| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 259, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 1.7394, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 1.7697, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-07, | |
| "loss": 1.7627, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.735, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.7082, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.2e-07, | |
| "loss": 1.6211, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 1.6883, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.5584, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.08e-06, | |
| "loss": 1.5924, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 1.5434, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.32e-06, | |
| "loss": 1.5874, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.5128, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.56e-06, | |
| "loss": 1.476, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 1.4491, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.8e-06, | |
| "loss": 1.4457, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.5084, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.0400000000000004e-06, | |
| "loss": 1.4479, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.16e-06, | |
| "loss": 1.4411, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.28e-06, | |
| "loss": 1.4247, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.4575, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.52e-06, | |
| "loss": 1.4149, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.64e-06, | |
| "loss": 1.3653, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.7600000000000003e-06, | |
| "loss": 1.38, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.88e-06, | |
| "loss": 1.3726, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3e-06, | |
| "loss": 1.3391, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.12e-06, | |
| "loss": 1.3389, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 1.3534, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.3600000000000004e-06, | |
| "loss": 1.3254, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.4799999999999997e-06, | |
| "loss": 1.3376, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.6e-06, | |
| "loss": 1.3301, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.72e-06, | |
| "loss": 1.2873, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 1.3017, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.96e-06, | |
| "loss": 1.2682, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.080000000000001e-06, | |
| "loss": 1.2898, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.2e-06, | |
| "loss": 1.3131, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.32e-06, | |
| "loss": 1.2749, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.44e-06, | |
| "loss": 1.2548, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.56e-06, | |
| "loss": 1.3048, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.68e-06, | |
| "loss": 1.2622, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 1.2622, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9199999999999995e-06, | |
| "loss": 1.2856, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 5.04e-06, | |
| "loss": 1.2536, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.16e-06, | |
| "loss": 1.2843, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.28e-06, | |
| "loss": 1.2241, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.4e-06, | |
| "loss": 1.2465, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5.5200000000000005e-06, | |
| "loss": 1.2621, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5.64e-06, | |
| "loss": 1.2394, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.76e-06, | |
| "loss": 1.2315, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 1.1924, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6e-06, | |
| "loss": 1.2384, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9999719894583576e-06, | |
| "loss": 1.2289, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9998879583564896e-06, | |
| "loss": 1.241, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.999747908263568e-06, | |
| "loss": 1.1919, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.999551841794844e-06, | |
| "loss": 1.2528, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.999299762611606e-06, | |
| "loss": 1.2073, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.998991675421101e-06, | |
| "loss": 1.256, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.998627585976455e-06, | |
| "loss": 1.256, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.998207501076565e-06, | |
| "loss": 1.2178, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.997731428565966e-06, | |
| "loss": 1.226, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.99719937733469e-06, | |
| "loss": 1.2135, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.996611357318102e-06, | |
| "loss": 1.2189, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.995967379496706e-06, | |
| "loss": 1.2547, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.9952674558959485e-06, | |
| "loss": 1.1776, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.994511599585988e-06, | |
| "loss": 1.1828, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.993699824681454e-06, | |
| "loss": 1.2118, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.992832146341183e-06, | |
| "loss": 1.2142, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5.991908580767936e-06, | |
| "loss": 1.1621, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5.990929145208094e-06, | |
| "loss": 1.1942, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.989893857951337e-06, | |
| "loss": 1.22, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.988802738330303e-06, | |
| "loss": 1.1666, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.9876558067202265e-06, | |
| "loss": 1.2408, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.986453084538559e-06, | |
| "loss": 1.1819, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.985194594244564e-06, | |
| "loss": 1.183, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.983880359338908e-06, | |
| "loss": 1.2063, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.982510404363209e-06, | |
| "loss": 1.168, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.981084754899591e-06, | |
| "loss": 1.2035, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.979603437570195e-06, | |
| "loss": 1.1778, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.978066480036686e-06, | |
| "loss": 1.2116, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.976473910999742e-06, | |
| "loss": 1.1549, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.974825760198511e-06, | |
| "loss": 1.1782, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 5.973122058410054e-06, | |
| "loss": 1.2074, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.971362837448781e-06, | |
| "loss": 1.208, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.969548130165845e-06, | |
| "loss": 1.1907, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.967677970448536e-06, | |
| "loss": 1.1988, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.965752393219645e-06, | |
| "loss": 1.17, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.963771434436812e-06, | |
| "loss": 1.1714, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.961735131091857e-06, | |
| "loss": 1.1543, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.959643521210085e-06, | |
| "loss": 1.226, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.957496643849582e-06, | |
| "loss": 1.1877, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5.955294539100477e-06, | |
| "loss": 1.1982, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5.953037248084204e-06, | |
| "loss": 1.1428, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.950724812952725e-06, | |
| "loss": 1.1791, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.948357276887745e-06, | |
| "loss": 1.1913, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.945934684099912e-06, | |
| "loss": 1.1774, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.94345707982798e-06, | |
| "loss": 1.1502, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.9409245103379775e-06, | |
| "loss": 1.1519, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.938337022922332e-06, | |
| "loss": 1.1538, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.935694665898992e-06, | |
| "loss": 1.1838, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 5.932997488610526e-06, | |
| "loss": 1.1512, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.9302455414231975e-06, | |
| "loss": 1.1567, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.927438875726029e-06, | |
| "loss": 1.1236, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.924577543929837e-06, | |
| "loss": 1.1731, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.921661599466257e-06, | |
| "loss": 1.1685, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.918691096786745e-06, | |
| "loss": 1.1885, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.915666091361561e-06, | |
| "loss": 1.2044, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.912586639678732e-06, | |
| "loss": 1.1825, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.909452799242996e-06, | |
| "loss": 1.1341, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.906264628574734e-06, | |
| "loss": 1.1393, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.90302218720887e-06, | |
| "loss": 1.1612, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.899725535693762e-06, | |
| "loss": 1.1081, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.896374735590074e-06, | |
| "loss": 1.1474, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.892969849469624e-06, | |
| "loss": 1.2151, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.889510940914214e-06, | |
| "loss": 1.1313, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.885998074514447e-06, | |
| "loss": 1.1518, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.882431315868514e-06, | |
| "loss": 1.1827, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.878810731580978e-06, | |
| "loss": 1.1554, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.875136389261522e-06, | |
| "loss": 1.1353, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.871408357523693e-06, | |
| "loss": 1.2146, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.867626705983617e-06, | |
| "loss": 1.1441, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.863791505258699e-06, | |
| "loss": 1.1537, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.859902826966304e-06, | |
| "loss": 1.1486, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.855960743722423e-06, | |
| "loss": 1.1022, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.851965329140315e-06, | |
| "loss": 1.1632, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.847916657829129e-06, | |
| "loss": 1.1447, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.8438148053925185e-06, | |
| "loss": 1.1589, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.839659848427221e-06, | |
| "loss": 1.1597, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.835451864521632e-06, | |
| "loss": 1.1661, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.831190932254359e-06, | |
| "loss": 1.1426, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.826877131192748e-06, | |
| "loss": 1.1405, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.822510541891403e-06, | |
| "loss": 1.1292, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.818091245890677e-06, | |
| "loss": 1.1677, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.8136193257151535e-06, | |
| "loss": 1.1283, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.809094864872104e-06, | |
| "loss": 1.1596, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.804517947849927e-06, | |
| "loss": 1.1647, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.799888660116573e-06, | |
| "loss": 1.1448, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.795207088117946e-06, | |
| "loss": 1.1514, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.790473319276292e-06, | |
| "loss": 1.1466, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.785687441988563e-06, | |
| "loss": 1.15, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.780849545624768e-06, | |
| "loss": 1.1518, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.775959720526308e-06, | |
| "loss": 1.0718, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.7710180580042804e-06, | |
| "loss": 1.1932, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.766024650337782e-06, | |
| "loss": 1.1303, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.760979590772182e-06, | |
| "loss": 1.141, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.75588297351738e-06, | |
| "loss": 1.1325, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.750734893746051e-06, | |
| "loss": 1.1484, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.7455354475918615e-06, | |
| "loss": 1.1534, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.740284732147681e-06, | |
| "loss": 1.15, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.7349828454637654e-06, | |
| "loss": 1.0994, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.7296298865459275e-06, | |
| "loss": 1.1786, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.724225955353685e-06, | |
| "loss": 1.1118, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.718771152798398e-06, | |
| "loss": 1.1282, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.713265580741383e-06, | |
| "loss": 1.0907, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.70770934199201e-06, | |
| "loss": 1.1562, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.7021025403057835e-06, | |
| "loss": 1.182, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.696445280382404e-06, | |
| "loss": 1.1148, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.690737667863816e-06, | |
| "loss": 1.1172, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.684979809332232e-06, | |
| "loss": 1.1513, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.6791718123081425e-06, | |
| "loss": 1.1334, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.673313785248308e-06, | |
| "loss": 1.118, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.667405837543736e-06, | |
| "loss": 1.1427, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.661448079517636e-06, | |
| "loss": 1.1403, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.6554406224233634e-06, | |
| "loss": 1.1438, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.649383578442334e-06, | |
| "loss": 1.1738, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.643277060681938e-06, | |
| "loss": 1.1444, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.63712118317342e-06, | |
| "loss": 1.0963, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.630916060869758e-06, | |
| "loss": 1.1497, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.624661809643507e-06, | |
| "loss": 1.1658, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.618358546284646e-06, | |
| "loss": 1.1281, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.612006388498386e-06, | |
| "loss": 1.124, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.605605454902983e-06, | |
| "loss": 1.0731, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.599155865027513e-06, | |
| "loss": 1.1296, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.592657739309649e-06, | |
| "loss": 1.1144, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.5861111990934034e-06, | |
| "loss": 1.1359, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.579516366626867e-06, | |
| "loss": 1.1321, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.572873365059928e-06, | |
| "loss": 1.1087, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.5661823184419655e-06, | |
| "loss": 1.0883, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.559443351719542e-06, | |
| "loss": 1.1151, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.5526565907340604e-06, | |
| "loss": 1.1163, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.545822162219423e-06, | |
| "loss": 1.1344, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.538940193799659e-06, | |
| "loss": 1.1188, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.532010813986544e-06, | |
| "loss": 1.084, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.5250341521772e-06, | |
| "loss": 1.1209, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.518010338651676e-06, | |
| "loss": 1.1116, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.51093950457052e-06, | |
| "loss": 1.1396, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.5038217819723285e-06, | |
| "loss": 1.1254, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.496657303771276e-06, | |
| "loss": 1.1158, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 5.489446203754641e-06, | |
| "loss": 1.162, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.482188616580301e-06, | |
| "loss": 1.1284, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.4748846777742195e-06, | |
| "loss": 1.1187, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 5.467534523727921e-06, | |
| "loss": 1.1254, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.4601382916959355e-06, | |
| "loss": 1.1071, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 5.452696119793238e-06, | |
| "loss": 1.1149, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.445208146992674e-06, | |
| "loss": 1.1392, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.437674513122359e-06, | |
| "loss": 1.1233, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.43009535886307e-06, | |
| "loss": 1.0951, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.422470825745617e-06, | |
| "loss": 1.0691, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.414801056148203e-06, | |
| "loss": 1.0983, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.407086193293761e-06, | |
| "loss": 1.1608, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5.399326381247282e-06, | |
| "loss": 1.1355, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5.391521764913127e-06, | |
| "loss": 1.1247, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.383672490032313e-06, | |
| "loss": 1.0917, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.375778703179803e-06, | |
| "loss": 1.1515, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5.3678405517617595e-06, | |
| "loss": 1.1357, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5.359858184012799e-06, | |
| "loss": 1.1481, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5.3518317489932144e-06, | |
| "loss": 1.1765, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.343761396586203e-06, | |
| "loss": 1.1124, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.335647277495057e-06, | |
| "loss": 1.1391, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.327489543240359e-06, | |
| "loss": 1.113, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.3192883461571465e-06, | |
| "loss": 1.1074, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.311043839392064e-06, | |
| "loss": 1.1039, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.302756176900516e-06, | |
| "loss": 1.1237, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.294425513443777e-06, | |
| "loss": 1.1101, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.2860520045861104e-06, | |
| "loss": 1.0947, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.277635806691863e-06, | |
| "loss": 1.1155, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.2691770769225414e-06, | |
| "loss": 1.1326, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.260675973233882e-06, | |
| "loss": 1.1411, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5.252132654372897e-06, | |
| "loss": 1.112, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5.243547279874911e-06, | |
| "loss": 1.1079, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.234920010060586e-06, | |
| "loss": 1.0891, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.22625100603292e-06, | |
| "loss": 1.1221, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.217540429674246e-06, | |
| "loss": 1.1151, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.208788443643207e-06, | |
| "loss": 1.1116, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.1999952113717135e-06, | |
| "loss": 1.0972, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.191160897061898e-06, | |
| "loss": 1.124, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.1822856656830485e-06, | |
| "loss": 1.1053, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.173369682968522e-06, | |
| "loss": 1.1473, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.1644131154126544e-06, | |
| "loss": 1.1131, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.1554161302676544e-06, | |
| "loss": 1.0935, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.146378895540472e-06, | |
| "loss": 1.0677, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.137301579989665e-06, | |
| "loss": 1.1269, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.128184353122252e-06, | |
| "loss": 1.0783, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.119027385190542e-06, | |
| "loss": 1.1059, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.109830847188955e-06, | |
| "loss": 1.0468, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.100594910850832e-06, | |
| "loss": 1.0851, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.091319748645226e-06, | |
| "loss": 1.0387, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.082005533773682e-06, | |
| "loss": 1.1361, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.072652440167001e-06, | |
| "loss": 1.1285, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.063260642481997e-06, | |
| "loss": 1.0934, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.053830316098228e-06, | |
| "loss": 1.1059, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.044361637114729e-06, | |
| "loss": 1.1031, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.034854782346716e-06, | |
| "loss": 1.0757, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.0253099293222925e-06, | |
| "loss": 1.1218, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.015727256279126e-06, | |
| "loss": 1.12, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.0061069421611245e-06, | |
| "loss": 1.0985, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.996449166615094e-06, | |
| "loss": 1.1192, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.986754109987384e-06, | |
| "loss": 1.0662, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.97702195332052e-06, | |
| "loss": 1.0671, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.967252878349821e-06, | |
| "loss": 1.1043, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.9574470675000085e-06, | |
| "loss": 1.1036, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.947604703881797e-06, | |
| "loss": 1.0729, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.937725971288479e-06, | |
| "loss": 1.1265, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.927811054192486e-06, | |
| "loss": 1.0854, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.9178601377419506e-06, | |
| "loss": 1.1334, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.907873407757248e-06, | |
| "loss": 1.1184, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.89785105072752e-06, | |
| "loss": 1.105, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.8877932538072e-06, | |
| "loss": 1.1124, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.877700204812516e-06, | |
| "loss": 1.0862, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.867572092217977e-06, | |
| "loss": 1.0856, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.857409105152865e-06, | |
| "loss": 1.0905, | |
| "step": 259 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 777, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 5.793364463679701e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |