| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0229357798165137, | |
| "eval_steps": 500, | |
| "global_step": 223, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.6207, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.25e-07, | |
| "loss": 0.4019, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.4114, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.8750000000000003e-06, | |
| "loss": 0.6422, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.3463, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.125e-06, | |
| "loss": 0.3494, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.322, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.3750000000000005e-06, | |
| "loss": 0.3168, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3176, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9942129629629636e-06, | |
| "loss": 0.3002, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.988425925925927e-06, | |
| "loss": 0.3029, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.982638888888889e-06, | |
| "loss": 0.301, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.976851851851852e-06, | |
| "loss": 0.2916, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9710648148148146e-06, | |
| "loss": 0.2866, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9652777777777786e-06, | |
| "loss": 0.2779, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.959490740740741e-06, | |
| "loss": 0.261, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.953703703703704e-06, | |
| "loss": 0.2669, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.947916666666667e-06, | |
| "loss": 0.2777, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.94212962962963e-06, | |
| "loss": 0.2544, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.936342592592593e-06, | |
| "loss": 0.2569, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.930555555555556e-06, | |
| "loss": 0.255, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.924768518518519e-06, | |
| "loss": 0.2494, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.918981481481482e-06, | |
| "loss": 0.2537, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9131944444444445e-06, | |
| "loss": 0.2457, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.907407407407408e-06, | |
| "loss": 0.2472, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.901620370370371e-06, | |
| "loss": 0.2481, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.895833333333333e-06, | |
| "loss": 0.252, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.890046296296297e-06, | |
| "loss": 0.2382, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.8842592592592595e-06, | |
| "loss": 0.2409, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.878472222222223e-06, | |
| "loss": 0.2503, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.872685185185186e-06, | |
| "loss": 0.2448, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.866898148148149e-06, | |
| "loss": 0.2443, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.861111111111111e-06, | |
| "loss": 0.2544, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.8553240740740745e-06, | |
| "loss": 0.2515, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.849537037037038e-06, | |
| "loss": 0.2437, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.84375e-06, | |
| "loss": 0.2425, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.837962962962963e-06, | |
| "loss": 0.2434, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.832175925925926e-06, | |
| "loss": 0.2232, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8263888888888895e-06, | |
| "loss": 0.2253, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.820601851851852e-06, | |
| "loss": 0.2269, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.814814814814815e-06, | |
| "loss": 0.2374, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.809027777777778e-06, | |
| "loss": 0.2224, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.803240740740741e-06, | |
| "loss": 0.2216, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.797453703703704e-06, | |
| "loss": 0.2264, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.791666666666668e-06, | |
| "loss": 0.2124, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.78587962962963e-06, | |
| "loss": 0.2206, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.780092592592593e-06, | |
| "loss": 0.2378, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.774305555555556e-06, | |
| "loss": 0.2258, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.768518518518519e-06, | |
| "loss": 0.2239, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.762731481481482e-06, | |
| "loss": 0.2225, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.756944444444445e-06, | |
| "loss": 0.2119, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.751157407407408e-06, | |
| "loss": 0.2045, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7453703703703705e-06, | |
| "loss": 0.2191, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.739583333333334e-06, | |
| "loss": 0.2177, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.733796296296297e-06, | |
| "loss": 0.2188, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.72800925925926e-06, | |
| "loss": 0.2099, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.2202, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7164351851851854e-06, | |
| "loss": 0.2084, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.710648148148149e-06, | |
| "loss": 0.1997, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.704861111111112e-06, | |
| "loss": 0.2251, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.699074074074074e-06, | |
| "loss": 0.2121, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.693287037037037e-06, | |
| "loss": 0.2112, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.6875000000000004e-06, | |
| "loss": 0.2193, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.681712962962964e-06, | |
| "loss": 0.2226, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.675925925925927e-06, | |
| "loss": 0.2129, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.670138888888889e-06, | |
| "loss": 0.2171, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.664351851851852e-06, | |
| "loss": 0.2105, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.658564814814815e-06, | |
| "loss": 0.2033, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.652777777777779e-06, | |
| "loss": 0.2063, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.646990740740741e-06, | |
| "loss": 0.2066, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.641203703703704e-06, | |
| "loss": 0.204, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.635416666666667e-06, | |
| "loss": 0.1921, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.62962962962963e-06, | |
| "loss": 0.2085, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.623842592592593e-06, | |
| "loss": 0.2188, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.618055555555556e-06, | |
| "loss": 0.2054, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.612268518518519e-06, | |
| "loss": 0.1993, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.606481481481481e-06, | |
| "loss": 0.2009, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.6006944444444446e-06, | |
| "loss": 0.2034, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.594907407407408e-06, | |
| "loss": 0.208, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.589120370370371e-06, | |
| "loss": 0.2008, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.583333333333333e-06, | |
| "loss": 0.1916, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.577546296296297e-06, | |
| "loss": 0.2045, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.5717592592592595e-06, | |
| "loss": 0.1995, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.565972222222223e-06, | |
| "loss": 0.1993, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.560185185185186e-06, | |
| "loss": 0.2049, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.554398148148148e-06, | |
| "loss": 0.1935, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.548611111111111e-06, | |
| "loss": 0.2063, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.5428240740740745e-06, | |
| "loss": 0.1968, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.537037037037038e-06, | |
| "loss": 0.1828, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.53125e-06, | |
| "loss": 0.2076, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.525462962962963e-06, | |
| "loss": 0.1774, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.519675925925926e-06, | |
| "loss": 0.2002, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.5138888888888895e-06, | |
| "loss": 0.2018, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.508101851851852e-06, | |
| "loss": 0.1934, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.502314814814815e-06, | |
| "loss": 0.2002, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.496527777777778e-06, | |
| "loss": 0.1852, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.490740740740741e-06, | |
| "loss": 0.1987, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.484953703703704e-06, | |
| "loss": 0.1922, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.479166666666667e-06, | |
| "loss": 0.1882, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.47337962962963e-06, | |
| "loss": 0.1899, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.467592592592593e-06, | |
| "loss": 0.2025, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.461805555555556e-06, | |
| "loss": 0.1926, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.456018518518519e-06, | |
| "loss": 0.1937, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.450231481481482e-06, | |
| "loss": 0.1798, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.1931, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.438657407407408e-06, | |
| "loss": 0.1726, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.4328703703703705e-06, | |
| "loss": 0.173, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.427083333333334e-06, | |
| "loss": 0.1924, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.421296296296297e-06, | |
| "loss": 0.1873, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.41550925925926e-06, | |
| "loss": 0.1807, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.409722222222222e-06, | |
| "loss": 0.1829, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.4039351851851855e-06, | |
| "loss": 0.1826, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.398148148148149e-06, | |
| "loss": 0.1747, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.392361111111112e-06, | |
| "loss": 0.1801, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.386574074074074e-06, | |
| "loss": 0.1819, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.380787037037037e-06, | |
| "loss": 0.1884, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.3750000000000005e-06, | |
| "loss": 0.1809, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.369212962962963e-06, | |
| "loss": 0.1799, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.363425925925927e-06, | |
| "loss": 0.1831, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.357638888888889e-06, | |
| "loss": 0.1602, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.351851851851852e-06, | |
| "loss": 0.1758, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.346064814814815e-06, | |
| "loss": 0.1883, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.340277777777779e-06, | |
| "loss": 0.1739, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.334490740740741e-06, | |
| "loss": 0.1733, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.328703703703704e-06, | |
| "loss": 0.1734, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.322916666666667e-06, | |
| "loss": 0.1892, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.31712962962963e-06, | |
| "loss": 0.1732, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.311342592592593e-06, | |
| "loss": 0.17, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.305555555555556e-06, | |
| "loss": 0.1841, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.299768518518519e-06, | |
| "loss": 0.18, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.293981481481481e-06, | |
| "loss": 0.167, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.288194444444445e-06, | |
| "loss": 0.1812, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.282407407407408e-06, | |
| "loss": 0.1679, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.276620370370371e-06, | |
| "loss": 0.1745, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.270833333333333e-06, | |
| "loss": 0.1857, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.265046296296297e-06, | |
| "loss": 0.1717, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.2592592592592596e-06, | |
| "loss": 0.1735, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.253472222222223e-06, | |
| "loss": 0.1645, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.247685185185186e-06, | |
| "loss": 0.1756, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.241898148148148e-06, | |
| "loss": 0.1712, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.236111111111111e-06, | |
| "loss": 0.1681, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.2303240740740746e-06, | |
| "loss": 0.1746, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.224537037037038e-06, | |
| "loss": 0.1806, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.21875e-06, | |
| "loss": 0.1662, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.212962962962963e-06, | |
| "loss": 0.1618, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.207175925925926e-06, | |
| "loss": 0.1752, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.2013888888888896e-06, | |
| "loss": 0.1681, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.195601851851852e-06, | |
| "loss": 0.1635, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.189814814814815e-06, | |
| "loss": 0.1625, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.184027777777778e-06, | |
| "loss": 0.1525, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.178240740740741e-06, | |
| "loss": 0.1608, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.172453703703704e-06, | |
| "loss": 0.1547, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.1553, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.16087962962963e-06, | |
| "loss": 0.1579, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.155092592592593e-06, | |
| "loss": 0.1554, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.149305555555556e-06, | |
| "loss": 0.1592, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.143518518518519e-06, | |
| "loss": 0.1588, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.137731481481482e-06, | |
| "loss": 0.1674, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.131944444444444e-06, | |
| "loss": 0.1578, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.126157407407408e-06, | |
| "loss": 0.1577, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.1203703703703705e-06, | |
| "loss": 0.1626, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.114583333333334e-06, | |
| "loss": 0.1599, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.108796296296297e-06, | |
| "loss": 0.1635, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.10300925925926e-06, | |
| "loss": 0.1547, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.097222222222222e-06, | |
| "loss": 0.161, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.0914351851851855e-06, | |
| "loss": 0.1591, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.085648148148149e-06, | |
| "loss": 0.1597, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.079861111111111e-06, | |
| "loss": 0.1602, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.074074074074074e-06, | |
| "loss": 0.1518, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.068287037037037e-06, | |
| "loss": 0.1486, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.0625000000000005e-06, | |
| "loss": 0.1575, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.056712962962963e-06, | |
| "loss": 0.1633, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.050925925925927e-06, | |
| "loss": 0.1597, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.045138888888889e-06, | |
| "loss": 0.1623, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.039351851851852e-06, | |
| "loss": 0.1618, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.033564814814815e-06, | |
| "loss": 0.16, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.027777777777779e-06, | |
| "loss": 0.1481, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.021990740740741e-06, | |
| "loss": 0.1452, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.016203703703704e-06, | |
| "loss": 0.1626, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.010416666666667e-06, | |
| "loss": 0.158, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.00462962962963e-06, | |
| "loss": 0.1408, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.998842592592593e-06, | |
| "loss": 0.1468, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.993055555555556e-06, | |
| "loss": 0.1486, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.987268518518519e-06, | |
| "loss": 0.1613, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.9814814814814814e-06, | |
| "loss": 0.1525, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.975694444444445e-06, | |
| "loss": 0.1568, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.969907407407408e-06, | |
| "loss": 0.1393, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.964120370370371e-06, | |
| "loss": 0.1466, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.958333333333333e-06, | |
| "loss": 0.1472, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.9525462962962964e-06, | |
| "loss": 0.144, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.94675925925926e-06, | |
| "loss": 0.1435, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.940972222222223e-06, | |
| "loss": 0.1426, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.935185185185186e-06, | |
| "loss": 0.15, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.929398148148148e-06, | |
| "loss": 0.134, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.9236111111111114e-06, | |
| "loss": 0.1379, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.917824074074074e-06, | |
| "loss": 0.1496, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.912037037037038e-06, | |
| "loss": 0.1437, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.90625e-06, | |
| "loss": 0.1479, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.900462962962963e-06, | |
| "loss": 0.135, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.894675925925926e-06, | |
| "loss": 0.1521, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.141, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.883101851851852e-06, | |
| "loss": 0.1457, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.877314814814815e-06, | |
| "loss": 0.1364, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.871527777777778e-06, | |
| "loss": 0.147, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.865740740740741e-06, | |
| "loss": 0.1506, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.859953703703704e-06, | |
| "loss": 0.139, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.854166666666667e-06, | |
| "loss": 0.1434, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.84837962962963e-06, | |
| "loss": 0.1327, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.842592592592592e-06, | |
| "loss": 0.1402, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.836805555555556e-06, | |
| "loss": 0.146, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.831018518518519e-06, | |
| "loss": 0.144, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.825231481481482e-06, | |
| "loss": 0.1411, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.819444444444444e-06, | |
| "loss": 0.1485, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.813657407407408e-06, | |
| "loss": 0.152, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.8078703703703705e-06, | |
| "loss": 0.1448, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.8020833333333333e-06, | |
| "loss": 0.1381, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.796296296296297e-06, | |
| "loss": 0.1248, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.7905092592592596e-06, | |
| "loss": 0.1404, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.7847222222222224e-06, | |
| "loss": 0.1395, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.778935185185186e-06, | |
| "loss": 0.1295, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.7731481481481487e-06, | |
| "loss": 0.1347, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.7673611111111114e-06, | |
| "loss": 0.1427, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.761574074074074e-06, | |
| "loss": 0.1248, | |
| "step": 223 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 872, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 2.4928313253424005e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |