.0323, + "v": 0.7845, + "H_a": 0.053, + "d_bind": 0.01979 + } + ] + }, + { + "step": 20980, + "tok": 85934080, + "ce": 2.4306, + "ppl": 11.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3701, + "loop_stats": [ + { + "tau": 0.131, + "v": 0.797, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0579, + "v": 0.7575, + "H_a": 0.077, + "d_bind": 0.04518 + }, + { + "tau": 0.0402, + "v": 0.7513, + "H_a": 0.061, + "d_bind": 0.02166 + } + ] + }, + { + "step": 21000, + "tok": 86016000, + "ce": 3.3005, + "ppl": 27.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3687, + "loop_stats": [ + { + "tau": 0.0849, + "v": 0.8336, + "H_a": 0.099, + "d_bind": 0.0 + }, + { + "tau": 0.0317, + "v": 0.7935, + "H_a": 0.073, + "d_bind": 0.04161 + }, + { + "tau": 0.0212, + "v": 0.7745, + "H_a": 0.06, + "d_bind": 0.02213 + } + ] + }, + { + "step": 21020, + "tok": 86097920, + "ce": 2.8297, + "ppl": 16.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22492, + "peak_MiB": 5301, + "v_gain": 0.3691, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.8544, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0547, + "v": 0.8246, + "H_a": 0.074, + "d_bind": 0.03942 + }, + { + "tau": 0.0405, + "v": 0.8093, + "H_a": 0.06, + "d_bind": 0.02326 + } + ] + }, + { + "step": 21040, + "tok": 86179840, + "ce": 2.4278, + "ppl": 11.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.96, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3648, + "loop_stats": [ + { + "tau": 0.1282, + "v": 0.8238, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0585, + "v": 0.8066, + "H_a": 0.069, + "d_bind": 0.04687 + }, + { + "tau": 0.0524, + "v": 0.7991, + "H_a": 0.058, + "d_bind": 0.02377 + } + ] + }, + { + "step": 21060, + "tok": 86261760, + "ce": 3.1451, + "ppl": 23.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3654, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.8681, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8381, + "H_a": 0.073, + "d_bind": 0.04142 + }, + { + "tau": 0.0333, + "v": 0.8307, + "H_a": 0.063, + "d_bind": 0.02338 + } + ] + }, + { + "step": 21080, + "tok": 86343680, + "ce": 3.1773, + "ppl": 24.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.962, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.365, + "loop_stats": [ + { + "tau": 0.0881, + "v": 0.8305, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.7958, + "H_a": 0.074, + "d_bind": 0.04088 + }, + { + "tau": 0.0227, + "v": 0.7821, + "H_a": 0.062, + "d_bind": 0.02286 + } + ] + }, + { + "step": 21100, + "tok": 86425600, + "ce": 2.6226, + "ppl": 13.8, + "pref_transfer_pron": 0.87, + "pref_train_sv": 0.962, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3643, + "loop_stats": [ + { + "tau": 0.1331, + "v": 0.83, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0514, + "v": 0.8084, + "H_a": 0.071, + "d_bind": 0.04446 + }, + { + "tau": 0.0375, + "v": 0.8026, + "H_a": 0.065, + "d_bind": 0.02208 + } + ] + }, + { + "step": 21120, + "tok": 86507520, + "ce": 2.7665, + "ppl": 15.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3635, + "loop_stats": [ + { + "tau": 0.1224, + "v": 0.8397, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0552, + "v": 0.809, + "H_a": 0.082, + "d_bind": 0.04101 + }, + { + "tau": 0.0341, + "v": 0.7941, + "H_a": 0.065, + "d_bind": 0.025 + } + ] + }, + { + "step": 21140, + "tok": 86589440, + "ce": 3.0712, + "ppl": 21.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3615, + "loop_stats": [ + { + "tau": 0.0942, + "v": 0.875, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8383, + "H_a": 0.073, + "d_bind": 0.03784 + }, + { + "tau": 0.0172, + "v": 0.8195, + "H_a": 0.066, + "d_bind": 0.02381 + } + ] + }, + { + "step": 21160, + "tok": 86671360, + "ce": 2.7493, + "ppl": 15.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.962, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3628, + "loop_stats": [ + { + "tau": 0.1246, + "v": 0.8629, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8336, + "H_a": 0.074, + "d_bind": 0.04135 + }, + { + "tau": 0.0265, + "v": 0.8236, + "H_a": 0.059, + "d_bind": 0.02379 + } + ] + }, + { + "step": 21180, + "tok": 86753280, + "ce": 2.7963, + "ppl": 16.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3622, + "loop_stats": [ + { + "tau": 0.0927, + "v": 0.8351, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8026, + "H_a": 0.07, + "d_bind": 0.04148 + }, + { + "tau": 0.0373, + "v": 0.7965, + "H_a": 0.057, + "d_bind": 0.02066 + } + ] + }, + { + "step": 21200, + "tok": 86835200, + "ce": 3.0437, + "ppl": 21.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3621, + "loop_stats": [ + { + "tau": 0.0927, + "v": 0.8341, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.804, + "H_a": 0.083, + "d_bind": 0.04096 + }, + { + "tau": 0.0178, + "v": 0.7937, + "H_a": 0.062, + "d_bind": 0.02312 + } + ] + }, + { + "step": 21220, + "tok": 86917120, + "ce": 3.3733, + "ppl": 29.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3639, + "loop_stats": [ + { + "tau": 0.069, + "v": 0.849, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0213, + "v": 0.8206, + "H_a": 0.068, + "d_bind": 0.03827 + }, + { + "tau": 0.013, + "v": 0.8152, + "H_a": 0.054, + "d_bind": 0.01958 + } + ] + }, + { + "step": 21240, + "tok": 86999040, + "ce": 2.781, + "ppl": 16.1, + "pref_transfer_pron": 0.873, + "pref_train_sv": 0.955, + "tok_s": 22493, + "peak_MiB": 5301, + "v_gain": 0.3624, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8483, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8222, + "H_a": 0.075, + "d_bind": 0.04409 + }, + { + "tau": 0.028, + "v": 0.8198, + "H_a": 0.062, + "d_bind": 0.02391 + } + ] + }, + { + "step": 21260, + "tok": 87080960, + "ce": 2.9607, + "ppl": 19.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3631, + "loop_stats": [ + { + "tau": 0.1051, + "v": 0.8418, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0517, + "v": 0.8189, + "H_a": 0.066, + "d_bind": 0.04003 + }, + { + "tau": 0.0352, + "v": 0.812, + "H_a": 0.052, + "d_bind": 0.02032 + } + ] + }, + { + "step": 21280, + "tok": 87162880, + "ce": 2.2301, + "ppl": 9.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3654, + "loop_stats": [ + { + "tau": 0.1531, + "v": 0.8033, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0668, + "v": 0.7839, + "H_a": 0.071, + "d_bind": 0.0477 + }, + { + "tau": 0.0468, + "v": 0.7803, + "H_a": 0.054, + "d_bind": 0.02082 + } + ] + }, + { + "step": 21300, + "tok": 87244800, + "ce": 3.097, + "ppl": 22.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3679, + "loop_stats": [ + { + "tau": 0.0907, + "v": 0.818, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.7773, + "H_a": 0.075, + "d_bind": 0.0402 + }, + { + "tau": 0.0268, + "v": 0.7603, + "H_a": 0.06, + "d_bind": 0.02167 + } + ] + }, + { + "step": 21320, + "tok": 87326720, + "ce": 2.7428, + "ppl": 15.5, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.95, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.3682, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.8373, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.051, + "v": 0.7992, + "H_a": 0.07, + "d_bind": 0.03819 + }, + { + "tau": 0.0274, + "v": 0.7868, + "H_a": 0.054, + "d_bind": 0.02125 + } + ] + }, + { + "step": 21340, + "tok": 87408640, + "ce": 2.5886, + "ppl": 13.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3691, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.8177, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.7883, + "H_a": 0.066, + "d_bind": 0.04082 + }, + { + "tau": 0.0248, + "v": 0.7781, + "H_a": 0.055, + "d_bind": 0.0212 + } + ] + }, + { + "step": 21360, + "tok": 87490560, + "ce": 3.1864, + "ppl": 24.2, + "pref_transfer_pron": 0.86, + "pref_train_sv": 0.942, + "tok_s": 22494, + "peak_MiB": 5301, + "v_gain": 0.3668, + "loop_stats": [ + { + "tau": 0.1023, + "v": 0.8498, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8154, + "H_a": 0.071, + "d_bind": 0.03685 + }, + { + "tau": 0.022, + "v": 0.7993, + "H_a": 0.061, + "d_bind": 0.02215 + } + ] + }, + { + "step": 21380, + "tok": 87572480, + "ce": 2.4776, + "ppl": 11.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.367, + "loop_stats": [ + { + "tau": 0.132, + "v": 0.8601, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.058, + "v": 0.836, + "H_a": 0.073, + "d_bind": 0.04101 + }, + { + "tau": 0.0393, + "v": 0.8225, + "H_a": 0.062, + "d_bind": 0.02283 + } + ] + }, + { + "step": 21400, + "tok": 87654400, + "ce": 3.1728, + "ppl": 23.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3668, + "loop_stats": [ + { + "tau": 0.0957, + "v": 0.8264, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.7875, + "H_a": 0.072, + "d_bind": 0.03873 + }, + { + "tau": 0.0219, + "v": 0.775, + "H_a": 0.062, + "d_bind": 0.02193 + } + ] + }, + { + "step": 21420, + "tok": 87736320, + "ce": 2.8239, + "ppl": 16.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3684, + "loop_stats": [ + { + "tau": 0.0937, + "v": 0.8172, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0318, + "v": 0.7833, + "H_a": 0.079, + "d_bind": 0.0387 + }, + { + "tau": 0.0218, + "v": 0.767, + "H_a": 0.069, + "d_bind": 0.02333 + } + ] + }, + { + "step": 21440, + "tok": 87818240, + "ce": 3.1974, + "ppl": 24.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.369, + "loop_stats": [ + { + "tau": 0.0845, + "v": 0.8605, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0343, + "v": 0.8309, + "H_a": 0.072, + "d_bind": 0.03811 + }, + { + "tau": 0.0168, + "v": 0.8169, + "H_a": 0.063, + "d_bind": 0.02256 + } + ] + }, + { + "step": 21460, + "tok": 87900160, + "ce": 3.0792, + "ppl": 21.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.3694, + "loop_stats": [ + { + "tau": 0.0974, + "v": 0.8546, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.8244, + "H_a": 0.075, + "d_bind": 0.03855 + }, + { + "tau": 0.0193, + "v": 0.8073, + "H_a": 0.058, + "d_bind": 0.02136 + } + ] + }, + { + "step": 21480, + "tok": 87982080, + "ce": 3.0788, + "ppl": 21.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22495, + "peak_MiB": 5301, + "v_gain": 0.3688, + "loop_stats": [ + { + "tau": 0.095, + "v": 0.8428, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8223, + "H_a": 0.078, + "d_bind": 0.03975 + }, + { + "tau": 0.0229, + "v": 0.8136, + "H_a": 0.063, + "d_bind": 0.0223 + } + ] + }, + { + "step": 21500, + "tok": 88064000, + "ce": 2.5575, + "ppl": 12.9, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3692, + "loop_stats": [ + { + "tau": 0.145, + "v": 0.8051, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0642, + "v": 0.7811, + "H_a": 0.071, + "d_bind": 0.04314 + }, + { + "tau": 0.0604, + "v": 0.7776, + "H_a": 0.057, + "d_bind": 0.02277 + } + ] + }, + { + "step": 21520, + "tok": 88145920, + "ce": 2.691, + "ppl": 14.7, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.945, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3715, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.7898, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.7643, + "H_a": 0.071, + "d_bind": 0.03846 + }, + { + "tau": 0.0395, + "v": 0.7534, + "H_a": 0.063, + "d_bind": 0.02202 + } + ] + }, + { + "step": 21540, + "tok": 88227840, + "ce": 2.9036, + "ppl": 18.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3725, + "loop_stats": [ + { + "tau": 0.1392, + "v": 0.799, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0474, + "v": 0.764, + "H_a": 0.076, + "d_bind": 0.03965 + }, + { + "tau": 0.0413, + "v": 0.7518, + "H_a": 0.064, + "d_bind": 0.0232 + } + ] + }, + { + "step": 21560, + "tok": 88309760, + "ce": 3.2993, + "ppl": 27.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.935, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.372, + "loop_stats": [ + { + "tau": 0.0774, + "v": 0.8456, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8107, + "H_a": 0.08, + "d_bind": 0.03928 + }, + { + "tau": 0.0271, + "v": 0.7993, + "H_a": 0.06, + "d_bind": 0.02265 + } + ] + }, + { + "step": 21580, + "tok": 88391680, + "ce": 3.2504, + "ppl": 25.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.3715, + "loop_stats": [ + { + "tau": 0.089, + "v": 0.8371, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0253, + "v": 0.8073, + "H_a": 0.075, + "d_bind": 0.03834 + }, + { + "tau": 0.0225, + "v": 0.795, + "H_a": 0.057, + "d_bind": 0.02119 + } + ] + }, + { + "step": 21600, + "tok": 88473600, + "ce": 3.2629, + "ppl": 26.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22496, + "peak_MiB": 5301, + "v_gain": 0.3731, + "loop_stats": [ + { + "tau": 0.0942, + "v": 0.8569, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0464, + "v": 0.8321, + "H_a": 0.076, + "d_bind": 0.03618 + }, + { + "tau": 0.0332, + "v": 0.8189, + "H_a": 0.06, + "d_bind": 0.02374 + } + ] + }, + { + "step": 21620, + "tok": 88555520, + "ce": 2.9452, + "ppl": 19.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3738, + "loop_stats": [ + { + "tau": 0.1192, + "v": 0.8203, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.7889, + "H_a": 0.074, + "d_bind": 0.03869 + }, + { + "tau": 0.0385, + "v": 0.7768, + "H_a": 0.064, + "d_bind": 0.02229 + } + ] + }, + { + "step": 21640, + "tok": 88637440, + "ce": 2.923, + "ppl": 18.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3753, + "loop_stats": [ + { + "tau": 0.1175, + "v": 0.8247, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8041, + "H_a": 0.079, + "d_bind": 0.04025 + }, + { + "tau": 0.0317, + "v": 0.7966, + "H_a": 0.064, + "d_bind": 0.0228 + } + ] + }, + { + "step": 21660, + "tok": 88719360, + "ce": 2.6374, + "ppl": 14.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.947, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3753, + "loop_stats": [ + { + "tau": 0.1218, + "v": 0.7748, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0473, + "v": 0.7534, + "H_a": 0.078, + "d_bind": 0.0395 + }, + { + "tau": 0.0409, + "v": 0.7526, + "H_a": 0.063, + "d_bind": 0.02281 + } + ] + }, + { + "step": 21680, + "tok": 88801280, + "ce": 3.3525, + "ppl": 28.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.938, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.3754, + "loop_stats": [ + { + "tau": 0.0831, + "v": 0.8781, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0473, + "v": 0.8427, + "H_a": 0.076, + "d_bind": 0.03687 + }, + { + "tau": 0.0325, + "v": 0.8226, + "H_a": 0.061, + "d_bind": 0.02146 + } + ] + }, + { + "step": 21700, + "tok": 88883200, + "ce": 2.7018, + "ppl": 14.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3769, + "loop_stats": [ + { + "tau": 0.1279, + "v": 0.829, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0538, + "v": 0.7998, + "H_a": 0.065, + "d_bind": 0.03854 + }, + { + "tau": 0.0461, + "v": 0.7799, + "H_a": 0.054, + "d_bind": 0.02017 + } + ] + }, + { + "step": 21720, + "tok": 88965120, + "ce": 2.9852, + "ppl": 19.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22497, + "peak_MiB": 5301, + "v_gain": 0.3793, + "loop_stats": [ + { + "tau": 0.0998, + "v": 0.8354, + "H_a": 0.103, + "d_bind": 0.0 + }, + { + "tau": 0.0427, + "v": 0.8109, + "H_a": 0.076, + "d_bind": 0.04056 + }, + { + "tau": 0.0314, + "v": 0.8053, + "H_a": 0.06, + "d_bind": 0.02232 + } + ] + }, + { + "step": 21740, + "tok": 89047040, + "ce": 2.8434, + "ppl": 17.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3812, + "loop_stats": [ + { + "tau": 0.1038, + "v": 0.8434, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.819, + "H_a": 0.072, + "d_bind": 0.03848 + }, + { + "tau": 0.0239, + "v": 0.811, + "H_a": 0.06, + "d_bind": 0.0228 + } + ] + }, + { + "step": 21760, + "tok": 89128960, + "ce": 2.6196, + "ppl": 13.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.38, + "loop_stats": [ + { + "tau": 0.1385, + "v": 0.8334, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.8242, + "H_a": 0.072, + "d_bind": 0.04355 + }, + { + "tau": 0.0461, + "v": 0.8192, + "H_a": 0.061, + "d_bind": 0.02245 + } + ] + }, + { + "step": 21780, + "tok": 89210880, + "ce": 2.9338, + "ppl": 18.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.3817, + "loop_stats": [ + { + "tau": 0.0958, + "v": 0.8512, + "H_a": 0.101, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.8207, + "H_a": 0.074, + "d_bind": 0.04437 + }, + { + "tau": 0.0334, + "v": 0.8078, + "H_a": 0.059, + "d_bind": 0.02091 + } + ] + }, + { + "step": 21800, + "tok": 89292800, + "ce": 3.1631, + "ppl": 23.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.381, + "loop_stats": [ + { + "tau": 0.0922, + "v": 0.8861, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.8658, + "H_a": 0.075, + "d_bind": 0.03764 + }, + { + "tau": 0.0256, + "v": 0.8547, + "H_a": 0.06, + "d_bind": 0.02247 + } + ] + }, + { + "step": 21820, + "tok": 89374720, + "ce": 3.1018, + "ppl": 22.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3795, + "loop_stats": [ + { + "tau": 0.0851, + "v": 0.8339, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.8136, + "H_a": 0.078, + "d_bind": 0.03927 + }, + { + "tau": 0.0229, + "v": 0.8007, + "H_a": 0.07, + "d_bind": 0.02442 + } + ] + }, + { + "step": 21840, + "tok": 89456640, + "ce": 3.1465, + "ppl": 23.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3809, + "loop_stats": [ + { + "tau": 0.0957, + "v": 0.8793, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.843, + "H_a": 0.076, + "d_bind": 0.03795 + }, + { + "tau": 0.0209, + "v": 0.8234, + "H_a": 0.061, + "d_bind": 0.02164 + } + ] + }, + { + "step": 21860, + "tok": 89538560, + "ce": 2.6944, + "ppl": 14.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3805, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.8066, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.778, + "H_a": 0.062, + "d_bind": 0.04181 + }, + { + "tau": 0.038, + "v": 0.7703, + "H_a": 0.056, + "d_bind": 0.02038 + } + ] + }, + { + "step": 21880, + "tok": 89620480, + "ce": 3.0724, + "ppl": 21.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.384, + "loop_stats": [ + { + "tau": 0.0799, + "v": 0.8341, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.021, + "v": 0.8015, + "H_a": 0.072, + "d_bind": 0.04045 + }, + { + "tau": 0.0147, + "v": 0.7857, + "H_a": 0.056, + "d_bind": 0.02224 + } + ] + }, + { + "step": 21900, + "tok": 89702400, + "ce": 2.7529, + "ppl": 15.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.3857, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.8454, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0472, + "v": 0.8179, + "H_a": 0.069, + "d_bind": 0.03992 + }, + { + "tau": 0.029, + "v": 0.8088, + "H_a": 0.054, + "d_bind": 0.0214 + } + ] + }, + { + "step": 21920, + "tok": 89784320, + "ce": 2.59, + "ppl": 13.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.935, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3887, + "loop_stats": [ + { + "tau": 0.1446, + "v": 0.8498, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8251, + "H_a": 0.067, + "d_bind": 0.04179 + }, + { + "tau": 0.0188, + "v": 0.8111, + "H_a": 0.06, + "d_bind": 0.02002 + } + ] + }, + { + "step": 21940, + "tok": 89866240, + "ce": 2.5215, + "ppl": 12.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.3891, + "loop_stats": [ + { + "tau": 0.1335, + "v": 0.8109, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0474, + "v": 0.7823, + "H_a": 0.067, + "d_bind": 0.04273 + }, + { + "tau": 0.0349, + "v": 0.7683, + "H_a": 0.055, + "d_bind": 0.02011 + } + ] + }, + { + "step": 21960, + "tok": 89948160, + "ce": 2.7659, + "ppl": 15.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22498, + "peak_MiB": 5301, + "v_gain": 0.3893, + "loop_stats": [ + { + "tau": 0.1404, + "v": 0.7997, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.051, + "v": 0.7668, + "H_a": 0.077, + "d_bind": 0.04335 + }, + { + "tau": 0.0361, + "v": 0.7563, + "H_a": 0.06, + "d_bind": 0.02076 + } + ] + }, + { + "step": 21980, + "tok": 90030080, + "ce": 2.6295, + "ppl": 13.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.3879, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.8253, + "H_a": 0.093, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.798, + "H_a": 0.072, + "d_bind": 0.04459 + }, + { + "tau": 0.0312, + "v": 0.7842, + "H_a": 0.053, + "d_bind": 0.02124 + } + ] + }, + { + "step": 22000, + "tok": 90112000, + "ce": 2.9737, + "ppl": 19.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.3887, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.8174, + "H_a": 0.098, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.7905, + "H_a": 0.071, + "d_bind": 0.04166 + }, + { + "tau": 0.0336, + "v": 0.7809, + "H_a": 0.061, + "d_bind": 0.02229 + } + ] + }, + { + "step": 22020, + "tok": 90193920, + "ce": 2.8772, + "ppl": 17.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3869, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.794, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.056, + "v": 0.7638, + "H_a": 0.071, + "d_bind": 0.0426 + }, + { + "tau": 0.0397, + "v": 0.7555, + "H_a": 0.06, + "d_bind": 0.02211 + } + ] + }, + { + "step": 22040, + "tok": 90275840, + "ce": 3.1781, + "ppl": 24.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.96, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3859, + "loop_stats": [ + { + "tau": 0.079, + "v": 0.8228, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.7804, + "H_a": 0.065, + "d_bind": 0.04041 + }, + { + "tau": 0.0205, + "v": 0.7636, + "H_a": 0.052, + "d_bind": 0.02019 + } + ] + }, + { + "step": 22060, + "tok": 90357760, + "ce": 2.6488, + "ppl": 14.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.962, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3855, + "loop_stats": [ + { + "tau": 0.12, + "v": 0.8669, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0598, + "v": 0.8372, + "H_a": 0.073, + "d_bind": 0.0421 + }, + { + "tau": 0.0308, + "v": 0.8212, + "H_a": 0.056, + "d_bind": 0.02335 + } + ] + }, + { + "step": 22080, + "tok": 90439680, + "ce": 2.8475, + "ppl": 17.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22499, + "peak_MiB": 5301, + "v_gain": 0.3854, + "loop_stats": [ + { + "tau": 0.1191, + "v": 0.862, + "H_a": 0.097, + "d_bind": 0.0 + }, + { + "tau": 0.0368, + "v": 0.8504, + "H_a": 0.072, + "d_bind": 0.0427 + }, + { + "tau": 0.0282, + "v": 0.8436, + "H_a": 0.059, + "d_bind": 0.02148 + } + ] + }, + { + "step": 22100, + "tok": 90521600, + "ce": 3.3768, + "ppl": 29.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.3873, + "loop_stats": [ + { + "tau": 0.0864, + "v": 0.8391, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0319, + "v": 0.8067, + "H_a": 0.072, + "d_bind": 0.0381 + }, + { + "tau": 0.0217, + "v": 0.7977, + "H_a": 0.058, + "d_bind": 0.02104 + } + ] + }, + { + "step": 22120, + "tok": 90603520, + "ce": 2.9111, + "ppl": 18.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3903, + "loop_stats": [ + { + "tau": 0.1247, + "v": 0.8176, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0473, + "v": 0.7938, + "H_a": 0.074, + "d_bind": 0.04234 + }, + { + "tau": 0.0349, + "v": 0.7914, + "H_a": 0.062, + "d_bind": 0.0226 + } + ] + }, + { + "step": 22140, + "tok": 90685440, + "ce": 2.9866, + "ppl": 19.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.957, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3911, + "loop_stats": [ + { + "tau": 0.1099, + "v": 0.8419, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0406, + "v": 0.82, + "H_a": 0.064, + "d_bind": 0.03709 + }, + { + "tau": 0.02, + "v": 0.8105, + "H_a": 0.061, + "d_bind": 0.02035 + } + ] + }, + { + "step": 22160, + "tok": 90767360, + "ce": 2.7018, + "ppl": 14.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3912, + "loop_stats": [ + { + "tau": 0.1227, + "v": 0.8236, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0506, + "v": 0.8074, + "H_a": 0.065, + "d_bind": 0.04159 + }, + { + "tau": 0.0378, + "v": 0.8129, + "H_a": 0.055, + "d_bind": 0.02145 + } + ] + }, + { + "step": 22180, + "tok": 90849280, + "ce": 3.3527, + "ppl": 28.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3932, + "loop_stats": [ + { + "tau": 0.0848, + "v": 0.8389, + "H_a": 0.1, + "d_bind": 0.0 + }, + { + "tau": 0.0284, + "v": 0.8147, + "H_a": 0.074, + "d_bind": 0.04045 + }, + { + "tau": 0.0181, + "v": 0.8071, + "H_a": 0.06, + "d_bind": 0.021 + } + ] + }, + { + "step": 22200, + "tok": 90931200, + "ce": 2.8431, + "ppl": 17.2, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.942, + "tok_s": 22500, + "peak_MiB": 5301, + "v_gain": 0.3935, + "loop_stats": [ + { + "tau": 0.1324, + "v": 0.8408, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0558, + "v": 0.8139, + "H_a": 0.065, + "d_bind": 0.04544 + }, + { + "tau": 0.0371, + "v": 0.8056, + "H_a": 0.051, + "d_bind": 0.01846 + } + ] + }, + { + "step": 22220, + "tok": 91013120, + "ce": 2.802, + "ppl": 16.5, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.945, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3965, + "loop_stats": [ + { + "tau": 0.1263, + "v": 0.8792, + "H_a": 0.096, + "d_bind": 0.0 + }, + { + "tau": 0.0376, + "v": 0.8473, + "H_a": 0.064, + "d_bind": 0.04175 + }, + { + "tau": 0.027, + "v": 0.8342, + "H_a": 0.051, + "d_bind": 0.0204 + } + ] + }, + { + "step": 22240, + "tok": 91095040, + "ce": 3.0302, + "ppl": 20.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3956, + "loop_stats": [ + { + "tau": 0.096, + "v": 0.8206, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.7934, + "H_a": 0.076, + "d_bind": 0.0409 + }, + { + "tau": 0.023, + "v": 0.7872, + "H_a": 0.057, + "d_bind": 0.02199 + } + ] + }, + { + "step": 22260, + "tok": 91176960, + "ce": 3.118, + "ppl": 22.6, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.952, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.3954, + "loop_stats": [ + { + "tau": 0.1039, + "v": 0.8539, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8155, + "H_a": 0.074, + "d_bind": 0.03873 + }, + { + "tau": 0.0248, + "v": 0.7988, + "H_a": 0.057, + "d_bind": 0.02179 + } + ] + }, + { + "step": 22280, + "tok": 91258880, + "ce": 2.7527, + "ppl": 15.7, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.95, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.3946, + "loop_stats": [ + { + "tau": 0.11, + "v": 0.8334, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.7964, + "H_a": 0.063, + "d_bind": 0.04147 + }, + { + "tau": 0.0248, + "v": 0.7808, + "H_a": 0.053, + "d_bind": 0.02125 + } + ] + }, + { + "step": 22300, + "tok": 91340800, + "ce": 2.7437, + "ppl": 15.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3942, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.7892, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.7632, + "H_a": 0.078, + "d_bind": 0.04471 + }, + { + "tau": 0.0357, + "v": 0.7536, + "H_a": 0.06, + "d_bind": 0.02114 + } + ] + }, + { + "step": 22320, + "tok": 91422720, + "ce": 2.9993, + "ppl": 20.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3918, + "loop_stats": [ + { + "tau": 0.1052, + "v": 0.8671, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0487, + "v": 0.8306, + "H_a": 0.064, + "d_bind": 0.04119 + }, + { + "tau": 0.0266, + "v": 0.8133, + "H_a": 0.053, + "d_bind": 0.02 + } + ] + }, + { + "step": 22340, + "tok": 91504640, + "ce": 3.0085, + "ppl": 20.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3929, + "loop_stats": [ + { + "tau": 0.0957, + "v": 0.823, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0312, + "v": 0.8004, + "H_a": 0.07, + "d_bind": 0.04058 + }, + { + "tau": 0.0226, + "v": 0.7898, + "H_a": 0.058, + "d_bind": 0.02139 + } + ] + }, + { + "step": 22360, + "tok": 91586560, + "ce": 2.8267, + "ppl": 16.9, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.947, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.393, + "loop_stats": [ + { + "tau": 0.1109, + "v": 0.7967, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.779, + "H_a": 0.071, + "d_bind": 0.04215 + }, + { + "tau": 0.0416, + "v": 0.7735, + "H_a": 0.054, + "d_bind": 0.02094 + } + ] + }, + { + "step": 22380, + "tok": 91668480, + "ce": 3.0138, + "ppl": 20.4, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.952, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3905, + "loop_stats": [ + { + "tau": 0.0995, + "v": 0.8551, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0408, + "v": 0.833, + "H_a": 0.069, + "d_bind": 0.03952 + }, + { + "tau": 0.0342, + "v": 0.8243, + "H_a": 0.054, + "d_bind": 0.02263 + } + ] + }, + { + "step": 22400, + "tok": 91750400, + "ce": 2.9087, + "ppl": 18.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.96, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3885, + "loop_stats": [ + { + "tau": 0.1022, + "v": 0.8718, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.8436, + "H_a": 0.066, + "d_bind": 0.03792 + }, + { + "tau": 0.0308, + "v": 0.8347, + "H_a": 0.058, + "d_bind": 0.02131 + } + ] + }, + { + "step": 22420, + "tok": 91832320, + "ce": 2.9302, + "ppl": 18.7, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.957, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.3901, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8215, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.7852, + "H_a": 0.07, + "d_bind": 0.04337 + }, + { + "tau": 0.0376, + "v": 0.7803, + "H_a": 0.053, + "d_bind": 0.02074 + } + ] + }, + { + "step": 22440, + "tok": 91914240, + "ce": 3.1301, + "ppl": 22.9, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.945, + "tok_s": 22501, + "peak_MiB": 5301, + "v_gain": 0.3892, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8614, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0455, + "v": 0.8364, + "H_a": 0.066, + "d_bind": 0.03879 + }, + { + "tau": 0.0254, + "v": 0.8276, + "H_a": 0.058, + "d_bind": 0.02008 + } + ] + }, + { + "step": 22460, + "tok": 91996160, + "ce": 2.8339, + "ppl": 17.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.389, + "loop_stats": [ + { + "tau": 0.1014, + "v": 0.8403, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8154, + "H_a": 0.062, + "d_bind": 0.04073 + }, + { + "tau": 0.0352, + "v": 0.8019, + "H_a": 0.053, + "d_bind": 0.02061 + } + ] + }, + { + "step": 22480, + "tok": 92078080, + "ce": 2.8146, + "ppl": 16.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3898, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.875, + "H_a": 0.092, + "d_bind": 0.0 + }, + { + "tau": 0.0231, + "v": 0.8536, + "H_a": 0.068, + "d_bind": 0.03816 + }, + { + "tau": 0.0161, + "v": 0.846, + "H_a": 0.053, + "d_bind": 0.02111 + } + ] + }, + { + "step": 22500, + "tok": 92160000, + "ce": 2.7171, + "ppl": 15.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3884, + "loop_stats": [ + { + "tau": 0.1252, + "v": 0.81, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0435, + "v": 0.7761, + "H_a": 0.068, + "d_bind": 0.04301 + }, + { + "tau": 0.0356, + "v": 0.7698, + "H_a": 0.058, + "d_bind": 0.01978 + } + ] + }, + { + "step": 22520, + "tok": 92241920, + "ce": 2.7496, + "ppl": 15.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3877, + "loop_stats": [ + { + "tau": 0.0988, + "v": 0.861, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0329, + "v": 0.8205, + "H_a": 0.059, + "d_bind": 0.04054 + }, + { + "tau": 0.0235, + "v": 0.8104, + "H_a": 0.054, + "d_bind": 0.01924 + } + ] + }, + { + "step": 22540, + "tok": 92323840, + "ce": 3.3377, + "ppl": 28.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3871, + "loop_stats": [ + { + "tau": 0.0824, + "v": 0.854, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0284, + "v": 0.8253, + "H_a": 0.07, + "d_bind": 0.03951 + }, + { + "tau": 0.0214, + "v": 0.8187, + "H_a": 0.058, + "d_bind": 0.02187 + } + ] + }, + { + "step": 22560, + "tok": 92405760, + "ce": 3.1114, + "ppl": 22.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22502, + "peak_MiB": 5301, + "v_gain": 0.3842, + "loop_stats": [ + { + "tau": 0.1007, + "v": 0.8698, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8251, + "H_a": 0.072, + "d_bind": 0.0392 + }, + { + "tau": 0.039, + "v": 0.8155, + "H_a": 0.059, + "d_bind": 0.02177 + } + ] + }, + { + "step": 22580, + "tok": 92487680, + "ce": 3.1246, + "ppl": 22.8, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.965, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3816, + "loop_stats": [ + { + "tau": 0.1113, + "v": 0.8565, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0492, + "v": 0.8261, + "H_a": 0.07, + "d_bind": 0.04227 + }, + { + "tau": 0.0396, + "v": 0.8195, + "H_a": 0.059, + "d_bind": 0.02086 + } + ] + }, + { + "step": 22600, + "tok": 92569600, + "ce": 3.117, + "ppl": 22.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3809, + "loop_stats": [ + { + "tau": 0.0967, + "v": 0.8678, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.831, + "H_a": 0.062, + "d_bind": 0.04317 + }, + { + "tau": 0.028, + "v": 0.8247, + "H_a": 0.05, + "d_bind": 0.01814 + } + ] + }, + { + "step": 22620, + "tok": 92651520, + "ce": 3.2818, + "ppl": 26.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3809, + "loop_stats": [ + { + "tau": 0.0868, + "v": 0.9014, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0369, + "v": 0.8671, + "H_a": 0.067, + "d_bind": 0.03312 + }, + { + "tau": 0.0171, + "v": 0.8501, + "H_a": 0.05, + "d_bind": 0.02064 + } + ] + }, + { + "step": 22640, + "tok": 92733440, + "ce": 2.4171, + "ppl": 11.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3817, + "loop_stats": [ + { + "tau": 0.1272, + "v": 0.8071, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0535, + "v": 0.7521, + "H_a": 0.07, + "d_bind": 0.04698 + }, + { + "tau": 0.0459, + "v": 0.7483, + "H_a": 0.054, + "d_bind": 0.02075 + } + ] + }, + { + "step": 22660, + "tok": 92815360, + "ce": 2.9356, + "ppl": 18.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.965, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3837, + "loop_stats": [ + { + "tau": 0.1081, + "v": 0.8436, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8005, + "H_a": 0.068, + "d_bind": 0.04035 + }, + { + "tau": 0.0296, + "v": 0.7923, + "H_a": 0.054, + "d_bind": 0.02036 + } + ] + }, + { + "step": 22680, + "tok": 92897280, + "ce": 2.6141, + "ppl": 13.7, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.957, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3854, + "loop_stats": [ + { + "tau": 0.1332, + "v": 0.9006, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0553, + "v": 0.8458, + "H_a": 0.056, + "d_bind": 0.04267 + }, + { + "tau": 0.0432, + "v": 0.8304, + "H_a": 0.048, + "d_bind": 0.01834 + } + ] + }, + { + "step": 22700, + "tok": 92979200, + "ce": 3.0332, + "ppl": 20.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.945, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3857, + "loop_stats": [ + { + "tau": 0.0964, + "v": 0.8761, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.828, + "H_a": 0.066, + "d_bind": 0.04134 + }, + { + "tau": 0.0264, + "v": 0.8085, + "H_a": 0.052, + "d_bind": 0.01983 + } + ] + }, + { + "step": 22720, + "tok": 93061120, + "ce": 2.7128, + "ppl": 15.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.957, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3888, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.866, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0462, + "v": 0.8242, + "H_a": 0.064, + "d_bind": 0.04102 + }, + { + "tau": 0.0299, + "v": 0.8123, + "H_a": 0.052, + "d_bind": 0.0189 + } + ] + }, + { + "step": 22740, + "tok": 93143040, + "ce": 2.6483, + "ppl": 14.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.3901, + "loop_stats": [ + { + "tau": 0.136, + "v": 0.8815, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0494, + "v": 0.8248, + "H_a": 0.063, + "d_bind": 0.04219 + }, + { + "tau": 0.0369, + "v": 0.8139, + "H_a": 0.056, + "d_bind": 0.01929 + } + ] + }, + { + "step": 22760, + "tok": 93224960, + "ce": 3.3009, + "ppl": 27.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3915, + "loop_stats": [ + { + "tau": 0.0919, + "v": 0.8299, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.7885, + "H_a": 0.076, + "d_bind": 0.03912 + }, + { + "tau": 0.0201, + "v": 0.7852, + "H_a": 0.061, + "d_bind": 0.02201 + } + ] + }, + { + "step": 22780, + "tok": 93306880, + "ce": 2.8137, + "ppl": 16.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.96, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3911, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.8502, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.054, + "v": 0.8261, + "H_a": 0.065, + "d_bind": 0.04282 + }, + { + "tau": 0.0424, + "v": 0.8226, + "H_a": 0.056, + "d_bind": 0.02071 + } + ] + }, + { + "step": 22800, + "tok": 93388800, + "ce": 2.5328, + "ppl": 12.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.957, + "tok_s": 22503, + "peak_MiB": 5301, + "v_gain": 0.3897, + "loop_stats": [ + { + "tau": 0.1189, + "v": 0.8402, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0551, + "v": 0.8137, + "H_a": 0.071, + "d_bind": 0.04325 + }, + { + "tau": 0.0427, + "v": 0.8072, + "H_a": 0.057, + "d_bind": 0.02015 + } + ] + }, + { + "step": 22820, + "tok": 93470720, + "ce": 3.3521, + "ppl": 28.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3915, + "loop_stats": [ + { + "tau": 0.0877, + "v": 0.8412, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8137, + "H_a": 0.061, + "d_bind": 0.03504 + }, + { + "tau": 0.0297, + "v": 0.8053, + "H_a": 0.051, + "d_bind": 0.01893 + } + ] + }, + { + "step": 22840, + "tok": 93552640, + "ce": 3.1533, + "ppl": 23.4, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.3894, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.8506, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.8189, + "H_a": 0.065, + "d_bind": 0.03757 + }, + { + "tau": 0.0337, + "v": 0.8066, + "H_a": 0.057, + "d_bind": 0.02006 + } + ] + }, + { + "step": 22860, + "tok": 93634560, + "ce": 2.8191, + "ppl": 16.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3844, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.8627, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8295, + "H_a": 0.06, + "d_bind": 0.03954 + }, + { + "tau": 0.0388, + "v": 0.818, + "H_a": 0.054, + "d_bind": 0.02003 + } + ] + }, + { + "step": 22880, + "tok": 93716480, + "ce": 2.8578, + "ppl": 17.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3851, + "loop_stats": [ + { + "tau": 0.1354, + "v": 0.8883, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8439, + "H_a": 0.064, + "d_bind": 0.04127 + }, + { + "tau": 0.0427, + "v": 0.8307, + "H_a": 0.05, + "d_bind": 0.01927 + } + ] + }, + { + "step": 22900, + "tok": 93798400, + "ce": 2.5962, + "ppl": 13.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3836, + "loop_stats": [ + { + "tau": 0.1252, + "v": 0.8188, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0572, + "v": 0.7889, + "H_a": 0.068, + "d_bind": 0.03902 + }, + { + "tau": 0.0492, + "v": 0.7859, + "H_a": 0.052, + "d_bind": 0.01846 + } + ] + }, + { + "step": 22920, + "tok": 93880320, + "ce": 2.9749, + "ppl": 19.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22504, + "peak_MiB": 5301, + "v_gain": 0.3868, + "loop_stats": [ + { + "tau": 0.1135, + "v": 0.8772, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0408, + "v": 0.8474, + "H_a": 0.065, + "d_bind": 0.03718 + }, + { + "tau": 0.0325, + "v": 0.8373, + "H_a": 0.053, + "d_bind": 0.01993 + } + ] + }, + { + "step": 22940, + "tok": 93962240, + "ce": 2.6346, + "ppl": 13.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.3873, + "loop_stats": [ + { + "tau": 0.1271, + "v": 0.827, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0484, + "v": 0.7806, + "H_a": 0.067, + "d_bind": 0.03978 + }, + { + "tau": 0.0447, + "v": 0.7657, + "H_a": 0.049, + "d_bind": 0.01996 + } + ] + }, + { + "step": 22960, + "tok": 94044160, + "ce": 2.8092, + "ppl": 16.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.942, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3857, + "loop_stats": [ + { + "tau": 0.1121, + "v": 0.8479, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0516, + "v": 0.8038, + "H_a": 0.059, + "d_bind": 0.03558 + }, + { + "tau": 0.0456, + "v": 0.7871, + "H_a": 0.05, + "d_bind": 0.01839 + } + ] + }, + { + "step": 22980, + "tok": 94126080, + "ce": 2.9431, + "ppl": 19.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.938, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3859, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8377, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0469, + "v": 0.8018, + "H_a": 0.06, + "d_bind": 0.03553 + }, + { + "tau": 0.0426, + "v": 0.7859, + "H_a": 0.05, + "d_bind": 0.01937 + } + ] + }, + { + "step": 23000, + "tok": 94208000, + "ce": 2.7877, + "ppl": 16.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3851, + "loop_stats": [ + { + "tau": 0.1217, + "v": 0.8361, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0495, + "v": 0.7857, + "H_a": 0.062, + "d_bind": 0.03883 + }, + { + "tau": 0.0453, + "v": 0.7642, + "H_a": 0.049, + "d_bind": 0.01968 + } + ] + }, + { + "step": 23020, + "tok": 94289920, + "ce": 3.0826, + "ppl": 21.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.927, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3851, + "loop_stats": [ + { + "tau": 0.0961, + "v": 0.8552, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0472, + "v": 0.8164, + "H_a": 0.062, + "d_bind": 0.0341 + }, + { + "tau": 0.0355, + "v": 0.8066, + "H_a": 0.049, + "d_bind": 0.01885 + } + ] + }, + { + "step": 23040, + "tok": 94371840, + "ce": 2.7737, + "ppl": 16.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.94, + "tok_s": 22505, + "peak_MiB": 5301, + "v_gain": 0.386, + "loop_stats": [ + { + "tau": 0.0923, + "v": 0.825, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.784, + "H_a": 0.061, + "d_bind": 0.03625 + }, + { + "tau": 0.0317, + "v": 0.7691, + "H_a": 0.05, + "d_bind": 0.01986 + } + ] + }, + { + "step": 23060, + "tok": 94453760, + "ce": 2.8485, + "ppl": 17.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3875, + "loop_stats": [ + { + "tau": 0.0937, + "v": 0.8238, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.7989, + "H_a": 0.064, + "d_bind": 0.03699 + }, + { + "tau": 0.0296, + "v": 0.7909, + "H_a": 0.051, + "d_bind": 0.01901 + } + ] + }, + { + "step": 23080, + "tok": 94535680, + "ce": 2.6004, + "ppl": 13.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.933, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.384, + "loop_stats": [ + { + "tau": 0.1384, + "v": 0.8497, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0595, + "v": 0.8224, + "H_a": 0.063, + "d_bind": 0.03942 + }, + { + "tau": 0.0509, + "v": 0.8133, + "H_a": 0.05, + "d_bind": 0.01864 + } + ] + }, + { + "step": 23100, + "tok": 94617600, + "ce": 2.7075, + "ppl": 15.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3836, + "loop_stats": [ + { + "tau": 0.1476, + "v": 0.8312, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0657, + "v": 0.8021, + "H_a": 0.059, + "d_bind": 0.03774 + }, + { + "tau": 0.0498, + "v": 0.7934, + "H_a": 0.052, + "d_bind": 0.01819 + } + ] + }, + { + "step": 23120, + "tok": 94699520, + "ce": 3.192, + "ppl": 24.3, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.945, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.3845, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.8802, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8331, + "H_a": 0.061, + "d_bind": 0.03699 + }, + { + "tau": 0.0222, + "v": 0.8142, + "H_a": 0.053, + "d_bind": 0.02013 + } + ] + }, + { + "step": 23140, + "tok": 94781440, + "ce": 3.1265, + "ppl": 22.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3828, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.8466, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.042, + "v": 0.7949, + "H_a": 0.066, + "d_bind": 0.0361 + }, + { + "tau": 0.0338, + "v": 0.7771, + "H_a": 0.05, + "d_bind": 0.01822 + } + ] + }, + { + "step": 23160, + "tok": 94863360, + "ce": 2.9829, + "ppl": 19.7, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.952, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3842, + "loop_stats": [ + { + "tau": 0.1096, + "v": 0.8794, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8378, + "H_a": 0.057, + "d_bind": 0.03529 + }, + { + "tau": 0.0273, + "v": 0.8211, + "H_a": 0.047, + "d_bind": 0.01823 + } + ] + }, + { + "step": 23180, + "tok": 94945280, + "ce": 2.9974, + "ppl": 20.0, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.942, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3877, + "loop_stats": [ + { + "tau": 0.0859, + "v": 0.8661, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0219, + "v": 0.8419, + "H_a": 0.065, + "d_bind": 0.03578 + }, + { + "tau": 0.0127, + "v": 0.8353, + "H_a": 0.053, + "d_bind": 0.02105 + } + ] + }, + { + "step": 23200, + "tok": 95027200, + "ce": 2.7049, + "ppl": 15.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.39, + "loop_stats": [ + { + "tau": 0.1288, + "v": 0.8557, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0476, + "v": 0.8234, + "H_a": 0.062, + "d_bind": 0.0375 + }, + { + "tau": 0.0329, + "v": 0.8091, + "H_a": 0.047, + "d_bind": 0.01857 + } + ] + }, + { + "step": 23220, + "tok": 95109120, + "ce": 2.8504, + "ppl": 17.3, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.947, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3895, + "loop_stats": [ + { + "tau": 0.1226, + "v": 0.8516, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8183, + "H_a": 0.06, + "d_bind": 0.0347 + }, + { + "tau": 0.0328, + "v": 0.8087, + "H_a": 0.04, + "d_bind": 0.01824 + } + ] + }, + { + "step": 23240, + "tok": 95191040, + "ce": 2.9148, + "ppl": 18.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.3884, + "loop_stats": [ + { + "tau": 0.1205, + "v": 0.8813, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8575, + "H_a": 0.065, + "d_bind": 0.03572 + }, + { + "tau": 0.0363, + "v": 0.8509, + "H_a": 0.048, + "d_bind": 0.0179 + } + ] + }, + { + "step": 23260, + "tok": 95272960, + "ce": 2.8709, + "ppl": 17.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.3897, + "loop_stats": [ + { + "tau": 0.1153, + "v": 0.858, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0494, + "v": 0.8162, + "H_a": 0.066, + "d_bind": 0.03637 + }, + { + "tau": 0.0317, + "v": 0.8054, + "H_a": 0.052, + "d_bind": 0.01898 + } + ] + }, + { + "step": 23280, + "tok": 95354880, + "ce": 2.5578, + "ppl": 12.9, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.96, + "tok_s": 22506, + "peak_MiB": 5301, + "v_gain": 0.3877, + "loop_stats": [ + { + "tau": 0.1379, + "v": 0.841, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0535, + "v": 0.8034, + "H_a": 0.061, + "d_bind": 0.03837 + }, + { + "tau": 0.0444, + "v": 0.7918, + "H_a": 0.052, + "d_bind": 0.02003 + } + ] + }, + { + "step": 23300, + "tok": 95436800, + "ce": 2.7747, + "ppl": 16.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.96, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.387, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.866, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.8319, + "H_a": 0.063, + "d_bind": 0.03562 + }, + { + "tau": 0.0351, + "v": 0.8246, + "H_a": 0.05, + "d_bind": 0.02158 + } + ] + }, + { + "step": 23320, + "tok": 95518720, + "ce": 3.2539, + "ppl": 25.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.962, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3911, + "loop_stats": [ + { + "tau": 0.0737, + "v": 0.8903, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0286, + "v": 0.8613, + "H_a": 0.067, + "d_bind": 0.03413 + }, + { + "tau": 0.0161, + "v": 0.847, + "H_a": 0.054, + "d_bind": 0.02003 + } + ] + }, + { + "step": 23340, + "tok": 95600640, + "ce": 3.435, + "ppl": 31.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.967, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3917, + "loop_stats": [ + { + "tau": 0.0864, + "v": 0.8889, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0211, + "v": 0.8569, + "H_a": 0.065, + "d_bind": 0.03433 + }, + { + "tau": 0.0137, + "v": 0.845, + "H_a": 0.051, + "d_bind": 0.02037 + } + ] + }, + { + "step": 23360, + "tok": 95682560, + "ce": 2.9594, + "ppl": 19.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.962, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.3877, + "loop_stats": [ + { + "tau": 0.1086, + "v": 0.8758, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8359, + "H_a": 0.065, + "d_bind": 0.03916 + }, + { + "tau": 0.0337, + "v": 0.8222, + "H_a": 0.052, + "d_bind": 0.01993 + } + ] + }, + { + "step": 23380, + "tok": 95764480, + "ce": 2.834, + "ppl": 17.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.3871, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.8523, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.8195, + "H_a": 0.062, + "d_bind": 0.03547 + }, + { + "tau": 0.0402, + "v": 0.8142, + "H_a": 0.048, + "d_bind": 0.01969 + } + ] + }, + { + "step": 23400, + "tok": 95846400, + "ce": 2.8995, + "ppl": 18.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3865, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.8623, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8477, + "H_a": 0.067, + "d_bind": 0.03462 + }, + { + "tau": 0.0334, + "v": 0.8437, + "H_a": 0.054, + "d_bind": 0.02172 + } + ] + }, + { + "step": 23420, + "tok": 95928320, + "ce": 2.45, + "ppl": 11.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3859, + "loop_stats": [ + { + "tau": 0.1323, + "v": 0.8719, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0612, + "v": 0.827, + "H_a": 0.062, + "d_bind": 0.03873 + }, + { + "tau": 0.0514, + "v": 0.8183, + "H_a": 0.054, + "d_bind": 0.02033 + } + ] + }, + { + "step": 23440, + "tok": 96010240, + "ce": 2.9636, + "ppl": 19.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3855, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.8609, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0343, + "v": 0.8254, + "H_a": 0.067, + "d_bind": 0.03631 + }, + { + "tau": 0.0244, + "v": 0.8186, + "H_a": 0.051, + "d_bind": 0.02016 + } + ] + }, + { + "step": 23460, + "tok": 96092160, + "ce": 2.947, + "ppl": 19.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3855, + "loop_stats": [ + { + "tau": 0.115, + "v": 0.8734, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0466, + "v": 0.8452, + "H_a": 0.066, + "d_bind": 0.03584 + }, + { + "tau": 0.0337, + "v": 0.8321, + "H_a": 0.051, + "d_bind": 0.0205 + } + ] + }, + { + "step": 23480, + "tok": 96174080, + "ce": 2.9883, + "ppl": 19.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.3899, + "loop_stats": [ + { + "tau": 0.0859, + "v": 0.8736, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8372, + "H_a": 0.063, + "d_bind": 0.03452 + }, + { + "tau": 0.0206, + "v": 0.8168, + "H_a": 0.051, + "d_bind": 0.01932 + } + ] + }, + { + "step": 23500, + "tok": 96256000, + "ce": 2.6217, + "ppl": 13.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.3905, + "loop_stats": [ + { + "tau": 0.1342, + "v": 0.8294, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.7986, + "H_a": 0.062, + "d_bind": 0.03747 + }, + { + "tau": 0.0443, + "v": 0.7933, + "H_a": 0.05, + "d_bind": 0.01887 + } + ] + }, + { + "step": 23520, + "tok": 96337920, + "ce": 2.9224, + "ppl": 18.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22507, + "peak_MiB": 5301, + "v_gain": 0.3893, + "loop_stats": [ + { + "tau": 0.1071, + "v": 0.8707, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8444, + "H_a": 0.069, + "d_bind": 0.03519 + }, + { + "tau": 0.0351, + "v": 0.8356, + "H_a": 0.054, + "d_bind": 0.01955 + } + ] + }, + { + "step": 23540, + "tok": 96419840, + "ce": 2.8382, + "ppl": 17.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.95, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.3919, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8617, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.055, + "v": 0.8484, + "H_a": 0.066, + "d_bind": 0.03901 + }, + { + "tau": 0.0475, + "v": 0.8401, + "H_a": 0.058, + "d_bind": 0.02147 + } + ] + }, + { + "step": 23560, + "tok": 96501760, + "ce": 2.7901, + "ppl": 16.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.3942, + "loop_stats": [ + { + "tau": 0.1139, + "v": 0.8523, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0451, + "v": 0.8226, + "H_a": 0.067, + "d_bind": 0.03677 + }, + { + "tau": 0.0341, + "v": 0.8082, + "H_a": 0.052, + "d_bind": 0.01969 + } + ] + }, + { + "step": 23580, + "tok": 96583680, + "ce": 2.7949, + "ppl": 16.4, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.965, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.3958, + "loop_stats": [ + { + "tau": 0.1174, + "v": 0.8624, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.8275, + "H_a": 0.072, + "d_bind": 0.03764 + }, + { + "tau": 0.0319, + "v": 0.8174, + "H_a": 0.059, + "d_bind": 0.02112 + } + ] + }, + { + "step": 23600, + "tok": 96665600, + "ce": 2.9851, + "ppl": 19.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.962, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.3989, + "loop_stats": [ + { + "tau": 0.1131, + "v": 0.8475, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.8192, + "H_a": 0.065, + "d_bind": 0.03969 + }, + { + "tau": 0.0385, + "v": 0.8166, + "H_a": 0.052, + "d_bind": 0.01896 + } + ] + }, + { + "step": 23620, + "tok": 96747520, + "ce": 3.11, + "ppl": 22.4, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.957, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.4049, + "loop_stats": [ + { + "tau": 0.1116, + "v": 0.8767, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0272, + "v": 0.8535, + "H_a": 0.067, + "d_bind": 0.03785 + }, + { + "tau": 0.0179, + "v": 0.8488, + "H_a": 0.053, + "d_bind": 0.02063 + } + ] + }, + { + "step": 23640, + "tok": 96829440, + "ce": 2.7773, + "ppl": 16.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22508, + "peak_MiB": 5301, + "v_gain": 0.4076, + "loop_stats": [ + { + "tau": 0.1142, + "v": 0.8718, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0229, + "v": 0.8445, + "H_a": 0.068, + "d_bind": 0.03967 + }, + { + "tau": 0.0113, + "v": 0.8383, + "H_a": 0.057, + "d_bind": 0.02083 + } + ] + }, + { + "step": 23660, + "tok": 96911360, + "ce": 2.7343, + "ppl": 15.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.407, + "loop_stats": [ + { + "tau": 0.1155, + "v": 0.8494, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.8298, + "H_a": 0.066, + "d_bind": 0.03685 + }, + { + "tau": 0.019, + "v": 0.8237, + "H_a": 0.053, + "d_bind": 0.01874 + } + ] + }, + { + "step": 23680, + "tok": 96993280, + "ce": 2.9606, + "ppl": 19.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4079, + "loop_stats": [ + { + "tau": 0.1175, + "v": 0.8422, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8326, + "H_a": 0.065, + "d_bind": 0.03896 + }, + { + "tau": 0.0241, + "v": 0.8345, + "H_a": 0.052, + "d_bind": 0.02016 + } + ] + }, + { + "step": 23700, + "tok": 97075200, + "ce": 2.6927, + "ppl": 14.8, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.938, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4088, + "loop_stats": [ + { + "tau": 0.1069, + "v": 0.8338, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8137, + "H_a": 0.07, + "d_bind": 0.03992 + }, + { + "tau": 0.0267, + "v": 0.8093, + "H_a": 0.055, + "d_bind": 0.02041 + } + ] + }, + { + "step": 23720, + "tok": 97157120, + "ce": 2.6282, + "ppl": 13.8, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.95, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.4056, + "loop_stats": [ + { + "tau": 0.1172, + "v": 0.8094, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0492, + "v": 0.7827, + "H_a": 0.062, + "d_bind": 0.0382 + }, + { + "tau": 0.0378, + "v": 0.7743, + "H_a": 0.051, + "d_bind": 0.02001 + } + ] + }, + { + "step": 23740, + "tok": 97239040, + "ce": 3.0791, + "ppl": 21.7, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.952, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.4037, + "loop_stats": [ + { + "tau": 0.0902, + "v": 0.8836, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0255, + "v": 0.8602, + "H_a": 0.068, + "d_bind": 0.03902 + }, + { + "tau": 0.0164, + "v": 0.8471, + "H_a": 0.052, + "d_bind": 0.02101 + } + ] + }, + { + "step": 23760, + "tok": 97320960, + "ce": 2.8064, + "ppl": 16.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22509, + "peak_MiB": 5301, + "v_gain": 0.4039, + "loop_stats": [ + { + "tau": 0.1113, + "v": 0.8626, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0464, + "v": 0.8414, + "H_a": 0.063, + "d_bind": 0.036 + }, + { + "tau": 0.0282, + "v": 0.8374, + "H_a": 0.049, + "d_bind": 0.01992 + } + ] + }, + { + "step": 23780, + "tok": 97402880, + "ce": 3.1351, + "ppl": 23.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4004, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.8495, + "H_a": 0.09, + "d_bind": 0.0 + }, + { + "tau": 0.0383, + "v": 0.816, + "H_a": 0.067, + "d_bind": 0.03928 + }, + { + "tau": 0.026, + "v": 0.8044, + "H_a": 0.049, + "d_bind": 0.02043 + } + ] + }, + { + "step": 23800, + "tok": 97484800, + "ce": 2.8634, + "ppl": 17.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.3982, + "loop_stats": [ + { + "tau": 0.1018, + "v": 0.8861, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8653, + "H_a": 0.068, + "d_bind": 0.03876 + }, + { + "tau": 0.0272, + "v": 0.8645, + "H_a": 0.053, + "d_bind": 0.02157 + } + ] + }, + { + "step": 23820, + "tok": 97566720, + "ce": 2.7743, + "ppl": 16.0, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.3984, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.8689, + "H_a": 0.094, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8432, + "H_a": 0.069, + "d_bind": 0.03939 + }, + { + "tau": 0.0346, + "v": 0.8361, + "H_a": 0.06, + "d_bind": 0.02033 + } + ] + }, + { + "step": 23840, + "tok": 97648640, + "ce": 2.9064, + "ppl": 18.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4006, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.8738, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.848, + "H_a": 0.067, + "d_bind": 0.03819 + }, + { + "tau": 0.0277, + "v": 0.8395, + "H_a": 0.055, + "d_bind": 0.02112 + } + ] + }, + { + "step": 23860, + "tok": 97730560, + "ce": 3.0538, + "ppl": 21.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.402, + "loop_stats": [ + { + "tau": 0.0961, + "v": 0.8722, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0298, + "v": 0.8503, + "H_a": 0.066, + "d_bind": 0.03704 + }, + { + "tau": 0.0221, + "v": 0.8371, + "H_a": 0.05, + "d_bind": 0.0199 + } + ] + }, + { + "step": 23880, + "tok": 97812480, + "ce": 3.1933, + "ppl": 24.4, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.962, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4025, + "loop_stats": [ + { + "tau": 0.0768, + "v": 0.8888, + "H_a": 0.095, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8593, + "H_a": 0.062, + "d_bind": 0.03659 + }, + { + "tau": 0.0143, + "v": 0.8389, + "H_a": 0.051, + "d_bind": 0.02023 + } + ] + }, + { + "step": 23900, + "tok": 97894400, + "ce": 3.0845, + "ppl": 21.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.965, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4041, + "loop_stats": [ + { + "tau": 0.0952, + "v": 0.8965, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0264, + "v": 0.8636, + "H_a": 0.065, + "d_bind": 0.03923 + }, + { + "tau": 0.0126, + "v": 0.8509, + "H_a": 0.056, + "d_bind": 0.0224 + } + ] + }, + { + "step": 23920, + "tok": 97976320, + "ce": 3.0027, + "ppl": 20.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.957, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.4032, + "loop_stats": [ + { + "tau": 0.0928, + "v": 0.8652, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0341, + "v": 0.8263, + "H_a": 0.063, + "d_bind": 0.03576 + }, + { + "tau": 0.0213, + "v": 0.8206, + "H_a": 0.053, + "d_bind": 0.01931 + } + ] + }, + { + "step": 23940, + "tok": 98058240, + "ce": 3.045, + "ppl": 21.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.4029, + "loop_stats": [ + { + "tau": 0.0908, + "v": 0.865, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8223, + "H_a": 0.067, + "d_bind": 0.03564 + }, + { + "tau": 0.0166, + "v": 0.811, + "H_a": 0.056, + "d_bind": 0.02103 + } + ] + }, + { + "step": 23960, + "tok": 98140160, + "ce": 2.6743, + "ppl": 14.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4053, + "loop_stats": [ + { + "tau": 0.1361, + "v": 0.9037, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.8616, + "H_a": 0.059, + "d_bind": 0.03946 + }, + { + "tau": 0.0301, + "v": 0.8495, + "H_a": 0.046, + "d_bind": 0.01684 + } + ] + }, + { + "step": 23980, + "tok": 98222080, + "ce": 2.5577, + "ppl": 12.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.957, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.4061, + "loop_stats": [ + { + "tau": 0.1265, + "v": 0.8901, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.8574, + "H_a": 0.066, + "d_bind": 0.04232 + }, + { + "tau": 0.0232, + "v": 0.8486, + "H_a": 0.057, + "d_bind": 0.02153 + } + ] + }, + { + "step": 24000, + "tok": 98304000, + "ce": 3.3525, + "ppl": 28.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22510, + "peak_MiB": 5301, + "v_gain": 0.4035, + "loop_stats": [ + { + "tau": 0.0695, + "v": 0.8528, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0207, + "v": 0.8113, + "H_a": 0.064, + "d_bind": 0.0358 + }, + { + "tau": 0.0145, + "v": 0.7979, + "H_a": 0.049, + "d_bind": 0.01923 + } + ] + }, + { + "step": 24020, + "tok": 98385920, + "ce": 3.0585, + "ppl": 21.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.4017, + "loop_stats": [ + { + "tau": 0.1136, + "v": 0.8851, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8532, + "H_a": 0.064, + "d_bind": 0.03532 + }, + { + "tau": 0.0314, + "v": 0.8427, + "H_a": 0.051, + "d_bind": 0.01962 + } + ] + }, + { + "step": 24040, + "tok": 98467840, + "ce": 2.748, + "ppl": 15.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.967, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.3993, + "loop_stats": [ + { + "tau": 0.1174, + "v": 0.8671, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.047, + "v": 0.841, + "H_a": 0.059, + "d_bind": 0.03929 + }, + { + "tau": 0.0387, + "v": 0.8292, + "H_a": 0.053, + "d_bind": 0.0187 + } + ] + }, + { + "step": 24060, + "tok": 98549760, + "ce": 2.7906, + "ppl": 16.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.965, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.3998, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8305, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8041, + "H_a": 0.065, + "d_bind": 0.04114 + }, + { + "tau": 0.0354, + "v": 0.797, + "H_a": 0.054, + "d_bind": 0.01999 + } + ] + }, + { + "step": 24080, + "tok": 98631680, + "ce": 2.6807, + "ppl": 14.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.975, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.403, + "loop_stats": [ + { + "tau": 0.1018, + "v": 0.8268, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8001, + "H_a": 0.068, + "d_bind": 0.0405 + }, + { + "tau": 0.0394, + "v": 0.8012, + "H_a": 0.052, + "d_bind": 0.0197 + } + ] + }, + { + "step": 24100, + "tok": 98713600, + "ce": 3.2583, + "ppl": 26.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.957, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4062, + "loop_stats": [ + { + "tau": 0.0876, + "v": 0.9118, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.8893, + "H_a": 0.061, + "d_bind": 0.03605 + }, + { + "tau": 0.0175, + "v": 0.8775, + "H_a": 0.05, + "d_bind": 0.02034 + } + ] + }, + { + "step": 24120, + "tok": 98795520, + "ce": 2.9514, + "ppl": 19.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.967, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.41, + "loop_stats": [ + { + "tau": 0.088, + "v": 0.8649, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8286, + "H_a": 0.067, + "d_bind": 0.03588 + }, + { + "tau": 0.0137, + "v": 0.8122, + "H_a": 0.055, + "d_bind": 0.02061 + } + ] + }, + { + "step": 24140, + "tok": 98877440, + "ce": 3.0753, + "ppl": 21.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.975, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.4106, + "loop_stats": [ + { + "tau": 0.074, + "v": 0.8697, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.023, + "v": 0.8206, + "H_a": 0.06, + "d_bind": 0.03631 + }, + { + "tau": 0.0164, + "v": 0.8016, + "H_a": 0.047, + "d_bind": 0.01812 + } + ] + }, + { + "step": 24160, + "tok": 98959360, + "ce": 3.0896, + "ppl": 22.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.962, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.4088, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.8495, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0376, + "v": 0.8226, + "H_a": 0.061, + "d_bind": 0.03867 + }, + { + "tau": 0.0259, + "v": 0.8108, + "H_a": 0.051, + "d_bind": 0.01869 + } + ] + }, + { + "step": 24180, + "tok": 99041280, + "ce": 2.8421, + "ppl": 17.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.4059, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.906, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0486, + "v": 0.8767, + "H_a": 0.058, + "d_bind": 0.03847 + }, + { + "tau": 0.0309, + "v": 0.8698, + "H_a": 0.046, + "d_bind": 0.01911 + } + ] + }, + { + "step": 24200, + "tok": 99123200, + "ce": 3.0711, + "ppl": 21.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.965, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.404, + "loop_stats": [ + { + "tau": 0.0912, + "v": 0.8563, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.835, + "H_a": 0.068, + "d_bind": 0.03834 + }, + { + "tau": 0.0251, + "v": 0.8276, + "H_a": 0.048, + "d_bind": 0.01866 + } + ] + }, + { + "step": 24220, + "tok": 99205120, + "ce": 3.0777, + "ppl": 21.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4045, + "loop_stats": [ + { + "tau": 0.0874, + "v": 0.8928, + "H_a": 0.087, + "d_bind": 0.0 + }, + { + "tau": 0.0333, + "v": 0.8661, + "H_a": 0.069, + "d_bind": 0.03576 + }, + { + "tau": 0.0158, + "v": 0.8544, + "H_a": 0.058, + "d_bind": 0.02121 + } + ] + }, + { + "step": 24240, + "tok": 99287040, + "ce": 3.0436, + "ppl": 21.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.962, + "tok_s": 22511, + "peak_MiB": 5301, + "v_gain": 0.402, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.8831, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8584, + "H_a": 0.067, + "d_bind": 0.03549 + }, + { + "tau": 0.0266, + "v": 0.8497, + "H_a": 0.052, + "d_bind": 0.01955 + } + ] + }, + { + "step": 24260, + "tok": 99368960, + "ce": 2.8628, + "ppl": 17.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.962, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.3994, + "loop_stats": [ + { + "tau": 0.0974, + "v": 0.8807, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.841, + "H_a": 0.064, + "d_bind": 0.03618 + }, + { + "tau": 0.0272, + "v": 0.8272, + "H_a": 0.053, + "d_bind": 0.02133 + } + ] + }, + { + "step": 24280, + "tok": 99450880, + "ce": 2.9587, + "ppl": 19.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.3974, + "loop_stats": [ + { + "tau": 0.0919, + "v": 0.8786, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0406, + "v": 0.839, + "H_a": 0.058, + "d_bind": 0.03233 + }, + { + "tau": 0.0232, + "v": 0.8228, + "H_a": 0.047, + "d_bind": 0.01831 + } + ] + }, + { + "step": 24300, + "tok": 99532800, + "ce": 2.811, + "ppl": 16.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.3953, + "loop_stats": [ + { + "tau": 0.1032, + "v": 0.9066, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8803, + "H_a": 0.058, + "d_bind": 0.03466 + }, + { + "tau": 0.0307, + "v": 0.8686, + "H_a": 0.049, + "d_bind": 0.01787 + } + ] + }, + { + "step": 24320, + "tok": 99614720, + "ce": 3.2216, + "ppl": 25.1, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.397, + "loop_stats": [ + { + "tau": 0.0754, + "v": 0.8787, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.025, + "v": 0.8489, + "H_a": 0.066, + "d_bind": 0.03505 + }, + { + "tau": 0.0169, + "v": 0.8367, + "H_a": 0.048, + "d_bind": 0.0193 + } + ] + }, + { + "step": 24340, + "tok": 99696640, + "ce": 3.3195, + "ppl": 27.6, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.967, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.3996, + "loop_stats": [ + { + "tau": 0.0978, + "v": 0.8887, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0218, + "v": 0.8587, + "H_a": 0.058, + "d_bind": 0.03077 + }, + { + "tau": 0.0131, + "v": 0.8483, + "H_a": 0.046, + "d_bind": 0.01769 + } + ] + }, + { + "step": 24360, + "tok": 99778560, + "ce": 3.0286, + "ppl": 20.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.97, + "tok_s": 22512, + "peak_MiB": 5301, + "v_gain": 0.4003, + "loop_stats": [ + { + "tau": 0.0997, + "v": 0.8497, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.8151, + "H_a": 0.065, + "d_bind": 0.03468 + }, + { + "tau": 0.0232, + "v": 0.8075, + "H_a": 0.047, + "d_bind": 0.01836 + } + ] + }, + { + "step": 24380, + "tok": 99860480, + "ce": 2.931, + "ppl": 18.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.965, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.4016, + "loop_stats": [ + { + "tau": 0.1191, + "v": 0.8815, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0491, + "v": 0.8489, + "H_a": 0.057, + "d_bind": 0.03294 + }, + { + "tau": 0.0387, + "v": 0.8433, + "H_a": 0.045, + "d_bind": 0.01775 + } + ] + }, + { + "step": 24400, + "tok": 99942400, + "ce": 2.2782, + "ppl": 9.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.965, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4032, + "loop_stats": [ + { + "tau": 0.1384, + "v": 0.8615, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0521, + "v": 0.8295, + "H_a": 0.066, + "d_bind": 0.03948 + }, + { + "tau": 0.0487, + "v": 0.8216, + "H_a": 0.05, + "d_bind": 0.02161 + } + ] + }, + { + "step": 24420, + "tok": 100024320, + "ce": 2.8628, + "ppl": 17.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.952, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4044, + "loop_stats": [ + { + "tau": 0.1078, + "v": 0.8921, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8545, + "H_a": 0.058, + "d_bind": 0.03439 + }, + { + "tau": 0.0316, + "v": 0.8362, + "H_a": 0.052, + "d_bind": 0.01861 + } + ] + }, + { + "step": 24440, + "tok": 100106240, + "ce": 3.1931, + "ppl": 24.4, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.947, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4031, + "loop_stats": [ + { + "tau": 0.0865, + "v": 0.9004, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0309, + "v": 0.8606, + "H_a": 0.063, + "d_bind": 0.03145 + }, + { + "tau": 0.015, + "v": 0.849, + "H_a": 0.057, + "d_bind": 0.02088 + } + ] + }, + { + "step": 24460, + "tok": 100188160, + "ce": 2.7461, + "ppl": 15.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.952, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4039, + "loop_stats": [ + { + "tau": 0.1287, + "v": 0.8835, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0499, + "v": 0.8483, + "H_a": 0.062, + "d_bind": 0.03557 + }, + { + "tau": 0.0316, + "v": 0.8311, + "H_a": 0.056, + "d_bind": 0.01979 + } + ] + }, + { + "step": 24480, + "tok": 100270080, + "ce": 2.6251, + "ppl": 13.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.962, + "tok_s": 22513, + "peak_MiB": 5301, + "v_gain": 0.4063, + "loop_stats": [ + { + "tau": 0.1214, + "v": 0.865, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8342, + "H_a": 0.061, + "d_bind": 0.03749 + }, + { + "tau": 0.035, + "v": 0.8328, + "H_a": 0.044, + "d_bind": 0.01757 + } + ] + }, + { + "step": 24500, + "tok": 100352000, + "ce": 3.0771, + "ppl": 21.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.96, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4061, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8663, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8377, + "H_a": 0.059, + "d_bind": 0.03693 + }, + { + "tau": 0.0255, + "v": 0.8368, + "H_a": 0.05, + "d_bind": 0.0193 + } + ] + }, + { + "step": 24520, + "tok": 100433920, + "ce": 2.8291, + "ppl": 16.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.96, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4006, + "loop_stats": [ + { + "tau": 0.0941, + "v": 0.9121, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0445, + "v": 0.8889, + "H_a": 0.06, + "d_bind": 0.03564 + }, + { + "tau": 0.0339, + "v": 0.8775, + "H_a": 0.048, + "d_bind": 0.01973 + } + ] + }, + { + "step": 24540, + "tok": 100515840, + "ce": 2.9235, + "ppl": 18.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.955, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4006, + "loop_stats": [ + { + "tau": 0.1231, + "v": 0.8722, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8391, + "H_a": 0.054, + "d_bind": 0.03666 + }, + { + "tau": 0.0426, + "v": 0.8282, + "H_a": 0.049, + "d_bind": 0.01776 + } + ] + }, + { + "step": 24560, + "tok": 100597760, + "ce": 2.4804, + "ppl": 11.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.952, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4046, + "loop_stats": [ + { + "tau": 0.1305, + "v": 0.8707, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0524, + "v": 0.8475, + "H_a": 0.056, + "d_bind": 0.03838 + }, + { + "tau": 0.0501, + "v": 0.8455, + "H_a": 0.045, + "d_bind": 0.01681 + } + ] + }, + { + "step": 24580, + "tok": 100679680, + "ce": 2.3347, + "ppl": 10.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4076, + "loop_stats": [ + { + "tau": 0.144, + "v": 0.8481, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0614, + "v": 0.8284, + "H_a": 0.059, + "d_bind": 0.04065 + }, + { + "tau": 0.0536, + "v": 0.8216, + "H_a": 0.048, + "d_bind": 0.01807 + } + ] + }, + { + "step": 24600, + "tok": 100761600, + "ce": 2.7264, + "ppl": 15.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.947, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4084, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.8786, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0456, + "v": 0.8533, + "H_a": 0.056, + "d_bind": 0.03869 + }, + { + "tau": 0.0364, + "v": 0.8426, + "H_a": 0.05, + "d_bind": 0.019 + } + ] + }, + { + "step": 24620, + "tok": 100843520, + "ce": 2.7487, + "ppl": 15.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.4086, + "loop_stats": [ + { + "tau": 0.1011, + "v": 0.8987, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.8662, + "H_a": 0.066, + "d_bind": 0.0362 + }, + { + "tau": 0.0234, + "v": 0.8534, + "H_a": 0.061, + "d_bind": 0.02133 + } + ] + }, + { + "step": 24640, + "tok": 100925440, + "ce": 3.1567, + "ppl": 23.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4064, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8526, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8222, + "H_a": 0.053, + "d_bind": 0.03541 + }, + { + "tau": 0.0303, + "v": 0.8148, + "H_a": 0.046, + "d_bind": 0.01831 + } + ] + }, + { + "step": 24660, + "tok": 101007360, + "ce": 2.9756, + "ppl": 19.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4054, + "loop_stats": [ + { + "tau": 0.081, + "v": 0.88, + "H_a": 0.091, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.8381, + "H_a": 0.068, + "d_bind": 0.03685 + }, + { + "tau": 0.0161, + "v": 0.8239, + "H_a": 0.059, + "d_bind": 0.0202 + } + ] + }, + { + "step": 24680, + "tok": 101089280, + "ce": 3.1452, + "ppl": 23.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4094, + "loop_stats": [ + { + "tau": 0.093, + "v": 0.9027, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0289, + "v": 0.8779, + "H_a": 0.06, + "d_bind": 0.03484 + }, + { + "tau": 0.0259, + "v": 0.8613, + "H_a": 0.053, + "d_bind": 0.01952 + } + ] + }, + { + "step": 24700, + "tok": 101171200, + "ce": 3.2653, + "ppl": 26.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4109, + "loop_stats": [ + { + "tau": 0.0791, + "v": 0.9039, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0265, + "v": 0.879, + "H_a": 0.061, + "d_bind": 0.03537 + }, + { + "tau": 0.0169, + "v": 0.8729, + "H_a": 0.055, + "d_bind": 0.01983 + } + ] + }, + { + "step": 24720, + "tok": 101253120, + "ce": 3.4554, + "ppl": 31.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22514, + "peak_MiB": 5301, + "v_gain": 0.41, + "loop_stats": [ + { + "tau": 0.0999, + "v": 0.9148, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8929, + "H_a": 0.06, + "d_bind": 0.03184 + }, + { + "tau": 0.0243, + "v": 0.8845, + "H_a": 0.049, + "d_bind": 0.02017 + } + ] + }, + { + "step": 24740, + "tok": 101335040, + "ce": 2.3917, + "ppl": 10.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4095, + "loop_stats": [ + { + "tau": 0.1386, + "v": 0.8302, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0488, + "v": 0.8022, + "H_a": 0.064, + "d_bind": 0.04201 + }, + { + "tau": 0.0468, + "v": 0.7935, + "H_a": 0.048, + "d_bind": 0.02 + } + ] + }, + { + "step": 24760, + "tok": 101416960, + "ce": 3.0319, + "ppl": 20.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4087, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8908, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0556, + "v": 0.8571, + "H_a": 0.065, + "d_bind": 0.0339 + }, + { + "tau": 0.0367, + "v": 0.8409, + "H_a": 0.054, + "d_bind": 0.01792 + } + ] + }, + { + "step": 24780, + "tok": 101498880, + "ce": 2.6755, + "ppl": 14.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4091, + "loop_stats": [ + { + "tau": 0.1214, + "v": 0.8516, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0523, + "v": 0.8208, + "H_a": 0.065, + "d_bind": 0.03752 + }, + { + "tau": 0.0441, + "v": 0.8129, + "H_a": 0.051, + "d_bind": 0.0205 + } + ] + }, + { + "step": 24800, + "tok": 101580800, + "ce": 2.8222, + "ppl": 16.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4111, + "loop_stats": [ + { + "tau": 0.099, + "v": 0.8695, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.8141, + "H_a": 0.06, + "d_bind": 0.03607 + }, + { + "tau": 0.029, + "v": 0.7943, + "H_a": 0.052, + "d_bind": 0.02021 + } + ] + }, + { + "step": 24820, + "tok": 101662720, + "ce": 2.4176, + "ppl": 11.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4094, + "loop_stats": [ + { + "tau": 0.1436, + "v": 0.8876, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8472, + "H_a": 0.059, + "d_bind": 0.04029 + }, + { + "tau": 0.0466, + "v": 0.8364, + "H_a": 0.053, + "d_bind": 0.01849 + } + ] + }, + { + "step": 24840, + "tok": 101744640, + "ce": 2.4094, + "ppl": 11.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4076, + "loop_stats": [ + { + "tau": 0.1428, + "v": 0.8426, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0618, + "v": 0.8159, + "H_a": 0.065, + "d_bind": 0.04192 + }, + { + "tau": 0.0567, + "v": 0.8065, + "H_a": 0.053, + "d_bind": 0.01844 + } + ] + }, + { + "step": 24860, + "tok": 101826560, + "ce": 2.9022, + "ppl": 18.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4096, + "loop_stats": [ + { + "tau": 0.1071, + "v": 0.8826, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8544, + "H_a": 0.064, + "d_bind": 0.03473 + }, + { + "tau": 0.0305, + "v": 0.838, + "H_a": 0.052, + "d_bind": 0.02121 + } + ] + }, + { + "step": 24880, + "tok": 101908480, + "ce": 2.7045, + "ppl": 14.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.945, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4106, + "loop_stats": [ + { + "tau": 0.1274, + "v": 0.8718, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8515, + "H_a": 0.065, + "d_bind": 0.03943 + }, + { + "tau": 0.0342, + "v": 0.84, + "H_a": 0.055, + "d_bind": 0.02008 + } + ] + }, + { + "step": 24900, + "tok": 101990400, + "ce": 2.6011, + "ppl": 13.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4121, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.8389, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8033, + "H_a": 0.065, + "d_bind": 0.04221 + }, + { + "tau": 0.0315, + "v": 0.7905, + "H_a": 0.049, + "d_bind": 0.01895 + } + ] + }, + { + "step": 24920, + "tok": 102072320, + "ce": 3.0954, + "ppl": 22.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4119, + "loop_stats": [ + { + "tau": 0.0836, + "v": 0.8911, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.8662, + "H_a": 0.066, + "d_bind": 0.03474 + }, + { + "tau": 0.0237, + "v": 0.8589, + "H_a": 0.056, + "d_bind": 0.02156 + } + ] + }, + { + "step": 24940, + "tok": 102154240, + "ce": 3.4213, + "ppl": 30.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.41, + "loop_stats": [ + { + "tau": 0.0813, + "v": 0.8822, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.8435, + "H_a": 0.061, + "d_bind": 0.03435 + }, + { + "tau": 0.0287, + "v": 0.8302, + "H_a": 0.054, + "d_bind": 0.01908 + } + ] + }, + { + "step": 24960, + "tok": 102236160, + "ce": 2.5441, + "ppl": 12.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22515, + "peak_MiB": 5301, + "v_gain": 0.4106, + "loop_stats": [ + { + "tau": 0.1163, + "v": 0.8716, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8471, + "H_a": 0.068, + "d_bind": 0.03921 + }, + { + "tau": 0.036, + "v": 0.8447, + "H_a": 0.057, + "d_bind": 0.02322 + } + ] + }, + { + "step": 24980, + "tok": 102318080, + "ce": 2.6756, + "ppl": 14.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.962, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4092, + "loop_stats": [ + { + "tau": 0.1139, + "v": 0.9025, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0525, + "v": 0.8627, + "H_a": 0.058, + "d_bind": 0.03495 + }, + { + "tau": 0.0318, + "v": 0.853, + "H_a": 0.051, + "d_bind": 0.01847 + } + ] + }, + { + "step": 25000, + "tok": 102400000, + "ce": 2.6635, + "ppl": 14.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4111, + "loop_stats": [ + { + "tau": 0.1331, + "v": 0.8638, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0572, + "v": 0.8297, + "H_a": 0.064, + "d_bind": 0.03986 + }, + { + "tau": 0.036, + "v": 0.8174, + "H_a": 0.055, + "d_bind": 0.02103 + } + ] + }, + { + "step": 25020, + "tok": 102481920, + "ce": 2.9465, + "ppl": 19.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4114, + "loop_stats": [ + { + "tau": 0.1097, + "v": 0.8902, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.8539, + "H_a": 0.066, + "d_bind": 0.03646 + }, + { + "tau": 0.0248, + "v": 0.8418, + "H_a": 0.053, + "d_bind": 0.021 + } + ] + }, + { + "step": 25040, + "tok": 102563840, + "ce": 2.7552, + "ppl": 15.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.96, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4115, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.9092, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8711, + "H_a": 0.068, + "d_bind": 0.03828 + }, + { + "tau": 0.0238, + "v": 0.8583, + "H_a": 0.054, + "d_bind": 0.02282 + } + ] + }, + { + "step": 25060, + "tok": 102645760, + "ce": 2.6074, + "ppl": 13.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4078, + "loop_stats": [ + { + "tau": 0.1562, + "v": 0.9059, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.065, + "v": 0.8694, + "H_a": 0.059, + "d_bind": 0.03825 + }, + { + "tau": 0.0482, + "v": 0.8598, + "H_a": 0.054, + "d_bind": 0.01957 + } + ] + }, + { + "step": 25080, + "tok": 102727680, + "ce": 3.0128, + "ppl": 20.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.967, + "tok_s": 22516, + "peak_MiB": 5301, + "v_gain": 0.4075, + "loop_stats": [ + { + "tau": 0.0883, + "v": 0.8848, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8499, + "H_a": 0.06, + "d_bind": 0.03693 + }, + { + "tau": 0.0199, + "v": 0.836, + "H_a": 0.05, + "d_bind": 0.01954 + } + ] + }, + { + "step": 25100, + "tok": 102809600, + "ce": 2.7191, + "ppl": 15.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.972, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4089, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.8915, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8474, + "H_a": 0.066, + "d_bind": 0.03821 + }, + { + "tau": 0.0296, + "v": 0.8363, + "H_a": 0.057, + "d_bind": 0.02136 + } + ] + }, + { + "step": 25120, + "tok": 102891520, + "ce": 2.419, + "ppl": 11.2, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4095, + "loop_stats": [ + { + "tau": 0.1431, + "v": 0.9013, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.06, + "v": 0.8516, + "H_a": 0.066, + "d_bind": 0.03804 + }, + { + "tau": 0.0457, + "v": 0.8448, + "H_a": 0.058, + "d_bind": 0.02054 + } + ] + }, + { + "step": 25140, + "tok": 102973440, + "ce": 2.8664, + "ppl": 17.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4109, + "loop_stats": [ + { + "tau": 0.0918, + "v": 0.8805, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.8215, + "H_a": 0.067, + "d_bind": 0.03531 + }, + { + "tau": 0.024, + "v": 0.8031, + "H_a": 0.056, + "d_bind": 0.02297 + } + ] + }, + { + "step": 25160, + "tok": 103055360, + "ce": 2.6732, + "ppl": 14.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4132, + "loop_stats": [ + { + "tau": 0.1134, + "v": 0.8928, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0415, + "v": 0.8449, + "H_a": 0.067, + "d_bind": 0.03861 + }, + { + "tau": 0.0296, + "v": 0.8309, + "H_a": 0.051, + "d_bind": 0.0213 + } + ] + }, + { + "step": 25180, + "tok": 103137280, + "ce": 3.1451, + "ppl": 23.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.95, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4142, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.8822, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8409, + "H_a": 0.066, + "d_bind": 0.03625 + }, + { + "tau": 0.0157, + "v": 0.8244, + "H_a": 0.052, + "d_bind": 0.02032 + } + ] + }, + { + "step": 25200, + "tok": 103219200, + "ce": 2.7117, + "ppl": 15.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4158, + "loop_stats": [ + { + "tau": 0.1334, + "v": 0.9055, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8718, + "H_a": 0.064, + "d_bind": 0.03799 + }, + { + "tau": 0.0341, + "v": 0.8623, + "H_a": 0.055, + "d_bind": 0.02126 + } + ] + }, + { + "step": 25220, + "tok": 103301120, + "ce": 2.8266, + "ppl": 16.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4159, + "loop_stats": [ + { + "tau": 0.1065, + "v": 0.898, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8636, + "H_a": 0.068, + "d_bind": 0.03807 + }, + { + "tau": 0.0223, + "v": 0.8579, + "H_a": 0.054, + "d_bind": 0.02074 + } + ] + }, + { + "step": 25240, + "tok": 103383040, + "ce": 3.0692, + "ppl": 21.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4203, + "loop_stats": [ + { + "tau": 0.0918, + "v": 0.8908, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.8591, + "H_a": 0.066, + "d_bind": 0.03262 + }, + { + "tau": 0.0074, + "v": 0.8518, + "H_a": 0.05, + "d_bind": 0.02061 + } + ] + }, + { + "step": 25260, + "tok": 103464960, + "ce": 3.0792, + "ppl": 21.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.418, + "loop_stats": [ + { + "tau": 0.1039, + "v": 0.8665, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.835, + "H_a": 0.068, + "d_bind": 0.03937 + }, + { + "tau": 0.0225, + "v": 0.8295, + "H_a": 0.054, + "d_bind": 0.02019 + } + ] + }, + { + "step": 25280, + "tok": 103546880, + "ce": 2.7288, + "ppl": 15.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4185, + "loop_stats": [ + { + "tau": 0.1054, + "v": 0.8655, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.8407, + "H_a": 0.065, + "d_bind": 0.03824 + }, + { + "tau": 0.0243, + "v": 0.8336, + "H_a": 0.052, + "d_bind": 0.02134 + } + ] + }, + { + "step": 25300, + "tok": 103628800, + "ce": 2.9731, + "ppl": 19.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4197, + "loop_stats": [ + { + "tau": 0.1126, + "v": 0.8627, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8228, + "H_a": 0.065, + "d_bind": 0.03965 + }, + { + "tau": 0.0346, + "v": 0.8144, + "H_a": 0.054, + "d_bind": 0.02069 + } + ] + }, + { + "step": 25320, + "tok": 103710720, + "ce": 3.1352, + "ppl": 23.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.96, + "tok_s": 22517, + "peak_MiB": 5301, + "v_gain": 0.4195, + "loop_stats": [ + { + "tau": 0.1092, + "v": 0.9316, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.9088, + "H_a": 0.059, + "d_bind": 0.03227 + }, + { + "tau": 0.017, + "v": 0.8982, + "H_a": 0.05, + "d_bind": 0.01917 + } + ] + }, + { + "step": 25340, + "tok": 103792640, + "ce": 3.006, + "ppl": 20.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.977, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4183, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.9146, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0417, + "v": 0.8739, + "H_a": 0.064, + "d_bind": 0.03552 + }, + { + "tau": 0.0194, + "v": 0.8598, + "H_a": 0.054, + "d_bind": 0.02055 + } + ] + }, + { + "step": 25360, + "tok": 103874560, + "ce": 2.8143, + "ppl": 16.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.967, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4162, + "loop_stats": [ + { + "tau": 0.1042, + "v": 0.9071, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8605, + "H_a": 0.062, + "d_bind": 0.03749 + }, + { + "tau": 0.0315, + "v": 0.8524, + "H_a": 0.052, + "d_bind": 0.01936 + } + ] + }, + { + "step": 25380, + "tok": 103956480, + "ce": 2.4922, + "ppl": 12.1, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.975, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4158, + "loop_stats": [ + { + "tau": 0.1439, + "v": 0.8993, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0493, + "v": 0.855, + "H_a": 0.067, + "d_bind": 0.03852 + }, + { + "tau": 0.0405, + "v": 0.842, + "H_a": 0.055, + "d_bind": 0.02129 + } + ] + }, + { + "step": 25400, + "tok": 104038400, + "ce": 2.5861, + "ppl": 13.3, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.967, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.419, + "loop_stats": [ + { + "tau": 0.1291, + "v": 0.8976, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0278, + "v": 0.8422, + "H_a": 0.063, + "d_bind": 0.03842 + }, + { + "tau": 0.0242, + "v": 0.8282, + "H_a": 0.055, + "d_bind": 0.02089 + } + ] + }, + { + "step": 25420, + "tok": 104120320, + "ce": 2.6063, + "ppl": 13.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.957, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4228, + "loop_stats": [ + { + "tau": 0.1484, + "v": 0.8756, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.8233, + "H_a": 0.058, + "d_bind": 0.03826 + }, + { + "tau": 0.0258, + "v": 0.8061, + "H_a": 0.046, + "d_bind": 0.0189 + } + ] + }, + { + "step": 25440, + "tok": 104202240, + "ce": 2.6519, + "ppl": 14.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.962, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4228, + "loop_stats": [ + { + "tau": 0.0963, + "v": 0.9084, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0255, + "v": 0.8577, + "H_a": 0.058, + "d_bind": 0.03389 + }, + { + "tau": 0.0071, + "v": 0.8374, + "H_a": 0.049, + "d_bind": 0.0194 + } + ] + }, + { + "step": 25460, + "tok": 104284160, + "ce": 2.8807, + "ppl": 17.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4229, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.8812, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.8444, + "H_a": 0.068, + "d_bind": 0.04089 + }, + { + "tau": 0.0197, + "v": 0.8318, + "H_a": 0.052, + "d_bind": 0.01895 + } + ] + }, + { + "step": 25480, + "tok": 104366080, + "ce": 2.6322, + "ppl": 13.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4222, + "loop_stats": [ + { + "tau": 0.114, + "v": 0.9122, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.8542, + "H_a": 0.069, + "d_bind": 0.03839 + }, + { + "tau": 0.0201, + "v": 0.8393, + "H_a": 0.05, + "d_bind": 0.01919 + } + ] + }, + { + "step": 25500, + "tok": 104448000, + "ce": 3.1512, + "ppl": 23.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.962, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4214, + "loop_stats": [ + { + "tau": 0.0957, + "v": 0.8999, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.8492, + "H_a": 0.061, + "d_bind": 0.03731 + }, + { + "tau": 0.0203, + "v": 0.8286, + "H_a": 0.052, + "d_bind": 0.02015 + } + ] + }, + { + "step": 25520, + "tok": 104529920, + "ce": 2.9072, + "ppl": 18.3, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4213, + "loop_stats": [ + { + "tau": 0.106, + "v": 0.8732, + "H_a": 0.088, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.855, + "H_a": 0.067, + "d_bind": 0.03925 + }, + { + "tau": 0.0253, + "v": 0.8557, + "H_a": 0.053, + "d_bind": 0.0208 + } + ] + }, + { + "step": 25540, + "tok": 104611840, + "ce": 2.4861, + "ppl": 12.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.962, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4208, + "loop_stats": [ + { + "tau": 0.1382, + "v": 0.864, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.8333, + "H_a": 0.063, + "d_bind": 0.04191 + }, + { + "tau": 0.0441, + "v": 0.8292, + "H_a": 0.054, + "d_bind": 0.02031 + } + ] + }, + { + "step": 25560, + "tok": 104693760, + "ce": 2.7055, + "ppl": 15.0, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.952, + "tok_s": 22518, + "peak_MiB": 5301, + "v_gain": 0.4203, + "loop_stats": [ + { + "tau": 0.1317, + "v": 0.8959, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0538, + "v": 0.8631, + "H_a": 0.068, + "d_bind": 0.04055 + }, + { + "tau": 0.0426, + "v": 0.8561, + "H_a": 0.052, + "d_bind": 0.01989 + } + ] + }, + { + "step": 25580, + "tok": 104775680, + "ce": 2.9954, + "ppl": 20.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.96, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4197, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.9137, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.866, + "H_a": 0.067, + "d_bind": 0.03858 + }, + { + "tau": 0.0271, + "v": 0.8485, + "H_a": 0.049, + "d_bind": 0.01912 + } + ] + }, + { + "step": 25600, + "tok": 104857600, + "ce": 2.7055, + "ppl": 15.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4235, + "loop_stats": [ + { + "tau": 0.1141, + "v": 0.8533, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8185, + "H_a": 0.071, + "d_bind": 0.04331 + }, + { + "tau": 0.0315, + "v": 0.8135, + "H_a": 0.049, + "d_bind": 0.02089 + } + ] + }, + { + "step": 25620, + "tok": 104939520, + "ce": 2.5856, + "ppl": 13.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4276, + "loop_stats": [ + { + "tau": 0.1369, + "v": 0.8635, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8166, + "H_a": 0.059, + "d_bind": 0.04012 + }, + { + "tau": 0.036, + "v": 0.8012, + "H_a": 0.053, + "d_bind": 0.01926 + } + ] + }, + { + "step": 25640, + "tok": 105021440, + "ce": 2.6271, + "ppl": 13.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4253, + "loop_stats": [ + { + "tau": 0.1309, + "v": 0.9105, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.8789, + "H_a": 0.058, + "d_bind": 0.03742 + }, + { + "tau": 0.0357, + "v": 0.8707, + "H_a": 0.051, + "d_bind": 0.01804 + } + ] + }, + { + "step": 25660, + "tok": 105103360, + "ce": 3.0534, + "ppl": 21.2, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.95, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4244, + "loop_stats": [ + { + "tau": 0.1031, + "v": 0.8941, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8489, + "H_a": 0.062, + "d_bind": 0.03615 + }, + { + "tau": 0.0175, + "v": 0.8282, + "H_a": 0.052, + "d_bind": 0.01915 + } + ] + }, + { + "step": 25680, + "tok": 105185280, + "ce": 3.2253, + "ppl": 25.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4226, + "loop_stats": [ + { + "tau": 0.0955, + "v": 0.8886, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0212, + "v": 0.8581, + "H_a": 0.07, + "d_bind": 0.03426 + }, + { + "tau": 0.0142, + "v": 0.8482, + "H_a": 0.056, + "d_bind": 0.0213 + } + ] + }, + { + "step": 25700, + "tok": 105267200, + "ce": 2.756, + "ppl": 15.7, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4238, + "loop_stats": [ + { + "tau": 0.1223, + "v": 0.8792, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8444, + "H_a": 0.056, + "d_bind": 0.03429 + }, + { + "tau": 0.0212, + "v": 0.8182, + "H_a": 0.048, + "d_bind": 0.01885 + } + ] + }, + { + "step": 25720, + "tok": 105349120, + "ce": 2.6855, + "ppl": 14.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4252, + "loop_stats": [ + { + "tau": 0.1182, + "v": 0.9128, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.874, + "H_a": 0.062, + "d_bind": 0.03615 + }, + { + "tau": 0.0286, + "v": 0.8636, + "H_a": 0.055, + "d_bind": 0.02146 + } + ] + }, + { + "step": 25740, + "tok": 105431040, + "ce": 3.0853, + "ppl": 21.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4246, + "loop_stats": [ + { + "tau": 0.1149, + "v": 0.8931, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8605, + "H_a": 0.061, + "d_bind": 0.03387 + }, + { + "tau": 0.0259, + "v": 0.8426, + "H_a": 0.048, + "d_bind": 0.0182 + } + ] + }, + { + "step": 25760, + "tok": 105512960, + "ce": 3.3, + "ppl": 27.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4248, + "loop_stats": [ + { + "tau": 0.0911, + "v": 0.8896, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.857, + "H_a": 0.066, + "d_bind": 0.03627 + }, + { + "tau": 0.0183, + "v": 0.8489, + "H_a": 0.057, + "d_bind": 0.02171 + } + ] + }, + { + "step": 25780, + "tok": 105594880, + "ce": 3.0424, + "ppl": 21.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4234, + "loop_stats": [ + { + "tau": 0.1096, + "v": 0.8945, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0333, + "v": 0.8662, + "H_a": 0.072, + "d_bind": 0.038 + }, + { + "tau": 0.024, + "v": 0.8565, + "H_a": 0.06, + "d_bind": 0.02205 + } + ] + }, + { + "step": 25800, + "tok": 105676800, + "ce": 2.7161, + "ppl": 15.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22519, + "peak_MiB": 5301, + "v_gain": 0.4222, + "loop_stats": [ + { + "tau": 0.12, + "v": 0.8658, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8241, + "H_a": 0.06, + "d_bind": 0.03817 + }, + { + "tau": 0.0364, + "v": 0.8077, + "H_a": 0.052, + "d_bind": 0.02094 + } + ] + }, + { + "step": 25820, + "tok": 105758720, + "ce": 3.0344, + "ppl": 20.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4244, + "loop_stats": [ + { + "tau": 0.1085, + "v": 0.9075, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0303, + "v": 0.8789, + "H_a": 0.058, + "d_bind": 0.03386 + }, + { + "tau": 0.0154, + "v": 0.8628, + "H_a": 0.051, + "d_bind": 0.02042 + } + ] + }, + { + "step": 25840, + "tok": 105840640, + "ce": 3.2615, + "ppl": 26.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4285, + "loop_stats": [ + { + "tau": 0.078, + "v": 0.9021, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0122, + "v": 0.8648, + "H_a": 0.058, + "d_bind": 0.03298 + }, + { + "tau": 0.0042, + "v": 0.8473, + "H_a": 0.048, + "d_bind": 0.01824 + } + ] + }, + { + "step": 25860, + "tok": 105922560, + "ce": 2.5654, + "ppl": 13.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.43, + "loop_stats": [ + { + "tau": 0.1257, + "v": 0.8618, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0369, + "v": 0.8294, + "H_a": 0.065, + "d_bind": 0.04084 + }, + { + "tau": 0.0261, + "v": 0.8169, + "H_a": 0.052, + "d_bind": 0.01924 + } + ] + }, + { + "step": 25880, + "tok": 106004480, + "ce": 2.6394, + "ppl": 14.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4292, + "loop_stats": [ + { + "tau": 0.1375, + "v": 0.9037, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0368, + "v": 0.8648, + "H_a": 0.059, + "d_bind": 0.03924 + }, + { + "tau": 0.0241, + "v": 0.8495, + "H_a": 0.05, + "d_bind": 0.01906 + } + ] + }, + { + "step": 25900, + "tok": 106086400, + "ce": 2.7163, + "ppl": 15.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.945, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4257, + "loop_stats": [ + { + "tau": 0.133, + "v": 0.888, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8492, + "H_a": 0.057, + "d_bind": 0.03811 + }, + { + "tau": 0.0308, + "v": 0.8348, + "H_a": 0.052, + "d_bind": 0.01947 + } + ] + }, + { + "step": 25920, + "tok": 106168320, + "ce": 2.7911, + "ppl": 16.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4228, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.8764, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8361, + "H_a": 0.067, + "d_bind": 0.04109 + }, + { + "tau": 0.0262, + "v": 0.819, + "H_a": 0.051, + "d_bind": 0.02044 + } + ] + }, + { + "step": 25940, + "tok": 106250240, + "ce": 2.454, + "ppl": 11.6, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.962, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4211, + "loop_stats": [ + { + "tau": 0.1433, + "v": 0.8919, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0604, + "v": 0.86, + "H_a": 0.061, + "d_bind": 0.04296 + }, + { + "tau": 0.0427, + "v": 0.8516, + "H_a": 0.047, + "d_bind": 0.01975 + } + ] + }, + { + "step": 25960, + "tok": 106332160, + "ce": 2.5186, + "ppl": 12.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.965, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4199, + "loop_stats": [ + { + "tau": 0.1259, + "v": 0.896, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0348, + "v": 0.8679, + "H_a": 0.064, + "d_bind": 0.03645 + }, + { + "tau": 0.0231, + "v": 0.8567, + "H_a": 0.052, + "d_bind": 0.02324 + } + ] + }, + { + "step": 25980, + "tok": 106414080, + "ce": 2.9707, + "ppl": 19.5, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4197, + "loop_stats": [ + { + "tau": 0.0799, + "v": 0.9072, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0265, + "v": 0.8476, + "H_a": 0.06, + "d_bind": 0.0352 + }, + { + "tau": 0.0145, + "v": 0.8225, + "H_a": 0.05, + "d_bind": 0.02015 + } + ] + }, + { + "step": 26000, + "tok": 106496000, + "ce": 2.8495, + "ppl": 17.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4202, + "loop_stats": [ + { + "tau": 0.1074, + "v": 0.8902, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.034, + "v": 0.8382, + "H_a": 0.06, + "d_bind": 0.03757 + }, + { + "tau": 0.0221, + "v": 0.8291, + "H_a": 0.049, + "d_bind": 0.02131 + } + ] + }, + { + "step": 26020, + "tok": 106577920, + "ce": 2.5905, + "ppl": 13.3, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4189, + "loop_stats": [ + { + "tau": 0.1125, + "v": 0.8633, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0423, + "v": 0.8171, + "H_a": 0.07, + "d_bind": 0.04076 + }, + { + "tau": 0.029, + "v": 0.8072, + "H_a": 0.058, + "d_bind": 0.0236 + } + ] + }, + { + "step": 26040, + "tok": 106659840, + "ce": 3.0276, + "ppl": 20.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.962, + "tok_s": 22520, + "peak_MiB": 5301, + "v_gain": 0.4202, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8791, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0402, + "v": 0.8258, + "H_a": 0.064, + "d_bind": 0.03835 + }, + { + "tau": 0.0264, + "v": 0.8107, + "H_a": 0.056, + "d_bind": 0.0209 + } + ] + }, + { + "step": 26060, + "tok": 106741760, + "ce": 2.9746, + "ppl": 19.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4227, + "loop_stats": [ + { + "tau": 0.093, + "v": 0.9042, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0326, + "v": 0.8697, + "H_a": 0.061, + "d_bind": 0.03247 + }, + { + "tau": 0.0128, + "v": 0.8587, + "H_a": 0.059, + "d_bind": 0.02202 + } + ] + }, + { + "step": 26080, + "tok": 106823680, + "ce": 2.9641, + "ppl": 19.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.947, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4222, + "loop_stats": [ + { + "tau": 0.1061, + "v": 0.9025, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.8638, + "H_a": 0.071, + "d_bind": 0.03513 + }, + { + "tau": 0.0205, + "v": 0.856, + "H_a": 0.055, + "d_bind": 0.02185 + } + ] + }, + { + "step": 26100, + "tok": 106905600, + "ce": 2.4575, + "ppl": 11.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4221, + "loop_stats": [ + { + "tau": 0.1233, + "v": 0.8701, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8357, + "H_a": 0.069, + "d_bind": 0.038 + }, + { + "tau": 0.0259, + "v": 0.8314, + "H_a": 0.06, + "d_bind": 0.02377 + } + ] + }, + { + "step": 26120, + "tok": 106987520, + "ce": 2.7793, + "ppl": 16.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4222, + "loop_stats": [ + { + "tau": 0.1019, + "v": 0.9029, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.048, + "v": 0.87, + "H_a": 0.059, + "d_bind": 0.03259 + }, + { + "tau": 0.0187, + "v": 0.8533, + "H_a": 0.051, + "d_bind": 0.02104 + } + ] + }, + { + "step": 26140, + "tok": 107069440, + "ce": 2.6999, + "ppl": 14.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4223, + "loop_stats": [ + { + "tau": 0.1372, + "v": 0.8513, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0532, + "v": 0.8237, + "H_a": 0.066, + "d_bind": 0.03528 + }, + { + "tau": 0.0337, + "v": 0.814, + "H_a": 0.054, + "d_bind": 0.02167 + } + ] + }, + { + "step": 26160, + "tok": 107151360, + "ce": 3.1583, + "ppl": 23.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22521, + "peak_MiB": 5301, + "v_gain": 0.4221, + "loop_stats": [ + { + "tau": 0.098, + "v": 0.9042, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0265, + "v": 0.8613, + "H_a": 0.059, + "d_bind": 0.03385 + }, + { + "tau": 0.0141, + "v": 0.8417, + "H_a": 0.049, + "d_bind": 0.01856 + } + ] + }, + { + "step": 26180, + "tok": 107233280, + "ce": 2.6685, + "ppl": 14.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4174, + "loop_stats": [ + { + "tau": 0.122, + "v": 0.8834, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0461, + "v": 0.8417, + "H_a": 0.063, + "d_bind": 0.03764 + }, + { + "tau": 0.0318, + "v": 0.827, + "H_a": 0.05, + "d_bind": 0.0206 + } + ] + }, + { + "step": 26200, + "tok": 107315200, + "ce": 2.3993, + "ppl": 11.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4191, + "loop_stats": [ + { + "tau": 0.1549, + "v": 0.8901, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0613, + "v": 0.8408, + "H_a": 0.062, + "d_bind": 0.03649 + }, + { + "tau": 0.0452, + "v": 0.8331, + "H_a": 0.052, + "d_bind": 0.01955 + } + ] + }, + { + "step": 26220, + "tok": 107397120, + "ce": 2.4856, + "ppl": 12.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4212, + "loop_stats": [ + { + "tau": 0.1325, + "v": 0.8917, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0493, + "v": 0.844, + "H_a": 0.065, + "d_bind": 0.0372 + }, + { + "tau": 0.033, + "v": 0.8371, + "H_a": 0.049, + "d_bind": 0.0215 + } + ] + }, + { + "step": 26240, + "tok": 107479040, + "ce": 2.6941, + "ppl": 14.8, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.945, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4203, + "loop_stats": [ + { + "tau": 0.1138, + "v": 0.8955, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.8536, + "H_a": 0.065, + "d_bind": 0.0338 + }, + { + "tau": 0.0249, + "v": 0.8385, + "H_a": 0.055, + "d_bind": 0.02152 + } + ] + }, + { + "step": 26260, + "tok": 107560960, + "ce": 3.4202, + "ppl": 30.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4178, + "loop_stats": [ + { + "tau": 0.0928, + "v": 0.9131, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.8808, + "H_a": 0.056, + "d_bind": 0.03169 + }, + { + "tau": 0.0197, + "v": 0.8738, + "H_a": 0.05, + "d_bind": 0.01804 + } + ] + }, + { + "step": 26280, + "tok": 107642880, + "ce": 2.8531, + "ppl": 17.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4156, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.8877, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0479, + "v": 0.8468, + "H_a": 0.065, + "d_bind": 0.03411 + }, + { + "tau": 0.0296, + "v": 0.8289, + "H_a": 0.054, + "d_bind": 0.02022 + } + ] + }, + { + "step": 26300, + "tok": 107724800, + "ce": 3.0879, + "ppl": 21.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4144, + "loop_stats": [ + { + "tau": 0.101, + "v": 0.8819, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.8385, + "H_a": 0.068, + "d_bind": 0.03454 + }, + { + "tau": 0.0289, + "v": 0.8205, + "H_a": 0.053, + "d_bind": 0.02049 + } + ] + }, + { + "step": 26320, + "tok": 107806720, + "ce": 2.6802, + "ppl": 14.6, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.935, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4166, + "loop_stats": [ + { + "tau": 0.1052, + "v": 0.8932, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0317, + "v": 0.8515, + "H_a": 0.065, + "d_bind": 0.0352 + }, + { + "tau": 0.0195, + "v": 0.8388, + "H_a": 0.058, + "d_bind": 0.02432 + } + ] + }, + { + "step": 26340, + "tok": 107888640, + "ce": 2.5427, + "ppl": 12.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4168, + "loop_stats": [ + { + "tau": 0.1396, + "v": 0.8876, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0585, + "v": 0.834, + "H_a": 0.057, + "d_bind": 0.03499 + }, + { + "tau": 0.0419, + "v": 0.8246, + "H_a": 0.051, + "d_bind": 0.02096 + } + ] + }, + { + "step": 26360, + "tok": 107970560, + "ce": 2.5043, + "ppl": 12.2, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.935, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4166, + "loop_stats": [ + { + "tau": 0.12, + "v": 0.861, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0528, + "v": 0.8249, + "H_a": 0.066, + "d_bind": 0.03641 + }, + { + "tau": 0.036, + "v": 0.817, + "H_a": 0.058, + "d_bind": 0.02287 + } + ] + }, + { + "step": 26380, + "tok": 108052480, + "ce": 2.5787, + "ppl": 13.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.96, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4164, + "loop_stats": [ + { + "tau": 0.1239, + "v": 0.8838, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.84, + "H_a": 0.065, + "d_bind": 0.0378 + }, + { + "tau": 0.0359, + "v": 0.8277, + "H_a": 0.057, + "d_bind": 0.02156 + } + ] + }, + { + "step": 26400, + "tok": 108134400, + "ce": 2.7227, + "ppl": 15.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22522, + "peak_MiB": 5301, + "v_gain": 0.4151, + "loop_stats": [ + { + "tau": 0.1126, + "v": 0.889, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.8465, + "H_a": 0.058, + "d_bind": 0.03723 + }, + { + "tau": 0.037, + "v": 0.8374, + "H_a": 0.046, + "d_bind": 0.01826 + } + ] + }, + { + "step": 26420, + "tok": 108216320, + "ce": 2.565, + "ppl": 13.0, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.933, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4151, + "loop_stats": [ + { + "tau": 0.1419, + "v": 0.9015, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0706, + "v": 0.8477, + "H_a": 0.053, + "d_bind": 0.03634 + }, + { + "tau": 0.0548, + "v": 0.8329, + "H_a": 0.044, + "d_bind": 0.01823 + } + ] + }, + { + "step": 26440, + "tok": 108298240, + "ce": 2.898, + "ppl": 18.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4155, + "loop_stats": [ + { + "tau": 0.1025, + "v": 0.8864, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8383, + "H_a": 0.061, + "d_bind": 0.03371 + }, + { + "tau": 0.0253, + "v": 0.8204, + "H_a": 0.055, + "d_bind": 0.02098 + } + ] + }, + { + "step": 26460, + "tok": 108380160, + "ce": 2.8969, + "ppl": 18.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4135, + "loop_stats": [ + { + "tau": 0.0894, + "v": 0.8829, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8424, + "H_a": 0.06, + "d_bind": 0.03724 + }, + { + "tau": 0.0296, + "v": 0.8275, + "H_a": 0.046, + "d_bind": 0.01982 + } + ] + }, + { + "step": 26480, + "tok": 108462080, + "ce": 2.5789, + "ppl": 13.2, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.935, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4158, + "loop_stats": [ + { + "tau": 0.1355, + "v": 0.8981, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8513, + "H_a": 0.062, + "d_bind": 0.03569 + }, + { + "tau": 0.0368, + "v": 0.8396, + "H_a": 0.051, + "d_bind": 0.02045 + } + ] + }, + { + "step": 26500, + "tok": 108544000, + "ce": 3.0256, + "ppl": 20.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.927, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4197, + "loop_stats": [ + { + "tau": 0.1019, + "v": 0.8985, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0362, + "v": 0.8684, + "H_a": 0.056, + "d_bind": 0.03212 + }, + { + "tau": 0.0228, + "v": 0.8543, + "H_a": 0.048, + "d_bind": 0.01994 + } + ] + }, + { + "step": 26520, + "tok": 108625920, + "ce": 3.0881, + "ppl": 21.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.42, + "loop_stats": [ + { + "tau": 0.0883, + "v": 0.9125, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0263, + "v": 0.884, + "H_a": 0.062, + "d_bind": 0.03351 + }, + { + "tau": 0.0148, + "v": 0.8695, + "H_a": 0.058, + "d_bind": 0.02126 + } + ] + }, + { + "step": 26540, + "tok": 108707840, + "ce": 2.9314, + "ppl": 18.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4189, + "loop_stats": [ + { + "tau": 0.1119, + "v": 0.8744, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0562, + "v": 0.8384, + "H_a": 0.062, + "d_bind": 0.0356 + }, + { + "tau": 0.0411, + "v": 0.8266, + "H_a": 0.049, + "d_bind": 0.02035 + } + ] + }, + { + "step": 26560, + "tok": 108789760, + "ce": 2.4598, + "ppl": 11.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.925, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4216, + "loop_stats": [ + { + "tau": 0.1344, + "v": 0.8535, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8204, + "H_a": 0.063, + "d_bind": 0.03624 + }, + { + "tau": 0.0365, + "v": 0.8082, + "H_a": 0.053, + "d_bind": 0.02122 + } + ] + }, + { + "step": 26580, + "tok": 108871680, + "ce": 2.7198, + "ppl": 15.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4233, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.8754, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0428, + "v": 0.8308, + "H_a": 0.064, + "d_bind": 0.03985 + }, + { + "tau": 0.0368, + "v": 0.8214, + "H_a": 0.057, + "d_bind": 0.02075 + } + ] + }, + { + "step": 26600, + "tok": 108953600, + "ce": 2.6546, + "ppl": 14.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4223, + "loop_stats": [ + { + "tau": 0.1405, + "v": 0.8727, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.8353, + "H_a": 0.063, + "d_bind": 0.03816 + }, + { + "tau": 0.0423, + "v": 0.8264, + "H_a": 0.055, + "d_bind": 0.02074 + } + ] + }, + { + "step": 26620, + "tok": 109035520, + "ce": 2.8521, + "ppl": 17.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4211, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.876, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0571, + "v": 0.8227, + "H_a": 0.063, + "d_bind": 0.03448 + }, + { + "tau": 0.044, + "v": 0.8055, + "H_a": 0.053, + "d_bind": 0.02126 + } + ] + }, + { + "step": 26640, + "tok": 109117440, + "ce": 2.7079, + "ppl": 15.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4192, + "loop_stats": [ + { + "tau": 0.1371, + "v": 0.8433, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0566, + "v": 0.8112, + "H_a": 0.073, + "d_bind": 0.03549 + }, + { + "tau": 0.0468, + "v": 0.7959, + "H_a": 0.056, + "d_bind": 0.02594 + } + ] + }, + { + "step": 26660, + "tok": 109199360, + "ce": 2.44, + "ppl": 11.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.421, + "loop_stats": [ + { + "tau": 0.1401, + "v": 0.8478, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0605, + "v": 0.8178, + "H_a": 0.065, + "d_bind": 0.03658 + }, + { + "tau": 0.0486, + "v": 0.8061, + "H_a": 0.055, + "d_bind": 0.02287 + } + ] + }, + { + "step": 26680, + "tok": 109281280, + "ce": 2.6967, + "ppl": 14.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4249, + "loop_stats": [ + { + "tau": 0.1292, + "v": 0.8674, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8361, + "H_a": 0.067, + "d_bind": 0.03545 + }, + { + "tau": 0.0321, + "v": 0.8278, + "H_a": 0.058, + "d_bind": 0.02295 + } + ] + }, + { + "step": 26700, + "tok": 109363200, + "ce": 2.7989, + "ppl": 16.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.927, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4268, + "loop_stats": [ + { + "tau": 0.1164, + "v": 0.8771, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8442, + "H_a": 0.07, + "d_bind": 0.03671 + }, + { + "tau": 0.0253, + "v": 0.8295, + "H_a": 0.056, + "d_bind": 0.02176 + } + ] + }, + { + "step": 26720, + "tok": 109445120, + "ce": 2.5071, + "ppl": 12.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4255, + "loop_stats": [ + { + "tau": 0.1414, + "v": 0.8334, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.055, + "v": 0.7968, + "H_a": 0.067, + "d_bind": 0.03734 + }, + { + "tau": 0.0464, + "v": 0.7796, + "H_a": 0.054, + "d_bind": 0.02212 + } + ] + }, + { + "step": 26740, + "tok": 109527040, + "ce": 2.4639, + "ppl": 11.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4252, + "loop_stats": [ + { + "tau": 0.1342, + "v": 0.8633, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0602, + "v": 0.8495, + "H_a": 0.063, + "d_bind": 0.03567 + }, + { + "tau": 0.0451, + "v": 0.8405, + "H_a": 0.054, + "d_bind": 0.02275 + } + ] + }, + { + "step": 26760, + "tok": 109608960, + "ce": 3.0453, + "ppl": 21.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.938, + "tok_s": 22523, + "peak_MiB": 5301, + "v_gain": 0.4241, + "loop_stats": [ + { + "tau": 0.0921, + "v": 0.8752, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0472, + "v": 0.8359, + "H_a": 0.059, + "d_bind": 0.03106 + }, + { + "tau": 0.0291, + "v": 0.8146, + "H_a": 0.048, + "d_bind": 0.01859 + } + ] + }, + { + "step": 26780, + "tok": 109690880, + "ce": 3.2044, + "ppl": 24.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.938, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4248, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.8904, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8532, + "H_a": 0.064, + "d_bind": 0.03488 + }, + { + "tau": 0.0319, + "v": 0.8424, + "H_a": 0.055, + "d_bind": 0.0209 + } + ] + }, + { + "step": 26800, + "tok": 109772800, + "ce": 3.0451, + "ppl": 21.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4267, + "loop_stats": [ + { + "tau": 0.0863, + "v": 0.8993, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0434, + "v": 0.8683, + "H_a": 0.058, + "d_bind": 0.03255 + }, + { + "tau": 0.0286, + "v": 0.8494, + "H_a": 0.05, + "d_bind": 0.0191 + } + ] + }, + { + "step": 26820, + "tok": 109854720, + "ce": 2.86, + "ppl": 17.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4275, + "loop_stats": [ + { + "tau": 0.1196, + "v": 0.8927, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0469, + "v": 0.8446, + "H_a": 0.063, + "d_bind": 0.03435 + }, + { + "tau": 0.0316, + "v": 0.8215, + "H_a": 0.055, + "d_bind": 0.02085 + } + ] + }, + { + "step": 26840, + "tok": 109936640, + "ce": 2.8016, + "ppl": 16.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4286, + "loop_stats": [ + { + "tau": 0.111, + "v": 0.8968, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0456, + "v": 0.852, + "H_a": 0.059, + "d_bind": 0.03342 + }, + { + "tau": 0.0323, + "v": 0.835, + "H_a": 0.053, + "d_bind": 0.02091 + } + ] + }, + { + "step": 26860, + "tok": 110018560, + "ce": 2.9186, + "ppl": 18.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4291, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.8885, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.836, + "H_a": 0.065, + "d_bind": 0.03319 + }, + { + "tau": 0.0289, + "v": 0.8188, + "H_a": 0.051, + "d_bind": 0.02102 + } + ] + }, + { + "step": 26880, + "tok": 110100480, + "ce": 2.6298, + "ppl": 13.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4318, + "loop_stats": [ + { + "tau": 0.1226, + "v": 0.9055, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0423, + "v": 0.8734, + "H_a": 0.063, + "d_bind": 0.03426 + }, + { + "tau": 0.0301, + "v": 0.8636, + "H_a": 0.054, + "d_bind": 0.02228 + } + ] + }, + { + "step": 26900, + "tok": 110182400, + "ce": 2.8133, + "ppl": 16.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4332, + "loop_stats": [ + { + "tau": 0.1163, + "v": 0.8938, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8565, + "H_a": 0.064, + "d_bind": 0.03267 + }, + { + "tau": 0.0272, + "v": 0.8376, + "H_a": 0.051, + "d_bind": 0.02041 + } + ] + }, + { + "step": 26920, + "tok": 110264320, + "ce": 3.0373, + "ppl": 20.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4317, + "loop_stats": [ + { + "tau": 0.0848, + "v": 0.8945, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0312, + "v": 0.8743, + "H_a": 0.063, + "d_bind": 0.03399 + }, + { + "tau": 0.0236, + "v": 0.8649, + "H_a": 0.047, + "d_bind": 0.01904 + } + ] + }, + { + "step": 26940, + "tok": 110346240, + "ce": 2.5448, + "ppl": 12.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4298, + "loop_stats": [ + { + "tau": 0.121, + "v": 0.8876, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.8652, + "H_a": 0.062, + "d_bind": 0.03427 + }, + { + "tau": 0.036, + "v": 0.8593, + "H_a": 0.051, + "d_bind": 0.02215 + } + ] + }, + { + "step": 26960, + "tok": 110428160, + "ce": 2.9132, + "ppl": 18.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4295, + "loop_stats": [ + { + "tau": 0.1047, + "v": 0.8979, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0315, + "v": 0.8732, + "H_a": 0.069, + "d_bind": 0.0348 + }, + { + "tau": 0.0259, + "v": 0.8619, + "H_a": 0.052, + "d_bind": 0.02149 + } + ] + }, + { + "step": 26980, + "tok": 110510080, + "ce": 3.0597, + "ppl": 21.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4275, + "loop_stats": [ + { + "tau": 0.0961, + "v": 0.9037, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8588, + "H_a": 0.059, + "d_bind": 0.03164 + }, + { + "tau": 0.0311, + "v": 0.8435, + "H_a": 0.056, + "d_bind": 0.01867 + } + ] + }, + { + "step": 27000, + "tok": 110592000, + "ce": 2.8802, + "ppl": 17.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22524, + "peak_MiB": 5301, + "v_gain": 0.4273, + "loop_stats": [ + { + "tau": 0.1086, + "v": 0.9146, + "H_a": 0.082, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.8824, + "H_a": 0.067, + "d_bind": 0.03518 + }, + { + "tau": 0.0326, + "v": 0.8725, + "H_a": 0.053, + "d_bind": 0.02041 + } + ] + }, + { + "step": 27020, + "tok": 110673920, + "ce": 3.2786, + "ppl": 26.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.93, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4296, + "loop_stats": [ + { + "tau": 0.0792, + "v": 0.9046, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0291, + "v": 0.8708, + "H_a": 0.057, + "d_bind": 0.03222 + }, + { + "tau": 0.019, + "v": 0.8559, + "H_a": 0.053, + "d_bind": 0.01982 + } + ] + }, + { + "step": 27040, + "tok": 110755840, + "ce": 2.8041, + "ppl": 16.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4297, + "loop_stats": [ + { + "tau": 0.126, + "v": 0.9035, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0466, + "v": 0.8675, + "H_a": 0.062, + "d_bind": 0.03466 + }, + { + "tau": 0.0356, + "v": 0.8445, + "H_a": 0.054, + "d_bind": 0.0205 + } + ] + }, + { + "step": 27060, + "tok": 110837760, + "ce": 3.023, + "ppl": 20.6, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4353, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.9015, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.8632, + "H_a": 0.057, + "d_bind": 0.03292 + }, + { + "tau": 0.0227, + "v": 0.841, + "H_a": 0.052, + "d_bind": 0.02005 + } + ] + }, + { + "step": 27080, + "tok": 110919680, + "ce": 2.5554, + "ppl": 12.9, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.952, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4381, + "loop_stats": [ + { + "tau": 0.1219, + "v": 0.8418, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0472, + "v": 0.81, + "H_a": 0.058, + "d_bind": 0.0363 + }, + { + "tau": 0.0384, + "v": 0.797, + "H_a": 0.047, + "d_bind": 0.01986 + } + ] + }, + { + "step": 27100, + "tok": 111001600, + "ce": 2.45, + "ppl": 11.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.97, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4382, + "loop_stats": [ + { + "tau": 0.1153, + "v": 0.865, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8259, + "H_a": 0.061, + "d_bind": 0.03535 + }, + { + "tau": 0.0281, + "v": 0.8112, + "H_a": 0.052, + "d_bind": 0.0202 + } + ] + }, + { + "step": 27120, + "tok": 111083520, + "ce": 2.9777, + "ppl": 19.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4368, + "loop_stats": [ + { + "tau": 0.1208, + "v": 0.9077, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.8741, + "H_a": 0.066, + "d_bind": 0.03163 + }, + { + "tau": 0.0209, + "v": 0.8613, + "H_a": 0.057, + "d_bind": 0.02119 + } + ] + }, + { + "step": 27140, + "tok": 111165440, + "ce": 2.6624, + "ppl": 14.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.957, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.435, + "loop_stats": [ + { + "tau": 0.1211, + "v": 0.8678, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0524, + "v": 0.8316, + "H_a": 0.059, + "d_bind": 0.03627 + }, + { + "tau": 0.037, + "v": 0.8199, + "H_a": 0.052, + "d_bind": 0.02015 + } + ] + }, + { + "step": 27160, + "tok": 111247360, + "ce": 2.5846, + "ppl": 13.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4333, + "loop_stats": [ + { + "tau": 0.1306, + "v": 0.8734, + "H_a": 0.086, + "d_bind": 0.0 + }, + { + "tau": 0.0515, + "v": 0.8548, + "H_a": 0.061, + "d_bind": 0.03433 + }, + { + "tau": 0.0336, + "v": 0.8484, + "H_a": 0.054, + "d_bind": 0.02224 + } + ] + }, + { + "step": 27180, + "tok": 111329280, + "ce": 2.8446, + "ppl": 17.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.96, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4323, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.8812, + "H_a": 0.089, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8391, + "H_a": 0.061, + "d_bind": 0.03416 + }, + { + "tau": 0.0271, + "v": 0.8256, + "H_a": 0.054, + "d_bind": 0.02099 + } + ] + }, + { + "step": 27200, + "tok": 111411200, + "ce": 2.7986, + "ppl": 16.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4305, + "loop_stats": [ + { + "tau": 0.1179, + "v": 0.8699, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8305, + "H_a": 0.061, + "d_bind": 0.03654 + }, + { + "tau": 0.0341, + "v": 0.8179, + "H_a": 0.049, + "d_bind": 0.01914 + } + ] + }, + { + "step": 27220, + "tok": 111493120, + "ce": 2.9448, + "ppl": 19.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4318, + "loop_stats": [ + { + "tau": 0.1043, + "v": 0.8859, + "H_a": 0.085, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8587, + "H_a": 0.054, + "d_bind": 0.03199 + }, + { + "tau": 0.0227, + "v": 0.844, + "H_a": 0.048, + "d_bind": 0.01892 + } + ] + }, + { + "step": 27240, + "tok": 111575040, + "ce": 2.8004, + "ppl": 16.5, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.952, + "tok_s": 22525, + "peak_MiB": 5301, + "v_gain": 0.4332, + "loop_stats": [ + { + "tau": 0.091, + "v": 0.9035, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8681, + "H_a": 0.063, + "d_bind": 0.03221 + }, + { + "tau": 0.0155, + "v": 0.8593, + "H_a": 0.05, + "d_bind": 0.02097 + } + ] + }, + { + "step": 27260, + "tok": 111656960, + "ce": 3.136, + "ppl": 23.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4344, + "loop_stats": [ + { + "tau": 0.0944, + "v": 0.8734, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8371, + "H_a": 0.062, + "d_bind": 0.03403 + }, + { + "tau": 0.0244, + "v": 0.8262, + "H_a": 0.054, + "d_bind": 0.02059 + } + ] + }, + { + "step": 27280, + "tok": 111738880, + "ce": 2.5785, + "ppl": 13.2, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.942, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4374, + "loop_stats": [ + { + "tau": 0.125, + "v": 0.9144, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0416, + "v": 0.8733, + "H_a": 0.055, + "d_bind": 0.03551 + }, + { + "tau": 0.0257, + "v": 0.86, + "H_a": 0.045, + "d_bind": 0.01957 + } + ] + }, + { + "step": 27300, + "tok": 111820800, + "ce": 2.4205, + "ppl": 11.3, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.95, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4376, + "loop_stats": [ + { + "tau": 0.1487, + "v": 0.873, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0427, + "v": 0.8291, + "H_a": 0.061, + "d_bind": 0.03838 + }, + { + "tau": 0.0262, + "v": 0.8171, + "H_a": 0.053, + "d_bind": 0.01985 + } + ] + }, + { + "step": 27320, + "tok": 111902720, + "ce": 2.6274, + "ppl": 13.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4328, + "loop_stats": [ + { + "tau": 0.1325, + "v": 0.9191, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0513, + "v": 0.8816, + "H_a": 0.056, + "d_bind": 0.03523 + }, + { + "tau": 0.0308, + "v": 0.8701, + "H_a": 0.045, + "d_bind": 0.01867 + } + ] + }, + { + "step": 27340, + "tok": 111984640, + "ce": 2.7128, + "ppl": 15.1, + "pref_transfer_pron": 0.94, + "pref_train_sv": 0.952, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4316, + "loop_stats": [ + { + "tau": 0.1252, + "v": 0.8837, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0461, + "v": 0.8437, + "H_a": 0.052, + "d_bind": 0.03303 + }, + { + "tau": 0.0377, + "v": 0.8313, + "H_a": 0.043, + "d_bind": 0.02067 + } + ] + }, + { + "step": 27360, + "tok": 112066560, + "ce": 2.9913, + "ppl": 19.9, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.952, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4308, + "loop_stats": [ + { + "tau": 0.1131, + "v": 0.9251, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0475, + "v": 0.8833, + "H_a": 0.051, + "d_bind": 0.03434 + }, + { + "tau": 0.0256, + "v": 0.8641, + "H_a": 0.042, + "d_bind": 0.01719 + } + ] + }, + { + "step": 27380, + "tok": 112148480, + "ce": 2.7646, + "ppl": 15.9, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.942, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4325, + "loop_stats": [ + { + "tau": 0.1216, + "v": 0.9119, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8699, + "H_a": 0.056, + "d_bind": 0.0326 + }, + { + "tau": 0.0375, + "v": 0.8562, + "H_a": 0.044, + "d_bind": 0.01966 + } + ] + }, + { + "step": 27400, + "tok": 112230400, + "ce": 2.4882, + "ppl": 12.0, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.957, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4325, + "loop_stats": [ + { + "tau": 0.1213, + "v": 0.8724, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0475, + "v": 0.8238, + "H_a": 0.059, + "d_bind": 0.03335 + }, + { + "tau": 0.0356, + "v": 0.811, + "H_a": 0.051, + "d_bind": 0.02203 + } + ] + }, + { + "step": 27420, + "tok": 112312320, + "ce": 3.1696, + "ppl": 23.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4315, + "loop_stats": [ + { + "tau": 0.0932, + "v": 0.907, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0272, + "v": 0.8622, + "H_a": 0.067, + "d_bind": 0.03305 + }, + { + "tau": 0.0193, + "v": 0.8441, + "H_a": 0.053, + "d_bind": 0.02045 + } + ] + }, + { + "step": 27440, + "tok": 112394240, + "ce": 2.9194, + "ppl": 18.5, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.957, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.429, + "loop_stats": [ + { + "tau": 0.0895, + "v": 0.9321, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0413, + "v": 0.895, + "H_a": 0.059, + "d_bind": 0.03171 + }, + { + "tau": 0.0303, + "v": 0.8767, + "H_a": 0.048, + "d_bind": 0.01793 + } + ] + }, + { + "step": 27460, + "tok": 112476160, + "ce": 3.2636, + "ppl": 26.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4291, + "loop_stats": [ + { + "tau": 0.0805, + "v": 0.8653, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0309, + "v": 0.8214, + "H_a": 0.063, + "d_bind": 0.03395 + }, + { + "tau": 0.029, + "v": 0.8073, + "H_a": 0.053, + "d_bind": 0.01969 + } + ] + }, + { + "step": 27480, + "tok": 112558080, + "ce": 2.8389, + "ppl": 17.1, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.957, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4293, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8835, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0402, + "v": 0.8327, + "H_a": 0.064, + "d_bind": 0.03666 + }, + { + "tau": 0.0324, + "v": 0.8188, + "H_a": 0.051, + "d_bind": 0.02044 + } + ] + }, + { + "step": 27500, + "tok": 112640000, + "ce": 2.4523, + "ppl": 11.6, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.965, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4322, + "loop_stats": [ + { + "tau": 0.1116, + "v": 0.9006, + "H_a": 0.084, + "d_bind": 0.0 + }, + { + "tau": 0.0397, + "v": 0.8545, + "H_a": 0.068, + "d_bind": 0.03424 + }, + { + "tau": 0.0309, + "v": 0.8383, + "H_a": 0.055, + "d_bind": 0.02131 + } + ] + }, + { + "step": 27520, + "tok": 112721920, + "ce": 2.9971, + "ppl": 20.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.945, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4326, + "loop_stats": [ + { + "tau": 0.0886, + "v": 0.8942, + "H_a": 0.079, + "d_bind": 0.0 + }, + { + "tau": 0.0319, + "v": 0.8457, + "H_a": 0.062, + "d_bind": 0.03445 + }, + { + "tau": 0.0268, + "v": 0.8257, + "H_a": 0.05, + "d_bind": 0.01946 + } + ] + }, + { + "step": 27540, + "tok": 112803840, + "ce": 2.7743, + "ppl": 16.0, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.955, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4327, + "loop_stats": [ + { + "tau": 0.1211, + "v": 0.8929, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0453, + "v": 0.8527, + "H_a": 0.063, + "d_bind": 0.03559 + }, + { + "tau": 0.0342, + "v": 0.8418, + "H_a": 0.052, + "d_bind": 0.01946 + } + ] + }, + { + "step": 27560, + "tok": 112885760, + "ce": 3.2203, + "ppl": 25.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.957, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4336, + "loop_stats": [ + { + "tau": 0.0863, + "v": 0.8986, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0284, + "v": 0.8569, + "H_a": 0.065, + "d_bind": 0.03308 + }, + { + "tau": 0.0133, + "v": 0.8381, + "H_a": 0.043, + "d_bind": 0.01873 + } + ] + }, + { + "step": 27580, + "tok": 112967680, + "ce": 2.8412, + "ppl": 17.1, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.952, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4307, + "loop_stats": [ + { + "tau": 0.1131, + "v": 0.9265, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0424, + "v": 0.8776, + "H_a": 0.054, + "d_bind": 0.03285 + }, + { + "tau": 0.0296, + "v": 0.8596, + "H_a": 0.043, + "d_bind": 0.01847 + } + ] + }, + { + "step": 27600, + "tok": 113049600, + "ce": 2.9883, + "ppl": 19.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22526, + "peak_MiB": 5301, + "v_gain": 0.4314, + "loop_stats": [ + { + "tau": 0.0918, + "v": 0.9221, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8947, + "H_a": 0.063, + "d_bind": 0.03194 + }, + { + "tau": 0.0215, + "v": 0.882, + "H_a": 0.05, + "d_bind": 0.02052 + } + ] + }, + { + "step": 27620, + "tok": 113131520, + "ce": 3.2208, + "ppl": 25.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4331, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.902, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8614, + "H_a": 0.066, + "d_bind": 0.03125 + }, + { + "tau": 0.0214, + "v": 0.8429, + "H_a": 0.051, + "d_bind": 0.02055 + } + ] + }, + { + "step": 27640, + "tok": 113213440, + "ce": 2.6519, + "ppl": 14.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.942, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4365, + "loop_stats": [ + { + "tau": 0.1096, + "v": 0.8978, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0477, + "v": 0.8498, + "H_a": 0.06, + "d_bind": 0.03346 + }, + { + "tau": 0.0308, + "v": 0.8353, + "H_a": 0.049, + "d_bind": 0.02012 + } + ] + }, + { + "step": 27660, + "tok": 113295360, + "ce": 2.8908, + "ppl": 18.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4386, + "loop_stats": [ + { + "tau": 0.1039, + "v": 0.8987, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0318, + "v": 0.8527, + "H_a": 0.06, + "d_bind": 0.0332 + }, + { + "tau": 0.0222, + "v": 0.8359, + "H_a": 0.047, + "d_bind": 0.01906 + } + ] + }, + { + "step": 27680, + "tok": 113377280, + "ce": 3.0802, + "ppl": 21.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4366, + "loop_stats": [ + { + "tau": 0.0986, + "v": 0.9292, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8903, + "H_a": 0.059, + "d_bind": 0.03265 + }, + { + "tau": 0.0254, + "v": 0.8797, + "H_a": 0.049, + "d_bind": 0.01871 + } + ] + }, + { + "step": 27700, + "tok": 113459200, + "ce": 2.7718, + "ppl": 16.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4363, + "loop_stats": [ + { + "tau": 0.1203, + "v": 0.9081, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8766, + "H_a": 0.055, + "d_bind": 0.0346 + }, + { + "tau": 0.0265, + "v": 0.8684, + "H_a": 0.042, + "d_bind": 0.01895 + } + ] + }, + { + "step": 27720, + "tok": 113541120, + "ce": 2.9847, + "ppl": 19.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.962, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4371, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.9239, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.8832, + "H_a": 0.054, + "d_bind": 0.03148 + }, + { + "tau": 0.0108, + "v": 0.8583, + "H_a": 0.051, + "d_bind": 0.01867 + } + ] + }, + { + "step": 27740, + "tok": 113623040, + "ce": 2.6613, + "ppl": 14.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4385, + "loop_stats": [ + { + "tau": 0.1312, + "v": 0.8851, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8539, + "H_a": 0.055, + "d_bind": 0.03211 + }, + { + "tau": 0.0278, + "v": 0.8434, + "H_a": 0.042, + "d_bind": 0.01984 + } + ] + }, + { + "step": 27760, + "tok": 113704960, + "ce": 2.7331, + "ppl": 15.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.437, + "loop_stats": [ + { + "tau": 0.1216, + "v": 0.9021, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.8588, + "H_a": 0.059, + "d_bind": 0.03549 + }, + { + "tau": 0.033, + "v": 0.8511, + "H_a": 0.046, + "d_bind": 0.01901 + } + ] + }, + { + "step": 27780, + "tok": 113786880, + "ce": 3.0764, + "ppl": 21.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.933, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4369, + "loop_stats": [ + { + "tau": 0.1203, + "v": 0.9058, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.037, + "v": 0.8605, + "H_a": 0.053, + "d_bind": 0.02955 + }, + { + "tau": 0.0244, + "v": 0.8456, + "H_a": 0.048, + "d_bind": 0.01846 + } + ] + }, + { + "step": 27800, + "tok": 113868800, + "ce": 2.7206, + "ppl": 15.2, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.93, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.435, + "loop_stats": [ + { + "tau": 0.1204, + "v": 0.9229, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8881, + "H_a": 0.06, + "d_bind": 0.03274 + }, + { + "tau": 0.0311, + "v": 0.8773, + "H_a": 0.045, + "d_bind": 0.02019 + } + ] + }, + { + "step": 27820, + "tok": 113950720, + "ce": 2.9398, + "ppl": 18.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4349, + "loop_stats": [ + { + "tau": 0.1143, + "v": 0.9181, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0429, + "v": 0.8719, + "H_a": 0.057, + "d_bind": 0.03332 + }, + { + "tau": 0.0255, + "v": 0.8535, + "H_a": 0.049, + "d_bind": 0.01889 + } + ] + }, + { + "step": 27840, + "tok": 114032640, + "ce": 3.2037, + "ppl": 24.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4345, + "loop_stats": [ + { + "tau": 0.0748, + "v": 0.9165, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.028, + "v": 0.8715, + "H_a": 0.057, + "d_bind": 0.03228 + }, + { + "tau": 0.0187, + "v": 0.8518, + "H_a": 0.048, + "d_bind": 0.01984 + } + ] + }, + { + "step": 27860, + "tok": 114114560, + "ce": 2.7579, + "ppl": 15.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.965, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4339, + "loop_stats": [ + { + "tau": 0.1118, + "v": 0.8895, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8455, + "H_a": 0.059, + "d_bind": 0.03546 + }, + { + "tau": 0.0305, + "v": 0.8303, + "H_a": 0.048, + "d_bind": 0.01853 + } + ] + }, + { + "step": 27880, + "tok": 114196480, + "ce": 3.1711, + "ppl": 23.8, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.96, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.433, + "loop_stats": [ + { + "tau": 0.1006, + "v": 0.9045, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0427, + "v": 0.8538, + "H_a": 0.05, + "d_bind": 0.0323 + }, + { + "tau": 0.0311, + "v": 0.8335, + "H_a": 0.044, + "d_bind": 0.01706 + } + ] + }, + { + "step": 27900, + "tok": 114278400, + "ce": 2.8632, + "ppl": 17.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4365, + "loop_stats": [ + { + "tau": 0.0995, + "v": 0.9131, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.8704, + "H_a": 0.057, + "d_bind": 0.03226 + }, + { + "tau": 0.0167, + "v": 0.8507, + "H_a": 0.052, + "d_bind": 0.01922 + } + ] + }, + { + "step": 27920, + "tok": 114360320, + "ce": 2.432, + "ppl": 11.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.935, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4384, + "loop_stats": [ + { + "tau": 0.1329, + "v": 0.8755, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.05, + "v": 0.8186, + "H_a": 0.067, + "d_bind": 0.03797 + }, + { + "tau": 0.0431, + "v": 0.8051, + "H_a": 0.05, + "d_bind": 0.02177 + } + ] + }, + { + "step": 27940, + "tok": 114442240, + "ce": 2.6854, + "ppl": 14.7, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.933, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4384, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.9122, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.046, + "v": 0.8561, + "H_a": 0.055, + "d_bind": 0.03262 + }, + { + "tau": 0.0352, + "v": 0.8338, + "H_a": 0.046, + "d_bind": 0.0203 + } + ] + }, + { + "step": 27960, + "tok": 114524160, + "ce": 2.9939, + "ppl": 20.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22527, + "peak_MiB": 5301, + "v_gain": 0.4425, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.9108, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0331, + "v": 0.8683, + "H_a": 0.06, + "d_bind": 0.03211 + }, + { + "tau": 0.0228, + "v": 0.853, + "H_a": 0.049, + "d_bind": 0.01904 + } + ] + }, + { + "step": 27980, + "tok": 114606080, + "ce": 2.6911, + "ppl": 14.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.94, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4424, + "loop_stats": [ + { + "tau": 0.124, + "v": 0.8803, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0493, + "v": 0.8328, + "H_a": 0.061, + "d_bind": 0.03542 + }, + { + "tau": 0.0403, + "v": 0.8221, + "H_a": 0.049, + "d_bind": 0.02073 + } + ] + }, + { + "step": 28000, + "tok": 114688000, + "ce": 2.5267, + "ppl": 12.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.935, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4402, + "loop_stats": [ + { + "tau": 0.1347, + "v": 0.9086, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0611, + "v": 0.8545, + "H_a": 0.06, + "d_bind": 0.03365 + }, + { + "tau": 0.0436, + "v": 0.8397, + "H_a": 0.051, + "d_bind": 0.02136 + } + ] + }, + { + "step": 28020, + "tok": 114769920, + "ce": 2.4976, + "ppl": 12.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4398, + "loop_stats": [ + { + "tau": 0.1397, + "v": 0.8751, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0529, + "v": 0.8416, + "H_a": 0.063, + "d_bind": 0.03444 + }, + { + "tau": 0.0476, + "v": 0.8322, + "H_a": 0.049, + "d_bind": 0.02165 + } + ] + }, + { + "step": 28040, + "tok": 114851840, + "ce": 2.5499, + "ppl": 12.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4394, + "loop_stats": [ + { + "tau": 0.1329, + "v": 0.8897, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0544, + "v": 0.8509, + "H_a": 0.06, + "d_bind": 0.0349 + }, + { + "tau": 0.0408, + "v": 0.8392, + "H_a": 0.043, + "d_bind": 0.01994 + } + ] + }, + { + "step": 28060, + "tok": 114933760, + "ce": 2.7621, + "ppl": 15.8, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.957, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4377, + "loop_stats": [ + { + "tau": 0.1311, + "v": 0.8925, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0518, + "v": 0.8656, + "H_a": 0.057, + "d_bind": 0.02917 + }, + { + "tau": 0.0328, + "v": 0.8537, + "H_a": 0.05, + "d_bind": 0.02236 + } + ] + }, + { + "step": 28080, + "tok": 115015680, + "ce": 3.0801, + "ppl": 21.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4369, + "loop_stats": [ + { + "tau": 0.0909, + "v": 0.9018, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8648, + "H_a": 0.06, + "d_bind": 0.03168 + }, + { + "tau": 0.0182, + "v": 0.8453, + "H_a": 0.048, + "d_bind": 0.02021 + } + ] + }, + { + "step": 28100, + "tok": 115097600, + "ce": 2.3517, + "ppl": 10.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4347, + "loop_stats": [ + { + "tau": 0.152, + "v": 0.8714, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0569, + "v": 0.8419, + "H_a": 0.058, + "d_bind": 0.03485 + }, + { + "tau": 0.045, + "v": 0.8333, + "H_a": 0.046, + "d_bind": 0.02061 + } + ] + }, + { + "step": 28120, + "tok": 115179520, + "ce": 3.0833, + "ppl": 21.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4333, + "loop_stats": [ + { + "tau": 0.0947, + "v": 0.8933, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0416, + "v": 0.8427, + "H_a": 0.058, + "d_bind": 0.03148 + }, + { + "tau": 0.0285, + "v": 0.8275, + "H_a": 0.047, + "d_bind": 0.02034 + } + ] + }, + { + "step": 28140, + "tok": 115261440, + "ce": 2.558, + "ppl": 12.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4363, + "loop_stats": [ + { + "tau": 0.1216, + "v": 0.8798, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0425, + "v": 0.8349, + "H_a": 0.063, + "d_bind": 0.0361 + }, + { + "tau": 0.0332, + "v": 0.8186, + "H_a": 0.05, + "d_bind": 0.02026 + } + ] + }, + { + "step": 28160, + "tok": 115343360, + "ce": 2.9642, + "ppl": 19.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.967, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4364, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.918, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.8853, + "H_a": 0.061, + "d_bind": 0.03468 + }, + { + "tau": 0.0214, + "v": 0.876, + "H_a": 0.052, + "d_bind": 0.01986 + } + ] + }, + { + "step": 28180, + "tok": 115425280, + "ce": 3.0656, + "ppl": 21.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.957, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4369, + "loop_stats": [ + { + "tau": 0.1061, + "v": 0.903, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0401, + "v": 0.8551, + "H_a": 0.061, + "d_bind": 0.03517 + }, + { + "tau": 0.031, + "v": 0.8392, + "H_a": 0.057, + "d_bind": 0.02192 + } + ] + }, + { + "step": 28200, + "tok": 115507200, + "ce": 2.8706, + "ppl": 17.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.965, + "tok_s": 22528, + "peak_MiB": 5301, + "v_gain": 0.4357, + "loop_stats": [ + { + "tau": 0.1099, + "v": 0.8886, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8405, + "H_a": 0.059, + "d_bind": 0.03458 + }, + { + "tau": 0.029, + "v": 0.8305, + "H_a": 0.045, + "d_bind": 0.01878 + } + ] + }, + { + "step": 28220, + "tok": 115589120, + "ce": 3.1566, + "ppl": 23.5, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.952, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.437, + "loop_stats": [ + { + "tau": 0.0906, + "v": 0.8904, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8465, + "H_a": 0.055, + "d_bind": 0.03497 + }, + { + "tau": 0.0197, + "v": 0.8317, + "H_a": 0.046, + "d_bind": 0.01807 + } + ] + }, + { + "step": 28240, + "tok": 115671040, + "ce": 2.5944, + "ppl": 13.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4346, + "loop_stats": [ + { + "tau": 0.131, + "v": 0.8752, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0521, + "v": 0.8249, + "H_a": 0.056, + "d_bind": 0.03525 + }, + { + "tau": 0.0425, + "v": 0.8075, + "H_a": 0.049, + "d_bind": 0.02019 + } + ] + }, + { + "step": 28260, + "tok": 115752960, + "ce": 2.5576, + "ppl": 12.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.96, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.432, + "loop_stats": [ + { + "tau": 0.1256, + "v": 0.8952, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.8511, + "H_a": 0.061, + "d_bind": 0.03487 + }, + { + "tau": 0.0367, + "v": 0.8339, + "H_a": 0.047, + "d_bind": 0.01888 + } + ] + }, + { + "step": 28280, + "tok": 115834880, + "ce": 3.1226, + "ppl": 22.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.947, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4303, + "loop_stats": [ + { + "tau": 0.0868, + "v": 0.8964, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.04, + "v": 0.8583, + "H_a": 0.06, + "d_bind": 0.03172 + }, + { + "tau": 0.0238, + "v": 0.8459, + "H_a": 0.046, + "d_bind": 0.01865 + } + ] + }, + { + "step": 28300, + "tok": 115916800, + "ce": 2.8899, + "ppl": 18.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.927, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4285, + "loop_stats": [ + { + "tau": 0.1022, + "v": 0.9308, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8854, + "H_a": 0.057, + "d_bind": 0.0342 + }, + { + "tau": 0.0289, + "v": 0.86, + "H_a": 0.047, + "d_bind": 0.01811 + } + ] + }, + { + "step": 28320, + "tok": 115998720, + "ce": 2.5214, + "ppl": 12.4, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.933, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4274, + "loop_stats": [ + { + "tau": 0.1281, + "v": 0.8871, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0547, + "v": 0.8172, + "H_a": 0.063, + "d_bind": 0.03777 + }, + { + "tau": 0.0487, + "v": 0.7961, + "H_a": 0.048, + "d_bind": 0.01987 + } + ] + }, + { + "step": 28340, + "tok": 116080640, + "ce": 2.9393, + "ppl": 18.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4269, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.8874, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8382, + "H_a": 0.062, + "d_bind": 0.03387 + }, + { + "tau": 0.031, + "v": 0.8279, + "H_a": 0.048, + "d_bind": 0.02106 + } + ] + }, + { + "step": 28360, + "tok": 116162560, + "ce": 2.9668, + "ppl": 19.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4281, + "loop_stats": [ + { + "tau": 0.0956, + "v": 0.9196, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8823, + "H_a": 0.052, + "d_bind": 0.02869 + }, + { + "tau": 0.0242, + "v": 0.8678, + "H_a": 0.043, + "d_bind": 0.01744 + } + ] + }, + { + "step": 28380, + "tok": 116244480, + "ce": 2.4394, + "ppl": 11.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.952, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4288, + "loop_stats": [ + { + "tau": 0.1453, + "v": 0.8789, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0598, + "v": 0.8399, + "H_a": 0.058, + "d_bind": 0.03207 + }, + { + "tau": 0.0456, + "v": 0.8338, + "H_a": 0.046, + "d_bind": 0.02142 + } + ] + }, + { + "step": 28400, + "tok": 116326400, + "ce": 2.6879, + "ppl": 14.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4285, + "loop_stats": [ + { + "tau": 0.1257, + "v": 0.9, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.8619, + "H_a": 0.055, + "d_bind": 0.03593 + }, + { + "tau": 0.0478, + "v": 0.8612, + "H_a": 0.042, + "d_bind": 0.0185 + } + ] + }, + { + "step": 28420, + "tok": 116408320, + "ce": 2.9352, + "ppl": 18.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4296, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.8768, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0345, + "v": 0.8415, + "H_a": 0.058, + "d_bind": 0.03201 + }, + { + "tau": 0.0247, + "v": 0.8342, + "H_a": 0.047, + "d_bind": 0.01999 + } + ] + }, + { + "step": 28440, + "tok": 116490240, + "ce": 2.6087, + "ppl": 13.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.938, + "tok_s": 22529, + "peak_MiB": 5301, + "v_gain": 0.4309, + "loop_stats": [ + { + "tau": 0.1356, + "v": 0.8847, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0517, + "v": 0.8584, + "H_a": 0.054, + "d_bind": 0.03378 + }, + { + "tau": 0.0336, + "v": 0.8516, + "H_a": 0.046, + "d_bind": 0.02028 + } + ] + }, + { + "step": 28460, + "tok": 116572160, + "ce": 2.6503, + "ppl": 14.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.952, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4303, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.8812, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8582, + "H_a": 0.059, + "d_bind": 0.03545 + }, + { + "tau": 0.0367, + "v": 0.8502, + "H_a": 0.048, + "d_bind": 0.02022 + } + ] + }, + { + "step": 28480, + "tok": 116654080, + "ce": 2.7555, + "ppl": 15.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4349, + "loop_stats": [ + { + "tau": 0.1082, + "v": 0.9085, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8789, + "H_a": 0.058, + "d_bind": 0.03236 + }, + { + "tau": 0.0271, + "v": 0.8692, + "H_a": 0.048, + "d_bind": 0.01909 + } + ] + }, + { + "step": 28500, + "tok": 116736000, + "ce": 3.3158, + "ppl": 27.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4348, + "loop_stats": [ + { + "tau": 0.0894, + "v": 0.8764, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0239, + "v": 0.8337, + "H_a": 0.058, + "d_bind": 0.03221 + }, + { + "tau": 0.016, + "v": 0.8131, + "H_a": 0.046, + "d_bind": 0.02072 + } + ] + }, + { + "step": 28520, + "tok": 116817920, + "ce": 3.1536, + "ppl": 23.4, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.947, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4357, + "loop_stats": [ + { + "tau": 0.0899, + "v": 0.9071, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0239, + "v": 0.8721, + "H_a": 0.058, + "d_bind": 0.03364 + }, + { + "tau": 0.0124, + "v": 0.8525, + "H_a": 0.045, + "d_bind": 0.01992 + } + ] + }, + { + "step": 28540, + "tok": 116899840, + "ce": 3.0494, + "ppl": 21.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4365, + "loop_stats": [ + { + "tau": 0.0823, + "v": 0.8499, + "H_a": 0.083, + "d_bind": 0.0 + }, + { + "tau": 0.0219, + "v": 0.7963, + "H_a": 0.057, + "d_bind": 0.03173 + }, + { + "tau": 0.01, + "v": 0.7728, + "H_a": 0.045, + "d_bind": 0.01752 + } + ] + }, + { + "step": 28560, + "tok": 116981760, + "ce": 2.9144, + "ppl": 18.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4334, + "loop_stats": [ + { + "tau": 0.1088, + "v": 0.9085, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.8673, + "H_a": 0.055, + "d_bind": 0.02976 + }, + { + "tau": 0.0219, + "v": 0.8526, + "H_a": 0.045, + "d_bind": 0.01942 + } + ] + }, + { + "step": 28580, + "tok": 117063680, + "ce": 2.6474, + "ppl": 14.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4343, + "loop_stats": [ + { + "tau": 0.1355, + "v": 0.9069, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.059, + "v": 0.8613, + "H_a": 0.057, + "d_bind": 0.03402 + }, + { + "tau": 0.0397, + "v": 0.8504, + "H_a": 0.042, + "d_bind": 0.01783 + } + ] + }, + { + "step": 28600, + "tok": 117145600, + "ce": 2.5204, + "ppl": 12.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4353, + "loop_stats": [ + { + "tau": 0.1102, + "v": 0.9203, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8678, + "H_a": 0.054, + "d_bind": 0.03566 + }, + { + "tau": 0.0278, + "v": 0.8512, + "H_a": 0.044, + "d_bind": 0.01876 + } + ] + }, + { + "step": 28620, + "tok": 117227520, + "ce": 2.8591, + "ppl": 17.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.957, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4351, + "loop_stats": [ + { + "tau": 0.1163, + "v": 0.9242, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0439, + "v": 0.877, + "H_a": 0.057, + "d_bind": 0.03332 + }, + { + "tau": 0.0246, + "v": 0.8653, + "H_a": 0.049, + "d_bind": 0.02039 + } + ] + }, + { + "step": 28640, + "tok": 117309440, + "ce": 2.7604, + "ppl": 15.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4357, + "loop_stats": [ + { + "tau": 0.0877, + "v": 0.899, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.8568, + "H_a": 0.052, + "d_bind": 0.03422 + }, + { + "tau": 0.024, + "v": 0.8479, + "H_a": 0.043, + "d_bind": 0.0196 + } + ] + }, + { + "step": 28660, + "tok": 117391360, + "ce": 2.8296, + "ppl": 16.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4371, + "loop_stats": [ + { + "tau": 0.1146, + "v": 0.929, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0308, + "v": 0.889, + "H_a": 0.058, + "d_bind": 0.0331 + }, + { + "tau": 0.0232, + "v": 0.8794, + "H_a": 0.045, + "d_bind": 0.01882 + } + ] + }, + { + "step": 28680, + "tok": 117473280, + "ce": 2.4866, + "ppl": 12.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4362, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.8819, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0451, + "v": 0.8468, + "H_a": 0.062, + "d_bind": 0.03406 + }, + { + "tau": 0.0364, + "v": 0.8433, + "H_a": 0.046, + "d_bind": 0.02069 + } + ] + }, + { + "step": 28700, + "tok": 117555200, + "ce": 3.1363, + "ppl": 23.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.957, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.436, + "loop_stats": [ + { + "tau": 0.111, + "v": 0.902, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8565, + "H_a": 0.057, + "d_bind": 0.03281 + }, + { + "tau": 0.0283, + "v": 0.839, + "H_a": 0.047, + "d_bind": 0.02186 + } + ] + }, + { + "step": 28720, + "tok": 117637120, + "ce": 3.0163, + "ppl": 20.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4349, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.9103, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.8669, + "H_a": 0.054, + "d_bind": 0.03316 + }, + { + "tau": 0.0231, + "v": 0.8545, + "H_a": 0.046, + "d_bind": 0.01864 + } + ] + }, + { + "step": 28740, + "tok": 117719040, + "ce": 2.715, + "ppl": 15.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4314, + "loop_stats": [ + { + "tau": 0.1202, + "v": 0.9051, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.8569, + "H_a": 0.05, + "d_bind": 0.03255 + }, + { + "tau": 0.0393, + "v": 0.8435, + "H_a": 0.043, + "d_bind": 0.01811 + } + ] + }, + { + "step": 28760, + "tok": 117800960, + "ce": 3.0077, + "ppl": 20.2, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.955, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.431, + "loop_stats": [ + { + "tau": 0.0833, + "v": 0.9286, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.8997, + "H_a": 0.056, + "d_bind": 0.03276 + }, + { + "tau": 0.0192, + "v": 0.887, + "H_a": 0.048, + "d_bind": 0.01998 + } + ] + }, + { + "step": 28780, + "tok": 117882880, + "ce": 2.7005, + "ppl": 14.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.962, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4289, + "loop_stats": [ + { + "tau": 0.1034, + "v": 0.9399, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.9033, + "H_a": 0.053, + "d_bind": 0.03314 + }, + { + "tau": 0.0302, + "v": 0.8771, + "H_a": 0.047, + "d_bind": 0.02002 + } + ] + }, + { + "step": 28800, + "tok": 117964800, + "ce": 3.0062, + "ppl": 20.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.962, + "tok_s": 22530, + "peak_MiB": 5301, + "v_gain": 0.4296, + "loop_stats": [ + { + "tau": 0.1128, + "v": 0.888, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0505, + "v": 0.8208, + "H_a": 0.056, + "d_bind": 0.0352 + }, + { + "tau": 0.0413, + "v": 0.8037, + "H_a": 0.042, + "d_bind": 0.01902 + } + ] + }, + { + "step": 28820, + "tok": 118046720, + "ce": 2.7367, + "ppl": 15.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.957, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4318, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.9067, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0383, + "v": 0.8783, + "H_a": 0.055, + "d_bind": 0.03223 + }, + { + "tau": 0.0299, + "v": 0.8674, + "H_a": 0.044, + "d_bind": 0.01917 + } + ] + }, + { + "step": 28840, + "tok": 118128640, + "ce": 2.8964, + "ppl": 18.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4337, + "loop_stats": [ + { + "tau": 0.0966, + "v": 0.8802, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.8338, + "H_a": 0.055, + "d_bind": 0.03271 + }, + { + "tau": 0.029, + "v": 0.8262, + "H_a": 0.045, + "d_bind": 0.02126 + } + ] + }, + { + "step": 28860, + "tok": 118210560, + "ce": 3.1591, + "ppl": 23.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4382, + "loop_stats": [ + { + "tau": 0.0681, + "v": 0.902, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0236, + "v": 0.8508, + "H_a": 0.058, + "d_bind": 0.03412 + }, + { + "tau": 0.0132, + "v": 0.8281, + "H_a": 0.039, + "d_bind": 0.01908 + } + ] + }, + { + "step": 28880, + "tok": 118292480, + "ce": 2.9059, + "ppl": 18.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4391, + "loop_stats": [ + { + "tau": 0.1036, + "v": 0.8876, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.835, + "H_a": 0.062, + "d_bind": 0.03377 + }, + { + "tau": 0.0266, + "v": 0.8168, + "H_a": 0.047, + "d_bind": 0.02119 + } + ] + }, + { + "step": 28900, + "tok": 118374400, + "ce": 2.9263, + "ppl": 18.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4405, + "loop_stats": [ + { + "tau": 0.0969, + "v": 0.9236, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0279, + "v": 0.8928, + "H_a": 0.056, + "d_bind": 0.03146 + }, + { + "tau": 0.0147, + "v": 0.882, + "H_a": 0.049, + "d_bind": 0.02104 + } + ] + }, + { + "step": 28920, + "tok": 118456320, + "ce": 2.9887, + "ppl": 19.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4393, + "loop_stats": [ + { + "tau": 0.1254, + "v": 0.8945, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0461, + "v": 0.8618, + "H_a": 0.055, + "d_bind": 0.03253 + }, + { + "tau": 0.0335, + "v": 0.8555, + "H_a": 0.042, + "d_bind": 0.02019 + } + ] + }, + { + "step": 28940, + "tok": 118538240, + "ce": 3.1288, + "ppl": 22.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4363, + "loop_stats": [ + { + "tau": 0.1018, + "v": 0.8938, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8598, + "H_a": 0.051, + "d_bind": 0.03016 + }, + { + "tau": 0.0256, + "v": 0.8448, + "H_a": 0.043, + "d_bind": 0.01772 + } + ] + }, + { + "step": 28960, + "tok": 118620160, + "ce": 2.6497, + "ppl": 14.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.942, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4336, + "loop_stats": [ + { + "tau": 0.1311, + "v": 0.8904, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8526, + "H_a": 0.05, + "d_bind": 0.03497 + }, + { + "tau": 0.0391, + "v": 0.8482, + "H_a": 0.045, + "d_bind": 0.0205 + } + ] + }, + { + "step": 28980, + "tok": 118702080, + "ce": 3.1764, + "ppl": 24.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4344, + "loop_stats": [ + { + "tau": 0.0848, + "v": 0.9061, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.8743, + "H_a": 0.057, + "d_bind": 0.03149 + }, + { + "tau": 0.0185, + "v": 0.861, + "H_a": 0.045, + "d_bind": 0.02077 + } + ] + }, + { + "step": 29000, + "tok": 118784000, + "ce": 2.818, + "ppl": 16.7, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.95, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4363, + "loop_stats": [ + { + "tau": 0.1144, + "v": 0.9105, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8802, + "H_a": 0.058, + "d_bind": 0.03488 + }, + { + "tau": 0.0197, + "v": 0.8713, + "H_a": 0.047, + "d_bind": 0.01986 + } + ] + }, + { + "step": 29020, + "tok": 118865920, + "ce": 3.1533, + "ppl": 23.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4377, + "loop_stats": [ + { + "tau": 0.1003, + "v": 0.9176, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0303, + "v": 0.8919, + "H_a": 0.055, + "d_bind": 0.03175 + }, + { + "tau": 0.0164, + "v": 0.8816, + "H_a": 0.048, + "d_bind": 0.0208 + } + ] + }, + { + "step": 29040, + "tok": 118947840, + "ce": 2.0983, + "ppl": 8.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.947, + "tok_s": 22531, + "peak_MiB": 5301, + "v_gain": 0.4356, + "loop_stats": [ + { + "tau": 0.1495, + "v": 0.8784, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0552, + "v": 0.8335, + "H_a": 0.055, + "d_bind": 0.03546 + }, + { + "tau": 0.0487, + "v": 0.832, + "H_a": 0.045, + "d_bind": 0.0224 + } + ] + }, + { + "step": 29060, + "tok": 119029760, + "ce": 2.5611, + "ppl": 13.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.93, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.434, + "loop_stats": [ + { + "tau": 0.1463, + "v": 0.8808, + "H_a": 0.053, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.836, + "H_a": 0.05, + "d_bind": 0.03132 + }, + { + "tau": 0.0448, + "v": 0.8285, + "H_a": 0.047, + "d_bind": 0.02088 + } + ] + }, + { + "step": 29080, + "tok": 119111680, + "ce": 2.5343, + "ppl": 12.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4334, + "loop_stats": [ + { + "tau": 0.1347, + "v": 0.9002, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8481, + "H_a": 0.044, + "d_bind": 0.03231 + }, + { + "tau": 0.0321, + "v": 0.8311, + "H_a": 0.042, + "d_bind": 0.01887 + } + ] + }, + { + "step": 29100, + "tok": 119193600, + "ce": 2.7515, + "ppl": 15.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4372, + "loop_stats": [ + { + "tau": 0.1129, + "v": 0.9015, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0392, + "v": 0.8603, + "H_a": 0.055, + "d_bind": 0.03281 + }, + { + "tau": 0.0313, + "v": 0.8459, + "H_a": 0.048, + "d_bind": 0.0204 + } + ] + }, + { + "step": 29120, + "tok": 119275520, + "ce": 2.8647, + "ppl": 17.5, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4372, + "loop_stats": [ + { + "tau": 0.1229, + "v": 0.9017, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8626, + "H_a": 0.052, + "d_bind": 0.03449 + }, + { + "tau": 0.0274, + "v": 0.8415, + "H_a": 0.047, + "d_bind": 0.02047 + } + ] + }, + { + "step": 29140, + "tok": 119357440, + "ce": 2.3991, + "ppl": 11.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4344, + "loop_stats": [ + { + "tau": 0.1274, + "v": 0.8898, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0543, + "v": 0.8535, + "H_a": 0.058, + "d_bind": 0.03404 + }, + { + "tau": 0.0431, + "v": 0.8473, + "H_a": 0.038, + "d_bind": 0.02031 + } + ] + }, + { + "step": 29160, + "tok": 119439360, + "ce": 2.7262, + "ppl": 15.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.945, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4353, + "loop_stats": [ + { + "tau": 0.0984, + "v": 0.8988, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0382, + "v": 0.847, + "H_a": 0.057, + "d_bind": 0.03242 + }, + { + "tau": 0.0293, + "v": 0.8391, + "H_a": 0.046, + "d_bind": 0.02208 + } + ] + }, + { + "step": 29180, + "tok": 119521280, + "ce": 2.8345, + "ppl": 17.0, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4367, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.9143, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.8701, + "H_a": 0.054, + "d_bind": 0.03604 + }, + { + "tau": 0.0188, + "v": 0.8609, + "H_a": 0.045, + "d_bind": 0.01991 + } + ] + }, + { + "step": 29200, + "tok": 119603200, + "ce": 2.993, + "ppl": 19.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4355, + "loop_stats": [ + { + "tau": 0.1015, + "v": 0.8765, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0432, + "v": 0.8344, + "H_a": 0.059, + "d_bind": 0.0341 + }, + { + "tau": 0.0309, + "v": 0.8222, + "H_a": 0.046, + "d_bind": 0.02027 + } + ] + }, + { + "step": 29220, + "tok": 119685120, + "ce": 2.7969, + "ppl": 16.4, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.942, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4344, + "loop_stats": [ + { + "tau": 0.1073, + "v": 0.9108, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0315, + "v": 0.8774, + "H_a": 0.059, + "d_bind": 0.03476 + }, + { + "tau": 0.0213, + "v": 0.8641, + "H_a": 0.046, + "d_bind": 0.02053 + } + ] + }, + { + "step": 29240, + "tok": 119767040, + "ce": 2.8831, + "ppl": 17.9, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4317, + "loop_stats": [ + { + "tau": 0.0939, + "v": 0.9143, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8705, + "H_a": 0.052, + "d_bind": 0.04009 + }, + { + "tau": 0.0279, + "v": 0.8536, + "H_a": 0.045, + "d_bind": 0.01768 + } + ] + }, + { + "step": 29260, + "tok": 119848960, + "ce": 2.7029, + "ppl": 14.9, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.955, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4325, + "loop_stats": [ + { + "tau": 0.1152, + "v": 0.8991, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0465, + "v": 0.8628, + "H_a": 0.048, + "d_bind": 0.03485 + }, + { + "tau": 0.0353, + "v": 0.8531, + "H_a": 0.046, + "d_bind": 0.01981 + } + ] + }, + { + "step": 29280, + "tok": 119930880, + "ce": 3.0858, + "ppl": 21.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4347, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.8993, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8503, + "H_a": 0.052, + "d_bind": 0.03672 + }, + { + "tau": 0.0226, + "v": 0.8355, + "H_a": 0.038, + "d_bind": 0.01911 + } + ] + }, + { + "step": 29300, + "tok": 120012800, + "ce": 2.6884, + "ppl": 14.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.955, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4348, + "loop_stats": [ + { + "tau": 0.1386, + "v": 0.9243, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0475, + "v": 0.8827, + "H_a": 0.056, + "d_bind": 0.03543 + }, + { + "tau": 0.0314, + "v": 0.871, + "H_a": 0.04, + "d_bind": 0.02079 + } + ] + }, + { + "step": 29320, + "tok": 120094720, + "ce": 2.9982, + "ppl": 20.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.945, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4355, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.9174, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.042, + "v": 0.8679, + "H_a": 0.053, + "d_bind": 0.0356 + }, + { + "tau": 0.0311, + "v": 0.8489, + "H_a": 0.045, + "d_bind": 0.02161 + } + ] + }, + { + "step": 29340, + "tok": 120176640, + "ce": 2.8181, + "ppl": 16.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.436, + "loop_stats": [ + { + "tau": 0.0887, + "v": 0.8816, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.8461, + "H_a": 0.059, + "d_bind": 0.03562 + }, + { + "tau": 0.0199, + "v": 0.8346, + "H_a": 0.043, + "d_bind": 0.02064 + } + ] + }, + { + "step": 29360, + "tok": 120258560, + "ce": 3.1248, + "ppl": 22.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.437, + "loop_stats": [ + { + "tau": 0.0902, + "v": 0.8985, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.848, + "H_a": 0.061, + "d_bind": 0.03533 + }, + { + "tau": 0.0287, + "v": 0.8196, + "H_a": 0.051, + "d_bind": 0.02261 + } + ] + }, + { + "step": 29380, + "tok": 120340480, + "ce": 2.7756, + "ppl": 16.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.947, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4376, + "loop_stats": [ + { + "tau": 0.1325, + "v": 0.9159, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.8696, + "H_a": 0.055, + "d_bind": 0.03564 + }, + { + "tau": 0.0415, + "v": 0.852, + "H_a": 0.044, + "d_bind": 0.02009 + } + ] + }, + { + "step": 29400, + "tok": 120422400, + "ce": 2.5295, + "ppl": 12.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22532, + "peak_MiB": 5301, + "v_gain": 0.4353, + "loop_stats": [ + { + "tau": 0.1372, + "v": 0.8683, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0576, + "v": 0.8341, + "H_a": 0.057, + "d_bind": 0.03562 + }, + { + "tau": 0.0474, + "v": 0.8225, + "H_a": 0.043, + "d_bind": 0.0201 + } + ] + }, + { + "step": 29420, + "tok": 120504320, + "ce": 2.6834, + "ppl": 14.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4344, + "loop_stats": [ + { + "tau": 0.1028, + "v": 0.8655, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.8302, + "H_a": 0.055, + "d_bind": 0.03357 + }, + { + "tau": 0.0366, + "v": 0.8137, + "H_a": 0.048, + "d_bind": 0.02093 + } + ] + }, + { + "step": 29440, + "tok": 120586240, + "ce": 3.146, + "ppl": 23.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4374, + "loop_stats": [ + { + "tau": 0.0907, + "v": 0.9035, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0365, + "v": 0.8495, + "H_a": 0.057, + "d_bind": 0.0364 + }, + { + "tau": 0.0243, + "v": 0.833, + "H_a": 0.045, + "d_bind": 0.01795 + } + ] + }, + { + "step": 29460, + "tok": 120668160, + "ce": 2.6672, + "ppl": 14.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.437, + "loop_stats": [ + { + "tau": 0.1106, + "v": 0.8717, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0415, + "v": 0.8367, + "H_a": 0.058, + "d_bind": 0.03451 + }, + { + "tau": 0.0341, + "v": 0.8275, + "H_a": 0.045, + "d_bind": 0.02207 + } + ] + }, + { + "step": 29480, + "tok": 120750080, + "ce": 2.914, + "ppl": 18.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4383, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.8943, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8598, + "H_a": 0.056, + "d_bind": 0.03651 + }, + { + "tau": 0.0275, + "v": 0.8424, + "H_a": 0.042, + "d_bind": 0.02058 + } + ] + }, + { + "step": 29500, + "tok": 120832000, + "ce": 2.8637, + "ppl": 17.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4389, + "loop_stats": [ + { + "tau": 0.1177, + "v": 0.9198, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.8796, + "H_a": 0.054, + "d_bind": 0.035 + }, + { + "tau": 0.0198, + "v": 0.8609, + "H_a": 0.042, + "d_bind": 0.0203 + } + ] + }, + { + "step": 29520, + "tok": 120913920, + "ce": 3.1943, + "ppl": 24.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4397, + "loop_stats": [ + { + "tau": 0.0823, + "v": 0.9182, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8759, + "H_a": 0.052, + "d_bind": 0.03299 + }, + { + "tau": 0.0215, + "v": 0.8613, + "H_a": 0.047, + "d_bind": 0.01932 + } + ] + }, + { + "step": 29540, + "tok": 120995840, + "ce": 3.0298, + "ppl": 20.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4408, + "loop_stats": [ + { + "tau": 0.0887, + "v": 0.8946, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.8531, + "H_a": 0.049, + "d_bind": 0.03247 + }, + { + "tau": 0.0192, + "v": 0.8464, + "H_a": 0.038, + "d_bind": 0.0182 + } + ] + }, + { + "step": 29560, + "tok": 121077760, + "ce": 3.1586, + "ppl": 23.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4416, + "loop_stats": [ + { + "tau": 0.103, + "v": 0.9035, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0383, + "v": 0.8684, + "H_a": 0.055, + "d_bind": 0.03388 + }, + { + "tau": 0.0238, + "v": 0.8486, + "H_a": 0.04, + "d_bind": 0.0191 + } + ] + }, + { + "step": 29580, + "tok": 121159680, + "ce": 2.7743, + "ppl": 16.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4418, + "loop_stats": [ + { + "tau": 0.0971, + "v": 0.8926, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0339, + "v": 0.8288, + "H_a": 0.057, + "d_bind": 0.03598 + }, + { + "tau": 0.0244, + "v": 0.8043, + "H_a": 0.047, + "d_bind": 0.02105 + } + ] + }, + { + "step": 29600, + "tok": 121241600, + "ce": 2.4954, + "ppl": 12.1, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4391, + "loop_stats": [ + { + "tau": 0.1252, + "v": 0.9146, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0435, + "v": 0.8639, + "H_a": 0.058, + "d_bind": 0.03579 + }, + { + "tau": 0.0328, + "v": 0.8511, + "H_a": 0.038, + "d_bind": 0.01969 + } + ] + }, + { + "step": 29620, + "tok": 121323520, + "ce": 2.9188, + "ppl": 18.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4417, + "loop_stats": [ + { + "tau": 0.0893, + "v": 0.9156, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0329, + "v": 0.8752, + "H_a": 0.052, + "d_bind": 0.03468 + }, + { + "tau": 0.0264, + "v": 0.8617, + "H_a": 0.042, + "d_bind": 0.01931 + } + ] + }, + { + "step": 29640, + "tok": 121405440, + "ce": 2.8811, + "ppl": 17.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22533, + "peak_MiB": 5301, + "v_gain": 0.4441, + "loop_stats": [ + { + "tau": 0.1185, + "v": 0.9183, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0527, + "v": 0.8673, + "H_a": 0.053, + "d_bind": 0.03644 + }, + { + "tau": 0.0328, + "v": 0.8473, + "H_a": 0.045, + "d_bind": 0.01977 + } + ] + }, + { + "step": 29660, + "tok": 121487360, + "ce": 2.8736, + "ppl": 17.7, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.962, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4419, + "loop_stats": [ + { + "tau": 0.12, + "v": 0.8816, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8205, + "H_a": 0.059, + "d_bind": 0.03663 + }, + { + "tau": 0.0353, + "v": 0.7994, + "H_a": 0.046, + "d_bind": 0.02064 + } + ] + }, + { + "step": 29680, + "tok": 121569280, + "ce": 2.8538, + "ppl": 17.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4393, + "loop_stats": [ + { + "tau": 0.105, + "v": 0.9168, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.867, + "H_a": 0.052, + "d_bind": 0.03474 + }, + { + "tau": 0.0232, + "v": 0.852, + "H_a": 0.046, + "d_bind": 0.02006 + } + ] + }, + { + "step": 29700, + "tok": 121651200, + "ce": 2.5497, + "ppl": 12.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.957, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4399, + "loop_stats": [ + { + "tau": 0.141, + "v": 0.9199, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0522, + "v": 0.8664, + "H_a": 0.056, + "d_bind": 0.03579 + }, + { + "tau": 0.04, + "v": 0.8581, + "H_a": 0.046, + "d_bind": 0.02173 + } + ] + }, + { + "step": 29720, + "tok": 121733120, + "ce": 3.0411, + "ppl": 20.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4419, + "loop_stats": [ + { + "tau": 0.105, + "v": 0.9159, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8721, + "H_a": 0.058, + "d_bind": 0.03447 + }, + { + "tau": 0.0249, + "v": 0.859, + "H_a": 0.042, + "d_bind": 0.01984 + } + ] + }, + { + "step": 29740, + "tok": 121815040, + "ce": 3.235, + "ppl": 25.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4444, + "loop_stats": [ + { + "tau": 0.094, + "v": 0.9375, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.8986, + "H_a": 0.058, + "d_bind": 0.03408 + }, + { + "tau": 0.0145, + "v": 0.8796, + "H_a": 0.048, + "d_bind": 0.02025 + } + ] + }, + { + "step": 29760, + "tok": 121896960, + "ce": 2.9683, + "ppl": 19.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4436, + "loop_stats": [ + { + "tau": 0.1091, + "v": 0.9135, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0479, + "v": 0.8762, + "H_a": 0.058, + "d_bind": 0.0329 + }, + { + "tau": 0.0231, + "v": 0.8635, + "H_a": 0.044, + "d_bind": 0.01965 + } + ] + }, + { + "step": 29780, + "tok": 121978880, + "ce": 2.6995, + "ppl": 14.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4446, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.9355, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.8969, + "H_a": 0.057, + "d_bind": 0.03394 + }, + { + "tau": 0.0111, + "v": 0.8856, + "H_a": 0.045, + "d_bind": 0.02016 + } + ] + }, + { + "step": 29800, + "tok": 122060800, + "ce": 2.7711, + "ppl": 16.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.942, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4431, + "loop_stats": [ + { + "tau": 0.1027, + "v": 0.9128, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8643, + "H_a": 0.056, + "d_bind": 0.03251 + }, + { + "tau": 0.024, + "v": 0.8502, + "H_a": 0.043, + "d_bind": 0.02073 + } + ] + }, + { + "step": 29820, + "tok": 122142720, + "ce": 2.9569, + "ppl": 19.2, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.955, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4417, + "loop_stats": [ + { + "tau": 0.1044, + "v": 0.8951, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8481, + "H_a": 0.051, + "d_bind": 0.03071 + }, + { + "tau": 0.0231, + "v": 0.8362, + "H_a": 0.042, + "d_bind": 0.02052 + } + ] + }, + { + "step": 29840, + "tok": 122224640, + "ce": 2.6453, + "ppl": 14.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.96, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4428, + "loop_stats": [ + { + "tau": 0.115, + "v": 0.9167, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.035, + "v": 0.8812, + "H_a": 0.055, + "d_bind": 0.03157 + }, + { + "tau": 0.021, + "v": 0.8654, + "H_a": 0.042, + "d_bind": 0.02041 + } + ] + }, + { + "step": 29860, + "tok": 122306560, + "ce": 2.8191, + "ppl": 16.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4439, + "loop_stats": [ + { + "tau": 0.1094, + "v": 0.9481, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.9122, + "H_a": 0.055, + "d_bind": 0.03433 + }, + { + "tau": 0.0203, + "v": 0.9023, + "H_a": 0.043, + "d_bind": 0.02124 + } + ] + }, + { + "step": 29880, + "tok": 122388480, + "ce": 3.2649, + "ppl": 26.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.95, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.446, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.9204, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0189, + "v": 0.8763, + "H_a": 0.054, + "d_bind": 0.03315 + }, + { + "tau": 0.0111, + "v": 0.8467, + "H_a": 0.045, + "d_bind": 0.01908 + } + ] + }, + { + "step": 29900, + "tok": 122470400, + "ce": 2.6606, + "ppl": 14.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4462, + "loop_stats": [ + { + "tau": 0.1357, + "v": 0.9143, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0527, + "v": 0.8594, + "H_a": 0.05, + "d_bind": 0.03508 + }, + { + "tau": 0.0399, + "v": 0.8394, + "H_a": 0.043, + "d_bind": 0.02145 + } + ] + }, + { + "step": 29920, + "tok": 122552320, + "ce": 2.8924, + "ppl": 18.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4488, + "loop_stats": [ + { + "tau": 0.1308, + "v": 0.9251, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8779, + "H_a": 0.053, + "d_bind": 0.03491 + }, + { + "tau": 0.0264, + "v": 0.8619, + "H_a": 0.046, + "d_bind": 0.02228 + } + ] + }, + { + "step": 29940, + "tok": 122634240, + "ce": 2.616, + "ppl": 13.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4491, + "loop_stats": [ + { + "tau": 0.1228, + "v": 0.8929, + "H_a": 0.057, + "d_bind": 0.0 + }, + { + "tau": 0.0566, + "v": 0.8377, + "H_a": 0.051, + "d_bind": 0.03653 + }, + { + "tau": 0.0393, + "v": 0.8223, + "H_a": 0.043, + "d_bind": 0.02043 + } + ] + }, + { + "step": 29960, + "tok": 122716160, + "ce": 2.9723, + "ppl": 19.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4495, + "loop_stats": [ + { + "tau": 0.1061, + "v": 0.9295, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0511, + "v": 0.8778, + "H_a": 0.051, + "d_bind": 0.03465 + }, + { + "tau": 0.029, + "v": 0.8497, + "H_a": 0.043, + "d_bind": 0.01906 + } + ] + }, + { + "step": 29980, + "tok": 122798080, + "ce": 3.077, + "ppl": 21.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.933, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4488, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.9116, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0389, + "v": 0.8625, + "H_a": 0.052, + "d_bind": 0.03579 + }, + { + "tau": 0.027, + "v": 0.85, + "H_a": 0.046, + "d_bind": 0.02024 + } + ] + }, + { + "step": 30000, + "tok": 122880000, + "ce": 3.1035, + "ppl": 22.3, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4523, + "loop_stats": [ + { + "tau": 0.0878, + "v": 0.9034, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0243, + "v": 0.8679, + "H_a": 0.056, + "d_bind": 0.03602 + }, + { + "tau": 0.0121, + "v": 0.8464, + "H_a": 0.045, + "d_bind": 0.01997 + } + ] + }, + { + "step": 30020, + "tok": 122961920, + "ce": 2.75, + "ppl": 15.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4518, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.9256, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.045, + "v": 0.8823, + "H_a": 0.063, + "d_bind": 0.03658 + }, + { + "tau": 0.0282, + "v": 0.8691, + "H_a": 0.046, + "d_bind": 0.01991 + } + ] + }, + { + "step": 30040, + "tok": 123043840, + "ce": 3.0301, + "ppl": 20.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.927, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.449, + "loop_stats": [ + { + "tau": 0.0925, + "v": 0.9215, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0244, + "v": 0.8802, + "H_a": 0.064, + "d_bind": 0.03893 + }, + { + "tau": 0.017, + "v": 0.8705, + "H_a": 0.047, + "d_bind": 0.02269 + } + ] + }, + { + "step": 30060, + "tok": 123125760, + "ce": 2.7696, + "ppl": 16.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4461, + "loop_stats": [ + { + "tau": 0.0994, + "v": 0.9072, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.8719, + "H_a": 0.058, + "d_bind": 0.03471 + }, + { + "tau": 0.0198, + "v": 0.8656, + "H_a": 0.049, + "d_bind": 0.02052 + } + ] + }, + { + "step": 30080, + "tok": 123207680, + "ce": 2.6854, + "ppl": 14.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4478, + "loop_stats": [ + { + "tau": 0.1308, + "v": 0.8986, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8449, + "H_a": 0.055, + "d_bind": 0.03933 + }, + { + "tau": 0.0379, + "v": 0.8315, + "H_a": 0.046, + "d_bind": 0.01969 + } + ] + }, + { + "step": 30100, + "tok": 123289600, + "ce": 2.8337, + "ppl": 17.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.933, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4485, + "loop_stats": [ + { + "tau": 0.0925, + "v": 0.9316, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0376, + "v": 0.894, + "H_a": 0.058, + "d_bind": 0.03531 + }, + { + "tau": 0.0203, + "v": 0.8767, + "H_a": 0.044, + "d_bind": 0.02021 + } + ] + }, + { + "step": 30120, + "tok": 123371520, + "ce": 3.0093, + "ppl": 20.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22534, + "peak_MiB": 5301, + "v_gain": 0.4467, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.9306, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.8918, + "H_a": 0.055, + "d_bind": 0.03743 + }, + { + "tau": 0.0203, + "v": 0.8798, + "H_a": 0.041, + "d_bind": 0.02069 + } + ] + }, + { + "step": 30140, + "tok": 123453440, + "ce": 2.3764, + "ppl": 10.8, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4456, + "loop_stats": [ + { + "tau": 0.1365, + "v": 0.9201, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8773, + "H_a": 0.058, + "d_bind": 0.03625 + }, + { + "tau": 0.0376, + "v": 0.862, + "H_a": 0.045, + "d_bind": 0.02087 + } + ] + }, + { + "step": 30160, + "tok": 123535360, + "ce": 2.5222, + "ppl": 12.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4463, + "loop_stats": [ + { + "tau": 0.1392, + "v": 0.9113, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0518, + "v": 0.8559, + "H_a": 0.059, + "d_bind": 0.0363 + }, + { + "tau": 0.0408, + "v": 0.8444, + "H_a": 0.047, + "d_bind": 0.02272 + } + ] + }, + { + "step": 30180, + "tok": 123617280, + "ce": 3.2164, + "ppl": 24.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4483, + "loop_stats": [ + { + "tau": 0.0979, + "v": 0.9357, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0312, + "v": 0.8956, + "H_a": 0.054, + "d_bind": 0.03199 + }, + { + "tau": 0.0177, + "v": 0.8829, + "H_a": 0.046, + "d_bind": 0.02106 + } + ] + }, + { + "step": 30200, + "tok": 123699200, + "ce": 2.9312, + "ppl": 18.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4498, + "loop_stats": [ + { + "tau": 0.099, + "v": 0.9275, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8842, + "H_a": 0.057, + "d_bind": 0.03695 + }, + { + "tau": 0.0222, + "v": 0.8685, + "H_a": 0.045, + "d_bind": 0.01936 + } + ] + }, + { + "step": 30220, + "tok": 123781120, + "ce": 2.9828, + "ppl": 19.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.955, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4514, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.9097, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0446, + "v": 0.8621, + "H_a": 0.06, + "d_bind": 0.03443 + }, + { + "tau": 0.0301, + "v": 0.847, + "H_a": 0.04, + "d_bind": 0.02123 + } + ] + }, + { + "step": 30240, + "tok": 123863040, + "ce": 3.3214, + "ppl": 27.7, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.933, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.45, + "loop_stats": [ + { + "tau": 0.0866, + "v": 0.9146, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0335, + "v": 0.8828, + "H_a": 0.056, + "d_bind": 0.03603 + }, + { + "tau": 0.0201, + "v": 0.8698, + "H_a": 0.044, + "d_bind": 0.01859 + } + ] + }, + { + "step": 30260, + "tok": 123944960, + "ce": 3.1214, + "ppl": 22.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.448, + "loop_stats": [ + { + "tau": 0.094, + "v": 0.9078, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8771, + "H_a": 0.052, + "d_bind": 0.03205 + }, + { + "tau": 0.0318, + "v": 0.8695, + "H_a": 0.045, + "d_bind": 0.0203 + } + ] + }, + { + "step": 30280, + "tok": 124026880, + "ce": 2.9038, + "ppl": 18.2, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.965, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4502, + "loop_stats": [ + { + "tau": 0.0951, + "v": 0.9084, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.03, + "v": 0.878, + "H_a": 0.056, + "d_bind": 0.03371 + }, + { + "tau": 0.0212, + "v": 0.8703, + "H_a": 0.043, + "d_bind": 0.01922 + } + ] + }, + { + "step": 30300, + "tok": 124108800, + "ce": 2.8824, + "ppl": 17.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4473, + "loop_stats": [ + { + "tau": 0.1091, + "v": 0.8926, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.8649, + "H_a": 0.057, + "d_bind": 0.03429 + }, + { + "tau": 0.0278, + "v": 0.8573, + "H_a": 0.048, + "d_bind": 0.02039 + } + ] + }, + { + "step": 30320, + "tok": 124190720, + "ce": 2.7434, + "ppl": 15.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4494, + "loop_stats": [ + { + "tau": 0.1191, + "v": 0.9227, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.906, + "H_a": 0.057, + "d_bind": 0.03548 + }, + { + "tau": 0.0192, + "v": 0.9011, + "H_a": 0.047, + "d_bind": 0.02083 + } + ] + }, + { + "step": 30340, + "tok": 124272640, + "ce": 2.7791, + "ppl": 16.1, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.955, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4495, + "loop_stats": [ + { + "tau": 0.1207, + "v": 0.9162, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.8778, + "H_a": 0.057, + "d_bind": 0.03379 + }, + { + "tau": 0.0342, + "v": 0.8672, + "H_a": 0.043, + "d_bind": 0.01952 + } + ] + }, + { + "step": 30360, + "tok": 124354560, + "ce": 2.5264, + "ppl": 12.5, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.967, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4526, + "loop_stats": [ + { + "tau": 0.153, + "v": 0.8944, + "H_a": 0.055, + "d_bind": 0.0 + }, + { + "tau": 0.0596, + "v": 0.8389, + "H_a": 0.045, + "d_bind": 0.03769 + }, + { + "tau": 0.0458, + "v": 0.8282, + "H_a": 0.04, + "d_bind": 0.01737 + } + ] + }, + { + "step": 30380, + "tok": 124436480, + "ce": 2.9228, + "ppl": 18.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.957, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.452, + "loop_stats": [ + { + "tau": 0.0907, + "v": 0.8966, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0281, + "v": 0.8555, + "H_a": 0.052, + "d_bind": 0.03343 + }, + { + "tau": 0.0219, + "v": 0.8425, + "H_a": 0.04, + "d_bind": 0.02024 + } + ] + }, + { + "step": 30400, + "tok": 124518400, + "ce": 3.0142, + "ppl": 20.4, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.952, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4523, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.9272, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0384, + "v": 0.8804, + "H_a": 0.059, + "d_bind": 0.03555 + }, + { + "tau": 0.0215, + "v": 0.8657, + "H_a": 0.041, + "d_bind": 0.01936 + } + ] + }, + { + "step": 30420, + "tok": 124600320, + "ce": 3.0269, + "ppl": 20.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.96, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4503, + "loop_stats": [ + { + "tau": 0.0874, + "v": 0.9365, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0339, + "v": 0.8991, + "H_a": 0.055, + "d_bind": 0.0333 + }, + { + "tau": 0.0233, + "v": 0.8806, + "H_a": 0.044, + "d_bind": 0.02 + } + ] + }, + { + "step": 30440, + "tok": 124682240, + "ce": 3.2143, + "ppl": 24.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.957, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4512, + "loop_stats": [ + { + "tau": 0.0986, + "v": 0.9201, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.869, + "H_a": 0.063, + "d_bind": 0.03597 + }, + { + "tau": 0.0261, + "v": 0.8483, + "H_a": 0.052, + "d_bind": 0.02315 + } + ] + }, + { + "step": 30460, + "tok": 124764160, + "ce": 2.3146, + "ppl": 10.1, + "pref_transfer_pron": 0.88, + "pref_train_sv": 0.947, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4553, + "loop_stats": [ + { + "tau": 0.1416, + "v": 0.9392, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.054, + "v": 0.898, + "H_a": 0.055, + "d_bind": 0.03171 + }, + { + "tau": 0.0313, + "v": 0.8872, + "H_a": 0.043, + "d_bind": 0.02126 + } + ] + }, + { + "step": 30480, + "tok": 124846080, + "ce": 2.9062, + "ppl": 18.3, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4587, + "loop_stats": [ + { + "tau": 0.1111, + "v": 0.9027, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.8652, + "H_a": 0.059, + "d_bind": 0.03243 + }, + { + "tau": 0.0171, + "v": 0.8521, + "H_a": 0.043, + "d_bind": 0.01931 + } + ] + }, + { + "step": 30500, + "tok": 124928000, + "ce": 2.2706, + "ppl": 9.7, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.461, + "loop_stats": [ + { + "tau": 0.1313, + "v": 0.9183, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0375, + "v": 0.8775, + "H_a": 0.058, + "d_bind": 0.03466 + }, + { + "tau": 0.0226, + "v": 0.8604, + "H_a": 0.047, + "d_bind": 0.02199 + } + ] + }, + { + "step": 30520, + "tok": 125009920, + "ce": 3.1184, + "ppl": 22.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.955, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4624, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.8983, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0262, + "v": 0.8635, + "H_a": 0.064, + "d_bind": 0.03614 + }, + { + "tau": 0.0135, + "v": 0.8512, + "H_a": 0.044, + "d_bind": 0.02053 + } + ] + }, + { + "step": 30540, + "tok": 125091840, + "ce": 2.9227, + "ppl": 18.6, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4619, + "loop_stats": [ + { + "tau": 0.0908, + "v": 0.9026, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0264, + "v": 0.8609, + "H_a": 0.056, + "d_bind": 0.03608 + }, + { + "tau": 0.0177, + "v": 0.8449, + "H_a": 0.048, + "d_bind": 0.01967 + } + ] + }, + { + "step": 30560, + "tok": 125173760, + "ce": 3.162, + "ppl": 23.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4607, + "loop_stats": [ + { + "tau": 0.0974, + "v": 0.9248, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.029, + "v": 0.8845, + "H_a": 0.054, + "d_bind": 0.03498 + }, + { + "tau": 0.0146, + "v": 0.8614, + "H_a": 0.046, + "d_bind": 0.01955 + } + ] + }, + { + "step": 30580, + "tok": 125255680, + "ce": 2.8785, + "ppl": 17.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4573, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.9266, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.03, + "v": 0.8907, + "H_a": 0.058, + "d_bind": 0.0359 + }, + { + "tau": 0.0187, + "v": 0.8792, + "H_a": 0.044, + "d_bind": 0.0189 + } + ] + }, + { + "step": 30600, + "tok": 125337600, + "ce": 3.0555, + "ppl": 21.2, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.955, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4572, + "loop_stats": [ + { + "tau": 0.0928, + "v": 0.9193, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0326, + "v": 0.8872, + "H_a": 0.054, + "d_bind": 0.03617 + }, + { + "tau": 0.0233, + "v": 0.8761, + "H_a": 0.051, + "d_bind": 0.02099 + } + ] + }, + { + "step": 30620, + "tok": 125419520, + "ce": 2.2562, + "ppl": 9.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.96, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.46, + "loop_stats": [ + { + "tau": 0.1466, + "v": 0.9125, + "H_a": 0.052, + "d_bind": 0.0 + }, + { + "tau": 0.0667, + "v": 0.8545, + "H_a": 0.057, + "d_bind": 0.03442 + }, + { + "tau": 0.0498, + "v": 0.8418, + "H_a": 0.046, + "d_bind": 0.02057 + } + ] + }, + { + "step": 30640, + "tok": 125501440, + "ce": 2.6936, + "ppl": 14.8, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.967, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4621, + "loop_stats": [ + { + "tau": 0.1218, + "v": 0.9026, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0421, + "v": 0.8573, + "H_a": 0.054, + "d_bind": 0.03166 + }, + { + "tau": 0.0236, + "v": 0.8407, + "H_a": 0.041, + "d_bind": 0.01963 + } + ] + }, + { + "step": 30660, + "tok": 125583360, + "ce": 2.7576, + "ppl": 15.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.957, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4629, + "loop_stats": [ + { + "tau": 0.0749, + "v": 0.9008, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0224, + "v": 0.8558, + "H_a": 0.057, + "d_bind": 0.03625 + }, + { + "tau": 0.0138, + "v": 0.8334, + "H_a": 0.05, + "d_bind": 0.02251 + } + ] + }, + { + "step": 30680, + "tok": 125665280, + "ce": 2.9675, + "ppl": 19.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.96, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4628, + "loop_stats": [ + { + "tau": 0.0872, + "v": 0.9118, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0259, + "v": 0.8606, + "H_a": 0.059, + "d_bind": 0.03707 + }, + { + "tau": 0.0136, + "v": 0.8389, + "H_a": 0.047, + "d_bind": 0.02108 + } + ] + }, + { + "step": 30700, + "tok": 125747200, + "ce": 2.5894, + "ppl": 13.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.957, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4628, + "loop_stats": [ + { + "tau": 0.1278, + "v": 0.9104, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0525, + "v": 0.8607, + "H_a": 0.054, + "d_bind": 0.0349 + }, + { + "tau": 0.0339, + "v": 0.8406, + "H_a": 0.046, + "d_bind": 0.0201 + } + ] + }, + { + "step": 30720, + "tok": 125829120, + "ce": 2.9858, + "ppl": 19.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.957, + "tok_s": 22535, + "peak_MiB": 5301, + "v_gain": 0.4619, + "loop_stats": [ + { + "tau": 0.0922, + "v": 0.8975, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0281, + "v": 0.8537, + "H_a": 0.06, + "d_bind": 0.03475 + }, + { + "tau": 0.0179, + "v": 0.8386, + "H_a": 0.05, + "d_bind": 0.02168 + } + ] + }, + { + "step": 30740, + "tok": 125911040, + "ce": 3.1421, + "ppl": 23.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4647, + "loop_stats": [ + { + "tau": 0.0812, + "v": 0.9369, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0258, + "v": 0.9111, + "H_a": 0.055, + "d_bind": 0.03326 + }, + { + "tau": 0.0097, + "v": 0.8959, + "H_a": 0.045, + "d_bind": 0.01942 + } + ] + }, + { + "step": 30760, + "tok": 125992960, + "ce": 2.813, + "ppl": 16.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.947, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4651, + "loop_stats": [ + { + "tau": 0.1241, + "v": 0.9258, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8878, + "H_a": 0.052, + "d_bind": 0.03341 + }, + { + "tau": 0.0239, + "v": 0.8723, + "H_a": 0.048, + "d_bind": 0.0206 + } + ] + }, + { + "step": 30780, + "tok": 126074880, + "ce": 2.5923, + "ppl": 13.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4629, + "loop_stats": [ + { + "tau": 0.1278, + "v": 0.9171, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.8769, + "H_a": 0.058, + "d_bind": 0.03594 + }, + { + "tau": 0.0307, + "v": 0.8649, + "H_a": 0.049, + "d_bind": 0.02233 + } + ] + }, + { + "step": 30800, + "tok": 126156800, + "ce": 2.7037, + "ppl": 14.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4643, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.9258, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0282, + "v": 0.9002, + "H_a": 0.056, + "d_bind": 0.0369 + }, + { + "tau": 0.02, + "v": 0.8896, + "H_a": 0.044, + "d_bind": 0.02054 + } + ] + }, + { + "step": 30820, + "tok": 126238720, + "ce": 2.7016, + "ppl": 14.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4659, + "loop_stats": [ + { + "tau": 0.1264, + "v": 0.9222, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8815, + "H_a": 0.053, + "d_bind": 0.03684 + }, + { + "tau": 0.0334, + "v": 0.8721, + "H_a": 0.046, + "d_bind": 0.02027 + } + ] + }, + { + "step": 30840, + "tok": 126320640, + "ce": 3.2088, + "ppl": 24.7, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4623, + "loop_stats": [ + { + "tau": 0.0923, + "v": 0.901, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0263, + "v": 0.8592, + "H_a": 0.054, + "d_bind": 0.03469 + }, + { + "tau": 0.0213, + "v": 0.8418, + "H_a": 0.046, + "d_bind": 0.02155 + } + ] + }, + { + "step": 30860, + "tok": 126402560, + "ce": 2.3817, + "ppl": 10.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4609, + "loop_stats": [ + { + "tau": 0.1402, + "v": 0.8917, + "H_a": 0.057, + "d_bind": 0.0 + }, + { + "tau": 0.0614, + "v": 0.835, + "H_a": 0.054, + "d_bind": 0.03506 + }, + { + "tau": 0.0499, + "v": 0.8195, + "H_a": 0.044, + "d_bind": 0.02104 + } + ] + }, + { + "step": 30880, + "tok": 126484480, + "ce": 2.6651, + "ppl": 14.4, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.955, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4601, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.9161, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0329, + "v": 0.8918, + "H_a": 0.06, + "d_bind": 0.03472 + }, + { + "tau": 0.0218, + "v": 0.8852, + "H_a": 0.045, + "d_bind": 0.02029 + } + ] + }, + { + "step": 30900, + "tok": 126566400, + "ce": 2.8503, + "ppl": 17.3, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.955, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4586, + "loop_stats": [ + { + "tau": 0.0981, + "v": 0.9212, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8769, + "H_a": 0.055, + "d_bind": 0.03425 + }, + { + "tau": 0.0309, + "v": 0.8612, + "H_a": 0.039, + "d_bind": 0.01937 + } + ] + }, + { + "step": 30920, + "tok": 126648320, + "ce": 2.97, + "ppl": 19.5, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.96, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4583, + "loop_stats": [ + { + "tau": 0.0865, + "v": 0.8998, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0354, + "v": 0.8587, + "H_a": 0.054, + "d_bind": 0.03557 + }, + { + "tau": 0.0254, + "v": 0.8453, + "H_a": 0.048, + "d_bind": 0.02032 + } + ] + }, + { + "step": 30940, + "tok": 126730240, + "ce": 3.1599, + "ppl": 23.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.962, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4571, + "loop_stats": [ + { + "tau": 0.1075, + "v": 0.9104, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8685, + "H_a": 0.06, + "d_bind": 0.03432 + }, + { + "tau": 0.0226, + "v": 0.8568, + "H_a": 0.047, + "d_bind": 0.02013 + } + ] + }, + { + "step": 30960, + "tok": 126812160, + "ce": 2.6307, + "ppl": 13.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4579, + "loop_stats": [ + { + "tau": 0.1242, + "v": 0.8919, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0484, + "v": 0.8417, + "H_a": 0.059, + "d_bind": 0.04047 + }, + { + "tau": 0.0308, + "v": 0.827, + "H_a": 0.048, + "d_bind": 0.02001 + } + ] + }, + { + "step": 30980, + "tok": 126894080, + "ce": 3.2133, + "ppl": 24.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4597, + "loop_stats": [ + { + "tau": 0.1208, + "v": 0.8983, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0386, + "v": 0.8495, + "H_a": 0.058, + "d_bind": 0.03634 + }, + { + "tau": 0.0292, + "v": 0.8386, + "H_a": 0.042, + "d_bind": 0.01947 + } + ] + }, + { + "step": 31000, + "tok": 126976000, + "ce": 2.7221, + "ppl": 15.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4595, + "loop_stats": [ + { + "tau": 0.1295, + "v": 0.9307, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0501, + "v": 0.8879, + "H_a": 0.062, + "d_bind": 0.03841 + }, + { + "tau": 0.0369, + "v": 0.8751, + "H_a": 0.048, + "d_bind": 0.02082 + } + ] + }, + { + "step": 31020, + "tok": 127057920, + "ce": 2.6334, + "ppl": 13.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.96, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4573, + "loop_stats": [ + { + "tau": 0.1385, + "v": 0.9002, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8557, + "H_a": 0.061, + "d_bind": 0.03548 + }, + { + "tau": 0.0382, + "v": 0.8464, + "H_a": 0.05, + "d_bind": 0.02334 + } + ] + }, + { + "step": 31040, + "tok": 127139840, + "ce": 3.0441, + "ppl": 21.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4575, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.91, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.039, + "v": 0.8684, + "H_a": 0.058, + "d_bind": 0.03551 + }, + { + "tau": 0.025, + "v": 0.8458, + "H_a": 0.044, + "d_bind": 0.01877 + } + ] + }, + { + "step": 31060, + "tok": 127221760, + "ce": 3.0769, + "ppl": 21.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4603, + "loop_stats": [ + { + "tau": 0.0691, + "v": 0.8857, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0258, + "v": 0.8349, + "H_a": 0.056, + "d_bind": 0.03453 + }, + { + "tau": 0.0175, + "v": 0.8137, + "H_a": 0.049, + "d_bind": 0.01976 + } + ] + }, + { + "step": 31080, + "tok": 127303680, + "ce": 3.1448, + "ppl": 23.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4634, + "loop_stats": [ + { + "tau": 0.1048, + "v": 0.9257, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.8743, + "H_a": 0.06, + "d_bind": 0.03887 + }, + { + "tau": 0.0237, + "v": 0.8592, + "H_a": 0.044, + "d_bind": 0.02057 + } + ] + }, + { + "step": 31100, + "tok": 127385600, + "ce": 3.0379, + "ppl": 20.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4655, + "loop_stats": [ + { + "tau": 0.1014, + "v": 0.9093, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.8696, + "H_a": 0.059, + "d_bind": 0.03335 + }, + { + "tau": 0.0235, + "v": 0.8534, + "H_a": 0.046, + "d_bind": 0.01869 + } + ] + }, + { + "step": 31120, + "tok": 127467520, + "ce": 3.1797, + "ppl": 24.0, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4681, + "loop_stats": [ + { + "tau": 0.0803, + "v": 0.905, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0214, + "v": 0.8579, + "H_a": 0.054, + "d_bind": 0.03423 + }, + { + "tau": 0.0116, + "v": 0.8446, + "H_a": 0.046, + "d_bind": 0.01964 + } + ] + }, + { + "step": 31140, + "tok": 127549440, + "ce": 2.8105, + "ppl": 16.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4695, + "loop_stats": [ + { + "tau": 0.1238, + "v": 0.8895, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.8448, + "H_a": 0.062, + "d_bind": 0.03793 + }, + { + "tau": 0.0304, + "v": 0.8344, + "H_a": 0.045, + "d_bind": 0.02045 + } + ] + }, + { + "step": 31160, + "tok": 127631360, + "ce": 2.8853, + "ppl": 17.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4667, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.8799, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0385, + "v": 0.8486, + "H_a": 0.06, + "d_bind": 0.03319 + }, + { + "tau": 0.0214, + "v": 0.8338, + "H_a": 0.047, + "d_bind": 0.01988 + } + ] + }, + { + "step": 31180, + "tok": 127713280, + "ce": 2.6049, + "ppl": 13.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.95, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4642, + "loop_stats": [ + { + "tau": 0.1304, + "v": 0.8869, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.8415, + "H_a": 0.056, + "d_bind": 0.03624 + }, + { + "tau": 0.0287, + "v": 0.823, + "H_a": 0.045, + "d_bind": 0.02216 + } + ] + }, + { + "step": 31200, + "tok": 127795200, + "ce": 2.6843, + "ppl": 14.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.96, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4634, + "loop_stats": [ + { + "tau": 0.1267, + "v": 0.8879, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8471, + "H_a": 0.055, + "d_bind": 0.03277 + }, + { + "tau": 0.0272, + "v": 0.8343, + "H_a": 0.043, + "d_bind": 0.02022 + } + ] + }, + { + "step": 31220, + "tok": 127877120, + "ce": 2.7612, + "ppl": 15.8, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.96, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4622, + "loop_stats": [ + { + "tau": 0.0964, + "v": 0.8793, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.8465, + "H_a": 0.058, + "d_bind": 0.03555 + }, + { + "tau": 0.0204, + "v": 0.839, + "H_a": 0.052, + "d_bind": 0.02134 + } + ] + }, + { + "step": 31240, + "tok": 127959040, + "ce": 3.3, + "ppl": 27.1, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4581, + "loop_stats": [ + { + "tau": 0.0743, + "v": 0.8979, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.8554, + "H_a": 0.057, + "d_bind": 0.03462 + }, + { + "tau": 0.0213, + "v": 0.8374, + "H_a": 0.05, + "d_bind": 0.01969 + } + ] + }, + { + "step": 31260, + "tok": 128040960, + "ce": 3.1004, + "ppl": 22.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.96, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4569, + "loop_stats": [ + { + "tau": 0.0975, + "v": 0.8854, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.8284, + "H_a": 0.062, + "d_bind": 0.03251 + }, + { + "tau": 0.0274, + "v": 0.8206, + "H_a": 0.049, + "d_bind": 0.02053 + } + ] + }, + { + "step": 31280, + "tok": 128122880, + "ce": 2.839, + "ppl": 17.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4551, + "loop_stats": [ + { + "tau": 0.1155, + "v": 0.9318, + "H_a": 0.077, + "d_bind": 0.0 + }, + { + "tau": 0.0393, + "v": 0.8942, + "H_a": 0.057, + "d_bind": 0.03706 + }, + { + "tau": 0.0245, + "v": 0.8833, + "H_a": 0.047, + "d_bind": 0.01863 + } + ] + }, + { + "step": 31300, + "tok": 128204800, + "ce": 2.71, + "ppl": 15.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4555, + "loop_stats": [ + { + "tau": 0.1072, + "v": 0.9088, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0502, + "v": 0.8706, + "H_a": 0.063, + "d_bind": 0.03647 + }, + { + "tau": 0.0317, + "v": 0.8571, + "H_a": 0.05, + "d_bind": 0.02231 + } + ] + }, + { + "step": 31320, + "tok": 128286720, + "ce": 2.5881, + "ppl": 13.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4582, + "loop_stats": [ + { + "tau": 0.1117, + "v": 0.9099, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0368, + "v": 0.8784, + "H_a": 0.061, + "d_bind": 0.03821 + }, + { + "tau": 0.024, + "v": 0.8632, + "H_a": 0.058, + "d_bind": 0.02354 + } + ] + }, + { + "step": 31340, + "tok": 128368640, + "ce": 2.699, + "ppl": 14.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.459, + "loop_stats": [ + { + "tau": 0.1299, + "v": 0.9057, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8717, + "H_a": 0.063, + "d_bind": 0.03776 + }, + { + "tau": 0.0355, + "v": 0.8605, + "H_a": 0.05, + "d_bind": 0.02261 + } + ] + }, + { + "step": 31360, + "tok": 128450560, + "ce": 3.075, + "ppl": 21.6, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.957, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4576, + "loop_stats": [ + { + "tau": 0.0936, + "v": 0.9277, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8978, + "H_a": 0.06, + "d_bind": 0.03688 + }, + { + "tau": 0.0235, + "v": 0.8771, + "H_a": 0.05, + "d_bind": 0.0201 + } + ] + }, + { + "step": 31380, + "tok": 128532480, + "ce": 2.6059, + "ppl": 13.5, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.46, + "loop_stats": [ + { + "tau": 0.1043, + "v": 0.9051, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8732, + "H_a": 0.06, + "d_bind": 0.03898 + }, + { + "tau": 0.0267, + "v": 0.8612, + "H_a": 0.048, + "d_bind": 0.02011 + } + ] + }, + { + "step": 31400, + "tok": 128614400, + "ce": 2.7236, + "ppl": 15.2, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.933, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4622, + "loop_stats": [ + { + "tau": 0.1234, + "v": 0.9165, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0358, + "v": 0.8735, + "H_a": 0.057, + "d_bind": 0.03893 + }, + { + "tau": 0.0237, + "v": 0.8644, + "H_a": 0.043, + "d_bind": 0.02017 + } + ] + }, + { + "step": 31420, + "tok": 128696320, + "ce": 2.6947, + "ppl": 14.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4614, + "loop_stats": [ + { + "tau": 0.1322, + "v": 0.9178, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8675, + "H_a": 0.062, + "d_bind": 0.03776 + }, + { + "tau": 0.0299, + "v": 0.853, + "H_a": 0.044, + "d_bind": 0.01953 + } + ] + }, + { + "step": 31440, + "tok": 128778240, + "ce": 3.1671, + "ppl": 23.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4587, + "loop_stats": [ + { + "tau": 0.0982, + "v": 0.915, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0485, + "v": 0.8701, + "H_a": 0.061, + "d_bind": 0.03725 + }, + { + "tau": 0.0323, + "v": 0.8533, + "H_a": 0.049, + "d_bind": 0.01893 + } + ] + }, + { + "step": 31460, + "tok": 128860160, + "ce": 2.8956, + "ppl": 18.1, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.459, + "loop_stats": [ + { + "tau": 0.1011, + "v": 0.9076, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0398, + "v": 0.8753, + "H_a": 0.063, + "d_bind": 0.03695 + }, + { + "tau": 0.0312, + "v": 0.8645, + "H_a": 0.049, + "d_bind": 0.02301 + } + ] + }, + { + "step": 31480, + "tok": 128942080, + "ce": 2.8977, + "ppl": 18.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.459, + "loop_stats": [ + { + "tau": 0.1077, + "v": 0.8734, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8153, + "H_a": 0.069, + "d_bind": 0.03449 + }, + { + "tau": 0.0301, + "v": 0.802, + "H_a": 0.055, + "d_bind": 0.02418 + } + ] + }, + { + "step": 31500, + "tok": 129024000, + "ce": 3.2321, + "ppl": 25.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.463, + "loop_stats": [ + { + "tau": 0.0751, + "v": 0.9378, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.9149, + "H_a": 0.061, + "d_bind": 0.03496 + }, + { + "tau": 0.0079, + "v": 0.9012, + "H_a": 0.048, + "d_bind": 0.01963 + } + ] + }, + { + "step": 31520, + "tok": 129105920, + "ce": 3.153, + "ppl": 23.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4621, + "loop_stats": [ + { + "tau": 0.0789, + "v": 0.9162, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0291, + "v": 0.8786, + "H_a": 0.062, + "d_bind": 0.03644 + }, + { + "tau": 0.009, + "v": 0.8608, + "H_a": 0.05, + "d_bind": 0.02095 + } + ] + }, + { + "step": 31540, + "tok": 129187840, + "ce": 2.8107, + "ppl": 16.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.957, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4609, + "loop_stats": [ + { + "tau": 0.1075, + "v": 0.8955, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0333, + "v": 0.8604, + "H_a": 0.052, + "d_bind": 0.03462 + }, + { + "tau": 0.0169, + "v": 0.8456, + "H_a": 0.038, + "d_bind": 0.01935 + } + ] + }, + { + "step": 31560, + "tok": 129269760, + "ce": 3.3143, + "ppl": 27.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4614, + "loop_stats": [ + { + "tau": 0.0815, + "v": 0.9064, + "H_a": 0.078, + "d_bind": 0.0 + }, + { + "tau": 0.0257, + "v": 0.8712, + "H_a": 0.067, + "d_bind": 0.0354 + }, + { + "tau": 0.0098, + "v": 0.8544, + "H_a": 0.045, + "d_bind": 0.02001 + } + ] + }, + { + "step": 31580, + "tok": 129351680, + "ce": 2.6917, + "ppl": 14.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4591, + "loop_stats": [ + { + "tau": 0.1259, + "v": 0.8921, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.8529, + "H_a": 0.06, + "d_bind": 0.03565 + }, + { + "tau": 0.0339, + "v": 0.8447, + "H_a": 0.046, + "d_bind": 0.02282 + } + ] + }, + { + "step": 31600, + "tok": 129433600, + "ce": 3.0155, + "ppl": 20.4, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.952, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4595, + "loop_stats": [ + { + "tau": 0.0917, + "v": 0.9047, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.03, + "v": 0.8743, + "H_a": 0.065, + "d_bind": 0.03592 + }, + { + "tau": 0.0206, + "v": 0.8683, + "H_a": 0.05, + "d_bind": 0.02111 + } + ] + }, + { + "step": 31620, + "tok": 129515520, + "ce": 3.058, + "ppl": 21.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4583, + "loop_stats": [ + { + "tau": 0.098, + "v": 0.8887, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0359, + "v": 0.8461, + "H_a": 0.061, + "d_bind": 0.03704 + }, + { + "tau": 0.0233, + "v": 0.8362, + "H_a": 0.044, + "d_bind": 0.02018 + } + ] + }, + { + "step": 31640, + "tok": 129597440, + "ce": 3.2291, + "ppl": 25.3, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4571, + "loop_stats": [ + { + "tau": 0.078, + "v": 0.9237, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0229, + "v": 0.8945, + "H_a": 0.06, + "d_bind": 0.03657 + }, + { + "tau": 0.0141, + "v": 0.8832, + "H_a": 0.046, + "d_bind": 0.01989 + } + ] + }, + { + "step": 31660, + "tok": 129679360, + "ce": 2.4122, + "ppl": 11.2, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.957, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4589, + "loop_stats": [ + { + "tau": 0.1288, + "v": 0.8816, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8464, + "H_a": 0.063, + "d_bind": 0.03701 + }, + { + "tau": 0.0369, + "v": 0.8387, + "H_a": 0.049, + "d_bind": 0.02331 + } + ] + }, + { + "step": 31680, + "tok": 129761280, + "ce": 2.8576, + "ppl": 17.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.962, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4598, + "loop_stats": [ + { + "tau": 0.121, + "v": 0.9155, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0579, + "v": 0.8826, + "H_a": 0.062, + "d_bind": 0.0373 + }, + { + "tau": 0.0299, + "v": 0.8751, + "H_a": 0.047, + "d_bind": 0.02022 + } + ] + }, + { + "step": 31700, + "tok": 129843200, + "ce": 2.5224, + "ppl": 12.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.955, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.462, + "loop_stats": [ + { + "tau": 0.1186, + "v": 0.912, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0417, + "v": 0.8891, + "H_a": 0.055, + "d_bind": 0.03069 + }, + { + "tau": 0.0235, + "v": 0.8832, + "H_a": 0.048, + "d_bind": 0.02063 + } + ] + }, + { + "step": 31720, + "tok": 129925120, + "ce": 2.3439, + "ppl": 10.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.962, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4628, + "loop_stats": [ + { + "tau": 0.1138, + "v": 0.8839, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0361, + "v": 0.8542, + "H_a": 0.067, + "d_bind": 0.03526 + }, + { + "tau": 0.0246, + "v": 0.8449, + "H_a": 0.054, + "d_bind": 0.02244 + } + ] + }, + { + "step": 31740, + "tok": 130007040, + "ce": 2.6929, + "ppl": 14.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4623, + "loop_stats": [ + { + "tau": 0.096, + "v": 0.8931, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8716, + "H_a": 0.052, + "d_bind": 0.03453 + }, + { + "tau": 0.0234, + "v": 0.8673, + "H_a": 0.046, + "d_bind": 0.02086 + } + ] + }, + { + "step": 31760, + "tok": 130088960, + "ce": 2.6719, + "ppl": 14.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.962, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4648, + "loop_stats": [ + { + "tau": 0.1219, + "v": 0.892, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0399, + "v": 0.8651, + "H_a": 0.057, + "d_bind": 0.03457 + }, + { + "tau": 0.022, + "v": 0.8565, + "H_a": 0.046, + "d_bind": 0.02119 + } + ] + }, + { + "step": 31780, + "tok": 130170880, + "ce": 2.7678, + "ppl": 15.9, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.955, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4664, + "loop_stats": [ + { + "tau": 0.113, + "v": 0.8926, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.033, + "v": 0.8647, + "H_a": 0.058, + "d_bind": 0.03485 + }, + { + "tau": 0.0149, + "v": 0.8567, + "H_a": 0.049, + "d_bind": 0.02167 + } + ] + }, + { + "step": 31800, + "tok": 130252800, + "ce": 3.047, + "ppl": 21.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22536, + "peak_MiB": 5301, + "v_gain": 0.4651, + "loop_stats": [ + { + "tau": 0.0929, + "v": 0.8838, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8519, + "H_a": 0.055, + "d_bind": 0.03413 + }, + { + "tau": 0.0238, + "v": 0.8432, + "H_a": 0.043, + "d_bind": 0.01977 + } + ] + }, + { + "step": 31820, + "tok": 130334720, + "ce": 3.0179, + "ppl": 20.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4658, + "loop_stats": [ + { + "tau": 0.0979, + "v": 0.8879, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0298, + "v": 0.8535, + "H_a": 0.059, + "d_bind": 0.03372 + }, + { + "tau": 0.019, + "v": 0.8415, + "H_a": 0.047, + "d_bind": 0.02072 + } + ] + }, + { + "step": 31840, + "tok": 130416640, + "ce": 3.0604, + "ppl": 21.3, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.945, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4639, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.9125, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0362, + "v": 0.8809, + "H_a": 0.051, + "d_bind": 0.0353 + }, + { + "tau": 0.0119, + "v": 0.868, + "H_a": 0.046, + "d_bind": 0.01938 + } + ] + }, + { + "step": 31860, + "tok": 130498560, + "ce": 2.7766, + "ppl": 16.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.467, + "loop_stats": [ + { + "tau": 0.0982, + "v": 0.9061, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0318, + "v": 0.8644, + "H_a": 0.061, + "d_bind": 0.03679 + }, + { + "tau": 0.0187, + "v": 0.8588, + "H_a": 0.045, + "d_bind": 0.0212 + } + ] + }, + { + "step": 31880, + "tok": 130580480, + "ce": 2.4008, + "ppl": 11.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.47, + "loop_stats": [ + { + "tau": 0.1345, + "v": 0.895, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0469, + "v": 0.8514, + "H_a": 0.058, + "d_bind": 0.034 + }, + { + "tau": 0.0302, + "v": 0.8455, + "H_a": 0.043, + "d_bind": 0.02091 + } + ] + }, + { + "step": 31900, + "tok": 130662400, + "ce": 2.3553, + "ppl": 10.5, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4685, + "loop_stats": [ + { + "tau": 0.1339, + "v": 0.8888, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.8506, + "H_a": 0.059, + "d_bind": 0.03598 + }, + { + "tau": 0.0266, + "v": 0.8414, + "H_a": 0.048, + "d_bind": 0.02024 + } + ] + }, + { + "step": 31920, + "tok": 130744320, + "ce": 2.9603, + "ppl": 19.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.96, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4679, + "loop_stats": [ + { + "tau": 0.0893, + "v": 0.8855, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0255, + "v": 0.8418, + "H_a": 0.062, + "d_bind": 0.03664 + }, + { + "tau": 0.0141, + "v": 0.8324, + "H_a": 0.043, + "d_bind": 0.01996 + } + ] + }, + { + "step": 31940, + "tok": 130826240, + "ce": 2.8658, + "ppl": 17.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.95, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4663, + "loop_stats": [ + { + "tau": 0.0884, + "v": 0.9189, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0242, + "v": 0.8924, + "H_a": 0.055, + "d_bind": 0.03445 + }, + { + "tau": 0.0157, + "v": 0.883, + "H_a": 0.05, + "d_bind": 0.02127 + } + ] + }, + { + "step": 31960, + "tok": 130908160, + "ce": 2.5512, + "ppl": 12.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.95, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4674, + "loop_stats": [ + { + "tau": 0.1421, + "v": 0.8728, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0508, + "v": 0.8271, + "H_a": 0.059, + "d_bind": 0.03392 + }, + { + "tau": 0.0329, + "v": 0.8124, + "H_a": 0.047, + "d_bind": 0.02101 + } + ] + }, + { + "step": 31980, + "tok": 130990080, + "ce": 2.9653, + "ppl": 19.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4657, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.8749, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8298, + "H_a": 0.055, + "d_bind": 0.03107 + }, + { + "tau": 0.0241, + "v": 0.8215, + "H_a": 0.05, + "d_bind": 0.02211 + } + ] + }, + { + "step": 32000, + "tok": 131072000, + "ce": 2.514, + "ppl": 12.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4646, + "loop_stats": [ + { + "tau": 0.1283, + "v": 0.8793, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0567, + "v": 0.8312, + "H_a": 0.052, + "d_bind": 0.03807 + }, + { + "tau": 0.045, + "v": 0.8247, + "H_a": 0.047, + "d_bind": 0.02074 + } + ] + }, + { + "step": 32020, + "tok": 131153920, + "ce": 2.6097, + "ppl": 13.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.464, + "loop_stats": [ + { + "tau": 0.1106, + "v": 0.8706, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.8319, + "H_a": 0.055, + "d_bind": 0.03536 + }, + { + "tau": 0.0348, + "v": 0.8263, + "H_a": 0.05, + "d_bind": 0.0219 + } + ] + }, + { + "step": 32040, + "tok": 131235840, + "ce": 2.4484, + "ppl": 11.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4665, + "loop_stats": [ + { + "tau": 0.1309, + "v": 0.8833, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0512, + "v": 0.8393, + "H_a": 0.048, + "d_bind": 0.03582 + }, + { + "tau": 0.0363, + "v": 0.8274, + "H_a": 0.047, + "d_bind": 0.02263 + } + ] + }, + { + "step": 32060, + "tok": 131317760, + "ce": 3.0446, + "ppl": 21.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.935, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4651, + "loop_stats": [ + { + "tau": 0.0926, + "v": 0.9268, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0225, + "v": 0.8882, + "H_a": 0.055, + "d_bind": 0.03677 + }, + { + "tau": 0.0146, + "v": 0.8735, + "H_a": 0.044, + "d_bind": 0.01939 + } + ] + }, + { + "step": 32080, + "tok": 131399680, + "ce": 2.5635, + "ppl": 13.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.93, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4627, + "loop_stats": [ + { + "tau": 0.1339, + "v": 0.864, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0596, + "v": 0.8147, + "H_a": 0.056, + "d_bind": 0.03653 + }, + { + "tau": 0.0503, + "v": 0.8095, + "H_a": 0.055, + "d_bind": 0.0221 + } + ] + }, + { + "step": 32100, + "tok": 131481600, + "ce": 2.8427, + "ppl": 17.2, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4618, + "loop_stats": [ + { + "tau": 0.1091, + "v": 0.9122, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8648, + "H_a": 0.056, + "d_bind": 0.0394 + }, + { + "tau": 0.0313, + "v": 0.8539, + "H_a": 0.044, + "d_bind": 0.02104 + } + ] + }, + { + "step": 32120, + "tok": 131563520, + "ce": 2.8833, + "ppl": 17.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.927, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4643, + "loop_stats": [ + { + "tau": 0.1103, + "v": 0.9136, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8685, + "H_a": 0.063, + "d_bind": 0.03588 + }, + { + "tau": 0.0231, + "v": 0.8481, + "H_a": 0.045, + "d_bind": 0.02116 + } + ] + }, + { + "step": 32140, + "tok": 131645440, + "ce": 2.6311, + "ppl": 13.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.464, + "loop_stats": [ + { + "tau": 0.1226, + "v": 0.8826, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0463, + "v": 0.8499, + "H_a": 0.054, + "d_bind": 0.03415 + }, + { + "tau": 0.0328, + "v": 0.8416, + "H_a": 0.044, + "d_bind": 0.02133 + } + ] + }, + { + "step": 32160, + "tok": 131727360, + "ce": 3.0469, + "ppl": 21.0, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.952, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4647, + "loop_stats": [ + { + "tau": 0.0884, + "v": 0.9152, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0288, + "v": 0.8805, + "H_a": 0.051, + "d_bind": 0.03529 + }, + { + "tau": 0.0157, + "v": 0.8611, + "H_a": 0.041, + "d_bind": 0.01812 + } + ] + }, + { + "step": 32180, + "tok": 131809280, + "ce": 2.3904, + "ppl": 10.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4664, + "loop_stats": [ + { + "tau": 0.151, + "v": 0.9009, + "H_a": 0.053, + "d_bind": 0.0 + }, + { + "tau": 0.0672, + "v": 0.8418, + "H_a": 0.054, + "d_bind": 0.03297 + }, + { + "tau": 0.049, + "v": 0.8273, + "H_a": 0.049, + "d_bind": 0.02296 + } + ] + }, + { + "step": 32200, + "tok": 131891200, + "ce": 2.8158, + "ppl": 16.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4662, + "loop_stats": [ + { + "tau": 0.1022, + "v": 0.9227, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8861, + "H_a": 0.052, + "d_bind": 0.03773 + }, + { + "tau": 0.0197, + "v": 0.8664, + "H_a": 0.044, + "d_bind": 0.01856 + } + ] + }, + { + "step": 32220, + "tok": 131973120, + "ce": 2.8815, + "ppl": 17.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4675, + "loop_stats": [ + { + "tau": 0.0831, + "v": 0.9272, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.8827, + "H_a": 0.057, + "d_bind": 0.03771 + }, + { + "tau": 0.0149, + "v": 0.8652, + "H_a": 0.048, + "d_bind": 0.022 + } + ] + }, + { + "step": 32240, + "tok": 132055040, + "ce": 2.7584, + "ppl": 15.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4691, + "loop_stats": [ + { + "tau": 0.1, + "v": 0.8977, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8621, + "H_a": 0.061, + "d_bind": 0.03737 + }, + { + "tau": 0.0281, + "v": 0.8538, + "H_a": 0.046, + "d_bind": 0.02262 + } + ] + }, + { + "step": 32260, + "tok": 132136960, + "ce": 2.8656, + "ppl": 17.6, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4681, + "loop_stats": [ + { + "tau": 0.0992, + "v": 0.8952, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0381, + "v": 0.8454, + "H_a": 0.062, + "d_bind": 0.03623 + }, + { + "tau": 0.0293, + "v": 0.8301, + "H_a": 0.044, + "d_bind": 0.02038 + } + ] + }, + { + "step": 32280, + "tok": 132218880, + "ce": 2.2525, + "ppl": 9.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4686, + "loop_stats": [ + { + "tau": 0.1515, + "v": 0.9109, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0565, + "v": 0.8555, + "H_a": 0.056, + "d_bind": 0.0386 + }, + { + "tau": 0.0428, + "v": 0.8396, + "H_a": 0.044, + "d_bind": 0.02015 + } + ] + }, + { + "step": 32300, + "tok": 132300800, + "ce": 2.7659, + "ppl": 15.9, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.942, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4684, + "loop_stats": [ + { + "tau": 0.0949, + "v": 0.8995, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.861, + "H_a": 0.056, + "d_bind": 0.03515 + }, + { + "tau": 0.0226, + "v": 0.8541, + "H_a": 0.04, + "d_bind": 0.01937 + } + ] + }, + { + "step": 32320, + "tok": 132382720, + "ce": 2.6534, + "ppl": 14.2, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4671, + "loop_stats": [ + { + "tau": 0.1224, + "v": 0.8733, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8103, + "H_a": 0.057, + "d_bind": 0.03427 + }, + { + "tau": 0.0374, + "v": 0.7878, + "H_a": 0.049, + "d_bind": 0.02196 + } + ] + }, + { + "step": 32340, + "tok": 132464640, + "ce": 2.39, + "ppl": 10.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4667, + "loop_stats": [ + { + "tau": 0.1298, + "v": 0.882, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0493, + "v": 0.8237, + "H_a": 0.068, + "d_bind": 0.03686 + }, + { + "tau": 0.041, + "v": 0.8094, + "H_a": 0.048, + "d_bind": 0.02346 + } + ] + }, + { + "step": 32360, + "tok": 132546560, + "ce": 2.7226, + "ppl": 15.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.466, + "loop_stats": [ + { + "tau": 0.0987, + "v": 0.8797, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0373, + "v": 0.8389, + "H_a": 0.06, + "d_bind": 0.0342 + }, + { + "tau": 0.0275, + "v": 0.8315, + "H_a": 0.047, + "d_bind": 0.02048 + } + ] + }, + { + "step": 32380, + "tok": 132628480, + "ce": 2.5254, + "ppl": 12.5, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4646, + "loop_stats": [ + { + "tau": 0.1263, + "v": 0.8737, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0368, + "v": 0.8258, + "H_a": 0.055, + "d_bind": 0.03432 + }, + { + "tau": 0.0304, + "v": 0.8129, + "H_a": 0.048, + "d_bind": 0.02023 + } + ] + }, + { + "step": 32400, + "tok": 132710400, + "ce": 2.8382, + "ppl": 17.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.933, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4649, + "loop_stats": [ + { + "tau": 0.1132, + "v": 0.8953, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.037, + "v": 0.8601, + "H_a": 0.064, + "d_bind": 0.03854 + }, + { + "tau": 0.0244, + "v": 0.8538, + "H_a": 0.044, + "d_bind": 0.02091 + } + ] + }, + { + "step": 32420, + "tok": 132792320, + "ce": 3.0485, + "ppl": 21.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4647, + "loop_stats": [ + { + "tau": 0.0933, + "v": 0.907, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8796, + "H_a": 0.051, + "d_bind": 0.03447 + }, + { + "tau": 0.0206, + "v": 0.8702, + "H_a": 0.043, + "d_bind": 0.01829 + } + ] + }, + { + "step": 32440, + "tok": 132874240, + "ce": 2.2764, + "ppl": 9.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4649, + "loop_stats": [ + { + "tau": 0.1294, + "v": 0.8661, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.8376, + "H_a": 0.059, + "d_bind": 0.03774 + }, + { + "tau": 0.0237, + "v": 0.8337, + "H_a": 0.044, + "d_bind": 0.02169 + } + ] + }, + { + "step": 32460, + "tok": 132956160, + "ce": 3.1197, + "ppl": 22.6, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4684, + "loop_stats": [ + { + "tau": 0.0879, + "v": 0.9068, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0273, + "v": 0.8672, + "H_a": 0.06, + "d_bind": 0.03658 + }, + { + "tau": 0.0161, + "v": 0.8513, + "H_a": 0.05, + "d_bind": 0.02253 + } + ] + }, + { + "step": 32480, + "tok": 133038080, + "ce": 2.913, + "ppl": 18.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.952, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4722, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.895, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.031, + "v": 0.8614, + "H_a": 0.055, + "d_bind": 0.03505 + }, + { + "tau": 0.0147, + "v": 0.8484, + "H_a": 0.043, + "d_bind": 0.01906 + } + ] + }, + { + "step": 32500, + "tok": 133120000, + "ce": 2.3631, + "ppl": 10.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4712, + "loop_stats": [ + { + "tau": 0.1499, + "v": 0.9075, + "H_a": 0.057, + "d_bind": 0.0 + }, + { + "tau": 0.0629, + "v": 0.8722, + "H_a": 0.052, + "d_bind": 0.03591 + }, + { + "tau": 0.043, + "v": 0.8629, + "H_a": 0.039, + "d_bind": 0.01868 + } + ] + }, + { + "step": 32520, + "tok": 133201920, + "ce": 2.5206, + "ppl": 12.4, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.955, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.471, + "loop_stats": [ + { + "tau": 0.1313, + "v": 0.8957, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8443, + "H_a": 0.057, + "d_bind": 0.03928 + }, + { + "tau": 0.0392, + "v": 0.8323, + "H_a": 0.042, + "d_bind": 0.02066 + } + ] + }, + { + "step": 32540, + "tok": 133283840, + "ce": 3.1467, + "ppl": 23.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4713, + "loop_stats": [ + { + "tau": 0.0936, + "v": 0.9086, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0237, + "v": 0.8727, + "H_a": 0.056, + "d_bind": 0.03565 + }, + { + "tau": 0.0135, + "v": 0.8625, + "H_a": 0.05, + "d_bind": 0.02004 + } + ] + }, + { + "step": 32560, + "tok": 133365760, + "ce": 2.4496, + "ppl": 11.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.471, + "loop_stats": [ + { + "tau": 0.1436, + "v": 0.8975, + "H_a": 0.056, + "d_bind": 0.0 + }, + { + "tau": 0.0581, + "v": 0.8634, + "H_a": 0.057, + "d_bind": 0.03598 + }, + { + "tau": 0.0385, + "v": 0.8578, + "H_a": 0.047, + "d_bind": 0.0195 + } + ] + }, + { + "step": 32580, + "tok": 133447680, + "ce": 3.1065, + "ppl": 22.3, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4704, + "loop_stats": [ + { + "tau": 0.0824, + "v": 0.9082, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8686, + "H_a": 0.053, + "d_bind": 0.03388 + }, + { + "tau": 0.0218, + "v": 0.8543, + "H_a": 0.042, + "d_bind": 0.01766 + } + ] + }, + { + "step": 32600, + "tok": 133529600, + "ce": 2.656, + "ppl": 14.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4686, + "loop_stats": [ + { + "tau": 0.1176, + "v": 0.9031, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0542, + "v": 0.8516, + "H_a": 0.061, + "d_bind": 0.03699 + }, + { + "tau": 0.0359, + "v": 0.8368, + "H_a": 0.046, + "d_bind": 0.02059 + } + ] + }, + { + "step": 32620, + "tok": 133611520, + "ce": 3.0002, + "ppl": 20.1, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4693, + "loop_stats": [ + { + "tau": 0.1168, + "v": 0.9198, + "H_a": 0.055, + "d_bind": 0.0 + }, + { + "tau": 0.051, + "v": 0.8788, + "H_a": 0.051, + "d_bind": 0.0359 + }, + { + "tau": 0.0283, + "v": 0.8638, + "H_a": 0.046, + "d_bind": 0.01641 + } + ] + }, + { + "step": 32640, + "tok": 133693440, + "ce": 2.645, + "ppl": 14.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4698, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.8839, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.839, + "H_a": 0.054, + "d_bind": 0.03754 + }, + { + "tau": 0.0345, + "v": 0.8238, + "H_a": 0.048, + "d_bind": 0.02074 + } + ] + }, + { + "step": 32660, + "tok": 133775360, + "ce": 3.0969, + "ppl": 22.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4699, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.8948, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0343, + "v": 0.8546, + "H_a": 0.054, + "d_bind": 0.03325 + }, + { + "tau": 0.0259, + "v": 0.848, + "H_a": 0.046, + "d_bind": 0.0202 + } + ] + }, + { + "step": 32680, + "tok": 133857280, + "ce": 3.0403, + "ppl": 20.9, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4709, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.92, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0337, + "v": 0.8844, + "H_a": 0.055, + "d_bind": 0.03428 + }, + { + "tau": 0.0237, + "v": 0.8677, + "H_a": 0.046, + "d_bind": 0.01945 + } + ] + }, + { + "step": 32700, + "tok": 133939200, + "ce": 2.3856, + "ppl": 10.9, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4724, + "loop_stats": [ + { + "tau": 0.1333, + "v": 0.883, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.047, + "v": 0.8398, + "H_a": 0.054, + "d_bind": 0.03681 + }, + { + "tau": 0.0417, + "v": 0.8287, + "H_a": 0.042, + "d_bind": 0.01965 + } + ] + }, + { + "step": 32720, + "tok": 134021120, + "ce": 3.2029, + "ppl": 24.6, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.947, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4713, + "loop_stats": [ + { + "tau": 0.0771, + "v": 0.908, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0299, + "v": 0.8671, + "H_a": 0.063, + "d_bind": 0.03571 + }, + { + "tau": 0.017, + "v": 0.8403, + "H_a": 0.049, + "d_bind": 0.02024 + } + ] + }, + { + "step": 32740, + "tok": 134103040, + "ce": 2.7873, + "ppl": 16.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.95, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4703, + "loop_stats": [ + { + "tau": 0.13, + "v": 0.8985, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0426, + "v": 0.8618, + "H_a": 0.055, + "d_bind": 0.03586 + }, + { + "tau": 0.0342, + "v": 0.8515, + "H_a": 0.044, + "d_bind": 0.01888 + } + ] + }, + { + "step": 32760, + "tok": 134184960, + "ce": 2.6359, + "ppl": 14.0, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.93, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4683, + "loop_stats": [ + { + "tau": 0.1156, + "v": 0.8783, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0468, + "v": 0.8507, + "H_a": 0.057, + "d_bind": 0.03423 + }, + { + "tau": 0.0351, + "v": 0.8404, + "H_a": 0.045, + "d_bind": 0.01968 + } + ] + }, + { + "step": 32780, + "tok": 134266880, + "ce": 2.9917, + "ppl": 19.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4652, + "loop_stats": [ + { + "tau": 0.108, + "v": 0.9044, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8665, + "H_a": 0.053, + "d_bind": 0.03516 + }, + { + "tau": 0.0341, + "v": 0.8516, + "H_a": 0.042, + "d_bind": 0.01919 + } + ] + }, + { + "step": 32800, + "tok": 134348800, + "ce": 2.5331, + "ppl": 12.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.464, + "loop_stats": [ + { + "tau": 0.1392, + "v": 0.8917, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0606, + "v": 0.8399, + "H_a": 0.059, + "d_bind": 0.03892 + }, + { + "tau": 0.0525, + "v": 0.836, + "H_a": 0.042, + "d_bind": 0.01934 + } + ] + }, + { + "step": 32820, + "tok": 134430720, + "ce": 2.7077, + "ppl": 15.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.93, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4652, + "loop_stats": [ + { + "tau": 0.1142, + "v": 0.9, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0414, + "v": 0.8559, + "H_a": 0.058, + "d_bind": 0.03804 + }, + { + "tau": 0.038, + "v": 0.8408, + "H_a": 0.048, + "d_bind": 0.01984 + } + ] + }, + { + "step": 32840, + "tok": 134512640, + "ce": 2.8945, + "ppl": 18.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.93, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4673, + "loop_stats": [ + { + "tau": 0.1114, + "v": 0.904, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.049, + "v": 0.853, + "H_a": 0.052, + "d_bind": 0.03366 + }, + { + "tau": 0.0382, + "v": 0.8364, + "H_a": 0.048, + "d_bind": 0.01952 + } + ] + }, + { + "step": 32860, + "tok": 134594560, + "ce": 2.6382, + "ppl": 14.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.93, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4683, + "loop_stats": [ + { + "tau": 0.106, + "v": 0.8841, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.0389, + "v": 0.8398, + "H_a": 0.055, + "d_bind": 0.03549 + }, + { + "tau": 0.0304, + "v": 0.8304, + "H_a": 0.048, + "d_bind": 0.01897 + } + ] + }, + { + "step": 32880, + "tok": 134676480, + "ce": 2.7402, + "ppl": 15.5, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.945, + "tok_s": 22537, + "peak_MiB": 5301, + "v_gain": 0.4686, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.9044, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0409, + "v": 0.8578, + "H_a": 0.053, + "d_bind": 0.03286 + }, + { + "tau": 0.0338, + "v": 0.8354, + "H_a": 0.052, + "d_bind": 0.01965 + } + ] + }, + { + "step": 32900, + "tok": 134758400, + "ce": 2.8015, + "ppl": 16.5, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.935, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4674, + "loop_stats": [ + { + "tau": 0.1141, + "v": 0.8839, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0527, + "v": 0.8304, + "H_a": 0.057, + "d_bind": 0.03552 + }, + { + "tau": 0.0452, + "v": 0.8139, + "H_a": 0.05, + "d_bind": 0.02012 + } + ] + }, + { + "step": 32920, + "tok": 134840320, + "ce": 2.152, + "ppl": 8.6, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.942, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4669, + "loop_stats": [ + { + "tau": 0.1435, + "v": 0.871, + "H_a": 0.056, + "d_bind": 0.0 + }, + { + "tau": 0.062, + "v": 0.8222, + "H_a": 0.061, + "d_bind": 0.03964 + }, + { + "tau": 0.0594, + "v": 0.8153, + "H_a": 0.045, + "d_bind": 0.0197 + } + ] + }, + { + "step": 32940, + "tok": 134922240, + "ce": 2.5659, + "ppl": 13.0, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.925, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4705, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8862, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0523, + "v": 0.8336, + "H_a": 0.054, + "d_bind": 0.03914 + }, + { + "tau": 0.0467, + "v": 0.8184, + "H_a": 0.044, + "d_bind": 0.0185 + } + ] + }, + { + "step": 32960, + "tok": 135004160, + "ce": 2.9228, + "ppl": 18.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.93, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4724, + "loop_stats": [ + { + "tau": 0.1014, + "v": 0.8958, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0283, + "v": 0.842, + "H_a": 0.054, + "d_bind": 0.03748 + }, + { + "tau": 0.0244, + "v": 0.8212, + "H_a": 0.037, + "d_bind": 0.01804 + } + ] + }, + { + "step": 32980, + "tok": 135086080, + "ce": 2.998, + "ppl": 20.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4737, + "loop_stats": [ + { + "tau": 0.0993, + "v": 0.9107, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0372, + "v": 0.876, + "H_a": 0.053, + "d_bind": 0.03441 + }, + { + "tau": 0.0213, + "v": 0.8603, + "H_a": 0.044, + "d_bind": 0.01735 + } + ] + }, + { + "step": 33000, + "tok": 135168000, + "ce": 2.9498, + "ppl": 19.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4767, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.8907, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0316, + "v": 0.8547, + "H_a": 0.06, + "d_bind": 0.03568 + }, + { + "tau": 0.0213, + "v": 0.8372, + "H_a": 0.048, + "d_bind": 0.02063 + } + ] + }, + { + "step": 33020, + "tok": 135249920, + "ce": 2.8054, + "ppl": 16.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.935, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4799, + "loop_stats": [ + { + "tau": 0.1158, + "v": 0.9164, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0459, + "v": 0.8642, + "H_a": 0.059, + "d_bind": 0.03941 + }, + { + "tau": 0.0286, + "v": 0.8448, + "H_a": 0.045, + "d_bind": 0.02025 + } + ] + }, + { + "step": 33040, + "tok": 135331840, + "ce": 3.067, + "ppl": 21.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4783, + "loop_stats": [ + { + "tau": 0.0887, + "v": 0.9045, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8528, + "H_a": 0.061, + "d_bind": 0.03729 + }, + { + "tau": 0.0194, + "v": 0.8371, + "H_a": 0.049, + "d_bind": 0.02066 + } + ] + }, + { + "step": 33060, + "tok": 135413760, + "ce": 2.8026, + "ppl": 16.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.92, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4756, + "loop_stats": [ + { + "tau": 0.1113, + "v": 0.9075, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0394, + "v": 0.869, + "H_a": 0.057, + "d_bind": 0.03944 + }, + { + "tau": 0.0295, + "v": 0.8523, + "H_a": 0.05, + "d_bind": 0.02151 + } + ] + }, + { + "step": 33080, + "tok": 135495680, + "ce": 2.4926, + "ppl": 12.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4781, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.8795, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0437, + "v": 0.8476, + "H_a": 0.06, + "d_bind": 0.03714 + }, + { + "tau": 0.0318, + "v": 0.8404, + "H_a": 0.048, + "d_bind": 0.02056 + } + ] + }, + { + "step": 33100, + "tok": 135577600, + "ce": 2.5445, + "ppl": 12.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.935, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4803, + "loop_stats": [ + { + "tau": 0.1298, + "v": 0.8893, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0504, + "v": 0.8444, + "H_a": 0.052, + "d_bind": 0.03544 + }, + { + "tau": 0.0325, + "v": 0.8256, + "H_a": 0.044, + "d_bind": 0.01899 + } + ] + }, + { + "step": 33120, + "tok": 135659520, + "ce": 2.3455, + "ppl": 10.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4779, + "loop_stats": [ + { + "tau": 0.1203, + "v": 0.8676, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8354, + "H_a": 0.06, + "d_bind": 0.03889 + }, + { + "tau": 0.0321, + "v": 0.8301, + "H_a": 0.049, + "d_bind": 0.02053 + } + ] + }, + { + "step": 33140, + "tok": 135741440, + "ce": 2.944, + "ppl": 19.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.947, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.476, + "loop_stats": [ + { + "tau": 0.0866, + "v": 0.9008, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0288, + "v": 0.8677, + "H_a": 0.053, + "d_bind": 0.0358 + }, + { + "tau": 0.0215, + "v": 0.8528, + "H_a": 0.041, + "d_bind": 0.01802 + } + ] + }, + { + "step": 33160, + "tok": 135823360, + "ce": 2.8511, + "ppl": 17.3, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4753, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.9131, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0443, + "v": 0.8825, + "H_a": 0.054, + "d_bind": 0.037 + }, + { + "tau": 0.0352, + "v": 0.8732, + "H_a": 0.041, + "d_bind": 0.0187 + } + ] + }, + { + "step": 33180, + "tok": 135905280, + "ce": 2.8002, + "ppl": 16.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4756, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.8734, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.8293, + "H_a": 0.054, + "d_bind": 0.03304 + }, + { + "tau": 0.0399, + "v": 0.8146, + "H_a": 0.046, + "d_bind": 0.02022 + } + ] + }, + { + "step": 33200, + "tok": 135987200, + "ce": 3.1893, + "ppl": 24.3, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4757, + "loop_stats": [ + { + "tau": 0.0858, + "v": 0.904, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0278, + "v": 0.8723, + "H_a": 0.052, + "d_bind": 0.03627 + }, + { + "tau": 0.0191, + "v": 0.8578, + "H_a": 0.046, + "d_bind": 0.01897 + } + ] + }, + { + "step": 33220, + "tok": 136069120, + "ce": 2.5505, + "ppl": 12.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.927, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4795, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8943, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0447, + "v": 0.8478, + "H_a": 0.054, + "d_bind": 0.03793 + }, + { + "tau": 0.0316, + "v": 0.8366, + "H_a": 0.046, + "d_bind": 0.01915 + } + ] + }, + { + "step": 33240, + "tok": 136151040, + "ce": 2.7374, + "ppl": 15.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4807, + "loop_stats": [ + { + "tau": 0.1161, + "v": 0.8629, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.8243, + "H_a": 0.055, + "d_bind": 0.03646 + }, + { + "tau": 0.0237, + "v": 0.8136, + "H_a": 0.046, + "d_bind": 0.01989 + } + ] + }, + { + "step": 33260, + "tok": 136232960, + "ce": 2.644, + "ppl": 14.1, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.945, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4778, + "loop_stats": [ + { + "tau": 0.1346, + "v": 0.8849, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0634, + "v": 0.8436, + "H_a": 0.055, + "d_bind": 0.03382 + }, + { + "tau": 0.0508, + "v": 0.8356, + "H_a": 0.047, + "d_bind": 0.01955 + } + ] + }, + { + "step": 33280, + "tok": 136314880, + "ce": 2.636, + "ppl": 14.0, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4762, + "loop_stats": [ + { + "tau": 0.1169, + "v": 0.9077, + "H_a": 0.073, + "d_bind": 0.0 + }, + { + "tau": 0.0496, + "v": 0.8668, + "H_a": 0.057, + "d_bind": 0.03523 + }, + { + "tau": 0.0376, + "v": 0.8542, + "H_a": 0.046, + "d_bind": 0.01838 + } + ] + }, + { + "step": 33300, + "tok": 136396800, + "ce": 3.2545, + "ppl": 25.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4765, + "loop_stats": [ + { + "tau": 0.0967, + "v": 0.9145, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0296, + "v": 0.8788, + "H_a": 0.053, + "d_bind": 0.03355 + }, + { + "tau": 0.0213, + "v": 0.8622, + "H_a": 0.047, + "d_bind": 0.02116 + } + ] + }, + { + "step": 33320, + "tok": 136478720, + "ce": 2.7761, + "ppl": 16.1, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.952, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4759, + "loop_stats": [ + { + "tau": 0.1116, + "v": 0.8953, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0454, + "v": 0.8512, + "H_a": 0.056, + "d_bind": 0.0343 + }, + { + "tau": 0.0332, + "v": 0.8381, + "H_a": 0.048, + "d_bind": 0.01891 + } + ] + }, + { + "step": 33340, + "tok": 136560640, + "ce": 2.5806, + "ppl": 13.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4749, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.9089, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0571, + "v": 0.8695, + "H_a": 0.056, + "d_bind": 0.03306 + }, + { + "tau": 0.0483, + "v": 0.8617, + "H_a": 0.047, + "d_bind": 0.01998 + } + ] + }, + { + "step": 33360, + "tok": 136642560, + "ce": 2.9489, + "ppl": 19.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4747, + "loop_stats": [ + { + "tau": 0.0994, + "v": 0.901, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0408, + "v": 0.8527, + "H_a": 0.054, + "d_bind": 0.03368 + }, + { + "tau": 0.0299, + "v": 0.8329, + "H_a": 0.048, + "d_bind": 0.01986 + } + ] + }, + { + "step": 33380, + "tok": 136724480, + "ce": 2.8457, + "ppl": 17.2, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.95, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.476, + "loop_stats": [ + { + "tau": 0.0956, + "v": 0.8864, + "H_a": 0.076, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8393, + "H_a": 0.056, + "d_bind": 0.03629 + }, + { + "tau": 0.0253, + "v": 0.8196, + "H_a": 0.047, + "d_bind": 0.01944 + } + ] + }, + { + "step": 33400, + "tok": 136806400, + "ce": 2.8223, + "ppl": 16.8, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4718, + "loop_stats": [ + { + "tau": 0.1138, + "v": 0.8779, + "H_a": 0.075, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8449, + "H_a": 0.048, + "d_bind": 0.03392 + }, + { + "tau": 0.0405, + "v": 0.8328, + "H_a": 0.044, + "d_bind": 0.01815 + } + ] + }, + { + "step": 33420, + "tok": 136888320, + "ce": 2.9951, + "ppl": 20.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4695, + "loop_stats": [ + { + "tau": 0.0846, + "v": 0.8969, + "H_a": 0.081, + "d_bind": 0.0 + }, + { + "tau": 0.0289, + "v": 0.854, + "H_a": 0.06, + "d_bind": 0.03531 + }, + { + "tau": 0.0262, + "v": 0.8398, + "H_a": 0.041, + "d_bind": 0.01764 + } + ] + }, + { + "step": 33440, + "tok": 136970240, + "ce": 3.029, + "ppl": 20.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.947, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.471, + "loop_stats": [ + { + "tau": 0.0944, + "v": 0.8729, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.8353, + "H_a": 0.05, + "d_bind": 0.03598 + }, + { + "tau": 0.0306, + "v": 0.8178, + "H_a": 0.046, + "d_bind": 0.01812 + } + ] + }, + { + "step": 33460, + "tok": 137052160, + "ce": 2.2905, + "ppl": 9.9, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.471, + "loop_stats": [ + { + "tau": 0.1495, + "v": 0.891, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0701, + "v": 0.8385, + "H_a": 0.056, + "d_bind": 0.03805 + }, + { + "tau": 0.0593, + "v": 0.8289, + "H_a": 0.047, + "d_bind": 0.02016 + } + ] + }, + { + "step": 33480, + "tok": 137134080, + "ce": 2.6761, + "ppl": 14.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.957, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4733, + "loop_stats": [ + { + "tau": 0.109, + "v": 0.9003, + "H_a": 0.071, + "d_bind": 0.0 + }, + { + "tau": 0.0429, + "v": 0.8599, + "H_a": 0.053, + "d_bind": 0.0343 + }, + { + "tau": 0.0325, + "v": 0.8402, + "H_a": 0.044, + "d_bind": 0.01953 + } + ] + }, + { + "step": 33500, + "tok": 137216000, + "ce": 2.6879, + "ppl": 14.7, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.962, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4727, + "loop_stats": [ + { + "tau": 0.1266, + "v": 0.8595, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0476, + "v": 0.8147, + "H_a": 0.049, + "d_bind": 0.03411 + }, + { + "tau": 0.0418, + "v": 0.8071, + "H_a": 0.045, + "d_bind": 0.01946 + } + ] + }, + { + "step": 33520, + "tok": 137297920, + "ce": 2.6452, + "ppl": 14.1, + "pref_transfer_pron": 0.938, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4718, + "loop_stats": [ + { + "tau": 0.1243, + "v": 0.8778, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0471, + "v": 0.8294, + "H_a": 0.055, + "d_bind": 0.03631 + }, + { + "tau": 0.039, + "v": 0.8186, + "H_a": 0.044, + "d_bind": 0.0184 + } + ] + }, + { + "step": 33540, + "tok": 137379840, + "ce": 2.5435, + "ppl": 12.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4702, + "loop_stats": [ + { + "tau": 0.1213, + "v": 0.9074, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0553, + "v": 0.851, + "H_a": 0.051, + "d_bind": 0.03572 + }, + { + "tau": 0.0415, + "v": 0.8387, + "H_a": 0.045, + "d_bind": 0.02137 + } + ] + }, + { + "step": 33560, + "tok": 137461760, + "ce": 2.7826, + "ppl": 16.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.942, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4712, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.9224, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.8827, + "H_a": 0.048, + "d_bind": 0.03187 + }, + { + "tau": 0.028, + "v": 0.8628, + "H_a": 0.043, + "d_bind": 0.01999 + } + ] + }, + { + "step": 33580, + "tok": 137543680, + "ce": 2.7392, + "ppl": 15.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4717, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.9021, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0485, + "v": 0.8607, + "H_a": 0.052, + "d_bind": 0.03241 + }, + { + "tau": 0.0392, + "v": 0.8475, + "H_a": 0.042, + "d_bind": 0.01943 + } + ] + }, + { + "step": 33600, + "tok": 137625600, + "ce": 3.0002, + "ppl": 20.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.93, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4697, + "loop_stats": [ + { + "tau": 0.0993, + "v": 0.8835, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.046, + "v": 0.8397, + "H_a": 0.051, + "d_bind": 0.03308 + }, + { + "tau": 0.0372, + "v": 0.8293, + "H_a": 0.04, + "d_bind": 0.01833 + } + ] + }, + { + "step": 33620, + "tok": 137707520, + "ce": 2.8311, + "ppl": 17.0, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.938, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4714, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.9215, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0332, + "v": 0.8828, + "H_a": 0.051, + "d_bind": 0.02943 + }, + { + "tau": 0.0195, + "v": 0.8673, + "H_a": 0.045, + "d_bind": 0.01955 + } + ] + }, + { + "step": 33640, + "tok": 137789440, + "ce": 2.7588, + "ppl": 15.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4745, + "loop_stats": [ + { + "tau": 0.107, + "v": 0.8667, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.8299, + "H_a": 0.051, + "d_bind": 0.03154 + }, + { + "tau": 0.0321, + "v": 0.8284, + "H_a": 0.045, + "d_bind": 0.01891 + } + ] + }, + { + "step": 33660, + "tok": 137871360, + "ce": 2.4208, + "ppl": 11.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.935, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4762, + "loop_stats": [ + { + "tau": 0.1319, + "v": 0.8883, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0503, + "v": 0.8456, + "H_a": 0.047, + "d_bind": 0.03624 + }, + { + "tau": 0.0384, + "v": 0.8252, + "H_a": 0.039, + "d_bind": 0.01861 + } + ] + }, + { + "step": 33680, + "tok": 137953280, + "ce": 3.4261, + "ppl": 30.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4781, + "loop_stats": [ + { + "tau": 0.0684, + "v": 0.8935, + "H_a": 0.074, + "d_bind": 0.0 + }, + { + "tau": 0.0217, + "v": 0.8466, + "H_a": 0.053, + "d_bind": 0.02974 + }, + { + "tau": 0.0088, + "v": 0.8247, + "H_a": 0.042, + "d_bind": 0.01792 + } + ] + }, + { + "step": 33700, + "tok": 138035200, + "ce": 2.7247, + "ppl": 15.3, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.93, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4821, + "loop_stats": [ + { + "tau": 0.106, + "v": 0.895, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.032, + "v": 0.8609, + "H_a": 0.056, + "d_bind": 0.03561 + }, + { + "tau": 0.017, + "v": 0.8489, + "H_a": 0.046, + "d_bind": 0.02098 + } + ] + }, + { + "step": 33720, + "tok": 138117120, + "ce": 3.0198, + "ppl": 20.5, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4878, + "loop_stats": [ + { + "tau": 0.1015, + "v": 0.9028, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.022, + "v": 0.8541, + "H_a": 0.055, + "d_bind": 0.033 + }, + { + "tau": 0.012, + "v": 0.8411, + "H_a": 0.043, + "d_bind": 0.01921 + } + ] + }, + { + "step": 33740, + "tok": 138199040, + "ce": 3.0207, + "ppl": 20.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4878, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.9376, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0227, + "v": 0.9027, + "H_a": 0.052, + "d_bind": 0.0346 + }, + { + "tau": 0.012, + "v": 0.8904, + "H_a": 0.048, + "d_bind": 0.02097 + } + ] + }, + { + "step": 33760, + "tok": 138280960, + "ce": 2.9131, + "ppl": 18.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4854, + "loop_stats": [ + { + "tau": 0.1115, + "v": 0.9053, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0298, + "v": 0.8666, + "H_a": 0.056, + "d_bind": 0.03247 + }, + { + "tau": 0.0112, + "v": 0.8498, + "H_a": 0.043, + "d_bind": 0.01957 + } + ] + }, + { + "step": 33780, + "tok": 138362880, + "ce": 2.8672, + "ppl": 17.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.925, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4836, + "loop_stats": [ + { + "tau": 0.0838, + "v": 0.8866, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0277, + "v": 0.8415, + "H_a": 0.052, + "d_bind": 0.03467 + }, + { + "tau": 0.0145, + "v": 0.8275, + "H_a": 0.042, + "d_bind": 0.01913 + } + ] + }, + { + "step": 33800, + "tok": 138444800, + "ce": 3.1002, + "ppl": 22.2, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.935, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4828, + "loop_stats": [ + { + "tau": 0.0961, + "v": 0.9154, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0322, + "v": 0.8822, + "H_a": 0.057, + "d_bind": 0.0339 + }, + { + "tau": 0.0174, + "v": 0.8717, + "H_a": 0.048, + "d_bind": 0.02144 + } + ] + }, + { + "step": 33820, + "tok": 138526720, + "ce": 2.597, + "ppl": 13.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.933, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4844, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.9128, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0209, + "v": 0.8559, + "H_a": 0.057, + "d_bind": 0.03426 + }, + { + "tau": 0.0132, + "v": 0.8371, + "H_a": 0.046, + "d_bind": 0.02145 + } + ] + }, + { + "step": 33840, + "tok": 138608640, + "ce": 3.0657, + "ppl": 21.4, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4827, + "loop_stats": [ + { + "tau": 0.0826, + "v": 0.8855, + "H_a": 0.08, + "d_bind": 0.0 + }, + { + "tau": 0.0235, + "v": 0.8374, + "H_a": 0.057, + "d_bind": 0.03426 + }, + { + "tau": 0.0155, + "v": 0.8192, + "H_a": 0.045, + "d_bind": 0.0196 + } + ] + }, + { + "step": 33860, + "tok": 138690560, + "ce": 2.8819, + "ppl": 17.8, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.93, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4836, + "loop_stats": [ + { + "tau": 0.0931, + "v": 0.895, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0301, + "v": 0.8355, + "H_a": 0.055, + "d_bind": 0.03248 + }, + { + "tau": 0.0193, + "v": 0.816, + "H_a": 0.047, + "d_bind": 0.01956 + } + ] + }, + { + "step": 33880, + "tok": 138772480, + "ce": 2.6333, + "ppl": 13.9, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4839, + "loop_stats": [ + { + "tau": 0.1261, + "v": 0.8793, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0526, + "v": 0.8328, + "H_a": 0.06, + "d_bind": 0.03726 + }, + { + "tau": 0.0337, + "v": 0.8132, + "H_a": 0.048, + "d_bind": 0.02113 + } + ] + }, + { + "step": 33900, + "tok": 138854400, + "ce": 2.8216, + "ppl": 16.8, + "pref_transfer_pron": 0.877, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4838, + "loop_stats": [ + { + "tau": 0.1187, + "v": 0.8876, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0366, + "v": 0.856, + "H_a": 0.049, + "d_bind": 0.03271 + }, + { + "tau": 0.0238, + "v": 0.8504, + "H_a": 0.047, + "d_bind": 0.01838 + } + ] + }, + { + "step": 33920, + "tok": 138936320, + "ce": 3.0958, + "ppl": 22.1, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.92, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4828, + "loop_stats": [ + { + "tau": 0.081, + "v": 0.8887, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0256, + "v": 0.8373, + "H_a": 0.058, + "d_bind": 0.03428 + }, + { + "tau": 0.0138, + "v": 0.8203, + "H_a": 0.045, + "d_bind": 0.02073 + } + ] + }, + { + "step": 33940, + "tok": 139018240, + "ce": 2.7223, + "ppl": 15.2, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.927, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4835, + "loop_stats": [ + { + "tau": 0.1085, + "v": 0.8732, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.8216, + "H_a": 0.06, + "d_bind": 0.0373 + }, + { + "tau": 0.0255, + "v": 0.8091, + "H_a": 0.049, + "d_bind": 0.02032 + } + ] + }, + { + "step": 33960, + "tok": 139100160, + "ce": 2.2906, + "ppl": 9.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4824, + "loop_stats": [ + { + "tau": 0.1304, + "v": 0.8793, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0618, + "v": 0.8345, + "H_a": 0.055, + "d_bind": 0.03524 + }, + { + "tau": 0.0426, + "v": 0.8225, + "H_a": 0.043, + "d_bind": 0.01905 + } + ] + }, + { + "step": 33980, + "tok": 139182080, + "ce": 2.8718, + "ppl": 17.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4838, + "loop_stats": [ + { + "tau": 0.1024, + "v": 0.8809, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0334, + "v": 0.8392, + "H_a": 0.052, + "d_bind": 0.03458 + }, + { + "tau": 0.0211, + "v": 0.8273, + "H_a": 0.043, + "d_bind": 0.01792 + } + ] + }, + { + "step": 34000, + "tok": 139264000, + "ce": 2.9381, + "ppl": 18.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.92, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4815, + "loop_stats": [ + { + "tau": 0.1021, + "v": 0.8855, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.044, + "v": 0.8539, + "H_a": 0.058, + "d_bind": 0.03556 + }, + { + "tau": 0.0315, + "v": 0.838, + "H_a": 0.044, + "d_bind": 0.01996 + } + ] + }, + { + "step": 34020, + "tok": 139345920, + "ce": 3.362, + "ppl": 28.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4815, + "loop_stats": [ + { + "tau": 0.0714, + "v": 0.8997, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0219, + "v": 0.8577, + "H_a": 0.054, + "d_bind": 0.03434 + }, + { + "tau": 0.0168, + "v": 0.8388, + "H_a": 0.039, + "d_bind": 0.01697 + } + ] + }, + { + "step": 34040, + "tok": 139427840, + "ce": 2.528, + "ppl": 12.5, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.922, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4832, + "loop_stats": [ + { + "tau": 0.1105, + "v": 0.8752, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8401, + "H_a": 0.053, + "d_bind": 0.03686 + }, + { + "tau": 0.0271, + "v": 0.8271, + "H_a": 0.043, + "d_bind": 0.01894 + } + ] + }, + { + "step": 34060, + "tok": 139509760, + "ce": 2.5281, + "ppl": 12.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4836, + "loop_stats": [ + { + "tau": 0.1316, + "v": 0.8587, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0457, + "v": 0.8211, + "H_a": 0.058, + "d_bind": 0.03501 + }, + { + "tau": 0.0329, + "v": 0.8146, + "H_a": 0.045, + "d_bind": 0.02005 + } + ] + }, + { + "step": 34080, + "tok": 139591680, + "ce": 2.8536, + "ppl": 17.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.93, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4828, + "loop_stats": [ + { + "tau": 0.1089, + "v": 0.8862, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0536, + "v": 0.8399, + "H_a": 0.054, + "d_bind": 0.0347 + }, + { + "tau": 0.033, + "v": 0.8255, + "H_a": 0.05, + "d_bind": 0.02013 + } + ] + }, + { + "step": 34100, + "tok": 139673600, + "ce": 2.6902, + "ppl": 14.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.938, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4854, + "loop_stats": [ + { + "tau": 0.1377, + "v": 0.871, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0547, + "v": 0.8209, + "H_a": 0.049, + "d_bind": 0.03302 + }, + { + "tau": 0.0431, + "v": 0.802, + "H_a": 0.038, + "d_bind": 0.01764 + } + ] + }, + { + "step": 34120, + "tok": 139755520, + "ce": 2.7949, + "ppl": 16.4, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.933, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.485, + "loop_stats": [ + { + "tau": 0.1296, + "v": 0.8839, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0486, + "v": 0.8464, + "H_a": 0.053, + "d_bind": 0.03308 + }, + { + "tau": 0.0346, + "v": 0.8326, + "H_a": 0.041, + "d_bind": 0.01795 + } + ] + }, + { + "step": 34140, + "tok": 139837440, + "ce": 2.834, + "ppl": 17.0, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.922, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.486, + "loop_stats": [ + { + "tau": 0.1056, + "v": 0.8874, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0314, + "v": 0.847, + "H_a": 0.062, + "d_bind": 0.03523 + }, + { + "tau": 0.0236, + "v": 0.8369, + "H_a": 0.044, + "d_bind": 0.02042 + } + ] + }, + { + "step": 34160, + "tok": 139919360, + "ce": 2.6732, + "ppl": 14.5, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.938, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4891, + "loop_stats": [ + { + "tau": 0.1198, + "v": 0.9147, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0326, + "v": 0.8771, + "H_a": 0.054, + "d_bind": 0.03339 + }, + { + "tau": 0.0166, + "v": 0.8671, + "H_a": 0.042, + "d_bind": 0.01857 + } + ] + }, + { + "step": 34180, + "tok": 140001280, + "ce": 3.2023, + "ppl": 24.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4896, + "loop_stats": [ + { + "tau": 0.1025, + "v": 0.8943, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8433, + "H_a": 0.051, + "d_bind": 0.03319 + }, + { + "tau": 0.0192, + "v": 0.8342, + "H_a": 0.04, + "d_bind": 0.01872 + } + ] + }, + { + "step": 34200, + "tok": 140083200, + "ce": 3.0338, + "ppl": 20.8, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22538, + "peak_MiB": 5301, + "v_gain": 0.4917, + "loop_stats": [ + { + "tau": 0.0905, + "v": 0.9121, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0332, + "v": 0.8697, + "H_a": 0.059, + "d_bind": 0.03519 + }, + { + "tau": 0.017, + "v": 0.8527, + "H_a": 0.052, + "d_bind": 0.02199 + } + ] + }, + { + "step": 34220, + "tok": 140165120, + "ce": 2.6385, + "ppl": 14.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.935, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4948, + "loop_stats": [ + { + "tau": 0.117, + "v": 0.9151, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0356, + "v": 0.8621, + "H_a": 0.054, + "d_bind": 0.03585 + }, + { + "tau": 0.0179, + "v": 0.852, + "H_a": 0.045, + "d_bind": 0.01923 + } + ] + }, + { + "step": 34240, + "tok": 140247040, + "ce": 2.7452, + "ppl": 15.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4958, + "loop_stats": [ + { + "tau": 0.1215, + "v": 0.8548, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0306, + "v": 0.804, + "H_a": 0.049, + "d_bind": 0.03295 + }, + { + "tau": 0.0152, + "v": 0.7921, + "H_a": 0.041, + "d_bind": 0.01712 + } + ] + }, + { + "step": 34260, + "tok": 140328960, + "ce": 2.3926, + "ppl": 10.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4952, + "loop_stats": [ + { + "tau": 0.1318, + "v": 0.8721, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8227, + "H_a": 0.05, + "d_bind": 0.03517 + }, + { + "tau": 0.0252, + "v": 0.8092, + "H_a": 0.039, + "d_bind": 0.0182 + } + ] + }, + { + "step": 34280, + "tok": 140410880, + "ce": 3.0906, + "ppl": 22.0, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.94, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4959, + "loop_stats": [ + { + "tau": 0.1127, + "v": 0.9093, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0274, + "v": 0.8614, + "H_a": 0.054, + "d_bind": 0.03093 + }, + { + "tau": 0.0099, + "v": 0.8381, + "H_a": 0.045, + "d_bind": 0.01883 + } + ] + }, + { + "step": 34300, + "tok": 140492800, + "ce": 2.8847, + "ppl": 17.9, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4971, + "loop_stats": [ + { + "tau": 0.1153, + "v": 0.8806, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.825, + "H_a": 0.057, + "d_bind": 0.03299 + }, + { + "tau": 0.0144, + "v": 0.8039, + "H_a": 0.046, + "d_bind": 0.01942 + } + ] + }, + { + "step": 34320, + "tok": 140574720, + "ce": 2.9047, + "ppl": 18.3, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4967, + "loop_stats": [ + { + "tau": 0.1192, + "v": 0.8928, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0338, + "v": 0.8504, + "H_a": 0.055, + "d_bind": 0.03766 + }, + { + "tau": 0.0168, + "v": 0.8353, + "H_a": 0.047, + "d_bind": 0.02037 + } + ] + }, + { + "step": 34340, + "tok": 140656640, + "ce": 2.9257, + "ppl": 18.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4949, + "loop_stats": [ + { + "tau": 0.0833, + "v": 0.8856, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0205, + "v": 0.8499, + "H_a": 0.056, + "d_bind": 0.03373 + }, + { + "tau": 0.0095, + "v": 0.8329, + "H_a": 0.05, + "d_bind": 0.02177 + } + ] + }, + { + "step": 34360, + "tok": 140738560, + "ce": 2.871, + "ppl": 17.7, + "pref_transfer_pron": 0.887, + "pref_train_sv": 0.938, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4938, + "loop_stats": [ + { + "tau": 0.119, + "v": 0.8759, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0457, + "v": 0.8312, + "H_a": 0.053, + "d_bind": 0.03457 + }, + { + "tau": 0.0324, + "v": 0.8142, + "H_a": 0.04, + "d_bind": 0.01812 + } + ] + }, + { + "step": 34380, + "tok": 140820480, + "ce": 2.6848, + "ppl": 14.7, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4928, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8961, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0287, + "v": 0.8521, + "H_a": 0.053, + "d_bind": 0.03269 + }, + { + "tau": 0.0154, + "v": 0.8306, + "H_a": 0.04, + "d_bind": 0.01705 + } + ] + }, + { + "step": 34400, + "tok": 140902400, + "ce": 2.8118, + "ppl": 16.6, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4908, + "loop_stats": [ + { + "tau": 0.1074, + "v": 0.9205, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0438, + "v": 0.8867, + "H_a": 0.062, + "d_bind": 0.03632 + }, + { + "tau": 0.0274, + "v": 0.8779, + "H_a": 0.049, + "d_bind": 0.02054 + } + ] + }, + { + "step": 34420, + "tok": 140984320, + "ce": 2.7432, + "ppl": 15.5, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4892, + "loop_stats": [ + { + "tau": 0.1279, + "v": 0.8775, + "H_a": 0.056, + "d_bind": 0.0 + }, + { + "tau": 0.0626, + "v": 0.8406, + "H_a": 0.059, + "d_bind": 0.03505 + }, + { + "tau": 0.0432, + "v": 0.8248, + "H_a": 0.048, + "d_bind": 0.0204 + } + ] + }, + { + "step": 34440, + "tok": 141066240, + "ce": 2.14, + "ppl": 8.5, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.94, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4901, + "loop_stats": [ + { + "tau": 0.1597, + "v": 0.8289, + "H_a": 0.056, + "d_bind": 0.0 + }, + { + "tau": 0.0627, + "v": 0.8019, + "H_a": 0.058, + "d_bind": 0.03661 + }, + { + "tau": 0.0473, + "v": 0.7917, + "H_a": 0.043, + "d_bind": 0.01968 + } + ] + }, + { + "step": 34460, + "tok": 141148160, + "ce": 2.763, + "ppl": 15.8, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4863, + "loop_stats": [ + { + "tau": 0.1101, + "v": 0.8717, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.8232, + "H_a": 0.057, + "d_bind": 0.03558 + }, + { + "tau": 0.0348, + "v": 0.797, + "H_a": 0.045, + "d_bind": 0.01763 + } + ] + }, + { + "step": 34480, + "tok": 141230080, + "ce": 2.5598, + "ppl": 12.9, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4863, + "loop_stats": [ + { + "tau": 0.1332, + "v": 0.8806, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0521, + "v": 0.8535, + "H_a": 0.048, + "d_bind": 0.03655 + }, + { + "tau": 0.0334, + "v": 0.8456, + "H_a": 0.037, + "d_bind": 0.01648 + } + ] + }, + { + "step": 34500, + "tok": 141312000, + "ce": 2.6085, + "ppl": 13.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4849, + "loop_stats": [ + { + "tau": 0.1134, + "v": 0.8772, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0311, + "v": 0.839, + "H_a": 0.057, + "d_bind": 0.03619 + }, + { + "tau": 0.0283, + "v": 0.8265, + "H_a": 0.046, + "d_bind": 0.0212 + } + ] + }, + { + "step": 34520, + "tok": 141393920, + "ce": 2.4307, + "ppl": 11.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4851, + "loop_stats": [ + { + "tau": 0.1297, + "v": 0.8766, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0544, + "v": 0.8431, + "H_a": 0.051, + "d_bind": 0.03598 + }, + { + "tau": 0.0427, + "v": 0.8316, + "H_a": 0.044, + "d_bind": 0.01993 + } + ] + }, + { + "step": 34540, + "tok": 141475840, + "ce": 2.8127, + "ppl": 16.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4873, + "loop_stats": [ + { + "tau": 0.1166, + "v": 0.8805, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0489, + "v": 0.8434, + "H_a": 0.051, + "d_bind": 0.03196 + }, + { + "tau": 0.0371, + "v": 0.8307, + "H_a": 0.049, + "d_bind": 0.02007 + } + ] + }, + { + "step": 34560, + "tok": 141557760, + "ce": 2.8696, + "ppl": 17.6, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.947, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4892, + "loop_stats": [ + { + "tau": 0.1033, + "v": 0.8854, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8422, + "H_a": 0.053, + "d_bind": 0.03325 + }, + { + "tau": 0.0318, + "v": 0.8342, + "H_a": 0.045, + "d_bind": 0.01953 + } + ] + }, + { + "step": 34580, + "tok": 141639680, + "ce": 3.1722, + "ppl": 23.9, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.491, + "loop_stats": [ + { + "tau": 0.1006, + "v": 0.88, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0405, + "v": 0.8287, + "H_a": 0.054, + "d_bind": 0.03092 + }, + { + "tau": 0.0214, + "v": 0.8075, + "H_a": 0.042, + "d_bind": 0.01849 + } + ] + }, + { + "step": 34600, + "tok": 141721600, + "ce": 2.5852, + "ppl": 13.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4903, + "loop_stats": [ + { + "tau": 0.1267, + "v": 0.9024, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0407, + "v": 0.8495, + "H_a": 0.055, + "d_bind": 0.03534 + }, + { + "tau": 0.0304, + "v": 0.8343, + "H_a": 0.04, + "d_bind": 0.01969 + } + ] + }, + { + "step": 34620, + "tok": 141803520, + "ce": 2.3898, + "ppl": 10.9, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4908, + "loop_stats": [ + { + "tau": 0.1285, + "v": 0.8592, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0497, + "v": 0.8301, + "H_a": 0.057, + "d_bind": 0.03271 + }, + { + "tau": 0.0313, + "v": 0.818, + "H_a": 0.041, + "d_bind": 0.01911 + } + ] + }, + { + "step": 34640, + "tok": 141885440, + "ce": 2.87, + "ppl": 17.6, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.947, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4869, + "loop_stats": [ + { + "tau": 0.1064, + "v": 0.8837, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8489, + "H_a": 0.056, + "d_bind": 0.03412 + }, + { + "tau": 0.0276, + "v": 0.8375, + "H_a": 0.043, + "d_bind": 0.02037 + } + ] + }, + { + "step": 34660, + "tok": 141967360, + "ce": 3.1075, + "ppl": 22.4, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.945, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4845, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.8649, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0371, + "v": 0.8206, + "H_a": 0.053, + "d_bind": 0.03257 + }, + { + "tau": 0.0291, + "v": 0.8007, + "H_a": 0.039, + "d_bind": 0.01744 + } + ] + }, + { + "step": 34680, + "tok": 142049280, + "ce": 3.1392, + "ppl": 23.1, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.938, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4816, + "loop_stats": [ + { + "tau": 0.0907, + "v": 0.9066, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0298, + "v": 0.87, + "H_a": 0.047, + "d_bind": 0.03212 + }, + { + "tau": 0.0215, + "v": 0.8553, + "H_a": 0.042, + "d_bind": 0.01949 + } + ] + }, + { + "step": 34700, + "tok": 142131200, + "ce": 2.8346, + "ppl": 17.0, + "pref_transfer_pron": 0.865, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4813, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.8905, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0328, + "v": 0.8416, + "H_a": 0.049, + "d_bind": 0.03225 + }, + { + "tau": 0.0256, + "v": 0.8206, + "H_a": 0.042, + "d_bind": 0.01971 + } + ] + }, + { + "step": 34720, + "tok": 142213120, + "ce": 2.8087, + "ppl": 16.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4836, + "loop_stats": [ + { + "tau": 0.0963, + "v": 0.8917, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0399, + "v": 0.8573, + "H_a": 0.056, + "d_bind": 0.03438 + }, + { + "tau": 0.0296, + "v": 0.8419, + "H_a": 0.043, + "d_bind": 0.01794 + } + ] + }, + { + "step": 34740, + "tok": 142295040, + "ce": 2.9578, + "ppl": 19.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4857, + "loop_stats": [ + { + "tau": 0.1081, + "v": 0.8478, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.05, + "v": 0.813, + "H_a": 0.058, + "d_bind": 0.03334 + }, + { + "tau": 0.0357, + "v": 0.8055, + "H_a": 0.047, + "d_bind": 0.021 + } + ] + }, + { + "step": 34760, + "tok": 142376960, + "ce": 2.7095, + "ppl": 15.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.938, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4883, + "loop_stats": [ + { + "tau": 0.1119, + "v": 0.857, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0507, + "v": 0.815, + "H_a": 0.057, + "d_bind": 0.03413 + }, + { + "tau": 0.0306, + "v": 0.8068, + "H_a": 0.046, + "d_bind": 0.01924 + } + ] + }, + { + "step": 34780, + "tok": 142458880, + "ce": 2.7826, + "ppl": 16.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4916, + "loop_stats": [ + { + "tau": 0.1111, + "v": 0.8862, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0478, + "v": 0.8664, + "H_a": 0.056, + "d_bind": 0.03634 + }, + { + "tau": 0.0217, + "v": 0.852, + "H_a": 0.047, + "d_bind": 0.02144 + } + ] + }, + { + "step": 34800, + "tok": 142540800, + "ce": 2.9404, + "ppl": 18.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.952, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4906, + "loop_stats": [ + { + "tau": 0.1231, + "v": 0.8499, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8125, + "H_a": 0.049, + "d_bind": 0.03432 + }, + { + "tau": 0.0296, + "v": 0.8029, + "H_a": 0.05, + "d_bind": 0.0191 + } + ] + }, + { + "step": 34820, + "tok": 142622720, + "ce": 2.6663, + "ppl": 14.4, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4942, + "loop_stats": [ + { + "tau": 0.1137, + "v": 0.9066, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0346, + "v": 0.8693, + "H_a": 0.053, + "d_bind": 0.03434 + }, + { + "tau": 0.0173, + "v": 0.8519, + "H_a": 0.045, + "d_bind": 0.01961 + } + ] + }, + { + "step": 34840, + "tok": 142704640, + "ce": 2.2708, + "ppl": 9.7, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.95, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4944, + "loop_stats": [ + { + "tau": 0.1184, + "v": 0.8558, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0502, + "v": 0.818, + "H_a": 0.052, + "d_bind": 0.03765 + }, + { + "tau": 0.0362, + "v": 0.8066, + "H_a": 0.047, + "d_bind": 0.01999 + } + ] + }, + { + "step": 34860, + "tok": 142786560, + "ce": 2.9368, + "ppl": 18.9, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4928, + "loop_stats": [ + { + "tau": 0.119, + "v": 0.8928, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8447, + "H_a": 0.056, + "d_bind": 0.03596 + }, + { + "tau": 0.0296, + "v": 0.8268, + "H_a": 0.044, + "d_bind": 0.01849 + } + ] + }, + { + "step": 34880, + "tok": 142868480, + "ce": 3.2442, + "ppl": 25.6, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.94, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4925, + "loop_stats": [ + { + "tau": 0.0845, + "v": 0.8866, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0315, + "v": 0.8553, + "H_a": 0.052, + "d_bind": 0.03166 + }, + { + "tau": 0.0209, + "v": 0.8496, + "H_a": 0.041, + "d_bind": 0.01738 + } + ] + }, + { + "step": 34900, + "tok": 142950400, + "ce": 2.9208, + "ppl": 18.6, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.935, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.49, + "loop_stats": [ + { + "tau": 0.1194, + "v": 0.9044, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0431, + "v": 0.8555, + "H_a": 0.059, + "d_bind": 0.03433 + }, + { + "tau": 0.0308, + "v": 0.8434, + "H_a": 0.05, + "d_bind": 0.02049 + } + ] + }, + { + "step": 34920, + "tok": 143032320, + "ce": 2.3679, + "ppl": 10.7, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.945, + "tok_s": 22539, + "peak_MiB": 5301, + "v_gain": 0.4887, + "loop_stats": [ + { + "tau": 0.1353, + "v": 0.919, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.052, + "v": 0.8794, + "H_a": 0.059, + "d_bind": 0.03735 + }, + { + "tau": 0.0417, + "v": 0.8707, + "H_a": 0.048, + "d_bind": 0.02142 + } + ] + }, + { + "step": 34940, + "tok": 143114240, + "ce": 2.6223, + "ppl": 13.8, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4885, + "loop_stats": [ + { + "tau": 0.1249, + "v": 0.8855, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.8294, + "H_a": 0.058, + "d_bind": 0.03769 + }, + { + "tau": 0.0475, + "v": 0.8166, + "H_a": 0.045, + "d_bind": 0.019 + } + ] + }, + { + "step": 34960, + "tok": 143196160, + "ce": 2.7744, + "ppl": 16.0, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.491, + "loop_stats": [ + { + "tau": 0.0984, + "v": 0.8842, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.04, + "v": 0.8265, + "H_a": 0.057, + "d_bind": 0.03516 + }, + { + "tau": 0.0304, + "v": 0.8016, + "H_a": 0.047, + "d_bind": 0.02133 + } + ] + }, + { + "step": 34980, + "tok": 143278080, + "ce": 3.2313, + "ppl": 25.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4941, + "loop_stats": [ + { + "tau": 0.0838, + "v": 0.908, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.027, + "v": 0.8645, + "H_a": 0.052, + "d_bind": 0.0362 + }, + { + "tau": 0.0187, + "v": 0.8442, + "H_a": 0.038, + "d_bind": 0.01796 + } + ] + }, + { + "step": 35000, + "tok": 143360000, + "ce": 2.5695, + "ppl": 13.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.935, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.497, + "loop_stats": [ + { + "tau": 0.1296, + "v": 0.9084, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0555, + "v": 0.8658, + "H_a": 0.059, + "d_bind": 0.03722 + }, + { + "tau": 0.0372, + "v": 0.8508, + "H_a": 0.048, + "d_bind": 0.02338 + } + ] + }, + { + "step": 35020, + "tok": 143441920, + "ce": 2.4871, + "ppl": 12.0, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.498, + "loop_stats": [ + { + "tau": 0.1131, + "v": 0.9197, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0444, + "v": 0.8681, + "H_a": 0.057, + "d_bind": 0.04093 + }, + { + "tau": 0.0379, + "v": 0.8513, + "H_a": 0.039, + "d_bind": 0.01968 + } + ] + }, + { + "step": 35040, + "tok": 143523840, + "ce": 2.7437, + "ppl": 15.5, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4981, + "loop_stats": [ + { + "tau": 0.1077, + "v": 0.9067, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8577, + "H_a": 0.061, + "d_bind": 0.03638 + }, + { + "tau": 0.0398, + "v": 0.8399, + "H_a": 0.049, + "d_bind": 0.01984 + } + ] + }, + { + "step": 35060, + "tok": 143605760, + "ce": 2.4823, + "ppl": 12.0, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5001, + "loop_stats": [ + { + "tau": 0.1157, + "v": 0.9014, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0465, + "v": 0.8548, + "H_a": 0.061, + "d_bind": 0.03945 + }, + { + "tau": 0.0395, + "v": 0.8392, + "H_a": 0.046, + "d_bind": 0.02032 + } + ] + }, + { + "step": 35080, + "tok": 143687680, + "ce": 3.0645, + "ppl": 21.4, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5004, + "loop_stats": [ + { + "tau": 0.1136, + "v": 0.9001, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0324, + "v": 0.875, + "H_a": 0.057, + "d_bind": 0.03643 + }, + { + "tau": 0.0243, + "v": 0.8672, + "H_a": 0.047, + "d_bind": 0.02072 + } + ] + }, + { + "step": 35100, + "tok": 143769600, + "ce": 2.646, + "ppl": 14.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.947, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5019, + "loop_stats": [ + { + "tau": 0.1186, + "v": 0.8819, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0452, + "v": 0.8407, + "H_a": 0.054, + "d_bind": 0.03741 + }, + { + "tau": 0.0334, + "v": 0.8332, + "H_a": 0.047, + "d_bind": 0.02016 + } + ] + }, + { + "step": 35120, + "tok": 143851520, + "ce": 2.5569, + "ppl": 12.9, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.933, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5019, + "loop_stats": [ + { + "tau": 0.1202, + "v": 0.8859, + "H_a": 0.057, + "d_bind": 0.0 + }, + { + "tau": 0.053, + "v": 0.8457, + "H_a": 0.058, + "d_bind": 0.03596 + }, + { + "tau": 0.0331, + "v": 0.8337, + "H_a": 0.055, + "d_bind": 0.02196 + } + ] + }, + { + "step": 35140, + "tok": 143933440, + "ce": 2.5972, + "ppl": 13.4, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.942, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.5016, + "loop_stats": [ + { + "tau": 0.1194, + "v": 0.9071, + "H_a": 0.057, + "d_bind": 0.0 + }, + { + "tau": 0.035, + "v": 0.8549, + "H_a": 0.053, + "d_bind": 0.03819 + }, + { + "tau": 0.0258, + "v": 0.8437, + "H_a": 0.046, + "d_bind": 0.01974 + } + ] + }, + { + "step": 35160, + "tok": 144015360, + "ce": 2.5545, + "ppl": 12.9, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.955, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5, + "loop_stats": [ + { + "tau": 0.138, + "v": 0.9039, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0539, + "v": 0.8506, + "H_a": 0.051, + "d_bind": 0.0385 + }, + { + "tau": 0.0429, + "v": 0.8385, + "H_a": 0.044, + "d_bind": 0.01787 + } + ] + }, + { + "step": 35180, + "tok": 144097280, + "ce": 2.936, + "ppl": 18.8, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.938, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4977, + "loop_stats": [ + { + "tau": 0.1063, + "v": 0.8871, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0403, + "v": 0.8392, + "H_a": 0.058, + "d_bind": 0.03733 + }, + { + "tau": 0.0229, + "v": 0.8253, + "H_a": 0.046, + "d_bind": 0.02028 + } + ] + }, + { + "step": 35200, + "tok": 144179200, + "ce": 2.7813, + "ppl": 16.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.955, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4953, + "loop_stats": [ + { + "tau": 0.1328, + "v": 0.8927, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0473, + "v": 0.8499, + "H_a": 0.052, + "d_bind": 0.03777 + }, + { + "tau": 0.0351, + "v": 0.8381, + "H_a": 0.046, + "d_bind": 0.01878 + } + ] + }, + { + "step": 35220, + "tok": 144261120, + "ce": 2.8841, + "ppl": 17.9, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4938, + "loop_stats": [ + { + "tau": 0.1225, + "v": 0.9182, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0436, + "v": 0.8721, + "H_a": 0.055, + "d_bind": 0.03762 + }, + { + "tau": 0.0274, + "v": 0.8545, + "H_a": 0.046, + "d_bind": 0.01893 + } + ] + }, + { + "step": 35240, + "tok": 144343040, + "ce": 2.9271, + "ppl": 18.7, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.947, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.4936, + "loop_stats": [ + { + "tau": 0.1095, + "v": 0.8845, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0336, + "v": 0.8433, + "H_a": 0.059, + "d_bind": 0.03742 + }, + { + "tau": 0.0249, + "v": 0.8234, + "H_a": 0.049, + "d_bind": 0.01886 + } + ] + }, + { + "step": 35260, + "tok": 144424960, + "ce": 3.2267, + "ppl": 25.2, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4918, + "loop_stats": [ + { + "tau": 0.0726, + "v": 0.8872, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0271, + "v": 0.8441, + "H_a": 0.063, + "d_bind": 0.03562 + }, + { + "tau": 0.0135, + "v": 0.8218, + "H_a": 0.046, + "d_bind": 0.02053 + } + ] + }, + { + "step": 35280, + "tok": 144506880, + "ce": 2.7368, + "ppl": 15.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4929, + "loop_stats": [ + { + "tau": 0.1248, + "v": 0.8553, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0467, + "v": 0.8026, + "H_a": 0.057, + "d_bind": 0.03625 + }, + { + "tau": 0.0335, + "v": 0.7939, + "H_a": 0.051, + "d_bind": 0.01933 + } + ] + }, + { + "step": 35300, + "tok": 144588800, + "ce": 2.678, + "ppl": 14.6, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4933, + "loop_stats": [ + { + "tau": 0.0994, + "v": 0.8837, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0347, + "v": 0.8457, + "H_a": 0.054, + "d_bind": 0.03556 + }, + { + "tau": 0.026, + "v": 0.8419, + "H_a": 0.048, + "d_bind": 0.01841 + } + ] + }, + { + "step": 35320, + "tok": 144670720, + "ce": 2.9512, + "ppl": 19.1, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4936, + "loop_stats": [ + { + "tau": 0.0983, + "v": 0.8983, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0351, + "v": 0.8644, + "H_a": 0.057, + "d_bind": 0.03648 + }, + { + "tau": 0.0223, + "v": 0.8509, + "H_a": 0.048, + "d_bind": 0.02042 + } + ] + }, + { + "step": 35340, + "tok": 144752640, + "ce": 2.8951, + "ppl": 18.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4907, + "loop_stats": [ + { + "tau": 0.1035, + "v": 0.8678, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0387, + "v": 0.8263, + "H_a": 0.059, + "d_bind": 0.03416 + }, + { + "tau": 0.0325, + "v": 0.8197, + "H_a": 0.047, + "d_bind": 0.01955 + } + ] + }, + { + "step": 35360, + "tok": 144834560, + "ce": 2.6805, + "ppl": 14.6, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.952, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.4894, + "loop_stats": [ + { + "tau": 0.1303, + "v": 0.8946, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0515, + "v": 0.8559, + "H_a": 0.055, + "d_bind": 0.03697 + }, + { + "tau": 0.034, + "v": 0.8433, + "H_a": 0.046, + "d_bind": 0.01807 + } + ] + }, + { + "step": 35380, + "tok": 144916480, + "ce": 2.1679, + "ppl": 8.7, + "pref_transfer_pron": 0.925, + "pref_train_sv": 0.95, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4878, + "loop_stats": [ + { + "tau": 0.1669, + "v": 0.8949, + "H_a": 0.055, + "d_bind": 0.0 + }, + { + "tau": 0.0548, + "v": 0.8472, + "H_a": 0.058, + "d_bind": 0.03947 + }, + { + "tau": 0.0401, + "v": 0.8422, + "H_a": 0.045, + "d_bind": 0.01935 + } + ] + }, + { + "step": 35400, + "tok": 144998400, + "ce": 2.8927, + "ppl": 18.0, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4879, + "loop_stats": [ + { + "tau": 0.0971, + "v": 0.8623, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.031, + "v": 0.8149, + "H_a": 0.056, + "d_bind": 0.03371 + }, + { + "tau": 0.0225, + "v": 0.8035, + "H_a": 0.05, + "d_bind": 0.01912 + } + ] + }, + { + "step": 35420, + "tok": 145080320, + "ce": 2.4215, + "ppl": 11.3, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.938, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4862, + "loop_stats": [ + { + "tau": 0.1067, + "v": 0.8883, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0422, + "v": 0.8568, + "H_a": 0.053, + "d_bind": 0.0356 + }, + { + "tau": 0.0327, + "v": 0.8483, + "H_a": 0.044, + "d_bind": 0.01926 + } + ] + }, + { + "step": 35440, + "tok": 145162240, + "ce": 2.7214, + "ppl": 15.2, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.938, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4876, + "loop_stats": [ + { + "tau": 0.1398, + "v": 0.8663, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0587, + "v": 0.8192, + "H_a": 0.055, + "d_bind": 0.03445 + }, + { + "tau": 0.0494, + "v": 0.8089, + "H_a": 0.041, + "d_bind": 0.01832 + } + ] + }, + { + "step": 35460, + "tok": 145244160, + "ce": 3.2203, + "ppl": 25.0, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.938, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4888, + "loop_stats": [ + { + "tau": 0.0778, + "v": 0.8952, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0191, + "v": 0.8618, + "H_a": 0.062, + "d_bind": 0.03349 + }, + { + "tau": 0.0109, + "v": 0.8444, + "H_a": 0.049, + "d_bind": 0.02092 + } + ] + }, + { + "step": 35480, + "tok": 145326080, + "ce": 3.076, + "ppl": 21.7, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.49, + "loop_stats": [ + { + "tau": 0.0882, + "v": 0.8843, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0342, + "v": 0.8426, + "H_a": 0.063, + "d_bind": 0.03445 + }, + { + "tau": 0.0207, + "v": 0.8291, + "H_a": 0.046, + "d_bind": 0.02146 + } + ] + }, + { + "step": 35500, + "tok": 145408000, + "ce": 2.5775, + "ppl": 13.2, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.933, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4906, + "loop_stats": [ + { + "tau": 0.1167, + "v": 0.8768, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.05, + "v": 0.8297, + "H_a": 0.061, + "d_bind": 0.03716 + }, + { + "tau": 0.035, + "v": 0.8145, + "H_a": 0.045, + "d_bind": 0.01943 + } + ] + }, + { + "step": 35520, + "tok": 145489920, + "ce": 2.844, + "ppl": 17.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4922, + "loop_stats": [ + { + "tau": 0.0878, + "v": 0.8798, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0344, + "v": 0.833, + "H_a": 0.059, + "d_bind": 0.03451 + }, + { + "tau": 0.0273, + "v": 0.8246, + "H_a": 0.041, + "d_bind": 0.01902 + } + ] + }, + { + "step": 35540, + "tok": 145571840, + "ce": 3.1276, + "ppl": 22.8, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.955, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4903, + "loop_stats": [ + { + "tau": 0.0674, + "v": 0.9156, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0222, + "v": 0.8775, + "H_a": 0.05, + "d_bind": 0.03194 + }, + { + "tau": 0.0135, + "v": 0.8644, + "H_a": 0.044, + "d_bind": 0.01819 + } + ] + }, + { + "step": 35560, + "tok": 145653760, + "ce": 2.7257, + "ppl": 15.3, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4926, + "loop_stats": [ + { + "tau": 0.0946, + "v": 0.8812, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0404, + "v": 0.8322, + "H_a": 0.057, + "d_bind": 0.03362 + }, + { + "tau": 0.0321, + "v": 0.8162, + "H_a": 0.043, + "d_bind": 0.01835 + } + ] + }, + { + "step": 35580, + "tok": 145735680, + "ce": 3.0816, + "ppl": 21.8, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.952, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4971, + "loop_stats": [ + { + "tau": 0.1087, + "v": 0.9042, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0445, + "v": 0.862, + "H_a": 0.059, + "d_bind": 0.03414 + }, + { + "tau": 0.034, + "v": 0.8543, + "H_a": 0.044, + "d_bind": 0.01946 + } + ] + }, + { + "step": 35600, + "tok": 145817600, + "ce": 3.3495, + "ppl": 28.5, + "pref_transfer_pron": 0.93, + "pref_train_sv": 0.942, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.4957, + "loop_stats": [ + { + "tau": 0.089, + "v": 0.9113, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0233, + "v": 0.8643, + "H_a": 0.05, + "d_bind": 0.03134 + }, + { + "tau": 0.0132, + "v": 0.836, + "H_a": 0.047, + "d_bind": 0.01998 + } + ] + }, + { + "step": 35620, + "tok": 145899520, + "ce": 2.8509, + "ppl": 17.3, + "pref_transfer_pron": 0.933, + "pref_train_sv": 0.942, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4976, + "loop_stats": [ + { + "tau": 0.0979, + "v": 0.876, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0367, + "v": 0.8534, + "H_a": 0.055, + "d_bind": 0.03442 + }, + { + "tau": 0.0286, + "v": 0.8424, + "H_a": 0.047, + "d_bind": 0.02069 + } + ] + }, + { + "step": 35640, + "tok": 145981440, + "ce": 2.7019, + "ppl": 14.9, + "pref_transfer_pron": 0.94, + "pref_train_sv": 0.938, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4984, + "loop_stats": [ + { + "tau": 0.1248, + "v": 0.8499, + "H_a": 0.064, + "d_bind": 0.0 + }, + { + "tau": 0.0509, + "v": 0.8057, + "H_a": 0.059, + "d_bind": 0.03578 + }, + { + "tau": 0.0379, + "v": 0.7857, + "H_a": 0.052, + "d_bind": 0.01897 + } + ] + }, + { + "step": 35660, + "tok": 146063360, + "ce": 2.6388, + "ppl": 14.0, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.95, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.4978, + "loop_stats": [ + { + "tau": 0.1357, + "v": 0.8718, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0517, + "v": 0.8344, + "H_a": 0.062, + "d_bind": 0.03702 + }, + { + "tau": 0.0388, + "v": 0.8274, + "H_a": 0.048, + "d_bind": 0.02 + } + ] + }, + { + "step": 35680, + "tok": 146145280, + "ce": 2.884, + "ppl": 17.9, + "pref_transfer_pron": 0.935, + "pref_train_sv": 0.938, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4969, + "loop_stats": [ + { + "tau": 0.1113, + "v": 0.9039, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0412, + "v": 0.8518, + "H_a": 0.059, + "d_bind": 0.03833 + }, + { + "tau": 0.0323, + "v": 0.8327, + "H_a": 0.048, + "d_bind": 0.02041 + } + ] + }, + { + "step": 35700, + "tok": 146227200, + "ce": 2.9778, + "ppl": 19.6, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.952, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4999, + "loop_stats": [ + { + "tau": 0.1215, + "v": 0.9011, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0458, + "v": 0.857, + "H_a": 0.057, + "d_bind": 0.03377 + }, + { + "tau": 0.0327, + "v": 0.8445, + "H_a": 0.045, + "d_bind": 0.02007 + } + ] + }, + { + "step": 35720, + "tok": 146309120, + "ce": 3.0623, + "ppl": 21.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5004, + "loop_stats": [ + { + "tau": 0.1148, + "v": 0.8764, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0395, + "v": 0.8179, + "H_a": 0.054, + "d_bind": 0.03521 + }, + { + "tau": 0.0343, + "v": 0.7996, + "H_a": 0.043, + "d_bind": 0.01882 + } + ] + }, + { + "step": 35740, + "tok": 146391040, + "ce": 2.7359, + "ppl": 15.4, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.94, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.4988, + "loop_stats": [ + { + "tau": 0.0995, + "v": 0.9292, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.035, + "v": 0.8844, + "H_a": 0.061, + "d_bind": 0.03686 + }, + { + "tau": 0.0253, + "v": 0.8703, + "H_a": 0.044, + "d_bind": 0.01929 + } + ] + }, + { + "step": 35760, + "tok": 146472960, + "ce": 2.7839, + "ppl": 16.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.947, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5014, + "loop_stats": [ + { + "tau": 0.1045, + "v": 0.886, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0433, + "v": 0.8392, + "H_a": 0.057, + "d_bind": 0.03686 + }, + { + "tau": 0.0273, + "v": 0.8214, + "H_a": 0.047, + "d_bind": 0.02043 + } + ] + }, + { + "step": 35780, + "tok": 146554880, + "ce": 2.7454, + "ppl": 15.6, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.952, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5032, + "loop_stats": [ + { + "tau": 0.1242, + "v": 0.8554, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0448, + "v": 0.7991, + "H_a": 0.058, + "d_bind": 0.03921 + }, + { + "tau": 0.0346, + "v": 0.7913, + "H_a": 0.047, + "d_bind": 0.01834 + } + ] + }, + { + "step": 35800, + "tok": 146636800, + "ce": 2.5032, + "ppl": 12.2, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.938, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5049, + "loop_stats": [ + { + "tau": 0.1201, + "v": 0.8936, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.041, + "v": 0.8654, + "H_a": 0.06, + "d_bind": 0.03719 + }, + { + "tau": 0.0208, + "v": 0.8541, + "H_a": 0.051, + "d_bind": 0.0227 + } + ] + }, + { + "step": 35820, + "tok": 146718720, + "ce": 2.3994, + "ppl": 11.0, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.935, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5047, + "loop_stats": [ + { + "tau": 0.1313, + "v": 0.8889, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0464, + "v": 0.8523, + "H_a": 0.057, + "d_bind": 0.03999 + }, + { + "tau": 0.0351, + "v": 0.8441, + "H_a": 0.05, + "d_bind": 0.01918 + } + ] + }, + { + "step": 35840, + "tok": 146800640, + "ce": 2.7403, + "ppl": 15.5, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.942, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5024, + "loop_stats": [ + { + "tau": 0.1189, + "v": 0.8749, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0571, + "v": 0.8428, + "H_a": 0.06, + "d_bind": 0.04013 + }, + { + "tau": 0.0386, + "v": 0.8276, + "H_a": 0.047, + "d_bind": 0.01821 + } + ] + }, + { + "step": 35860, + "tok": 146882560, + "ce": 2.6214, + "ppl": 13.8, + "pref_transfer_pron": 0.942, + "pref_train_sv": 0.945, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.5009, + "loop_stats": [ + { + "tau": 0.1366, + "v": 0.8904, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0587, + "v": 0.8511, + "H_a": 0.058, + "d_bind": 0.03574 + }, + { + "tau": 0.0446, + "v": 0.8403, + "H_a": 0.048, + "d_bind": 0.02016 + } + ] + }, + { + "step": 35880, + "tok": 146964480, + "ce": 3.5565, + "ppl": 35.0, + "pref_transfer_pron": 0.927, + "pref_train_sv": 0.94, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5025, + "loop_stats": [ + { + "tau": 0.0735, + "v": 0.8804, + "H_a": 0.072, + "d_bind": 0.0 + }, + { + "tau": 0.0285, + "v": 0.8421, + "H_a": 0.059, + "d_bind": 0.03304 + }, + { + "tau": 0.0175, + "v": 0.8308, + "H_a": 0.047, + "d_bind": 0.01944 + } + ] + }, + { + "step": 35900, + "tok": 147046400, + "ce": 3.1957, + "ppl": 24.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.938, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5004, + "loop_stats": [ + { + "tau": 0.0893, + "v": 0.8898, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.045, + "v": 0.8443, + "H_a": 0.063, + "d_bind": 0.03602 + }, + { + "tau": 0.0348, + "v": 0.8256, + "H_a": 0.043, + "d_bind": 0.02025 + } + ] + }, + { + "step": 35920, + "tok": 147128320, + "ce": 2.7289, + "ppl": 15.3, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5028, + "loop_stats": [ + { + "tau": 0.1072, + "v": 0.8718, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0449, + "v": 0.8479, + "H_a": 0.061, + "d_bind": 0.03719 + }, + { + "tau": 0.0299, + "v": 0.8419, + "H_a": 0.051, + "d_bind": 0.02061 + } + ] + }, + { + "step": 35940, + "tok": 147210240, + "ce": 2.8646, + "ppl": 17.5, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.503, + "loop_stats": [ + { + "tau": 0.1022, + "v": 0.8733, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0352, + "v": 0.8374, + "H_a": 0.058, + "d_bind": 0.03472 + }, + { + "tau": 0.022, + "v": 0.8265, + "H_a": 0.045, + "d_bind": 0.02118 + } + ] + }, + { + "step": 35960, + "tok": 147292160, + "ce": 2.6973, + "ppl": 14.8, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.93, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5015, + "loop_stats": [ + { + "tau": 0.1012, + "v": 0.8986, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0415, + "v": 0.8621, + "H_a": 0.056, + "d_bind": 0.03522 + }, + { + "tau": 0.0308, + "v": 0.8508, + "H_a": 0.049, + "d_bind": 0.02001 + } + ] + }, + { + "step": 35980, + "tok": 147374080, + "ce": 2.8066, + "ppl": 16.6, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.935, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.4997, + "loop_stats": [ + { + "tau": 0.1071, + "v": 0.8842, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8399, + "H_a": 0.061, + "d_bind": 0.03884 + }, + { + "tau": 0.0317, + "v": 0.8254, + "H_a": 0.044, + "d_bind": 0.02065 + } + ] + }, + { + "step": 36000, + "tok": 147456000, + "ce": 2.5106, + "ppl": 12.3, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.933, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5002, + "loop_stats": [ + { + "tau": 0.1321, + "v": 0.9002, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0547, + "v": 0.859, + "H_a": 0.061, + "d_bind": 0.03631 + }, + { + "tau": 0.0496, + "v": 0.8573, + "H_a": 0.046, + "d_bind": 0.02126 + } + ] + }, + { + "step": 36020, + "tok": 147537920, + "ce": 2.1307, + "ppl": 8.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.945, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5028, + "loop_stats": [ + { + "tau": 0.1503, + "v": 0.8656, + "H_a": 0.058, + "d_bind": 0.0 + }, + { + "tau": 0.0664, + "v": 0.8201, + "H_a": 0.05, + "d_bind": 0.0404 + }, + { + "tau": 0.0581, + "v": 0.808, + "H_a": 0.046, + "d_bind": 0.02136 + } + ] + }, + { + "step": 36040, + "tok": 147619840, + "ce": 2.9052, + "ppl": 18.3, + "pref_transfer_pron": 0.91, + "pref_train_sv": 0.935, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5031, + "loop_stats": [ + { + "tau": 0.0958, + "v": 0.9079, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0319, + "v": 0.868, + "H_a": 0.053, + "d_bind": 0.03518 + }, + { + "tau": 0.0211, + "v": 0.8506, + "H_a": 0.046, + "d_bind": 0.01974 + } + ] + }, + { + "step": 36060, + "tok": 147701760, + "ce": 2.6469, + "ppl": 14.1, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5028, + "loop_stats": [ + { + "tau": 0.0991, + "v": 0.8706, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0378, + "v": 0.8355, + "H_a": 0.06, + "d_bind": 0.03618 + }, + { + "tau": 0.0286, + "v": 0.8272, + "H_a": 0.046, + "d_bind": 0.02148 + } + ] + }, + { + "step": 36080, + "tok": 147783680, + "ce": 2.7365, + "ppl": 15.4, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.955, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5022, + "loop_stats": [ + { + "tau": 0.1153, + "v": 0.8894, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0349, + "v": 0.8441, + "H_a": 0.054, + "d_bind": 0.03271 + }, + { + "tau": 0.0298, + "v": 0.8272, + "H_a": 0.045, + "d_bind": 0.0186 + } + ] + }, + { + "step": 36100, + "tok": 147865600, + "ce": 2.7412, + "ppl": 15.5, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.938, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.5028, + "loop_stats": [ + { + "tau": 0.1037, + "v": 0.8835, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0355, + "v": 0.8378, + "H_a": 0.056, + "d_bind": 0.03389 + }, + { + "tau": 0.0275, + "v": 0.8233, + "H_a": 0.044, + "d_bind": 0.01972 + } + ] + }, + { + "step": 36120, + "tok": 147947520, + "ce": 2.8742, + "ppl": 17.7, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5052, + "loop_stats": [ + { + "tau": 0.1086, + "v": 0.9165, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0353, + "v": 0.8744, + "H_a": 0.056, + "d_bind": 0.03591 + }, + { + "tau": 0.0262, + "v": 0.8608, + "H_a": 0.046, + "d_bind": 0.0202 + } + ] + }, + { + "step": 36140, + "tok": 148029440, + "ce": 2.5027, + "ppl": 12.2, + "pref_transfer_pron": 0.897, + "pref_train_sv": 0.942, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5053, + "loop_stats": [ + { + "tau": 0.1215, + "v": 0.8937, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.043, + "v": 0.8559, + "H_a": 0.051, + "d_bind": 0.03759 + }, + { + "tau": 0.0342, + "v": 0.8478, + "H_a": 0.046, + "d_bind": 0.01949 + } + ] + }, + { + "step": 36160, + "tok": 148111360, + "ce": 2.5477, + "ppl": 12.8, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.947, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5073, + "loop_stats": [ + { + "tau": 0.1331, + "v": 0.8863, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0479, + "v": 0.8367, + "H_a": 0.06, + "d_bind": 0.03594 + }, + { + "tau": 0.0349, + "v": 0.8236, + "H_a": 0.046, + "d_bind": 0.01874 + } + ] + }, + { + "step": 36180, + "tok": 148193280, + "ce": 2.4891, + "ppl": 12.1, + "pref_transfer_pron": 0.895, + "pref_train_sv": 0.957, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5087, + "loop_stats": [ + { + "tau": 0.1342, + "v": 0.9156, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0483, + "v": 0.8669, + "H_a": 0.056, + "d_bind": 0.03619 + }, + { + "tau": 0.0305, + "v": 0.8497, + "H_a": 0.04, + "d_bind": 0.019 + } + ] + }, + { + "step": 36200, + "tok": 148275200, + "ce": 2.4013, + "ppl": 11.0, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.962, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5078, + "loop_stats": [ + { + "tau": 0.1363, + "v": 0.8908, + "H_a": 0.056, + "d_bind": 0.0 + }, + { + "tau": 0.0494, + "v": 0.8386, + "H_a": 0.055, + "d_bind": 0.03805 + }, + { + "tau": 0.0419, + "v": 0.8239, + "H_a": 0.042, + "d_bind": 0.01906 + } + ] + }, + { + "step": 36220, + "tok": 148357120, + "ce": 2.0882, + "ppl": 8.1, + "pref_transfer_pron": 0.922, + "pref_train_sv": 0.95, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5078, + "loop_stats": [ + { + "tau": 0.1625, + "v": 0.8424, + "H_a": 0.055, + "d_bind": 0.0 + }, + { + "tau": 0.0717, + "v": 0.8218, + "H_a": 0.06, + "d_bind": 0.03897 + }, + { + "tau": 0.0596, + "v": 0.8196, + "H_a": 0.051, + "d_bind": 0.02155 + } + ] + }, + { + "step": 36240, + "tok": 148439040, + "ce": 2.6825, + "ppl": 14.6, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.935, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5091, + "loop_stats": [ + { + "tau": 0.1053, + "v": 0.8708, + "H_a": 0.068, + "d_bind": 0.0 + }, + { + "tau": 0.0293, + "v": 0.8245, + "H_a": 0.057, + "d_bind": 0.03457 + }, + { + "tau": 0.0244, + "v": 0.8087, + "H_a": 0.048, + "d_bind": 0.02065 + } + ] + }, + { + "step": 36260, + "tok": 148520960, + "ce": 3.1474, + "ppl": 23.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.95, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5052, + "loop_stats": [ + { + "tau": 0.1005, + "v": 0.908, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.0364, + "v": 0.8729, + "H_a": 0.055, + "d_bind": 0.03147 + }, + { + "tau": 0.0263, + "v": 0.8559, + "H_a": 0.044, + "d_bind": 0.02049 + } + ] + }, + { + "step": 36280, + "tok": 148602880, + "ce": 3.047, + "ppl": 21.1, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.925, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5062, + "loop_stats": [ + { + "tau": 0.0989, + "v": 0.8896, + "H_a": 0.06, + "d_bind": 0.0 + }, + { + "tau": 0.0377, + "v": 0.8444, + "H_a": 0.057, + "d_bind": 0.0328 + }, + { + "tau": 0.0302, + "v": 0.8253, + "H_a": 0.046, + "d_bind": 0.0189 + } + ] + }, + { + "step": 36300, + "tok": 148684800, + "ce": 2.6952, + "ppl": 14.8, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5066, + "loop_stats": [ + { + "tau": 0.1165, + "v": 0.897, + "H_a": 0.062, + "d_bind": 0.0 + }, + { + "tau": 0.0482, + "v": 0.8524, + "H_a": 0.06, + "d_bind": 0.03641 + }, + { + "tau": 0.0377, + "v": 0.8375, + "H_a": 0.047, + "d_bind": 0.02124 + } + ] + }, + { + "step": 36320, + "tok": 148766720, + "ce": 2.428, + "ppl": 11.3, + "pref_transfer_pron": 0.907, + "pref_train_sv": 0.94, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.504, + "loop_stats": [ + { + "tau": 0.1335, + "v": 0.8831, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0532, + "v": 0.8436, + "H_a": 0.054, + "d_bind": 0.0356 + }, + { + "tau": 0.0419, + "v": 0.8341, + "H_a": 0.043, + "d_bind": 0.01867 + } + ] + }, + { + "step": 36340, + "tok": 148848640, + "ce": 2.838, + "ppl": 17.1, + "pref_transfer_pron": 0.892, + "pref_train_sv": 0.93, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5013, + "loop_stats": [ + { + "tau": 0.1017, + "v": 0.909, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0418, + "v": 0.8659, + "H_a": 0.055, + "d_bind": 0.03431 + }, + { + "tau": 0.03, + "v": 0.8528, + "H_a": 0.043, + "d_bind": 0.01823 + } + ] + }, + { + "step": 36360, + "tok": 148930560, + "ce": 2.9707, + "ppl": 19.5, + "pref_transfer_pron": 0.917, + "pref_train_sv": 0.938, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.499, + "loop_stats": [ + { + "tau": 0.1016, + "v": 0.8916, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0411, + "v": 0.8477, + "H_a": 0.053, + "d_bind": 0.03515 + }, + { + "tau": 0.0334, + "v": 0.8355, + "H_a": 0.044, + "d_bind": 0.01854 + } + ] + }, + { + "step": 36380, + "tok": 149012480, + "ce": 3.1075, + "ppl": 22.4, + "pref_transfer_pron": 0.912, + "pref_train_sv": 0.94, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.4976, + "loop_stats": [ + { + "tau": 0.0829, + "v": 0.9108, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0357, + "v": 0.8677, + "H_a": 0.055, + "d_bind": 0.03066 + }, + { + "tau": 0.0252, + "v": 0.8524, + "H_a": 0.05, + "d_bind": 0.0202 + } + ] + }, + { + "step": 36400, + "tok": 149094400, + "ce": 2.7469, + "ppl": 15.6, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.935, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5003, + "loop_stats": [ + { + "tau": 0.1068, + "v": 0.8745, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0419, + "v": 0.8222, + "H_a": 0.053, + "d_bind": 0.03447 + }, + { + "tau": 0.0326, + "v": 0.8051, + "H_a": 0.044, + "d_bind": 0.01739 + } + ] + }, + { + "step": 36420, + "tok": 149176320, + "ce": 2.9175, + "ppl": 18.5, + "pref_transfer_pron": 0.902, + "pref_train_sv": 0.942, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.5013, + "loop_stats": [ + { + "tau": 0.1128, + "v": 0.8894, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0441, + "v": 0.8378, + "H_a": 0.056, + "d_bind": 0.03264 + }, + { + "tau": 0.0336, + "v": 0.8238, + "H_a": 0.045, + "d_bind": 0.01917 + } + ] + }, + { + "step": 36440, + "tok": 149258240, + "ce": 2.852, + "ppl": 17.3, + "pref_transfer_pron": 0.882, + "pref_train_sv": 0.942, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5017, + "loop_stats": [ + { + "tau": 0.121, + "v": 0.8586, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0442, + "v": 0.8013, + "H_a": 0.051, + "d_bind": 0.03484 + }, + { + "tau": 0.0393, + "v": 0.7831, + "H_a": 0.042, + "d_bind": 0.01745 + } + ] + }, + { + "step": 36460, + "tok": 149340160, + "ce": 2.9644, + "ppl": 19.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.938, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5042, + "loop_stats": [ + { + "tau": 0.0916, + "v": 0.92, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.038, + "v": 0.8848, + "H_a": 0.056, + "d_bind": 0.03198 + }, + { + "tau": 0.0227, + "v": 0.8661, + "H_a": 0.047, + "d_bind": 0.01845 + } + ] + }, + { + "step": 36480, + "tok": 149422080, + "ce": 2.8728, + "ppl": 17.7, + "pref_transfer_pron": 0.885, + "pref_train_sv": 0.945, + "tok_s": 22540, + "peak_MiB": 5301, + "v_gain": 0.5057, + "loop_stats": [ + { + "tau": 0.0981, + "v": 0.9314, + "H_a": 0.07, + "d_bind": 0.0 + }, + { + "tau": 0.0236, + "v": 0.8838, + "H_a": 0.055, + "d_bind": 0.03304 + }, + { + "tau": 0.012, + "v": 0.8634, + "H_a": 0.046, + "d_bind": 0.02008 + } + ] + }, + { + "step": 36500, + "tok": 149504000, + "ce": 2.9815, + "ppl": 19.7, + "pref_transfer_pron": 0.915, + "pref_train_sv": 0.955, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5057, + "loop_stats": [ + { + "tau": 0.1178, + "v": 0.881, + "H_a": 0.066, + "d_bind": 0.0 + }, + { + "tau": 0.0506, + "v": 0.8345, + "H_a": 0.053, + "d_bind": 0.03485 + }, + { + "tau": 0.0402, + "v": 0.8216, + "H_a": 0.045, + "d_bind": 0.01918 + } + ] + }, + { + "step": 36520, + "tok": 149585920, + "ce": 2.888, + "ppl": 18.0, + "pref_transfer_pron": 0.92, + "pref_train_sv": 0.955, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.505, + "loop_stats": [ + { + "tau": 0.0911, + "v": 0.873, + "H_a": 0.065, + "d_bind": 0.0 + }, + { + "tau": 0.0381, + "v": 0.8285, + "H_a": 0.05, + "d_bind": 0.03522 + }, + { + "tau": 0.0323, + "v": 0.8108, + "H_a": 0.043, + "d_bind": 0.0191 + } + ] + }, + { + "step": 36540, + "tok": 149667840, + "ce": 2.8453, + "ppl": 17.2, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.947, + "tok_s": 22542, + "peak_MiB": 5301, + "v_gain": 0.5074, + "loop_stats": [ + { + "tau": 0.1049, + "v": 0.8848, + "H_a": 0.069, + "d_bind": 0.0 + }, + { + "tau": 0.0267, + "v": 0.8497, + "H_a": 0.053, + "d_bind": 0.03335 + }, + { + "tau": 0.0171, + "v": 0.8379, + "H_a": 0.05, + "d_bind": 0.02019 + } + ] + }, + { + "step": 36560, + "tok": 149749760, + "ce": 3.253, + "ppl": 25.9, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.945, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.507, + "loop_stats": [ + { + "tau": 0.0922, + "v": 0.9016, + "H_a": 0.067, + "d_bind": 0.0 + }, + { + "tau": 0.0268, + "v": 0.8597, + "H_a": 0.055, + "d_bind": 0.03261 + }, + { + "tau": 0.0176, + "v": 0.845, + "H_a": 0.046, + "d_bind": 0.02053 + } + ] + }, + { + "step": 36580, + "tok": 149831680, + "ce": 2.6641, + "ppl": 14.4, + "pref_transfer_pron": 0.9, + "pref_train_sv": 0.945, + "tok_s": 22543, + "peak_MiB": 5301, + "v_gain": 0.5079, + "loop_stats": [ + { + "tau": 0.1107, + "v": 0.8842, + "H_a": 0.063, + "d_bind": 0.0 + }, + { + "tau": 0.0388, + "v": 0.843, + "H_a": 0.058, + "d_bind": 0.03539 + }, + { + "tau": 0.0316, + "v": 0.8399, + "H_a": 0.049, + "d_bind": 0.02017 + } + ] + }, + { + "step": 36600, + "tok": 149913600, + "ce": 2.6378, + "ppl": 14.0, + "pref_transfer_pron": 0.89, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5076, + "loop_stats": [ + { + "tau": 0.1443, + "v": 0.8818, + "H_a": 0.061, + "d_bind": 0.0 + }, + { + "tau": 0.0472, + "v": 0.8465, + "H_a": 0.06, + "d_bind": 0.03653 + }, + { + "tau": 0.0397, + "v": 0.8441, + "H_a": 0.047, + "d_bind": 0.02038 + } + ] + }, + { + "step": 36620, + "tok": 149995520, + "ce": 2.24, + "ppl": 9.4, + "pref_transfer_pron": 0.905, + "pref_train_sv": 0.942, + "tok_s": 22541, + "peak_MiB": 5301, + "v_gain": 0.5099, + "loop_stats": [ + { + "tau": 0.1306, + "v": 0.8945, + "H_a": 0.059, + "d_bind": 0.0 + }, + { + "tau": 0.048, + "v": 0.8408, + "H_a": 0.054, + "d_bind": 0.03744 + }, + { + "tau": 0.039, + "v": 0.8223, + "H_a": 0.046, + "d_bind": 0.01905 + } + ] + } + ] +} \ No newline at end of file