| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0182835197862747, | |
| "eval_steps": 500, | |
| "global_step": 763, | |
| "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 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.986541706591961e-06, | |
| "loss": 0.7944, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.985688003068041e-06, | |
| "loss": 0.7792, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.984808104523189e-06, | |
| "loss": 0.7852, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.983902017569859e-06, | |
| "loss": 0.7866, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.982969749017318e-06, | |
| "loss": 0.7708, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.982011305871581e-06, | |
| "loss": 0.7979, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.981026695335371e-06, | |
| "loss": 0.7729, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.980015924808057e-06, | |
| "loss": 0.7846, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.978979001885602e-06, | |
| "loss": 0.8077, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.9779159343605024e-06, | |
| "loss": 0.7805, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.9768267302217355e-06, | |
| "loss": 0.7984, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.975711397654693e-06, | |
| "loss": 0.775, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.974569945041124e-06, | |
| "loss": 0.7771, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.97340238095907e-06, | |
| "loss": 0.7963, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.972208714182799e-06, | |
| "loss": 0.7831, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.970988953682744e-06, | |
| "loss": 0.7892, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.96974310862543e-06, | |
| "loss": 0.7852, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.9684711883734115e-06, | |
| "loss": 0.7965, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.9671732024851965e-06, | |
| "loss": 0.766, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.965849160715176e-06, | |
| "loss": 0.7918, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.964499073013553e-06, | |
| "loss": 0.767, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.963122949526267e-06, | |
| "loss": 0.7807, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.961720800594914e-06, | |
| "loss": 0.7976, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.960292636756675e-06, | |
| "loss": 0.7732, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.9588384687442315e-06, | |
| "loss": 0.7791, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.957358307485689e-06, | |
| "loss": 0.7776, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.9558521641044894e-06, | |
| "loss": 0.7885, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.954320049919333e-06, | |
| "loss": 0.7852, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.952761976444094e-06, | |
| "loss": 0.7919, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.9511779553877245e-06, | |
| "loss": 0.7863, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.949567998654181e-06, | |
| "loss": 0.7738, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.947932118342319e-06, | |
| "loss": 0.7629, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.946270326745818e-06, | |
| "loss": 0.7802, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.944582636353076e-06, | |
| "loss": 0.7803, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.942869059847123e-06, | |
| "loss": 0.8036, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.941129610105525e-06, | |
| "loss": 0.772, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.939364300200283e-06, | |
| "loss": 0.7958, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.93757314339774e-06, | |
| "loss": 0.7837, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.935756153158478e-06, | |
| "loss": 0.7632, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.933913343137219e-06, | |
| "loss": 0.7921, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.9320447271827186e-06, | |
| "loss": 0.793, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.930150319337667e-06, | |
| "loss": 0.789, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.92823013383858e-06, | |
| "loss": 0.7785, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.9262841851156935e-06, | |
| "loss": 0.7719, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.924312487792855e-06, | |
| "loss": 0.791, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.922315056687412e-06, | |
| "loss": 0.795, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.920291906810102e-06, | |
| "loss": 0.7837, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.918243053364941e-06, | |
| "loss": 0.8042, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.916168511749106e-06, | |
| "loss": 0.78, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.914068297552825e-06, | |
| "loss": 0.7781, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.9119424265592495e-06, | |
| "loss": 0.7679, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.909790914744349e-06, | |
| "loss": 0.761, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.907613778276779e-06, | |
| "loss": 0.7773, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.90541103351777e-06, | |
| "loss": 0.7817, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.903182697020994e-06, | |
| "loss": 0.7634, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.90092878553245e-06, | |
| "loss": 0.7739, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.898649315990332e-06, | |
| "loss": 0.7679, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.896344305524902e-06, | |
| "loss": 0.7755, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.894013771458362e-06, | |
| "loss": 0.7964, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.891657731304729e-06, | |
| "loss": 0.7756, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.889276202769695e-06, | |
| "loss": 0.7938, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.886869203750498e-06, | |
| "loss": 0.7752, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.884436752335787e-06, | |
| "loss": 0.7724, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.881978866805488e-06, | |
| "loss": 0.7867, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.879495565630666e-06, | |
| "loss": 0.7843, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.87698686747338e-06, | |
| "loss": 0.7912, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.8744527911865535e-06, | |
| "loss": 0.7851, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.871893355813823e-06, | |
| "loss": 0.7682, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.869308580589402e-06, | |
| "loss": 0.7538, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.866698484937932e-06, | |
| "loss": 0.7906, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.864063088474338e-06, | |
| "loss": 0.7771, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.861402411003682e-06, | |
| "loss": 0.7749, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.858716472521012e-06, | |
| "loss": 0.77, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.856005293211217e-06, | |
| "loss": 0.7723, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.853268893448865e-06, | |
| "loss": 0.7714, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.850507293798059e-06, | |
| "loss": 0.7736, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.847720515012284e-06, | |
| "loss": 0.7748, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.8449085780342395e-06, | |
| "loss": 0.7886, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.842071503995695e-06, | |
| "loss": 0.7771, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.839209314217322e-06, | |
| "loss": 0.8013, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.83632203020854e-06, | |
| "loss": 0.7785, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.8334096736673505e-06, | |
| "loss": 0.778, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.830472266480176e-06, | |
| "loss": 0.7811, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.827509830721694e-06, | |
| "loss": 0.7668, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.824522388654676e-06, | |
| "loss": 0.7761, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.821509962729811e-06, | |
| "loss": 0.7784, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.818472575585549e-06, | |
| "loss": 0.7844, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.8154102500479155e-06, | |
| "loss": 0.7892, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.812323009130357e-06, | |
| "loss": 0.7758, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.809210876033554e-06, | |
| "loss": 0.7931, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.806073874145257e-06, | |
| "loss": 0.771, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.802912027040099e-06, | |
| "loss": 0.7969, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.799725358479433e-06, | |
| "loss": 0.7696, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.79651389241114e-06, | |
| "loss": 0.7863, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.793277652969458e-06, | |
| "loss": 0.778, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.790016664474797e-06, | |
| "loss": 0.768, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.786730951433557e-06, | |
| "loss": 0.7675, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.78342053853794e-06, | |
| "loss": 0.7711, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.780085450665775e-06, | |
| "loss": 0.7677, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.776725712880315e-06, | |
| "loss": 0.7544, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.773341350430065e-06, | |
| "loss": 0.7687, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.769932388748583e-06, | |
| "loss": 0.7768, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.766498853454287e-06, | |
| "loss": 0.7783, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.763040770350272e-06, | |
| "loss": 0.7832, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.759558165424105e-06, | |
| "loss": 0.763, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.75605106484764e-06, | |
| "loss": 0.7656, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.752519494976812e-06, | |
| "loss": 0.7685, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.748963482351447e-06, | |
| "loss": 0.7724, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.745383053695056e-06, | |
| "loss": 0.7693, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.741778235914637e-06, | |
| "loss": 0.7815, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.738149056100475e-06, | |
| "loss": 0.7702, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.734495541525934e-06, | |
| "loss": 0.7744, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.730817719647257e-06, | |
| "loss": 0.766, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.727115618103354e-06, | |
| "loss": 0.7582, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.7233892647156e-06, | |
| "loss": 0.7821, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.719638687487618e-06, | |
| "loss": 0.7831, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.71586391460508e-06, | |
| "loss": 0.7501, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.712064974435485e-06, | |
| "loss": 0.783, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.708241895527952e-06, | |
| "loss": 0.7857, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.704394706613e-06, | |
| "loss": 0.7757, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.700523436602338e-06, | |
| "loss": 0.7651, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.696628114588645e-06, | |
| "loss": 0.7759, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.69270876984535e-06, | |
| "loss": 0.7801, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.688765431826413e-06, | |
| "loss": 0.7662, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.6847981301661046e-06, | |
| "loss": 0.7566, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.680806894678784e-06, | |
| "loss": 0.7733, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.676791755358671e-06, | |
| "loss": 0.754, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.6727527423796255e-06, | |
| "loss": 0.7871, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.668689886094918e-06, | |
| "loss": 0.7614, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.664603217037001e-06, | |
| "loss": 0.7691, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.66049276591728e-06, | |
| "loss": 0.7669, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.656358563625887e-06, | |
| "loss": 0.7487, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.6522006412314404e-06, | |
| "loss": 0.7689, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.648019029980816e-06, | |
| "loss": 0.7628, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.643813761298915e-06, | |
| "loss": 0.7426, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.6395848667884215e-06, | |
| "loss": 0.7693, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.635332378229571e-06, | |
| "loss": 0.7734, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.631056327579905e-06, | |
| "loss": 0.7633, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.626756746974038e-06, | |
| "loss": 0.7947, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.622433668723412e-06, | |
| "loss": 0.7693, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.6180871253160525e-06, | |
| "loss": 0.7843, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.613717149416329e-06, | |
| "loss": 0.7569, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.609323773864704e-06, | |
| "loss": 0.7652, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.60490703167749e-06, | |
| "loss": 0.7715, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.6004669560466e-06, | |
| "loss": 0.778, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.596003580339301e-06, | |
| "loss": 0.7888, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.5915169380979565e-06, | |
| "loss": 0.7567, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.58700706303978e-06, | |
| "loss": 0.768, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.58247398905658e-06, | |
| "loss": 0.7729, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.577917750214508e-06, | |
| "loss": 0.767, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.573338380753795e-06, | |
| "loss": 0.7601, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.568735915088504e-06, | |
| "loss": 0.7714, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.564110387806263e-06, | |
| "loss": 0.7579, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.5594618336680095e-06, | |
| "loss": 0.7693, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.55479028760773e-06, | |
| "loss": 0.7818, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.550095784732193e-06, | |
| "loss": 0.7641, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.54537836032069e-06, | |
| "loss": 0.7658, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.5406380498247695e-06, | |
| "loss": 0.7655, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.535874888867964e-06, | |
| "loss": 0.7664, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.531088913245536e-06, | |
| "loss": 0.7706, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.526280158924192e-06, | |
| "loss": 0.771, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.521448662041825e-06, | |
| "loss": 0.7582, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.516594458907243e-06, | |
| "loss": 0.7743, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.511717585999884e-06, | |
| "loss": 0.7657, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.5068180799695556e-06, | |
| "loss": 0.7699, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.501895977636151e-06, | |
| "loss": 0.7675, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.4969513159893755e-06, | |
| "loss": 0.7887, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.49198413218847e-06, | |
| "loss": 0.7647, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.4869944635619266e-06, | |
| "loss": 0.7672, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.4819823476072154e-06, | |
| "loss": 0.7506, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.476947821990495e-06, | |
| "loss": 0.7772, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.471890924546335e-06, | |
| "loss": 0.7484, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.466811693277429e-06, | |
| "loss": 0.7668, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.4617101663543115e-06, | |
| "loss": 0.7762, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.456586382115066e-06, | |
| "loss": 0.7797, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.4514403790650435e-06, | |
| "loss": 0.7401, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.4462721958765695e-06, | |
| "loss": 0.7589, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.441081871388649e-06, | |
| "loss": 0.7648, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.435869444606687e-06, | |
| "loss": 0.7714, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.430634954702182e-06, | |
| "loss": 0.7544, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.42537844101244e-06, | |
| "loss": 0.7706, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.420099943040274e-06, | |
| "loss": 0.7833, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.414799500453712e-06, | |
| "loss": 0.7769, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.409477153085693e-06, | |
| "loss": 0.7537, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.404132940933776e-06, | |
| "loss": 0.773, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.398766904159829e-06, | |
| "loss": 0.7794, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.393379083089736e-06, | |
| "loss": 0.7687, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.387969518213087e-06, | |
| "loss": 0.7838, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.382538250182883e-06, | |
| "loss": 0.7789, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.377085319815218e-06, | |
| "loss": 0.7616, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.371610768088981e-06, | |
| "loss": 0.772, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.366114636145552e-06, | |
| "loss": 0.7507, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.360596965288476e-06, | |
| "loss": 0.7573, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 6.3550577969831725e-06, | |
| "loss": 0.7706, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 6.3494971728566115e-06, | |
| "loss": 0.7783, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 6.343915134697002e-06, | |
| "loss": 0.7667, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 6.338311724453484e-06, | |
| "loss": 0.7681, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.332686984235804e-06, | |
| "loss": 0.7756, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.327040956314011e-06, | |
| "loss": 0.7616, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.321373683118123e-06, | |
| "loss": 0.7762, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.315685207237823e-06, | |
| "loss": 0.773, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.309975571422131e-06, | |
| "loss": 0.7471, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.304244818579083e-06, | |
| "loss": 0.7537, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.298492991775411e-06, | |
| "loss": 0.7712, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.292720134236218e-06, | |
| "loss": 0.7735, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.286926289344656e-06, | |
| "loss": 0.7456, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.2811115006415916e-06, | |
| "loss": 0.7729, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.275275811825288e-06, | |
| "loss": 0.7844, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.269419266751077e-06, | |
| "loss": 0.7748, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.263541909431018e-06, | |
| "loss": 0.7601, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.257643784033581e-06, | |
| "loss": 0.7783, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.251724934883305e-06, | |
| "loss": 0.7934, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.245785406460471e-06, | |
| "loss": 0.7638, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.239825243400762e-06, | |
| "loss": 0.7636, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.233844490494934e-06, | |
| "loss": 0.764, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.227843192688475e-06, | |
| "loss": 0.7677, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.221821395081267e-06, | |
| "loss": 0.7657, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.21577914292725e-06, | |
| "loss": 0.7745, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.209716481634081e-06, | |
| "loss": 0.7658, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.20363345676279e-06, | |
| "loss": 0.7726, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.1975301140274415e-06, | |
| "loss": 0.7627, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.1914064992947875e-06, | |
| "loss": 0.7601, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.1852626585839265e-06, | |
| "loss": 0.758, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.179098638065954e-06, | |
| "loss": 0.7636, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.172914484063619e-06, | |
| "loss": 0.7678, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.166710243050972e-06, | |
| "loss": 0.7489, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.160485961653018e-06, | |
| "loss": 0.7629, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.154241686645368e-06, | |
| "loss": 0.7684, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.147977464953884e-06, | |
| "loss": 0.7567, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.141693343654327e-06, | |
| "loss": 0.7699, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.135389369972006e-06, | |
| "loss": 0.7728, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.129065591281419e-06, | |
| "loss": 0.7455, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.1227220551059e-06, | |
| "loss": 0.7448, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 6.116358809117262e-06, | |
| "loss": 0.7585, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.109975901135435e-06, | |
| "loss": 0.7339, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.1035733791281125e-06, | |
| "loss": 0.7715, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.097151291210385e-06, | |
| "loss": 0.7648, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.090709685644382e-06, | |
| "loss": 0.7566, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.084248610838911e-06, | |
| "loss": 0.753, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.077768115349088e-06, | |
| "loss": 0.7709, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.071268247875976e-06, | |
| "loss": 0.756, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.064749057266223e-06, | |
| "loss": 0.7812, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.058210592511687e-06, | |
| "loss": 0.7728, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.05165290274907e-06, | |
| "loss": 0.7775, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.045076037259554e-06, | |
| "loss": 0.7655, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.038480045468426e-06, | |
| "loss": 0.7668, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.03186497694471e-06, | |
| "loss": 0.7593, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.025230881400785e-06, | |
| "loss": 0.7666, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.0185778086920245e-06, | |
| "loss": 0.7841, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.011905808816416e-06, | |
| "loss": 0.7709, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.0052149319141795e-06, | |
| "loss": 0.7763, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 5.998505228267403e-06, | |
| "loss": 0.7759, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 5.991776748299656e-06, | |
| "loss": 0.7607, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.985029542575608e-06, | |
| "loss": 0.7639, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.978263661800657e-06, | |
| "loss": 0.7745, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.971479156820547e-06, | |
| "loss": 0.7635, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.964676078620975e-06, | |
| "loss": 0.7504, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 5.957854478327225e-06, | |
| "loss": 0.7548, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 5.95101440720377e-06, | |
| "loss": 0.7677, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 5.944155916653893e-06, | |
| "loss": 0.7543, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.9372790582192985e-06, | |
| "loss": 0.7651, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.930383883579727e-06, | |
| "loss": 0.7578, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.923470444552567e-06, | |
| "loss": 0.7588, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.91653879309246e-06, | |
| "loss": 0.755, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.909588981290919e-06, | |
| "loss": 0.7546, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.90262106137593e-06, | |
| "loss": 0.752, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.895635085711562e-06, | |
| "loss": 0.7806, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.8886311067975745e-06, | |
| "loss": 0.7765, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.8816091772690205e-06, | |
| "loss": 0.7637, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.87456934989585e-06, | |
| "loss": 0.7792, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.867511677582522e-06, | |
| "loss": 0.7492, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.860436213367597e-06, | |
| "loss": 0.7509, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 5.85334301042334e-06, | |
| "loss": 0.7514, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 5.846232122055325e-06, | |
| "loss": 0.7695, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 5.839103601702033e-06, | |
| "loss": 0.7693, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.83195750293445e-06, | |
| "loss": 0.7466, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.824793879455662e-06, | |
| "loss": 0.7652, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.817612785100454e-06, | |
| "loss": 0.7558, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.8104142738349076e-06, | |
| "loss": 0.7638, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.803198399755989e-06, | |
| "loss": 0.7717, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.795965217091152e-06, | |
| "loss": 0.7551, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.788714780197918e-06, | |
| "loss": 0.7692, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.7814471435634775e-06, | |
| "loss": 0.7454, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.774162361804279e-06, | |
| "loss": 0.7591, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.7668604896656145e-06, | |
| "loss": 0.7511, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.7595415820212105e-06, | |
| "loss": 0.757, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.752205693872819e-06, | |
| "loss": 0.7593, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 5.7448528803497975e-06, | |
| "loss": 0.7667, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 5.737483196708698e-06, | |
| "loss": 0.766, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 5.730096698332854e-06, | |
| "loss": 0.7675, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.722693440731961e-06, | |
| "loss": 0.7549, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.715273479541661e-06, | |
| "loss": 0.7377, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.707836870523121e-06, | |
| "loss": 0.748, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.700383669562621e-06, | |
| "loss": 0.7765, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.692913932671127e-06, | |
| "loss": 0.7583, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.685427715983874e-06, | |
| "loss": 0.7747, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.677925075759944e-06, | |
| "loss": 0.7733, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.670406068381839e-06, | |
| "loss": 0.7632, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.662870750355062e-06, | |
| "loss": 0.7824, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.655319178307693e-06, | |
| "loss": 0.7623, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.647751408989961e-06, | |
| "loss": 0.7601, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.640167499273813e-06, | |
| "loss": 0.7556, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 5.632567506152498e-06, | |
| "loss": 0.7711, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 5.624951486740128e-06, | |
| "loss": 0.7567, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 5.617319498271254e-06, | |
| "loss": 0.7585, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.609671598100434e-06, | |
| "loss": 0.7467, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.6020078437018046e-06, | |
| "loss": 0.7504, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.594328292668647e-06, | |
| "loss": 0.7736, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.586633002712952e-06, | |
| "loss": 0.7626, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.578922031664992e-06, | |
| "loss": 0.7627, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.571195437472881e-06, | |
| "loss": 0.7522, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.563453278202141e-06, | |
| "loss": 0.7572, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.555695612035269e-06, | |
| "loss": 0.7501, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.547922497271293e-06, | |
| "loss": 0.7641, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.54013399232534e-06, | |
| "loss": 0.769, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.532330155728192e-06, | |
| "loss": 0.7741, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.524511046125852e-06, | |
| "loss": 0.7799, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 5.516676722279097e-06, | |
| "loss": 0.7538, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 5.50882724306304e-06, | |
| "loss": 0.769, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 5.500962667466689e-06, | |
| "loss": 0.7714, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.4930830545925e-06, | |
| "loss": 0.7627, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.485188463655933e-06, | |
| "loss": 0.7669, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.477278953985013e-06, | |
| "loss": 0.7449, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.469354585019874e-06, | |
| "loss": 0.7427, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.461415416312322e-06, | |
| "loss": 0.7405, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.453461507525382e-06, | |
| "loss": 0.7655, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.4454929184328505e-06, | |
| "loss": 0.7592, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 5.437509708918847e-06, | |
| "loss": 0.7678, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 5.429511938977365e-06, | |
| "loss": 0.7652, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 5.421499668711819e-06, | |
| "loss": 0.7534, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 5.4134729583345936e-06, | |
| "loss": 0.7434, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 5.405431868166591e-06, | |
| "loss": 0.7676, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.3973764586367775e-06, | |
| "loss": 0.7609, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.389306790281732e-06, | |
| "loss": 0.776, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.381222923745186e-06, | |
| "loss": 0.7525, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 5.3731249197775734e-06, | |
| "loss": 0.7476, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 5.365012839235568e-06, | |
| "loss": 0.7695, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 5.356886743081632e-06, | |
| "loss": 0.7484, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 5.348746692383553e-06, | |
| "loss": 0.7758, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 5.34059274831399e-06, | |
| "loss": 0.7764, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 5.33242497215001e-06, | |
| "loss": 0.7557, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 5.324243425272624e-06, | |
| "loss": 0.7578, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 5.316048169166339e-06, | |
| "loss": 0.7841, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.30783926541868e-06, | |
| "loss": 0.7584, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.299616775719737e-06, | |
| "loss": 0.7628, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.2913807618616975e-06, | |
| "loss": 0.7618, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.283131285738386e-06, | |
| "loss": 0.7585, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.274868409344793e-06, | |
| "loss": 0.7537, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.266592194776616e-06, | |
| "loss": 0.7753, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.258302704229784e-06, | |
| "loss": 0.7528, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.25e-06, | |
| "loss": 0.7833, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.241684144482266e-06, | |
| "loss": 0.7568, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.233355200170416e-06, | |
| "loss": 0.7591, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.225013229656649e-06, | |
| "loss": 0.7611, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.216658295631052e-06, | |
| "loss": 0.7563, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.208290460881136e-06, | |
| "loss": 0.7676, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.199909788291361e-06, | |
| "loss": 0.7736, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.1915163408426645e-06, | |
| "loss": 0.7544, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.183110181611983e-06, | |
| "loss": 0.7441, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.174691373771788e-06, | |
| "loss": 0.7579, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.166259980589603e-06, | |
| "loss": 0.7773, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.157816065427529e-06, | |
| "loss": 0.7492, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.149359691741772e-06, | |
| "loss": 0.7707, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.1408909230821635e-06, | |
| "loss": 0.7464, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.132409823091685e-06, | |
| "loss": 0.7513, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.123916455505986e-06, | |
| "loss": 0.7552, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.1154108841529075e-06, | |
| "loss": 0.7706, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.106893172952004e-06, | |
| "loss": 0.7581, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.098363385914059e-06, | |
| "loss": 0.7519, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.089821587140607e-06, | |
| "loss": 0.7698, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.081267840823451e-06, | |
| "loss": 0.747, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.0727022112441795e-06, | |
| "loss": 0.7374, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.064124762773687e-06, | |
| "loss": 0.7555, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.055535559871682e-06, | |
| "loss": 0.7688, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.046934667086212e-06, | |
| "loss": 0.75, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.038322149053172e-06, | |
| "loss": 0.7457, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.029698070495822e-06, | |
| "loss": 0.7543, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.021062496224301e-06, | |
| "loss": 0.7317, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.012415491135133e-06, | |
| "loss": 0.7653, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.0037571202107525e-06, | |
| "loss": 0.7702, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.995087448519005e-06, | |
| "loss": 0.7647, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.986406541212659e-06, | |
| "loss": 0.7484, | |
| "step": 763 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1496, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 1.2700841023456543e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |