| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0178660878276842, |
| "eval_steps": 500, |
| "global_step": 381, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 2e-08, |
| "loss": 1.1427, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4e-08, |
| "loss": 1.1555, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 6.000000000000001e-08, |
| "loss": 1.1633, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 8e-08, |
| "loss": 1.1405, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 1e-07, |
| "loss": 1.1607, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.2000000000000002e-07, |
| "loss": 1.1587, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.4e-07, |
| "loss": 1.1463, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.6e-07, |
| "loss": 1.1764, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.8e-07, |
| "loss": 1.1509, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2e-07, |
| "loss": 1.1516, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2.2e-07, |
| "loss": 1.1589, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2.4000000000000003e-07, |
| "loss": 1.1416, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2.6e-07, |
| "loss": 1.1474, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 2.8e-07, |
| "loss": 1.1785, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 3e-07, |
| "loss": 1.1259, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 3.2e-07, |
| "loss": 1.1578, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 3.4e-07, |
| "loss": 1.1406, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 3.6e-07, |
| "loss": 1.1533, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 3.7999999999999996e-07, |
| "loss": 1.1593, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4e-07, |
| "loss": 1.1401, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.2e-07, |
| "loss": 1.1595, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.4e-07, |
| "loss": 1.1482, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.6e-07, |
| "loss": 1.1682, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.800000000000001e-07, |
| "loss": 1.1453, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 5e-07, |
| "loss": 1.1329, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 5.2e-07, |
| "loss": 1.1488, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 5.399999999999999e-07, |
| "loss": 1.1528, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 5.6e-07, |
| "loss": 1.1463, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 5.8e-07, |
| "loss": 1.1445, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 6e-07, |
| "loss": 1.1578, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 6.2e-07, |
| "loss": 1.1523, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 6.4e-07, |
| "loss": 1.1547, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 6.6e-07, |
| "loss": 1.1309, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 6.8e-07, |
| "loss": 1.1249, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 7.000000000000001e-07, |
| "loss": 1.1774, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.2e-07, |
| "loss": 1.124, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.4e-07, |
| "loss": 1.1485, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.599999999999999e-07, |
| "loss": 1.1442, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.8e-07, |
| "loss": 1.1106, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 8e-07, |
| "loss": 1.1494, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 8.2e-07, |
| "loss": 1.1084, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 8.4e-07, |
| "loss": 1.1244, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 8.6e-07, |
| "loss": 1.1482, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 8.8e-07, |
| "loss": 1.1164, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 8.999999999999999e-07, |
| "loss": 1.1382, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.2e-07, |
| "loss": 1.131, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.4e-07, |
| "loss": 1.1241, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.600000000000001e-07, |
| "loss": 1.1159, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.800000000000001e-07, |
| "loss": 1.1291, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1e-06, |
| "loss": 1.1361, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.02e-06, |
| "loss": 1.126, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.04e-06, |
| "loss": 1.1184, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.06e-06, |
| "loss": 1.1193, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.0799999999999998e-06, |
| "loss": 1.1187, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.1e-06, |
| "loss": 1.1145, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.12e-06, |
| "loss": 1.1019, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.14e-06, |
| "loss": 1.1157, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.16e-06, |
| "loss": 1.1024, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.18e-06, |
| "loss": 1.098, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.2e-06, |
| "loss": 1.0946, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.22e-06, |
| "loss": 1.0818, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.24e-06, |
| "loss": 1.121, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.26e-06, |
| "loss": 1.09, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.28e-06, |
| "loss": 1.0958, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.3e-06, |
| "loss": 1.0888, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.32e-06, |
| "loss": 1.0793, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.3399999999999999e-06, |
| "loss": 1.0929, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.36e-06, |
| "loss": 1.0685, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.38e-06, |
| "loss": 1.0888, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.4000000000000001e-06, |
| "loss": 1.0688, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.42e-06, |
| "loss": 1.0627, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.44e-06, |
| "loss": 1.0627, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.46e-06, |
| "loss": 1.0858, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.48e-06, |
| "loss": 1.0517, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.4999999999999998e-06, |
| "loss": 1.0626, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.5199999999999998e-06, |
| "loss": 1.0708, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.54e-06, |
| "loss": 1.0391, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.56e-06, |
| "loss": 1.0436, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.58e-06, |
| "loss": 1.0396, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.6e-06, |
| "loss": 1.0334, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.62e-06, |
| "loss": 1.0291, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.64e-06, |
| "loss": 1.0421, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.66e-06, |
| "loss": 1.0493, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.68e-06, |
| "loss": 1.0293, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.7e-06, |
| "loss": 1.0766, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.72e-06, |
| "loss": 1.058, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.74e-06, |
| "loss": 1.0244, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.76e-06, |
| "loss": 1.0399, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.78e-06, |
| "loss": 1.0272, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7999999999999997e-06, |
| "loss": 1.029, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.82e-06, |
| "loss": 1.0477, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.84e-06, |
| "loss": 1.0166, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.8600000000000002e-06, |
| "loss": 1.0309, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.88e-06, |
| "loss": 1.0135, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.8999999999999998e-06, |
| "loss": 1.0226, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.9200000000000003e-06, |
| "loss": 1.0402, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.94e-06, |
| "loss": 1.0252, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.9600000000000003e-06, |
| "loss": 1.0064, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.98e-06, |
| "loss": 0.9967, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 2e-06, |
| "loss": 1.0094, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 2.02e-06, |
| "loss": 0.9938, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 2.04e-06, |
| "loss": 1.0199, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 2.0599999999999998e-06, |
| "loss": 1.005, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 2.08e-06, |
| "loss": 1.0016, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 2.1e-06, |
| "loss": 1.0169, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 2.12e-06, |
| "loss": 0.978, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 2.14e-06, |
| "loss": 0.9787, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 2.1599999999999996e-06, |
| "loss": 0.9992, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 2.18e-06, |
| "loss": 0.985, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 2.2e-06, |
| "loss": 0.9832, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 2.22e-06, |
| "loss": 0.9922, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 2.24e-06, |
| "loss": 0.9882, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 2.26e-06, |
| "loss": 0.979, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 2.28e-06, |
| "loss": 0.9866, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 2.3e-06, |
| "loss": 0.97, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 2.32e-06, |
| "loss": 1.0042, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 2.34e-06, |
| "loss": 0.9778, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 2.36e-06, |
| "loss": 0.9906, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 2.38e-06, |
| "loss": 0.9547, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 2.4e-06, |
| "loss": 0.9792, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 2.4199999999999997e-06, |
| "loss": 0.964, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 2.44e-06, |
| "loss": 0.9797, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 2.4599999999999997e-06, |
| "loss": 0.9576, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 2.48e-06, |
| "loss": 0.9795, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 2.5e-06, |
| "loss": 0.9619, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 2.52e-06, |
| "loss": 0.9756, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 2.5400000000000002e-06, |
| "loss": 0.9678, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 2.56e-06, |
| "loss": 0.973, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 2.58e-06, |
| "loss": 0.9581, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 2.6e-06, |
| "loss": 0.9559, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 2.62e-06, |
| "loss": 0.9623, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 2.64e-06, |
| "loss": 0.9608, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 2.66e-06, |
| "loss": 0.9599, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 2.6799999999999998e-06, |
| "loss": 0.9481, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 2.7e-06, |
| "loss": 0.951, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 2.72e-06, |
| "loss": 0.9553, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 2.7399999999999996e-06, |
| "loss": 0.9471, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 2.76e-06, |
| "loss": 0.9228, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 2.78e-06, |
| "loss": 0.9411, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 2.8000000000000003e-06, |
| "loss": 0.9541, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 2.82e-06, |
| "loss": 0.9318, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 2.84e-06, |
| "loss": 0.9113, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 2.86e-06, |
| "loss": 0.9316, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 2.88e-06, |
| "loss": 0.9326, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 2.9e-06, |
| "loss": 0.9471, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 2.92e-06, |
| "loss": 0.9234, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 2.94e-06, |
| "loss": 0.9384, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 2.96e-06, |
| "loss": 0.9317, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 2.98e-06, |
| "loss": 0.9157, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 2.9999999999999997e-06, |
| "loss": 0.9411, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 3.02e-06, |
| "loss": 0.9566, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 3.0399999999999997e-06, |
| "loss": 0.9443, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 3.06e-06, |
| "loss": 0.9215, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 3.08e-06, |
| "loss": 0.9129, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 3.1e-06, |
| "loss": 0.9175, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.12e-06, |
| "loss": 0.9309, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.14e-06, |
| "loss": 0.9185, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.16e-06, |
| "loss": 0.9196, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.18e-06, |
| "loss": 0.9038, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.2e-06, |
| "loss": 0.9205, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.22e-06, |
| "loss": 0.9308, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.24e-06, |
| "loss": 0.939, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.2599999999999997e-06, |
| "loss": 0.9211, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.28e-06, |
| "loss": 0.9214, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.2999999999999997e-06, |
| "loss": 0.9243, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.32e-06, |
| "loss": 0.9098, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.3399999999999998e-06, |
| "loss": 0.9223, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.36e-06, |
| "loss": 0.9213, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.3800000000000002e-06, |
| "loss": 0.9059, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.4e-06, |
| "loss": 0.8938, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.42e-06, |
| "loss": 0.9139, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.44e-06, |
| "loss": 0.898, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.46e-06, |
| "loss": 0.9103, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.48e-06, |
| "loss": 0.9042, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 3.5e-06, |
| "loss": 0.8833, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 3.52e-06, |
| "loss": 0.8933, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 3.5399999999999996e-06, |
| "loss": 0.9022, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 3.56e-06, |
| "loss": 0.9141, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 3.58e-06, |
| "loss": 0.9089, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 3.5999999999999994e-06, |
| "loss": 0.9007, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 3.6199999999999996e-06, |
| "loss": 0.8985, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 3.64e-06, |
| "loss": 0.8977, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 3.66e-06, |
| "loss": 0.8996, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 3.68e-06, |
| "loss": 0.8999, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 3.7e-06, |
| "loss": 0.9141, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 3.7200000000000004e-06, |
| "loss": 0.8918, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 3.7399999999999998e-06, |
| "loss": 0.9008, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 3.76e-06, |
| "loss": 0.9021, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 3.7800000000000002e-06, |
| "loss": 0.8951, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 3.7999999999999996e-06, |
| "loss": 0.9089, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 3.82e-06, |
| "loss": 0.8966, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 3.8400000000000005e-06, |
| "loss": 0.8937, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 3.8599999999999995e-06, |
| "loss": 0.8969, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 3.88e-06, |
| "loss": 0.8719, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 3.9e-06, |
| "loss": 0.884, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 3.920000000000001e-06, |
| "loss": 0.8809, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 3.9399999999999995e-06, |
| "loss": 0.8711, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 3.96e-06, |
| "loss": 0.877, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 3.98e-06, |
| "loss": 0.8892, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4e-06, |
| "loss": 0.8781, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.02e-06, |
| "loss": 0.8697, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.04e-06, |
| "loss": 0.8745, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.059999999999999e-06, |
| "loss": 0.8738, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.08e-06, |
| "loss": 0.8804, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.1000000000000006e-06, |
| "loss": 0.8843, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.1199999999999995e-06, |
| "loss": 0.8736, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.14e-06, |
| "loss": 0.8689, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.16e-06, |
| "loss": 0.8614, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.180000000000001e-06, |
| "loss": 0.866, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.2e-06, |
| "loss": 0.8598, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.22e-06, |
| "loss": 0.878, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.24e-06, |
| "loss": 0.8624, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.26e-06, |
| "loss": 0.8788, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.28e-06, |
| "loss": 0.8713, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.3e-06, |
| "loss": 0.8682, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.319999999999999e-06, |
| "loss": 0.8827, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.34e-06, |
| "loss": 0.8754, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.36e-06, |
| "loss": 0.8867, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.3799999999999996e-06, |
| "loss": 0.8707, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.4e-06, |
| "loss": 0.8655, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.42e-06, |
| "loss": 0.8503, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.44e-06, |
| "loss": 0.8674, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.46e-06, |
| "loss": 0.8507, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.48e-06, |
| "loss": 0.8668, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.5e-06, |
| "loss": 0.8548, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.52e-06, |
| "loss": 0.8756, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.54e-06, |
| "loss": 0.8424, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.56e-06, |
| "loss": 0.8501, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.579999999999999e-06, |
| "loss": 0.8633, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.6e-06, |
| "loss": 0.859, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.62e-06, |
| "loss": 0.8497, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.64e-06, |
| "loss": 0.8491, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.6599999999999994e-06, |
| "loss": 0.8596, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.68e-06, |
| "loss": 0.8636, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.7e-06, |
| "loss": 0.8677, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.72e-06, |
| "loss": 0.8605, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.74e-06, |
| "loss": 0.8493, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.76e-06, |
| "loss": 0.8367, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.78e-06, |
| "loss": 0.8443, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.8e-06, |
| "loss": 0.8464, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 4.8200000000000004e-06, |
| "loss": 0.8588, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.839999999999999e-06, |
| "loss": 0.8502, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.86e-06, |
| "loss": 0.849, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.88e-06, |
| "loss": 0.8576, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 4.9e-06, |
| "loss": 0.8418, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.9199999999999995e-06, |
| "loss": 0.8717, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.94e-06, |
| "loss": 0.8529, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 4.96e-06, |
| "loss": 0.8328, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 4.98e-06, |
| "loss": 0.85, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5e-06, |
| "loss": 0.8376, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.02e-06, |
| "loss": 0.8431, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.04e-06, |
| "loss": 0.8446, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.06e-06, |
| "loss": 0.8424, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.0800000000000005e-06, |
| "loss": 0.8377, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.0999999999999995e-06, |
| "loss": 0.8453, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.12e-06, |
| "loss": 0.8591, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 5.14e-06, |
| "loss": 0.8536, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 5.16e-06, |
| "loss": 0.8399, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 5.1799999999999995e-06, |
| "loss": 0.8259, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 5.2e-06, |
| "loss": 0.8399, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 5.219999999999999e-06, |
| "loss": 0.8443, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 5.24e-06, |
| "loss": 0.8331, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 5.2600000000000005e-06, |
| "loss": 0.8268, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 5.28e-06, |
| "loss": 0.8338, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 5.3e-06, |
| "loss": 0.8287, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 5.32e-06, |
| "loss": 0.8273, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 5.3400000000000005e-06, |
| "loss": 0.8471, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 5.3599999999999995e-06, |
| "loss": 0.8364, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 5.38e-06, |
| "loss": 0.8167, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 5.4e-06, |
| "loss": 0.8433, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 5.42e-06, |
| "loss": 0.8163, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 5.44e-06, |
| "loss": 0.8186, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 5.46e-06, |
| "loss": 0.8491, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 5.479999999999999e-06, |
| "loss": 0.8222, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 5.5e-06, |
| "loss": 0.8362, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 5.52e-06, |
| "loss": 0.8375, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 5.54e-06, |
| "loss": 0.7934, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 5.56e-06, |
| "loss": 0.8096, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 5.58e-06, |
| "loss": 0.8185, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 5.600000000000001e-06, |
| "loss": 0.8126, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 5.6199999999999996e-06, |
| "loss": 0.8297, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 5.64e-06, |
| "loss": 0.8196, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 5.66e-06, |
| "loss": 0.8251, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 5.68e-06, |
| "loss": 0.8291, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 5.7e-06, |
| "loss": 0.8067, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 5.72e-06, |
| "loss": 0.8171, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 5.739999999999999e-06, |
| "loss": 0.8091, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 5.76e-06, |
| "loss": 0.8149, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 5.78e-06, |
| "loss": 0.8393, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 5.8e-06, |
| "loss": 0.8157, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 5.819999999999999e-06, |
| "loss": 0.8043, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 5.84e-06, |
| "loss": 0.8166, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 5.860000000000001e-06, |
| "loss": 0.8022, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 5.88e-06, |
| "loss": 0.7949, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 5.9e-06, |
| "loss": 0.8102, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 5.92e-06, |
| "loss": 0.81, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 5.94e-06, |
| "loss": 0.8104, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 5.96e-06, |
| "loss": 0.8089, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 5.98e-06, |
| "loss": 0.824, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 5.999999999999999e-06, |
| "loss": 0.8145, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 6.02e-06, |
| "loss": 0.8085, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 6.04e-06, |
| "loss": 0.8183, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 6.0600000000000004e-06, |
| "loss": 0.8114, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 6.079999999999999e-06, |
| "loss": 0.8112, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 6.1e-06, |
| "loss": 0.8308, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 6.12e-06, |
| "loss": 0.8303, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 6.14e-06, |
| "loss": 0.8287, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 6.16e-06, |
| "loss": 0.7957, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 6.18e-06, |
| "loss": 0.817, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 6.2e-06, |
| "loss": 0.8209, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 6.22e-06, |
| "loss": 0.8019, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 6.24e-06, |
| "loss": 0.8099, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 6.259999999999999e-06, |
| "loss": 0.8275, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 6.28e-06, |
| "loss": 0.814, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 6.3e-06, |
| "loss": 0.7992, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 6.32e-06, |
| "loss": 0.7964, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 6.3399999999999994e-06, |
| "loss": 0.8013, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 6.36e-06, |
| "loss": 0.8023, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 6.38e-06, |
| "loss": 0.8074, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 6.4e-06, |
| "loss": 0.8154, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 6.42e-06, |
| "loss": 0.7984, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 6.44e-06, |
| "loss": 0.7987, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 6.46e-06, |
| "loss": 0.8007, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 6.48e-06, |
| "loss": 0.7999, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 6.5000000000000004e-06, |
| "loss": 0.8107, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 6.519999999999999e-06, |
| "loss": 0.7928, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 6.54e-06, |
| "loss": 0.8033, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 6.56e-06, |
| "loss": 0.802, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 6.58e-06, |
| "loss": 0.8056, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 6.5999999999999995e-06, |
| "loss": 0.7857, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 6.62e-06, |
| "loss": 0.8046, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.64e-06, |
| "loss": 0.7865, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.66e-06, |
| "loss": 0.814, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.6799999999999996e-06, |
| "loss": 0.7934, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.7e-06, |
| "loss": 0.7893, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 6.72e-06, |
| "loss": 0.7984, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 6.74e-06, |
| "loss": 0.7932, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 6.7600000000000005e-06, |
| "loss": 0.7886, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.7799999999999995e-06, |
| "loss": 0.7925, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.8e-06, |
| "loss": 0.7872, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.82e-06, |
| "loss": 0.8191, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 6.84e-06, |
| "loss": 0.7994, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.8599999999999995e-06, |
| "loss": 0.8014, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.88e-06, |
| "loss": 0.7998, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.9e-06, |
| "loss": 0.7809, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 6.92e-06, |
| "loss": 0.7845, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 6.94e-06, |
| "loss": 0.7795, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 6.96e-06, |
| "loss": 0.7959, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 6.979999999999999e-06, |
| "loss": 0.7848, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 7e-06, |
| "loss": 0.7915, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 6.999986848712049e-06, |
| "loss": 0.7891, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 6.999947394947031e-06, |
| "loss": 0.7838, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 6.999881639001441e-06, |
| "loss": 0.8101, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.999789581369437e-06, |
| "loss": 0.8062, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.999671222742834e-06, |
| "loss": 0.7923, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.999526564011098e-06, |
| "loss": 0.8033, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 6.999355606261343e-06, |
| "loss": 0.7905, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.999158350778321e-06, |
| "loss": 0.7833, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.998934799044413e-06, |
| "loss": 0.7916, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.9986849527396114e-06, |
| "loss": 0.7848, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.998408813741518e-06, |
| "loss": 0.7888, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 6.9981063841253256e-06, |
| "loss": 0.7971, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 6.997777666163796e-06, |
| "loss": 0.7947, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 6.997422662327253e-06, |
| "loss": 0.7943, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 6.9970413752835595e-06, |
| "loss": 0.7864, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 6.996633807898094e-06, |
| "loss": 0.8132, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 6.996199963233736e-06, |
| "loss": 0.7814, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 6.995739844550836e-06, |
| "loss": 0.8083, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 6.995253455307197e-06, |
| "loss": 0.7831, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 6.994740799158044e-06, |
| "loss": 0.7775, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 6.994201879955999e-06, |
| "loss": 0.7969, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 6.993636701751052e-06, |
| "loss": 0.7726, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 6.993045268790529e-06, |
| "loss": 0.783, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 6.9924275855190615e-06, |
| "loss": 0.7802, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 6.991783656578554e-06, |
| "loss": 0.7885, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 6.991113486808145e-06, |
| "loss": 0.7916, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 6.990417081244175e-06, |
| "loss": 0.7961, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 6.989694445120147e-06, |
| "loss": 0.7912, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 6.9889455838666875e-06, |
| "loss": 0.7711, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 6.988170503111504e-06, |
| "loss": 0.7695, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 6.987369208679347e-06, |
| "loss": 0.7693, |
| "step": 381 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1496, |
| "num_train_epochs": 4, |
| "save_steps": 500, |
| "total_flos": 6.35040259173974e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|