| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9987389659520807, | |
| "global_step": 594, | |
| "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": 9.2391, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 9.104, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0, | |
| "loss": 8.9021, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-06, | |
| "loss": 8.8682, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-06, | |
| "loss": 8.8262, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-06, | |
| "loss": 8.4047, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-06, | |
| "loss": 8.3502, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1e-05, | |
| "loss": 7.9771, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2e-05, | |
| "loss": 7.8414, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4e-05, | |
| "loss": 7.6606, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.6e-05, | |
| "loss": 7.5532, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.8e-05, | |
| "loss": 7.5309, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2e-05, | |
| "loss": 7.3389, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.2e-05, | |
| "loss": 7.5473, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4e-05, | |
| "loss": 7.3029, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.6e-05, | |
| "loss": 7.3217, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.8e-05, | |
| "loss": 7.3244, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-05, | |
| "loss": 7.2319, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.2e-05, | |
| "loss": 7.1778, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 7.257, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.6e-05, | |
| "loss": 7.3211, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.8e-05, | |
| "loss": 7.2754, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4e-05, | |
| "loss": 7.268, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.2000000000000004e-05, | |
| "loss": 7.2223, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-05, | |
| "loss": 7.0232, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.6e-05, | |
| "loss": 7.377, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.8e-05, | |
| "loss": 7.1471, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-05, | |
| "loss": 7.0311, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.2e-05, | |
| "loss": 7.1356, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.4e-05, | |
| "loss": 7.1118, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.6e-05, | |
| "loss": 7.2032, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.800000000000001e-05, | |
| "loss": 7.1625, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6e-05, | |
| "loss": 7.0952, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.2e-05, | |
| "loss": 7.0539, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.4e-05, | |
| "loss": 7.0215, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.6e-05, | |
| "loss": 7.1635, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 7.118, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7.000000000000001e-05, | |
| "loss": 7.1143, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.2e-05, | |
| "loss": 7.1712, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.4e-05, | |
| "loss": 7.0529, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.6e-05, | |
| "loss": 7.0855, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.8e-05, | |
| "loss": 7.1638, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8e-05, | |
| "loss": 6.9741, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.2e-05, | |
| "loss": 7.0521, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.400000000000001e-05, | |
| "loss": 7.1212, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.599999999999999e-05, | |
| "loss": 7.0775, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.8e-05, | |
| "loss": 6.9064, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.999999999999999e-05, | |
| "loss": 7.0588, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.2e-05, | |
| "loss": 6.7153, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.400000000000001e-05, | |
| "loss": 6.5244, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.6e-05, | |
| "loss": 7.3138, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.800000000000001e-05, | |
| "loss": 7.2796, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001, | |
| "loss": 7.0741, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000102, | |
| "loss": 7.1467, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000104, | |
| "loss": 7.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000106, | |
| "loss": 7.1375, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000108, | |
| "loss": 6.9579, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00011, | |
| "loss": 7.1664, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000112, | |
| "loss": 7.1162, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000114, | |
| "loss": 6.9863, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00011600000000000001, | |
| "loss": 7.1279, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000118, | |
| "loss": 7.0931, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00012, | |
| "loss": 7.0619, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000122, | |
| "loss": 7.0442, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000124, | |
| "loss": 7.0437, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000126, | |
| "loss": 7.0038, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000128, | |
| "loss": 7.05, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00013000000000000002, | |
| "loss": 6.863, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000132, | |
| "loss": 7.0607, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000134, | |
| "loss": 7.0143, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 7.041, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013800000000000002, | |
| "loss": 7.151, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00014000000000000001, | |
| "loss": 6.8324, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00014199999999999998, | |
| "loss": 6.9986, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000144, | |
| "loss": 7.0604, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000146, | |
| "loss": 7.0742, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000148, | |
| "loss": 7.0835, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00015, | |
| "loss": 7.0913, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000152, | |
| "loss": 6.9772, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000154, | |
| "loss": 7.025, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000156, | |
| "loss": 7.0169, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000158, | |
| "loss": 7.0588, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00016, | |
| "loss": 7.0161, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000162, | |
| "loss": 7.0474, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000164, | |
| "loss": 6.9451, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00016600000000000002, | |
| "loss": 6.8814, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00016800000000000002, | |
| "loss": 7.0686, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00017, | |
| "loss": 7.0055, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00017199999999999998, | |
| "loss": 6.9607, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000174, | |
| "loss": 7.0966, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000176, | |
| "loss": 6.9914, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000178, | |
| "loss": 7.0491, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 6.9837, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.000182, | |
| "loss": 6.9912, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.000184, | |
| "loss": 6.8828, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.000186, | |
| "loss": 6.8851, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00018800000000000002, | |
| "loss": 6.9958, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00019, | |
| "loss": 6.6052, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000192, | |
| "loss": 6.8389, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000194, | |
| "loss": 6.4098, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00019600000000000002, | |
| "loss": 7.1685, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00019800000000000002, | |
| "loss": 7.0531, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 7.0706, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000202, | |
| "loss": 6.9564, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000204, | |
| "loss": 6.9889, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000206, | |
| "loss": 7.0398, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000208, | |
| "loss": 7.1539, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00021, | |
| "loss": 7.0754, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000212, | |
| "loss": 7.0787, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000214, | |
| "loss": 7.044, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000216, | |
| "loss": 6.9558, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000218, | |
| "loss": 6.9957, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00022, | |
| "loss": 6.9355, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000222, | |
| "loss": 7.0677, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000224, | |
| "loss": 7.0284, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00022600000000000002, | |
| "loss": 6.9798, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000228, | |
| "loss": 7.0059, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00023, | |
| "loss": 6.9782, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00023200000000000003, | |
| "loss": 7.0379, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00023400000000000002, | |
| "loss": 7.0151, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000236, | |
| "loss": 7.1422, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00023799999999999998, | |
| "loss": 7.1631, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00024, | |
| "loss": 7.0311, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.000242, | |
| "loss": 7.0971, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.000244, | |
| "loss": 6.9513, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.000246, | |
| "loss": 7.0559, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.000248, | |
| "loss": 6.963, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00025, | |
| "loss": 7.0279, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000252, | |
| "loss": 7.0081, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000254, | |
| "loss": 7.1089, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000256, | |
| "loss": 7.0063, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00025800000000000004, | |
| "loss": 6.9755, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00026000000000000003, | |
| "loss": 7.0143, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000262, | |
| "loss": 7.1515, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000264, | |
| "loss": 7.0614, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000266, | |
| "loss": 7.1099, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000268, | |
| "loss": 6.9085, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00027, | |
| "loss": 7.0223, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00027200000000000005, | |
| "loss": 6.9545, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00027400000000000005, | |
| "loss": 6.993, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00027600000000000004, | |
| "loss": 7.0001, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00027800000000000004, | |
| "loss": 7.1251, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00028000000000000003, | |
| "loss": 7.1047, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00028199999999999997, | |
| "loss": 6.7781, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00028399999999999996, | |
| "loss": 6.8235, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00028599999999999996, | |
| "loss": 7.1905, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000288, | |
| "loss": 6.8899, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00029, | |
| "loss": 6.7851, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000292, | |
| "loss": 6.8372, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000294, | |
| "loss": 6.4942, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000296, | |
| "loss": 7.1939, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000298, | |
| "loss": 7.1799, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0003, | |
| "loss": 7.1226, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000302, | |
| "loss": 6.963, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000304, | |
| "loss": 7.0494, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000306, | |
| "loss": 7.0654, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000308, | |
| "loss": 7.0491, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00031, | |
| "loss": 7.0751, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000312, | |
| "loss": 6.9384, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000314, | |
| "loss": 7.1189, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000316, | |
| "loss": 6.9326, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00031800000000000003, | |
| "loss": 6.9468, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00032, | |
| "loss": 7.0409, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000322, | |
| "loss": 7.0507, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000324, | |
| "loss": 6.9408, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000326, | |
| "loss": 7.1385, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000328, | |
| "loss": 7.0965, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00033, | |
| "loss": 7.0572, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00033200000000000005, | |
| "loss": 6.9633, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00033400000000000004, | |
| "loss": 7.0541, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00033600000000000004, | |
| "loss": 6.9672, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00033800000000000003, | |
| "loss": 6.8879, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00034, | |
| "loss": 6.9559, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.000342, | |
| "loss": 6.8725, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00034399999999999996, | |
| "loss": 6.9745, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000346, | |
| "loss": 7.0163, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000348, | |
| "loss": 7.0641, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00035, | |
| "loss": 6.9334, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000352, | |
| "loss": 7.0079, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000354, | |
| "loss": 7.0947, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000356, | |
| "loss": 7.0143, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000358, | |
| "loss": 7.087, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00035999999999999997, | |
| "loss": 7.0131, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000362, | |
| "loss": 6.8973, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000364, | |
| "loss": 7.1537, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000366, | |
| "loss": 6.9218, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000368, | |
| "loss": 7.0113, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00037, | |
| "loss": 6.9827, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000372, | |
| "loss": 7.0805, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000374, | |
| "loss": 7.0587, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00037600000000000003, | |
| "loss": 7.0688, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000378, | |
| "loss": 7.0879, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00038, | |
| "loss": 6.9182, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000382, | |
| "loss": 6.9707, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000384, | |
| "loss": 6.9151, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000386, | |
| "loss": 6.7898, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000388, | |
| "loss": 6.9734, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00039000000000000005, | |
| "loss": 6.9121, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00039200000000000004, | |
| "loss": 6.6485, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00039400000000000004, | |
| "loss": 6.6154, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00039600000000000003, | |
| "loss": 7.0778, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000398, | |
| "loss": 7.0853, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0004, | |
| "loss": 7.1074, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000402, | |
| "loss": 6.8628, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000404, | |
| "loss": 6.9699, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00040600000000000006, | |
| "loss": 6.9795, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000408, | |
| "loss": 7.0912, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00041, | |
| "loss": 7.0847, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000412, | |
| "loss": 7.0106, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000414, | |
| "loss": 7.1056, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000416, | |
| "loss": 7.0423, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00041799999999999997, | |
| "loss": 7.0226, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00042, | |
| "loss": 7.0216, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000422, | |
| "loss": 7.0556, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000424, | |
| "loss": 7.0019, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000426, | |
| "loss": 7.0848, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000428, | |
| "loss": 6.8998, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00043, | |
| "loss": 6.9194, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000432, | |
| "loss": 6.9024, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00043400000000000003, | |
| "loss": 7.1083, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000436, | |
| "loss": 6.9852, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000438, | |
| "loss": 7.0352, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00044, | |
| "loss": 7.0216, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000442, | |
| "loss": 7.0737, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000444, | |
| "loss": 7.0748, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000446, | |
| "loss": 6.9234, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000448, | |
| "loss": 6.9104, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00045000000000000004, | |
| "loss": 7.0043, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00045200000000000004, | |
| "loss": 7.0571, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00045400000000000003, | |
| "loss": 7.0416, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.000456, | |
| "loss": 7.0239, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.000458, | |
| "loss": 6.9304, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00046, | |
| "loss": 7.0238, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.000462, | |
| "loss": 7.0501, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00046400000000000006, | |
| "loss": 7.0763, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00046600000000000005, | |
| "loss": 6.893, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00046800000000000005, | |
| "loss": 6.9541, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00047, | |
| "loss": 6.9877, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000472, | |
| "loss": 7.0107, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000474, | |
| "loss": 7.0396, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00047599999999999997, | |
| "loss": 7.1206, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00047799999999999996, | |
| "loss": 7.0092, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00048, | |
| "loss": 7.0429, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000482, | |
| "loss": 6.9709, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000484, | |
| "loss": 6.8753, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000486, | |
| "loss": 6.8831, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000488, | |
| "loss": 6.8629, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00049, | |
| "loss": 6.905, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000492, | |
| "loss": 6.7441, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000494, | |
| "loss": 6.5621, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000496, | |
| "loss": 7.2098, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000498, | |
| "loss": 7.2084, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005, | |
| "loss": 7.0214, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005020000000000001, | |
| "loss": 7.0158, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000504, | |
| "loss": 7.1246, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000506, | |
| "loss": 7.0502, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000508, | |
| "loss": 7.0624, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00051, | |
| "loss": 6.9986, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000512, | |
| "loss": 7.1072, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000514, | |
| "loss": 7.0238, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005160000000000001, | |
| "loss": 7.0155, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000518, | |
| "loss": 6.9996, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005200000000000001, | |
| "loss": 7.0465, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000522, | |
| "loss": 6.8355, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000524, | |
| "loss": 6.9596, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000526, | |
| "loss": 6.9608, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000528, | |
| "loss": 7.0486, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0005300000000000001, | |
| "loss": 6.9112, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000532, | |
| "loss": 7.0594, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0005340000000000001, | |
| "loss": 6.9925, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.000536, | |
| "loss": 7.0339, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005380000000000001, | |
| "loss": 7.0785, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00054, | |
| "loss": 7.0008, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005420000000000001, | |
| "loss": 6.976, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005440000000000001, | |
| "loss": 7.0215, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.000546, | |
| "loss": 7.0251, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005480000000000001, | |
| "loss": 7.0722, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00055, | |
| "loss": 7.0051, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005520000000000001, | |
| "loss": 7.0019, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.000554, | |
| "loss": 7.0538, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005560000000000001, | |
| "loss": 7.0292, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.000558, | |
| "loss": 7.0564, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005600000000000001, | |
| "loss": 6.8884, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005620000000000001, | |
| "loss": 7.0051, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005639999999999999, | |
| "loss": 7.0457, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.000566, | |
| "loss": 6.9338, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005679999999999999, | |
| "loss": 7.1379, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00057, | |
| "loss": 6.8807, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005719999999999999, | |
| "loss": 6.9174, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.000574, | |
| "loss": 7.0579, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.000576, | |
| "loss": 7.0882, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.000578, | |
| "loss": 6.9218, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00058, | |
| "loss": 6.9869, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005819999999999999, | |
| "loss": 7.0547, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.000584, | |
| "loss": 6.8407, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005859999999999999, | |
| "loss": 6.8458, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.000588, | |
| "loss": 6.7859, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00059, | |
| "loss": 6.8682, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.000592, | |
| "loss": 6.6934, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.000594, | |
| "loss": 6.5845, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000596, | |
| "loss": 7.236, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000598, | |
| "loss": 7.2218, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0006, | |
| "loss": 7.1656, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000602, | |
| "loss": 7.0663, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000604, | |
| "loss": 7.1259, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000606, | |
| "loss": 7.1706, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000608, | |
| "loss": 7.0302, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00061, | |
| "loss": 7.132, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000612, | |
| "loss": 7.0026, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000614, | |
| "loss": 7.0629, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000616, | |
| "loss": 6.9223, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0006180000000000001, | |
| "loss": 7.16, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00062, | |
| "loss": 6.9756, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.000622, | |
| "loss": 6.9914, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.000624, | |
| "loss": 7.018, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.000626, | |
| "loss": 6.948, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.000628, | |
| "loss": 6.9204, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00063, | |
| "loss": 7.1444, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000632, | |
| "loss": 7.0219, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000634, | |
| "loss": 7.0609, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0006360000000000001, | |
| "loss": 6.9684, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000638, | |
| "loss": 6.9689, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00064, | |
| "loss": 6.9515, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000642, | |
| "loss": 6.9158, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000644, | |
| "loss": 7.1327, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000646, | |
| "loss": 7.0171, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000648, | |
| "loss": 7.1963, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0006500000000000001, | |
| "loss": 7.0119, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000652, | |
| "loss": 6.9315, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0006540000000000001, | |
| "loss": 7.0296, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.000656, | |
| "loss": 6.9909, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0006580000000000001, | |
| "loss": 6.9743, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00066, | |
| "loss": 6.9316, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.000662, | |
| "loss": 7.0359, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0006640000000000001, | |
| "loss": 7.0403, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.000666, | |
| "loss": 6.9493, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0006680000000000001, | |
| "loss": 6.991, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00067, | |
| "loss": 6.9762, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0006720000000000001, | |
| "loss": 7.0749, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.000674, | |
| "loss": 7.1129, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0006760000000000001, | |
| "loss": 6.9891, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0006780000000000001, | |
| "loss": 6.9923, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00068, | |
| "loss": 6.8966, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0006820000000000001, | |
| "loss": 7.0609, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.000684, | |
| "loss": 6.9641, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0006860000000000001, | |
| "loss": 6.7934, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0006879999999999999, | |
| "loss": 6.814, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00069, | |
| "loss": 7.1134, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.000692, | |
| "loss": 6.7713, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.000694, | |
| "loss": 6.7095, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.000696, | |
| "loss": 7.2116, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0006979999999999999, | |
| "loss": 7.1415, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0007, | |
| "loss": 7.0851, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0007019999999999999, | |
| "loss": 7.0795, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.000704, | |
| "loss": 7.0815, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0007059999999999999, | |
| "loss": 7.1496, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.000708, | |
| "loss": 7.0391, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00071, | |
| "loss": 7.0427, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.000712, | |
| "loss": 7.006, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000714, | |
| "loss": 7.0213, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000716, | |
| "loss": 7.0563, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000718, | |
| "loss": 6.9326, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0007199999999999999, | |
| "loss": 285.9326, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000722, | |
| "loss": 7.0385, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000724, | |
| "loss": 7.1396, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000726, | |
| "loss": 6.8595, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000728, | |
| "loss": 7.1262, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00073, | |
| "loss": 6.9156, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000732, | |
| "loss": 6.9526, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000734, | |
| "loss": 7.0985, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000736, | |
| "loss": 6.993, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000738, | |
| "loss": 6.9229, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00074, | |
| "loss": 6.9861, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000742, | |
| "loss": 6.911, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000744, | |
| "loss": 7.0069, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000746, | |
| "loss": 6.9277, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000748, | |
| "loss": 6.956, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00075, | |
| "loss": 7.0714, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0007520000000000001, | |
| "loss": 6.9807, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.000754, | |
| "loss": 7.108, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.000756, | |
| "loss": 7.0734, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.000758, | |
| "loss": 7.0057, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00076, | |
| "loss": 6.9243, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.000762, | |
| "loss": 7.031, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.000764, | |
| "loss": 7.0799, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0007660000000000001, | |
| "loss": 7.0769, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.000768, | |
| "loss": 7.1266, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0007700000000000001, | |
| "loss": 7.0542, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.000772, | |
| "loss": 6.965, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0007740000000000001, | |
| "loss": 7.0061, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000776, | |
| "loss": 7.089, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000778, | |
| "loss": 7.0489, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0007800000000000001, | |
| "loss": 7.0362, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000782, | |
| "loss": 6.9978, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0007840000000000001, | |
| "loss": 6.7938, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.000786, | |
| "loss": 6.7785, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0007880000000000001, | |
| "loss": 6.9116, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00079, | |
| "loss": 7.0112, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0007920000000000001, | |
| "loss": 6.6276, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0007940000000000001, | |
| "loss": 6.4829, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.000796, | |
| "loss": 7.3835, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0007980000000000001, | |
| "loss": 7.2009, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0008, | |
| "loss": 7.1478, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0008020000000000001, | |
| "loss": 7.112, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.000804, | |
| "loss": 7.0844, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0008060000000000001, | |
| "loss": 6.9843, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.000808, | |
| "loss": 6.9488, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0008100000000000001, | |
| "loss": 7.1111, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0008120000000000001, | |
| "loss": 6.9722, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0008139999999999999, | |
| "loss": 6.9885, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.000816, | |
| "loss": 7.0014, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0008179999999999999, | |
| "loss": 7.0313, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00082, | |
| "loss": 6.8849, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0008219999999999999, | |
| "loss": 7.0015, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000824, | |
| "loss": 7.1869, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000826, | |
| "loss": 6.9322, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000828, | |
| "loss": 7.0145, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00083, | |
| "loss": 6.812, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000832, | |
| "loss": 6.9888, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.000834, | |
| "loss": 7.0677, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0008359999999999999, | |
| "loss": 6.9418, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.000838, | |
| "loss": 6.9906, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00084, | |
| "loss": 7.0385, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.000842, | |
| "loss": 7.0314, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.000844, | |
| "loss": 6.9856, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000846, | |
| "loss": 7.0841, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000848, | |
| "loss": 6.9897, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00085, | |
| "loss": 6.9838, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000852, | |
| "loss": 7.1931, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000854, | |
| "loss": 6.9795, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000856, | |
| "loss": 6.9566, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.000858, | |
| "loss": 6.8935, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00086, | |
| "loss": 7.0856, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.000862, | |
| "loss": 7.0538, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.000864, | |
| "loss": 7.1257, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.000866, | |
| "loss": 6.9539, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0008680000000000001, | |
| "loss": 6.9312, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00087, | |
| "loss": 7.0382, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.000872, | |
| "loss": 7.0195, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.000874, | |
| "loss": 7.1437, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.000876, | |
| "loss": 6.9629, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.000878, | |
| "loss": 7.1108, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00088, | |
| "loss": 6.9873, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000882, | |
| "loss": 6.8853, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000884, | |
| "loss": 6.8583, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0008860000000000001, | |
| "loss": 7.0631, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000888, | |
| "loss": 6.8263, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0008900000000000001, | |
| "loss": 6.6845, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000892, | |
| "loss": 6.8607, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.000894, | |
| "loss": 6.5526, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.000896, | |
| "loss": 7.4182, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.000898, | |
| "loss": 7.2457, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0009000000000000001, | |
| "loss": 7.1327, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.000902, | |
| "loss": 6.9056, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0009040000000000001, | |
| "loss": 7.174, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.000906, | |
| "loss": 7.0875, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0009080000000000001, | |
| "loss": 7.0482, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00091, | |
| "loss": 7.2428, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.000912, | |
| "loss": 6.9443, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0009140000000000001, | |
| "loss": 6.978, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.000916, | |
| "loss": 7.0511, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0009180000000000001, | |
| "loss": 6.9723, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00092, | |
| "loss": 6.9377, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0009220000000000001, | |
| "loss": 6.9787, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.000924, | |
| "loss": 7.0386, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0009260000000000001, | |
| "loss": 7.041, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0009280000000000001, | |
| "loss": 7.1201, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00093, | |
| "loss": 6.9869, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0009320000000000001, | |
| "loss": 6.9156, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.000934, | |
| "loss": 6.9832, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0009360000000000001, | |
| "loss": 6.9666, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 7.0176, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_loss": NaN, | |
| "eval_runtime": 772.1245, | |
| "eval_samples_per_second": 3.422, | |
| "eval_steps_per_second": 0.286, | |
| "eval_wer": 2.0097029585094353, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0009379999999999999, | |
| "loss": 0.0, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 594, | |
| "total_flos": 0.0, | |
| "train_loss": 6.071190901476927, | |
| "train_runtime": 5557.8497, | |
| "train_samples_per_second": 5.135, | |
| "train_steps_per_second": 0.107 | |
| } | |
| ], | |
| "max_steps": 594, | |
| "num_train_epochs": 1, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |